Research ArticleCancer

The kinase ALK stimulates the kinase ERK5 to promote the expression of the oncogene MYCN in neuroblastoma

See allHide authors and affiliations

Science Signaling  28 Oct 2014:
Vol. 7, Issue 349, pp. ra102
DOI: 10.1126/scisignal.2005470


Anaplastic lymphoma kinase (ALK) is an important molecular target in neuroblastoma. Although tyrosine kinase inhibitors abrogating ALK activity are currently in clinical use for the treatment of ALK-positive (ALK+) disease, monotherapy with ALK tyrosine kinase inhibitors may not be an adequate solution for ALK+ neuroblastoma patients. Increased expression of the gene encoding the transcription factor MYCN is common in neuroblastomas and correlates with poor prognosis. We found that the kinase ERK5 [also known as big mitogen-activated protein kinase (MAPK) 1 (BMK1)] is activated by ALK through a pathway mediated by phosphoinositide 3-kinase (PI3K), AKT, MAPK kinase kinase 3 (MEKK3), and MAPK kinase 5 (MEK5). ALK-induced transcription of MYCN and stimulation of cell proliferation required ERK5. Pharmacological or RNA interference–mediated inhibition of ERK5 suppressed the proliferation of neuroblastoma cells in culture and enhanced the antitumor efficacy of the ALK inhibitor crizotinib in both cells and xenograft models. Together, our results indicate that ERK5 mediates ALK-induced transcription of MYCN and proliferation of neuroblastoma, suggesting that targeting both ERK5 and ALK may be beneficial in neuroblastoma patients.


Over the past decade, anaplastic lymphoma kinase (ALK), a receptor tyrosine kinase (RTK) (1, 2), has been identified as one partner in a wide variety of translocation events that mediate oncogenesis in different cell types. These translocations result in more than 20 oncogenic fusion proteins, such as with echinoderm microtubule-associated protein-like 4 (EML4) or TRK-fused gene (TFG) to produce EML4-ALK and TFG-ALK in non–small cell lung cancer (NSCLC), with tropomyosin 3 (TPM3) or TPM4 to produce TPM3-ALK and TPM4-ALK in immunomyofibroblastic tumor (IMT), and with nucleophosmin (NPM) to produce NPM-ALK in anaplastic large cell lymphoma (ALCL) [(3) and references therein]. A number of small tyrosine kinase inhibitors (TKIs) have been developed that inhibit ALK activity. Among these, the best described is crizotinib, a small competitive ATP (adenosine 5′-triphosphate)–binding inhibitor that has a high response rate in the treatment of ALK-positive (ALK+) NSCLC patients (46).

In tumors such as neuroblastoma and thyroid cancer, in which ALK itself is mutated, the picture is less clear regarding the role of ALK as an oncogenic driver. However, the genetic evidence reported in both familial and sporadic neuroblastoma support ALK as a major driver of tumor formation (3, 711). Neuroblastoma is a complex and heterogeneous tumor that is derived from the neural crest of the postganglionic sympathetic nervous system and accounts for about 15% of all childhood deaths (12). Genetically, neuroblastoma is characterized by frequent deletion of parts of the chromosomes 1p and 11q, gain of parts of 17q, or MYCN gene amplification. Concomitant amplification of the MYCN locus and ALK has been observed in several cases together with gain-of-function mutations of ALK, which occurs in both familial and sporadic neuroblastoma (1317).

A phase 1 trial monitoring the safety and activity of crizotinib for both ALK+ and non-ALK+ pediatric patients with refractory solid tumor was recently reported (18). In this trial, ALK+ patients diagnosed with ALCL, NSCLC, or IMT were treated with crizotinib as a single therapeutic agent, with observed antitumoral activity in pediatric malignancies harboring ALK translocations (18). In the case of ALK+ neuroblastoma, responses in patients treated with crizotinib were not as impressive as those seen with ALK+ ALCL, NSCLC, or IMT (18). The data accumulated thus far suggest that ALK TKI monotherapy may not be an effective single agent solution for all ALK+ neuroblastoma patients and that individualized combinations of specific drugs will be required to address the disease. Along these lines, Berry and colleagues recently reported that combined treatment with crizotinib and an inhibitor against the phosphoinositide 3-kinase (PI3K)/mammalian target of rapamycin (mTOR) pathway overcame tumor resistance in a tyrosine hydroxylase–driven ALK gain-of-function/MYCN transgenic mouse model (19), suggesting that combinatorial therapies might improve outcome for ALK+ neuroblastoma patients with an otherwise poor prognosis.

