SEARCH

SEARCH BY CITATION

Keywords:

  • Neuroblastoma;
  • ERBB;
  • EGFR;
  • HER-2;
  • HER-4

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

BACKGROUND:

ERBB receptor tyrosine kinases can mediate proliferation, migration, adhesion, differentiation, and survival in many types of cells and play critical roles in many malignancies. Recent reports suggest a role for EGFR signaling in proliferation and survival of neuroblastoma, a common form of pediatric cancer that often has an extremely poor outcome.

METHODS:

The authors examined ERBB family expression in neuroblastoma cell lines and patient samples by flow cytometry, western blot, and quantitative real time polymerase chain reaction (Q-PCR). Response to ERBB inhibition was assessed in vitro by cell-cycle analysis and western blot and in vivo by serial tumor-size measurements.

RESULTS:

A panel of neuroblastoma cell lines and primary patient tumors expressed EGFR, HER-3, and HER-4, with HER-2 in some tumors. HER-4 mRNA was expressed predominantly in cleavable isoforms. Whereas EGFR inhibition with erlotinib and pan-ERBB inhibition with CI-1033 inhibited EGF-induced phosphorylation of EGFR, AKT, and ERK1/2, only CI-1033 induced growth inhibition and dose-dependent apoptosis in vitro. Both CI-1033 and erlotinib treatment of neuroblastoma xenograft tumors resulted in decreased tumor growth in vivo, although CI-1033 was more effective. In vivo expression of EGFR was observed predominantly in vascular endothelial cells.

CONCLUSIONS:

Pan-ERBB inhibition is required for ERBB-related neuroblastoma apoptosis in vitro, although EGFR contributes indirectly to tumor growth in vivo. Inhibition of EGFR in endothelial cells may be an important aspect of erlotinib's impact on neuroblastoma growth in vivo. Our results suggest that non-EGFR ERBB family members contribute directly to neuroblastoma growth and survival, and pan-ERBB inhibition represents a potential therapeutic target for treating neuroblastoma. Cancer 2010. © 2010 American Cancer Society.

Neuroblastoma is a severe cancer of childhood, causing >7% of malignancies in children younger than 15 years of age and approximately 15% of all pediatric cancer deaths. Outcome for children with high-risk neuroblastoma remains <40% with current treatment regimens.1 The poor survival rates of high-risk patients warrants investigation into novel treatment options.

Dependence on growth factors is one of the hallmarks of cancer.2 The ERBB family, comprised of EGFR, HER-2, HER-3, and HER-4, is a group of receptor tyrosine kinases with common and unique signaling properties3 that can provide these essential signals. Constitutive activation of EGFR and HER-2 through autocrine ligand production, receptor overexpression, or activating mutation occurs commonly in carcinomas and is correlated with poor prognosis. Less is known about HER-3 and HER-4 in cancer, and few reports correlate HER-3 and HER-4 expression or activation with prognosis.4

The role of the ERBB family has not been well established in neuroblastoma. EGFR expression and effectiveness of anti-EGFR treatment in vitro has been reported,5, 6 although others found no response to erlotinib in vivo.7 Surprisingly, exposure to EGF induced apoptosis in a dose-dependent manner.8-10 None of these reports evaluated contributions of other ERBB family members nor tested their importance in vivo.

We report that pan-ERBB inhibition is more effective than EGFR-specific inhibition in neuroblastoma. A pan-ERBB inhibitor caused significant growth inhibition in vitro and in vivo, whereas EGFR-specific inhibition was ineffective in vitro and produced a modest reduction in tumor growth in vivo. These results suggest that HER-4 signaling may be important in neuroblastoma and that a pan-ERBB inhibitor may be a novel therapeutic approach for children with neuroblastoma.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Chemicals

CI-1033 (Pfizer, NY, NY) was dissolved in water, 5 mM for in vitro use and at 4.5 mg/mL for in vivo use. Erlotinib (OSI Pharmaceuticals, Melville, NY) was dissolved in dimethyl sulfoxide, 10 mM for in vitro use and 7.5 mg/mL for in vivo use.