As part of a phosphoproteomic study reported earlier (20), we identified an additional potential target downstream of ALK and PI3K/mTOR that was worthy of exploration. Extracellular signal–regulated kinase 5 (ERK5), also known as big mitogen-activated protein kinase (MAPK) 1 (BMK1) or MAPK7, is a member of the MAPK family (2124). ERK5 is expressed in many different tissue types and has been described as being localized in both the cytoplasm and the nucleus, depending on its activation status (25). Although ERK5 has a dual phosphorylation site (TEY) in the activation loop, similar to that of ERK1/2, it differs from other MAPKs in its unique C-terminal extension (2326). This C-terminal portion of ERK5 contains a nuclear localization signal (NLS) domain, two proline-rich domains (PR1 and PR2), a myocyte enhancer factor 2 (MEF2)–interacting region, and a transcriptional activation domain (21, 22, 2529). ERK5 responds to mitogenic signals, such as growth factors and trophic factors, as well as to cellular stress through its upstream activators MAPK kinase kinase 2 (MEKK2), MEKK3, and dual-specificity MAPK kinase 5 (MEK5) (2124, 30, 31). ERK5 has autophosphorylation activity, can phosphorylate MEK5, and activates transcription factors, such as c-MYC, cyclic adenosine monophosphate (cAMP) response element–binding protein (CREB), SRF accessory protein 1 (SAP1), and those of the MEF2 family (23, 24).

Loss of ERK5 signaling, as studied by targeted deletion of either ERK5, MEK5, or MEKK2/3, results in embryonic death in mice because of abnormalities in cardiovascular development and vascular integrity (23, 24, 32). Ablation of ERK5 specifically in endothelial cells leads to angiogenic failure and embryonic death, similar to that of global ERK5 knockout (33). At the cellular level, ERK5 is required for epidermal growth factor (EGF)–induced cell proliferation, cell cycle promotion, and cellular transformation (23, 24). More recently, it has been shown that ERK5 controls the antitumor effect of promyelocytic leukemia (PML) protein, by increasing the expression of the gene encoding p21, enabling cells to overcome the G1-S transition checkpoint (34).

Here, we investigated the mechanism by which ERK5 mediates neuroblastoma growth and found that targeting ERK5 is a potential therapeutic strategy for ALK+ neuroblastoma patients with a constitutively active ALK.


Activation of ALK leads to ERK5 phosphorylation, which is required for MYCN expression in neuroblastoma cells

We identified ERK5 as a putative ALK signaling target from a phosphoproteomics analysis of neuroblastoma cell lines (20). Initial investigation in PC12 cells verified ERK5 as a target for ALK, and the activation of ERK5 was abrogated by the addition of inhibitors of either PI3K or its downstream target, protein kinase B (PKB, also known as Akt) (fig. S1, A and B). To follow up these initial observations in a more appropriate context, we decided to investigate the role of both ALK and ERK5 in neuroblastoma cell lines, in which the mutation and activation of ALK are described in both familial and somatic cases (3, 711). Because ALK activity is important to initiate the transcription of MYCN (35), we extended our analysis to address whether modulation of ERK5 activity impacts upon MYCN expression. In these experiments, three independent neuroblastoma cell lines were used: the first cell line, CLB-BAR, harbors MYCN and ALK amplifications as well as functionally activated ALK (36); the second, CLB-GE, exhibits both MYCN and ALK amplifications, as well as an ALKF1174V-activating mutation (37); the third, IMR32, has a MYCN amplification in addition to amplification of part of the extracellular domain of ALK (37, 38). Both the CLB-BAR and CLB-GE cell lines, which have activated ALK variants, exhibit constitutive downstream signaling, resulting in the activation of targets such as ERK1/2 and PKB/Akt, in addition to ERK5 (Fig. 1, A and B). In IMR32 cells, stimulation of ALK with an activating antibody (mAb46) also results in the phosphorylation (activation) of ERK5 (Fig. 1C).

Fig. 1 Activation of ERK5 in neuroblastoma cell lines expressing activated ALK.

(A to C) Immunoblotting for the indicated proteins in neuroblastoma cells CLB-BAR (A), CLB-GE (B), and IMR32 (C) cultured on six-well plates in complete growth medium and treated with inhibitors as indicated for 6 hours alone (A and B) or before (C) stimulation with mAb46 for 30 min. AZD, AZD8055; p, phosphorylated; S473, Ser473. (D) Immunoblotting for MYCN abundance in CLB-BAR and CLB-GE cells transfected with scrambled control or one of two ERK5 siRNAs. The abundance of total pan-ERK1/2 served as a loading control. Data are means ± SE from at least three independent experiments. SiC, control siRNA. *P < 0.005, **P < 0.006, Student’s t test.

Because PI3K activity is important for MYCN protein stability (39, 40), the PI3K/mTOR pathway inhibitors NVP-BEZ235 and AZD8055 were used for comparison to crizotinib (an ALK inhibitor) and XMD8-92 (an ERK5 inhibitor) (34, 41, 42). In agreement with earlier reports (35), treating CLB-BAR, CLB-GE, and IMR32 cells with crizotinib decreased the phosphorylation of ERK5, ERK1/2, and PKB/Akt, and decreased the protein abundance of MYCN (Fig. 1, A to C). Both NVP-BEZ235 and AZD8055 efficiently blocked ERK5 activation (Fig. 1, A to C), suggesting that ERK5 activity is dependent on PI3K/mTOR pathway activity in these cells and raising the question of whether ERK5 activity may be involved in the maintenance of cellular MYCN protein abundance. Indeed, inhibition of ERK5 caused a decrease in the abundance of MYCN in all three cell lines (Fig. 1, A to C), observed within 1 hour of XMD8-92 treatment, and was barely detectable 24 hours after treatment (fig. S2). The kinase selectivity and the antiproliferative effect of XMD8-92 were previously determined in various kinase assays (34, 42). The loss of MYCN observed in response to ERK5 inhibition was observed in the presence of active PKB/Akt (Fig. 1, A to C), thereby uncoupling the activity of the PI3K pathway from the stabilization of MYCN protein in the cell (39).

In agreement with our findings in PC12 cells (fig. S1, A and B), U0126, an ERK1/2 inhibitor, specifically abrogated phosphorylation of ERK1/2—without affecting ALK-mediated activation of either ERK5 or PKB/Akt—in all neuroblastoma cell lines examined (Fig. 1, A to C). In contrast to inhibition of PI3K/mTOR or ERK5, inhibition of ERK1/2 with U0126 did not significantly alter MYCN protein abundance in the CLB-BAR cell line, but did reduce MYCN abundance in CLB-GE and IMR32 cells (Fig. 1, A to C). The reason for these observed differences is currently unknown but may reflect the unique individual genetic backgrounds of these patient-derived neuroblastoma lines.

To further investigate the effect of loss of ERK5 on MYCN protein abundance, we complemented the ERK5 inhibition approach with small interfering RNA (siRNA) targeting ERK5. Compared with a scrambled control siRNA, two independent ERK5 siRNAs reduced the abundance of both ERK5 and MYCN (Fig. 1D). Thus, MYCN protein abundance in the three independent neuroblastoma cells examined here is dependent on the presence and activity of ERK5.

PI3K activity is required for ALK-mediated activation of ERK5

Given our finding that ERK5 activation is sensitive to PI3K pathway inhibition in neuroblastoma cells, we decided to investigate the importance of PI3K activity on ERK5. Constitutively active (p110CAAX) or kinase-deficient (p110KD) variants of p110 (43), the catalytic subunit of PI3K, were expressed in PC12, CLB-BAR, and CLB-GE cells (Fig. 2, A to C). We were unable to detect increased amounts of phosphorylated ERK5 protein upon transient transfection of p110CAAX (Fig. 2, A and B), which may be expected because this pathway is constitutively activated by ALK. However, expression of p110KD markedly reduced the amount of phosphorylated ERK5 in both neuroblastoma cell lines (Fig. 2, A and B). As an additional control in PC12 cells, we found that the phosphorylation of ERK5 was increased upon either stimulation with EGF or transfection of p110CAAX. Conversely, expression of p110KD before stimulation with EGF blocked the phosphorylation of ERK5 (Fig. 2C). Together, these results suggest that PI3K activity is important for the activation of ERK5.

Fig. 2 ALK activates ERK5 through the PI3K pathway.

(A and B) Western blotting and quantification of phosphorylated ERK5 or AKT (at Ser473) in (A) CLB-BAR or (B) CLB-GE cells transfected with Lipofectamine 2000, p110CAAX, or p110KD. Where indicated, p110CAAX-transfected cells were also treated with 250 nM crizotinib for 6 hours as indicated. (C) Western blotting and quantification as in (A) and (B) in lysates from PC12 cells transiently transfected with p110CAAX or p110KD. Where indicated, untransfected cells and p110KD-transfected cells were also stimulated with EGF. Tubulin was a loading control. Data are means ± SE from at least three independent experiments. *P < 0.0005, Student’s t test.

MEKK3 is a downstream target of PKB/Akt activity

It has been reported that the signaling core of the ERK5 pathway consists of the kinases MEKK2/MEKK3, MEK5, and ERK5 (44, 45). Our results above suggest that ERK5 is activated (phosphorylated) by the ALK/PI3K/PKB/Akt pathway. In an attempt to understand the connection between the two pathways, we examined the phosphorylation status of MEKK3 using PhosphoSitePlus and identified several phosphorylated Ser residues preceded by an Arg residue in position −3 and/or −5 (fig. S3) ( PKB/Akt is characterized as an Arg-directed kinase and has a minimum sequence motif required for the efficient phosphorylation of various substrates: Arg-Xaa-Arg-Yaa-Zaa-Ser/Thr-Hyd [where Xaa is any amino acid, Yaa and Zaa are small residues other than Gly, and Hyd is a bulky hydrophobic residue (Phe or Leu)] (46). To investigate whether MEKK3 could be a target for PKB/Akt, we used antibodies against the phosphorylated (Ser/Thr) PKB/Akt substrate motif to immunoprecipitate PKB/Akt substrate proteins from CLB-GE neuroblastoma cells treated with crizotinib, XMD8-92, or NVP-BEZ235 for 6 hours (Fig. 3A). Immunoprecipitates were immunoblotted with known PKB/Akt downstream targets, such as glycogen synthase kinase 3β (GSK3β) (46), as well as for MEKK3. Inhibition with either crizotinib or NVP-BEZ235 abrogated the phosphorylation of both MEKK3 and GSK3β, indicating that MEKK3 is a downstream target of PKB/Akt (Fig. 3A). On the basis of these results, we tested our hypothesis that MEKK3 is important for MYCN expression. CLB-GE and CLB-BAR neuroblastoma cells were transiently transfected with three independent MEKK3 siRNAs or a control scrambled siRNA. After 36 hours of siRNA transfection, the abundance of MEKK3 as well as MYCN was substantially reduced. In addition, the phosphorylation of ERK5 was also decreased compared to the control (Fig. 3, B and C). Thus, the data suggest that activation of ALK and PI3K–PKB/Akt mediates the phosphorylation of the upstream kinase of ERK5, MEKK3. In keeping with this hypothesis, silencing MEKK3 resulted in a loss of phosphorylated ERK5 and reduced abundance of MYCN protein.