Cell lines

Neuroblastoma cell lines SK-N-AS, SK-N-SH, SH-EP, SH-SY5Y, IMR-32, SMS-KCNR, LA1-55N, NGP, and CHP-134 were obtained from Dr. Susan Cohn (The University of Chicago Children's Hospital, Chicago, Ill) and Dr. John Maris (Children's Hospital of Philadelphia, Philadelphia, Pa) and have been previously described.11-18 Cells were incubated at 37°C and 5% CO2 in RPMI 1640 (Mediatech, Manassas, Va) with 10% fetal bovine serum (Hyclone, Logan, Utah), 1% L-glutamine (Lonza, Allendale, NJ), and 1% Penicillin/Streptomycin (Gemini, Woodland, Calif). A431 and SKOV-3 cells19, 20 were used as positive controls and grown in DMEM (Invitrogen, Carlsbad, Calif) with all other conditions the same. The breast cancer cell line T47D was obtained from Dr. Seth Corey (The Children's Cancer Hospital at M.D. Anderson Cancer Center, Houston, Tex) and was grown in DMEM/F12 (Mediatech), with all other conditions as above. For all experiments, cells were harvested when confluence was less than 80%.

Patient-Derived Tumor Samples

For primary tumors, approval for specimens was obtained from the Children's Oncology Group (COG) Neuroblastoma committee. Frozen primary tumor samples were then received from the Cooperative Human Tissue Network (CHTN) funded by the National Cancer Institute. Other investigators may have received specimens from the same subjects.

Flow Cytometry

Adherent neuroblastoma cells were washed with PBS and mobilized with enzyme-free cell-dissociation buffer (Invitrogen) and washed with cold PBS/1 % bovine serum albumin (Fisher, Pittsburgh, Pa). Antibodies to EGFR and HER-2 directly conjugated to phycoerythrin (PE) (Becton Dickinson, Franklin Lakes, NJ), HER-3 and HER-4 (Neomarkers, Fremont, Calif) antibodies, or the appropriate IgG isotype control (Becton Dickinson) were then incubated with the cells on ice for 30 minutes. The cells were then washed with PBS/1% BSA 3 times. For the HER-3 and HER-4, a secondary PE-conjugated anti-mouse (R&D Systems, Minneapolis, Minn) was incubated with the cells for 30 minutes on ice. ERBB expression was then analyzed by flow cytometer (FACSCalibur, Becton Dickinson).

HER-4 Isoform Analysis

Sample Preparation

RNA was extracted from neuroblastoma cells grown to 70% confluency with the RNeasy Mini Kit (Qiagen) and treated with DNase I (Qiagen) according to the manufacturer's instructions. RNA was reverse transcribed using the Omniscript Reverse Transcriptase Kit (Qiagen) with oligo-dTs (Invitrogen) according to the manufacturer's instructions.

Primer Design

Primers were designed to anneal uniquely to the 4 HER-4 JM a-d isoforms according to previously reported sequences.21 Forward primers: HER4 JM-a (5′-CTGCACCCAAGGGTGTAACG-3′), HER4 JM-b (5′-GGCCTGATGGATAGAACTCC-3′), HER4 JM-c (5′-CAAACTGCACCCAAGGAACTC-3′), and HER4 JM-d (5′-CGGCCTGATGGATAGGTGTAAC-3′). A common reverse primer was used for all 4 HER4 JM isoforms (5′-GCAAATGTCAGACCCACAATG-3′).

The HER4 JM isoform data was normalized by comparing the number of isoform copies to the number of GAPDH copies with the forward primer (5′-GCATCCTGGGCTACACTGAG-3′) and the reverse primer (5′-CCACCACCCTGTTGCTGTAG-3′).

Q-PCR

Quantitative real time polymerase chain reaction (Q-PCR) was performed using iQ SYBR Green Supermix (Biorad) according to the manufacturer's instructions. Thermal cycling of HER4 JM a-d consisted of 38 cycles of: 95°C for 30 seconds; 61, 63, 64, or 64°C for 30 seconds, respectively; and 72°C for 25, 15, 15, or 20 seconds, respectively. Expected amplification products for HER4JM a-d were: 155, 82, 84, and 158 bp, respectively. Thermal cycling of GAPDH consisted of 38 cycles of: 95°C for 30 seconds, 60°C for 30 seconds, and 72°C for 15 seconds with an expected product of 161 bp.

The mRNA transcript copy number was quantified absolutely by creating standard curves of the target message. HER4 JM isoforms standard curves were created by amplifying pcDNA3.1 (Invitrogen) containing HER4 JM isoform cDNA; GAPDH standard curve was created by amplifying pCR2.1 (Invitrogen) containing a 161 bp fragment of GAPDH. Products were evaluated for specificity by electrophoresis on a 1.8% agarose gels.