Fig. 3 PKB/Akt phosphorylates MEKK3.

(A to C) Western blotting in either whole-cell lysates (W.C.L.) or immunoprecipitates from an AKT substrate antibody (Ab I.P.) from (A) CLB-GE cells treated as indicated for 6 hours, or (B) CLB-GE or (C) CLB-BAR cells transfected with scrambled control siRNA (SiC) or one of three siRNAs against MEKK3 (Si1 to Si3). Tubulin was a loading control. Blots are representative, and data are means ± SE from at least three independent experiments. *P < 0.002, Student’s t test.

Phosphorylated ERK5 is observed in the nucleus of neuroblastoma cells

Previous work shows that activated (phosphorylated) ERK5 localizes in the nucleus upon activation (47). Treatment of CLB-BAR cells with either crizotinib or XMD8-92 or transfection with ERK5 siRNA decreased the abundance of phosphorylated ERK5 in the nucleus (Fig. 4A), whereas untreated cells were positive for the presence of nuclear phosphorylated ERK5 (Fig. 4, A and B). Nonphosphorylated ERK5 protein is detected in the nucleus upon inhibitor treatment (Fig. 4A); however, the abundance was significantly reduced after ERK5 siRNA transfection (Fig. 4, A to C). We detected reduced abundance of MYCN protein in the nuclear fraction of XMD8-92–treated CLB-BAR cells as compared with the nonnuclear fraction (Fig. 4A), in agreement with an earlier report (35). Together, our results thus far suggest that ERK5 plays an important role in the maintenance of MYCN protein abundance in neuroblastoma cells.

Fig. 4 Activated ERK5 is located in the nucleus.

(A) Western blotting for phosphorylated ERK5 (p-ERK5), MYCN, and total ERK5 (pan-ERK5) in fractionated nuclear or cytoplasmic extracts from neuroblastoma cell line CLB-BAR either untreated, transfected with ERK5-targeted siRNA #2, treated with 250 nM crizotinib, or treated with 5 μM XMD8-92 for 6 hours. Histone H1 was a loading control. (B) Immunofluorescent images of neuroblastoma cell line CLB-BAR, untreated, transfected with siRNA ERK5 #2, treated with 250 nM crizotinib, or treated with 5 μM XMD8-92 for 6 hours. Green, phalloidin; red, phosphorylated ERK5; blue, nuclei [4′,6-diamidino-2-phenylindole (DAPI)]. (C) Quantification of the number of cells with p-ERK5 detectable in the nucleus, represented in (B). Data are means ± SE from at least three independent experiments.

ERK5 activity regulates MYCN at the transcriptional level

Because ALK activity promotes the initiation of MYCN transcription (35), and it is also known that the PI3K pathway activity is important for the stability of MYCN protein (39, 40, 48), we wished to investigate the role of ERK5 in these two scenarios, which we hypothesized are not necessarily mutually exclusive. To address the role of ERK5 in MYCN protein stability, we treated CLB-BAR and CLB-GE cells with the MG-132 proteasome inhibitor in the presence or absence of XMD8-92. Treatment with MG-132 in both cell lines led to a robust increase in MYCN protein abundance (Fig. 5A). This increase was completely blocked by the addition of XMD8-92, suggesting that ERK5 activity promotes MYCN abundance at the transcriptional level.

Fig. 5 Activated ERK5 regulates transcription of MYCN.

(A) Western blotting for MYCN in whole-cell lysates from CLB-BAR or CLB-GE cells treated with ERK5 inhibitor XMD8-92 for 15 hours, proteasome inhibitor MG-132, or combined for the indicated time (h, hours). Tubulin, loading control. (B) Western blotting in whole-cell lysates from parental (P) or MYCN-expressing SHEP cells treated as indicated with EGF for 15 min, PI3K inhibitor NVP-BEZ235 for 6 hours, or XMD8-92 for 6 hours. Tubulin, loading control. Blots in (A) and (B) are representative of three independent experiments. (C and D) Quantitative reverse transcription polymerase chain reaction (RT-PCR) of MYCN mRNA in CLB-BAR and CLB-GE neuroblastoma cell lines. Primers amplifying part of the coding sequence of the ribosomal protein 19 (RPL19) (C) and RPL29 (D) were used to control for differences in cDNA (complementary DNA) input. Data are means ± SEM of the relative expression (calculated according to the ΔΔCt relative quantification method) from at least three experiments, each consisting of duplicates of each sample. *P < 0.01, **P < 0.05, Student’s t test.