Western Blotting

For analysis of target inhibition, cells were treated with CI-1033 or erlotinib for 24 hours, followed by 5 minutes of exposure to EGF (100 ng/mL), then washed in cold phosphate-buffered saline (PBS) and incubated on ice with protein lysis buffer (50 mM Hepes, 150 mM NaCl, 1 mM EGTA, 10 mM Sodium Pyrophosphate, 10 mM NaF, 10% glycerol, 1.5 mM MgCl2, 1% Triton X-100) containing phosphatase inhibitor (Sigma-Aldrich, St. Louis, Mo) and protease inhibitor (Roche, Basel, Switzerland) for 10 minutes. Lysate was then scraped from the plate and centrifuged at 12,000 rpm for 10 minutes at 4°C. Supernatant was collected and protein concentration determined with the BCA Protein Assay kit (Pierce, Rockford, Ill). Frozen patient samples were homogenized using a Dismembrator (Sartorius, Edgewood, NY) then lysed in the same lysis buffer. Protein lysate in the amount of 30 μg was loaded into an 8% SDS-PAGE gel and then transferred to nitrocellulose using standard methods. Membranes were blocked in 5% milk in TBST (100 mM Tris, 1.5 M NaCl, 0.1% Tween-20, pH 7.9) for 1 hour at room temperature and then washed with TBS-Tween 3 times for 5 minutes. Membranes were then incubated with primary antibody (EGFR, pEGFR, AKT, pAKT, ERK1/2, pERK1/2, and PARP at 1:1000 (Cell Signaling, Danvers, Mass), HER-3 and HER-4 at 1:50 (Neomarkers), HER-2 at 1:1500 (Epitomics, Burlingame, Calif), and actin at 1:1000 (Sigma)) overnight at 4°C. Membranes were washed 3 times for 10 minutes in TBST, then incubated with the appropriate secondary antibody/HRP for 1 hour at room temperature and then washed 3 times for 10 minutes in TBST. Chemiluminescence detection reagents (Pierce) were incubated with the membranes for 1 minute, followed by film development on a Kodak developer.

Growth Analysis

Cells were plated at 100,000 cells per well in a 6-well plate. After cells became adherent, drug was added and refreshed every 24 hours. After 3 or 7 days of triplicate treatments with 0, 0.01, 0.1, 1, 3, or 5 μM CI-1033 or 0, 0.1, 1, 3, 5, or 10 μM erlotinib, media and dead cells were removed by washing with PBS. Intact nuclei were extracted by addition of a 0.01M hepes-0.015 M MgCl2 buffer for 5 minutes, followed by addition of 5% Bretol (ethyl hexadecyldimethylammonium bromide) in water and agitation for 10 minutes. Nuclei were then fixed with a 0.9% NaCl and 0.5% formalin solution in water and counted using the ViCell Cell Viability Analyzer (Beckman Coulter, Fullerton, Calif) as previously described.22 IC50 values were calculated using Microsoft Excel with a best-fit trendline.

EGFR and K-Ras mutation analysis

Genomic DNA was isolated from neuroblastoma cell lines using the DNeasy tissue kit (Qiagen, GmbH, Hilden, Germany). The somatic status of the EGFR gene was investigated by PCR using primers specific for exons 18-21, encompassing the tyrosine kinase domain. For the K-RAS gene, PCR was performed using exon 2 specific primers. Subsequently, PCR fragments were sequenced and analyzed by direct sequencing in both sense and antisense directions. For the ease of sequencing, M13 tails were attached to every primer pair. Primer sequences were as follows:

  • EGFR exon 18: forward primer: CCTGAGGTGACCCTTGTCTCTGTGTTCTT,

  • reverse primer: GAGGCCTGTGCCAGGGACCTTA,

  • EGFR exon 19: forward primer: CGCACCATCTCACAATTGCCAGTTA,

  • reverse primer: AAAGGTGGGCCTGAGGTTCA,

  • EGFR exon 20, forward primer: CACACTGACGTGCCTCTCC,

  • reverse primer:TATCTCCCCTCCCCGTATCT,

  • EGFR exon 21: forward primer: CCCTCACAGCAGGGTCTTCTCTGT,

  • reverse primer: TCAGGAAAATGCTGGCTGACCTA,

  • K-RAS exon 2: forward primer:CGTCCTGCACCAGTAATATGC,

  • reverse primer: GTATTAACCTTATGTGTGACA.

The following PCR program was applied: 5 minutes 95°C, 30 seconds 95°C, 30 seconds 62°C, 30 seconds 68°C (last 3 steps were repeated 42 times), 7 minutes 68°C.

Cell-cycle analysis

Untreated and cells treated with CI-1033 or erlotinib for 48 hours were evaluated for changes in cell cycle by propidium iodide staining. Adherent and nonadherent cells were harvested using trypsin (Gibco), washed with PBS, and resuspended in a 0.005% propidium iodide and 0.1% Triton solution in PBS. Cells were incubated overnight and evaluated on the FACS Calibur Flow Cytometer (Becton Dickinson).