In an attempt to further clarify whether ERK5 regulates the initiation of MYCN transcription or protein stability, we used previously generated SHEP neuroblastoma cell lines, which have been engineered to stably express MYCN (Fig. 5B) (39). The parental cell line [SHEP(P)] lacks exogenous MYCN, whereas the MYCN-expressing SHEP cell line [SHEP(MYCN)] expresses exogenous MYCN under the control of the cytomegalovirus (CMV) promoter. We observed that stimulation of SHEP(P) and SHEP(MYCN) cells with EGF induced the phosphorylation of ERK5 (Fig. 5B). In the SHEP(MYCN) cells, the expression of MYCN is robust (Fig. 5B), but treatment with the PI3K/mTOR inhibitor NVP-BEZ235 markedly decreased the abundance of both phosphorylated and total MYCN protein (Fig. 5B), which is in agreement with an earlier report (39). Treatment with the ERK5 inhibitor XMD8-92 for 6 hours does not affect the abundance of exogenously expressed MYCN protein (Fig. 5B), suggesting that endogenous ERK5 activity is not required for MYCN protein stability. Further, CLB-GE and CLB-BAR neuroblastoma cell lines treated with XMD8-92 had a significantly reduced amount of MYCN mRNA (Fig. 5, C and D). These results support a role for ERK5 activity in the initiation of transcription of MYCN, rather than in the stabilization of MYCN protein.

ERK5 and ALK inhibitors act synergistically to reduce neuroblastoma cell growth

Having established that ERK5 is phosphorylated by ALK in neuroblastoma cells, we wanted to investigate the consequence of reducing the enzymatic activity of either ALK or ERK5 (or both) on cell proliferation. We previously showed that addition of 250 nM crizotinib reduces proliferation rates in CLB-BAR and CLB-GE cells (20). Here, neuroblastoma cell lines treated with either crizotinib (50 or 100 nM) or the ERK5 inhibitor XMD8-92 (1.5 or 3 μM) for 5 days had reduced proliferation (Fig. 6, A and B). However, in comparison to these single-agent treatments, which reduced proliferation by about 20 to 30% by day 5, a combinatorial treatment of crizotinib and XMD8-92 (either 50 nM and 1.5 μM, or 100 nM and 3 μM, respectively) showed a further decrease in proliferation, with the higher-dose combination reducing cell proliferation by 75 to 85% by day 5 in either cell line (Fig. 6, A and B). Both low- and high-dose combinations showed synergy (table S1), as defined by the combination index calculated according to the Chou and Talalay method (49). We also showed earlier that knocking down ALK reduces the expression of ALK and results in a decrease in cell number (35). Similarly, knocking down ERK5 significantly decreased both the abundance of ERK5 and cell proliferation compared with control siRNA-transfected cells (fig. S4). Thus, a combinatorial treatment abrogating the activity of both ALK and ERK5 was more effective than the single agents at inhibiting the proliferation of neuroblastoma cells in culture.

Fig. 6 XMD8-92–mediated inhibition of ERK5 activity impairs growth and survival of neuroblastoma cell lines.

(A and B) Proliferation assessed over 5 days using the resazurin cell proliferation assay in neuroblastoma cell lines CLB-BAR (A) and CLB-GE (B) treated with either XMD8-92 alone (1.5 or 3 μM) or in combination with crizotinib as indicated. Data are means ± SE of the fold relative fluorescence from treated cells relative to untreated cells from three independent experiments.

Combined treatment with XMD8-92 and crizotinib abrogates the growth of xenografts in mice

To evaluate the effectiveness of crizotinib and XMD8-92 as a combinatorial treatment in vivo, we investigated the aforementioned compounds in BalbC/NUDE mice subcutaneously injected with human neuroblastoma (CLB-BAR) cells. Single-agent treatment of mice (oral gavage) with XMD8-92 inhibited the growth of the xenografted human tumor (Fig. 7, A to C), similar to the observed inhibition of proliferation in our in vitro assays above. Single-agent treatment with crizotinib also reduced tumor growth compared with vehicle-treated mice (Fig. 7B). However, combinatorial treatment using both crizotinib and XMD8-92 resulted in significantly greater inhibition of tumor growth compared with vehicle-treated, tumor-bearing mice (Fig. 7, A to C). Both mono and combinatorial treatments in mice were well tolerated during the treatment period. No sign of distress was observed, and the mean body weights in the vehicle- and drug-treated mice were not significantly different (Fig. 7D). Excised tumors from mice receiving the combination therapy exhibited reduced staining for the proliferation marker Ki-67 (Fig. 7, E and F) and substantially reduced abundance of both phosphorylated ALK and ERK5 (Fig. 7G). Furthermore, the abundance of MYCN was decreased in mice receiving combination therapy (Fig. 7G), as would be predicted from our aforementioned observations in cultured cells. Thus, a combinatorial treatment of crizotinib and XMD8-92—inhibiting the activation of both ALK and ERK5—was superior to single-agent treatment of neuroblastomas in mice.

Fig. 7 Combined targeting of ALK and ERK5 is effective against CLB-BAR xenografts.

(A) Representative MRI images depicting the response of CLB-BAR xenografts to vehicle, crizotinib, XMD8-92, or the combination before and after 12 days of treatment. (B) MRI quantitation of changes in CLB-BAR xenograft tumor volume in animals treated with vehicle, crizotinib, XMD8-92, or the combination before and after treatment. Data are means ± SD from six or eight mice, as indicated by n. (C) Representative mice and tumors at the end of treatment. (D) Body weights relative to day 0 in mice for up to 12 days. No significant difference was detected between treatment and vehicle groups. (E and F) Immunohistochemical staining (E) and quantification (F) of the number of cells positive for Ki-67 in tumors treated as indicated. Scale bars, 50 μm. Data are means ± SE from five tumors each. P values indicated; Student’s t test. (G) Immunoblotting analysis of indicated proteins in lysates from tumors collected after 12 days of treatment with vehicle, crizotinib, XMD8-92, or the combination. β-Actin, loading control. Blots are representative of three experiments.