In vivo xenografts and administration of ERBB inhibitors

12-week-old NOD-SCID-IL2R gamma knockout mice (Stock # 005557, The Jackson Laboratories) were injected subcutaneously in the right flank with 5 million SK-N-SH cells. Mice were randomized into control, CI-1033 treatment, and erlotinib-treatment groups at the time of injection. Treatment began when tumors became palpable (approximately 50 mm3). CI-1033 (30 mg/kg) and erlotinib (10 mg/kg) were given by gavage daily. The dose of CI-1033, which is the murine maximally tolerated dose, was chosen to give a serum level exceeding the in vitro IC50 identified in our preliminary data for SH-N-SH.23 For erlotinib, we wished to use the minimally biologically effective dose, as determined by inhibition of the EGFR-driven tumor HN5.24 Note that at a dose of 10 mg/kg, the inhibition of EGFR phosphorylation is identical whether the drug is given intraperitoneally or orally.24 Tumor size was measured in triplicate every other day with electric calipers, and the average measurement was used for analysis. Tumor volume was determined with the formula for the area of an ellipse, 4/3π × R1 × R22. All mice were sacrificed 18 days after first tumor appearance.

Immunohistochemistry

Mice were euthanized, and tumors were excised and flash frozen in OCT (Sakura, Torrance, Calif). Frozen tumors were sectioned and fixed with ice-cold acetone. Hematoxylin and eosin (H&E) staining was performed by standard techniques. Slides were incubated in 4% fish gelatin in PBS for 20 minutes, followed by addition of anti-CD31 antibody (Becton Dickinson) overnight at 4°C. Slides were rinsed with PBS and incubated with Texas Red-labeled goat antirat secondary antibody (Jackson ImmunoResearch, West Grove, Pa) for 1 hour. Slides were then washed in PBS and incubated with anti-EGFR antibody (Santa Cruz) at 4°C overnight. Alexa488-labeled antirabbit secondary antibody was incubated for 1 hour. Slides were then rinsed, counter-stained with Hoechst 33,342 (Invitrogen) for 2 minutes, and coverslips were placed with antifade reagent (Invitrogen).

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Expression of ERBB receptors in neuroblastoma cell lines and primary tumors

We evaluated the cell surface expression of EGFR, HER-3, and HER-4 by flow cytometry in 9 human neuroblastoma tumor cell lines (Fig. 1A). EGFR was variably expressed on the cell surface in 6 of the cell lines, with highest expression in the SK-N-AS and SH-EP cell lines. Interestingly, SK-N-SH cells had bimodal expression, which may reflect their dual population of N- type (SH-SY5Y) and S-type (SHEP) cells. HER-2 was not detected (data not shown). Low HER-3 expression was detected in several cell lines by flow cytometry. HER-4 was the most consistently detected ERBB receptor, with expression in 7 of the 9 cell lines.

thumbnail image

Figure 1. Cell surface expression of ERBB is depicted. (A) Nine neuroblastoma cell lines were incubated with anti-ERBB antibody either directly conjugated to PE (EGFR) or then incubated with a secondary PE conjugated antibody (HER-3 and HER-4). Positive controls were A431 (EGFR) and T47D (HER-3 and HER-4). Filled histograms represent the isotype control and unfilled histograms represent ERBB-PE expression. (B) RNA from cell lines SK-N-AS, SK-N-SH, IMR32 and CHP134 was reverse-transcribed and assessed by Q-PCR for expression of each of the HER-4 juxtamembranous isoforms. Histograms represent copies of each of the 4 isoforms (JMa, JMb, JMc, and JMd) per million copies of GAPDH. Triplicate samples were analyzed; error bars represent standard deviation. Cleavable isoforms of HER-4 (JMa and JMd) were much more abundant than noncleavable isoforms (JMb and JMc) for all lines tested.

Download figure to PowerPoint

Four cell lines (SK-N-AS, SK-N-SH, IMR-32, and CHP134) were further evaluated by Q-PCR to determine the proportion of mRNA in each of the juxtamembranous isoforms21 (Fig. 1B). Expression of HER-4 mRNA was proportional to cell-surface expression, and cleavable isoforms of HER-4 (JMa and JMd) were expressed at much higher levels than noncleavable isoforms (JMb and JMc).