Phase 1 trial results suggest that crizotinib monotherapy is not favorable for ALK+ pediatric neuroblastoma patients compared with pediatric ALK+ ALCL, NSCLC, or IMT patients (18). Combination therapy might offer future therapeutic options in these cases. Here, we found that combination therapy using the ERK5 inhibitor XMD8-92 together with the Food and Drug Administration–approved ALK inhibitor crizotinib is an effective approach in a preclinical setting. It was previously shown that pharmacological inhibition of ERK5 suppresses tumor growth through PML proteins (34). Here, we also show that components of the ALK/PI3K/PKB/MEKK3/MEK5/ERK5 pathway are important for the transcription of MYCN, an oncogene frequently amplified in neuroblastomas (fig. S5) (50). MYCN amplification is strongly associated with poor outcome and, until recently, was the only genetic factor integrated into risk stratification and treatment planning (17). Furthermore, MYCN is a tumorigenic driver in several different neuroblastoma model systems (51, 52). ALK and MYCN are located in close proximity to each other on chromosome 2p, and amplification of MYCN can also involve amplification of the ALK locus. Patients harboring a combination of gain-of-function ALK mutations together with MYCN amplification have an unfavorable aggressive neuroblastoma phenotype. ALK promotes MYCN transcription (35), and a recent report identified ALK as a direct transcriptional target of MYCN (53), implying a positive feedback loop between these oncogenic drivers. This reciprocal regulation offers an opportunity for new therapeutic approaches, an area of considerable difficulty when considering therapeutically targeting MYCN (14, 54).

Our approach was to identify potential “druggable” targets of ALK activity, thereby offering a channel to additionally target MYCN expression in neuroblastoma cells (35). The observed synergistic therapeutic effect of combined ALK/ERK5 inhibition in a preclinical neuroblastoma model suggests a novel approach. ERK5 has several downstream targets, including the MYCN gene (23, 24). It is currently unclear whether ERK5 regulates MYCN transcription directly, or through one or several other regulatory proteins, and this will be an important subject for future studies. Whereas previous work has reported that MYCN protein is stabilized by the activity of the PI3K pathway (39, 40, 48), our data suggest that abrogation of ERK5 activity affects the initiation of MYCN transcription rather than protein stability.

Our results highlight a potential novel therapy strategy for ALK+ neuroblastoma patients with constitutive active ALK mutations. Crizotinib displays significant antitumor activity in ALK+ pediatric patients in childhood malignancies (18). In these cases, ALK is active as a result of both overexpression and dimerization, which leads to the activation of the kinase domain of ALK by the fusion partner protein. However, less response is observed in patients with neuroblastoma, even those harboring defined ALK oncogenic mutations (18). It should also be noted that neuroblastoma is a very heterogeneous disease and includes many chromosomal aberrations, which is observed in both familiar and sporadic neuroblastoma (12, 55). ALK mutations secondary to the original disease have been reported, as in the case presented by Martinsson et al. (56), and a recent report suggests that subclonal ALK hotspot mutations present at diagnosis subsequently exhibit clonal expansion at relapse (57). Further, we may anticipate secondary ALK mutations arising as a consequence of crizotinib treatment, as are generated through drug resistance in NSCLC (58, 59). Therefore, there is a clear demand for new therapeutic strategies to overcome crizotinib resistance in neuroblastoma patients, and we find that this may be achieved through combinatorial treatment with ALK and ERK5 inhibitors.



Antibody against pan-ERK1/2 (1:5000; catalog no. 610123) was purchased from BD Transduction Laboratories. Antibodies for AKT substrate (catalog no. 9646), phosphorylated ALK (Tyr1278, catalog no. 6941), phosphorylated AKT (Ser473, catalog no. 9271), ERK5 (catalog no. 3552), phosphorylated ERK5 (catalog no. 3371), MEKK3 (catalog no. 5727), and MYCN (catalog no. 9405) were from Cell Signaling Technology. Antibodies for phosphorylated MYC (catalog no. sc-8000-R) and histone H1 (catalog no. sc-8030) were from Santa Cruz Biotechnology, and that for tubulin (1:5000; catalog no. T5168) was from Sigma Aldrich. Horseradish peroxidase (HRP)–conjugated secondary antibody, goat anti-mouse immunoglobulin G (IgG), and goat anti-rabbit IgG (all 1:5000) were from Thermo Scientific. Activating monoclonal antibody (mAb46) was previously described (35). Crizotinib and XMD8-92 were from Haoyuan Chemexpress Co. Ltd.; NVP-BEZ235 and AZD8055 were from Selleckchem; U0126, LY249002, and ALKi were from Calbiochem; and MG-132 was from Sigma (35). Stable PC12 Tet-On clones expressing human wild-type pTTP-ALK or pTTP-ALKF1174L have been previously described (35, 60). PC12 and neuroblastoma cell lines used in this study (CLB-BAR, CLB-GE, and IMR32) were cultured as previously described (35), as was the SHEP cell line (39). Immunoprecipitation and immunoblotting were performed as previously described (61).

RNA interference

For ERK5 and MEKK3 siRNA, CLB-BAR and CLB-GE cultures were transfected with duplex siRNA (Stealth RNAi, Invitrogen) targeting ERK5 and MEKK3 according to the manufacturer’s protocols. Cells transfected with scrambled siRNA (Invitrogen) served as negative controls.

Cell transfection

PC12 cells (2 × 106) were transfected with 0.5 μg of pmax GFP (green fluorescent protein) and 0.75 μg of DNA (p110aCAAX or p110aCAAX R916) (43) with the Nucleofector Kit (Amaxa Biosystems) according to the manufacturer’s protocol. Transfected cells were serum-starved for 36 hours and stimulated with EGF (1 μg/ml) for 15 min. Cell lysis and immunoblotting were performed as described previously (61).

Subcellular fractionation

Cytoplasmic extracts were prepared by incubating cells with 125 μl of TD buffer [25 mM tris (pH 8) and 2 mM MgCl2 containing protease inhibitors] on ice for 5 min and lysed by addition of 5% NP-40 on ice for an additional 5 min. Nuclei were pelleted (5000 rpm, 5 min, 4°C), and supernatants were stored. Nuclear pellets were washed once with 1 ml of 1× phosphate-buffered saline (PBS), centrifuged, and incubated in BL buffer [10 mM tris (pH 8), 0.4 M LiCl, and 20% glycerol] on ice for 15 min. Cellular debris was removed by centrifugation (13,000 rpm, 5 min, 4°C), and supernatants and nuclear extracts were stored before analysis.


Collected tumors were fixed in 4% paraformaldehyde (PFA) overnight. After two washes in PBS, they were dehydrated through ethanol series and then embedded in paraffin (Tissue-Tek VIP processor, Sakura). Tissue sections (5 to 7 μm thick) were mounted on SuperFrost Plus slides (Menzel-Gläser, Thermo Scientific). Slides were deparaffinized in xylene and rehydrated through ethanol series. Epitope retrieval was achieved by heating samples in a 10 mM sodium citrate (pH 6.0) solution in the microwave for 3 min. After washing with PBS and blocking for 15 min in a 5% bovine serum albumin (BSA) solution in PBS with 0.1% Triton X-100, samples were incubated with a primary antibody against Ki-67 (1:500, Cell Signaling Technology, catalog no. 12202S). Primary antibody was detected with the appropriate HRP-conjugated secondary antibody, and HRP activity was detected with the chromogenic substrate ImmPACT DAB (SK-4105, Vector Labs).


Cells were grown on cover slips coated with collagen (PureCol, Advanced BioMatrix) in 24-well plates. Before staining, cells were fixed with 4% PFA. Subsequently, cells were rinsed three times with PBS, and permeabilized with 5% normal goat serum in 0.3% Triton X-100 in PBS. Cells were incubated overnight at 4°C with antibody for phosphorylated ERK5 in a solution of 1% BSA and 0.3% Triton X-100 in PBS. After rinsing with PBS, cells were incubated with Cy3-conjugated anti-rabbit IgG (Jackson ImmunoResearch) for 2 hours at room temperature in the dark. Cells were rinsed with PBS and incubated with Alexa Fluor 488 phalloidin (Invitrogen) and DAPI (1 μg/ml) for 20 and 4 min, respectively, at room temperature. Finally, cells were rinsed with PBS, and cover slips were mounted on glass slides using Fluoromount-G (SouthernBiotech). Cells were visualized with Zeiss ApoTome.

Proliferation assay

To assess the role of ERK5 in cellular proliferation, CLB-BAR and CLB-GE neuroblastoma cells (0.25 × 106) were plated on collagen-coated 48-well plates. Cells were treated with inhibitors as indicated in the figures and monitored for 5 days. To assess cell proliferation, cells were incubated with resazurin (55 μM/liter) (Sigma) for 3 hours at 37°C (62, 63). Metabolized resazurin was analyzed by a plate reader (TEKAN) as relative fluorescence. Three independent experiments were performed.

Quantitative RT-PCR

Cells were starved and treated with either 1 or 5 μM XMD8-92 (ERK5 inhibitor) for 12 hours. RNA was isolated with the NucleoSpin RNA II Kit (Macherey-Nagel). Total RNA (1 μg) was reverse-transcribed using the iScript cDNA Synthesis Kit (Bio-Rad). For PCR amplification in an iCycler iQ5 (Bio-Rad), 25 ng of cDNA was used in a total reaction mix of 20 μl containing 10 μl of Quantimix Easy SYG Kit (BioTools), 250 nM forward and reverse primer, and 0.08 μl of fluorescein (USB, Affymetrix). Primers amplifying part of the coding sequence of RPL29 were used to control for differences in cDNA input. The following primers were used: human MYCN (forward, 5′-ACCACAAGGCCCTCAGTACC-3′; reverse, 5′-TCTCCACAGTGACCACGTCGATTT-3′), human RPL19 (forward, 5′-AACACATCCACAAGCTGAAGGCAG-3′; reverse, 5′-TCTTCACGGCGCTTGCGT-3′), and human RPL29 (forward, 5′-ATGGCCAAGTCCAAGAACCACA-3′; reverse, 5′-TTGGCATTGTTGGCCTGCAT-3′). Relative expression was calculated according to the ΔΔCt relative quantification method. Each sample within an experiment was performed in duplicate, and the experiment was carried out at least three times. Results are presented as MYCN mRNA expression relative to untreated samples.

Xenograft tumor models

Male immunodeficient BalbC/NUDE mice (Taconic) at 9 to 13 weeks of age were maintained in a pathogen-free environment, and all in vivo procedures were approved by the Institutional Animal Care and Use Committee at Umeå University, Sweden (ethical approval: A51-13). CLB-BAR (amplified MYCN/ALKΔexon4–12, 1p deletion, 17q gain) neuroblastoma cells (2.5 × 106) were subcutaneously implanted into the animals. Mice bearing small (20 to 30 mm3) and medium (30 to 70 mm3) established tumors were equally distributed among treatment groups and treated with either vehicle, or XMD8-92 (50 mg/kg) twice per day (intraperitoneal; 8.00 and 18.00), or crizotinib (80 mg/kg) once per day (oral gavage; 12.00), or both treatment administrated alternatively for 12 days. XMD8-92 was formulated in 30% hydroxypropyl-β-cyclodextrin, whereas crizotinib is formulated in 90% poly(ethylene glycol) and 10% 1-methyl-2-pyrrolidinone solution. Tumor volumes (V) were calculated by caliper measurements of the width (W), length (L), and thickness (T) of each tumor, using the following formula: V = 0.5236 (LWT), as well as by magnetic resonance imaging (MRI) calculation. Tumor growth inhibition was monitored by tumor volume measurements at days 0, 7, and 12. As a measurement of in vivo efficacy, the T/C% value was determined as the change in average tumor volumes of each treated group relative to the vehicle-treated group.

MRI and analysis

Respiration-gated T2 sequences were obtained using a 9.4-T BioSpec 94/20 USR micro MRI scanner (Bruker) using a Turbo RARE sequence with the following parameters: TR = 553.7 ms, TE = 8.7 ms, averages = 4, flip angle = 180°, RARE partitions = 2, field of view = 3.00 cm, matrix = 256 × 256, and 4 averages. A 72-mm quadrupolar volume coil was used in transmit/receive mode. For mice with respiration rates lower than 60/min, the amount of RARE partitions was increased to 4 to optimize scanning times. Fifteen axial slices were obtained with 0.7-mm thickness and 1.0-mm interslice distance; additional slices were added as needed to cover tumors longer than 15 mm in their entirety. Images were obtained from each mouse on days 0, 7, and 12 after treatment. For scanning, mice were anesthetized using isoflurane in oxygen with an induction dose of 4% and a maintenance dose of 2.5%. Respiration was monitored using a pillow-type pressure sensor. Body temperatures were monitored using a rectal thermo-probe and held as closely to 37°C as possible using a heating pad with circulating water.

For quantitative analysis of subcutaneous mouse tumors, MRI sequences were imported into the open source ImageJ scientific image analysis toolbox ( Sequences were converted into montages, and regions of interest were drawn around tumors. Because subcutaneous tumors were solid and well defined, the volume of each tumor was estimated by multiplying the area of each region of interest with the slice thickness and adding the results together.


Fig. S1. Activation of ERK5 in PC12 cells expressing wild-type hALK or hALKF1174S.

Fig. S2. Abundance of MYCN in neuroblastoma cells is decreased in response to XMD8-92.

Fig. S3. Putative AKT phosphorylation sites in MEKK3.

Fig. S4. siRNA-mediated suppression of ERK5 decreases the proliferation of neuroblastoma cells.

Fig. S5. ERK5 and MYCN are activated as downstream targets of ALK.

Table S1. XMD8-92 acts synergistically with crizotinib to inhibit neuroblastoma cell line proliferation.


Funding: This study was supported by grants from the Swedish Cancer Society (grant nos. 12-0722 to B.H. and 12-0796 to R.H.P.), the Children’s Cancer Foundation (grant nos. 11/020 to B.H. and 13/049 to R.H.P.), the Swedish Research Council (grant nos. 621-2011-5181 to R.H.P. and 521-2012-2831 to B.H.), the Lions Cancer Society, Umeå (grant nos. LP13-2012 to B.H., LP12-1946 to R.H.P., and AMP13-720 to A.E.W.), and the JC Kempe Foundation to B.H. and R.H.P. Author contributions: G.U. carried out most of the cell and biochemical analyses together with K.R. and C.S. With contribution from L.C., L.D., M.J., and S.K., A.E.W. and B.W. executed the animal experiments, MRI, and ultrasound analyses. A.E.W. and B.W. performed the immunohistological and immunofluorescence experiments. N.S.G. and X.D. designed the compound XMD8-92. R.H.P. and B.H. designed the study, supervised the project, and wrote the manuscript. All authors were involved in the revision of the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: XMD8-92 is part of a patent application.
View Abstract

Stay Connected to Science Signaling

Navigate This Article