Western blots were used to evaluate total ERBB expression (Fig. 2A). EGFR was detected in all cell lines by western blot analysis, with highest levels in the SK-N-AS and SH-EP cell lines. The detection of EGFR by western blot in surface-negative lines suggests internalization of activated protein in these samples.25 HER-3 was variably expressed in all cell lines, with highest levels in SK-N-SH, SH-SY5Y, and LA1-55N cells. HER-4 was expressed in all 9 neuroblastoma cell lines, with highest expression in SH-SY5Y, IMR-32, and KCNR cells. Consistent with the flow-cytometry results, HER-2 expression was below the threshold of detection for western blot analysis (data not shown).

thumbnail image

Figure 2. Shown is the expression of ERBB in neuroblastoma cell lines and patient samples. (A) Protein expression of EGFR, HER-3, and HER-4 of 9 neuroblastoma cell lines was evaluated by western blot analysis. (B) Protein lysates from primary patient tumors were evaluated for expression of EGFR, HER-2, HER-3, and HER-4. Actin was used as the loading control.

Download figure to PowerPoint

Expression of the ERBB receptors in primary neuroblastoma tumor samples was determined by western blot. All 4 receptors, EGFR, HER-2, HER-3, and HER-4, were variably expressed in 20 patient samples (Fig. 2B). EGFR and HER-4 were the predominant ERBB receptors expressed, whereas HER-2 and HER-3 had variable expression. These results suggest that the role of HER-4 in neuroblastoma should be explored.

Sensitivity of neuroblastoma to EGFR and Pan-ERBB Inhibition

Given the expression of ERBB family members in neuroblastoma cell lines and patient samples, we compared the effects of the selective EGFR inhibitor erlotinib,24 with a pan-ERBB inhibitor, CI-1033,26 on a panel of neuroblastoma cell lines. Cells were treated with inhibitors for 3 days, and proliferation and survival were assessed (Table 1). CI-1033 treatment caused a dramatic reduction in cell yield for all lines, with IC50 values ranging from 0.94 μM to 2.45 μM. (CI-1033 is specific for the ERBB family of receptor tyrosine kinases in whole cells up to concentrations of 20 micromolar.27) Expression of EGFR tended to correlate with CI-1033 sensitivity, in that the lines with the highest EGFR expression by either flow cytometry or western blot analysis had the lowest IC50 values with CI-1033. The level of HER-4 expression did not appear to correlate with CI-1033 responsiveness. In contrast, erlotinib induced minimal growth inhibition, achieving an IC50 <10 μM in only 2 of the 9 cell lines (SK-N-AS and LA1-55N). All neuroblastoma cell lines examined were negative for mutations in all investigated EGFR exons and K-RAS exon 2 (data not shown).

Table 1. IC50 Values for CI-1033 and Erlotinib Treatment for 72 Hours
Cell LineCI-1033 IC50 (μM)Erlotinib IC50 (μM)
SK-N-AS1.04
LA1-55N1.010
SK-N-SH1.0>10
SH-EP1.3>10
CHP-1341.7>10
SH-SY5Y1.8>10
IMR-321.9>10
SMS-KCNR2.0>10
NGP2.5>10

Differential induction of apoptosis with CI-1033 and erlotinib treatment in vitro

Cell-cycle effects of CI-1033 and erlotinib in SK-N-SH cells were determined by propidium iodide staining (Fig. 3A). CI-1033 treatment at 3 μM for 48 hours increased subdiploid SK-N-SH cells from 3% to 56% (P < .005). In contrast, erlotinib only slightly increased the proportion of subdiploid cells, even at 10 μM (P < .05). Similarly, in SK-N-AS cells, significant apoptosis was induced with CI-1033 at 5 μM (P < .005) but not with erlotinib (data not shown).

thumbnail image

Figure 3. Shown are cell-cycle analysis and apoptosis of SK-N-SH cells treated with CI-1033 or erlotinib. (A) SK-N-SH cells were treated with a range of concentrations of CI-1033 or erlotinib for 24 hours, and DNA content was measured by flow cytometry. The percentage of cells in the subdiploid population are shown. Error bars represent standard deviation (*P < .005). (B) SK-N-SH cells were treated with CI-1033 or erlotinib for 24 hours and analyzed for cleavage of PARP by western blot analysis. Actin was used as the loading control.

Download figure to PowerPoint

To confirm that CI-1033 treatment was inducing apoptosis, we assessed for PARP cleavage by western blot. A dose-dependent increase in PARP cleavage was evident in SK-N-SH cells by 24 hours of CI-1033 treatment. In contrast, erlotinib caused no PARP cleavage (Fig. 3B). We obtained similar results with SK-N-AS (data not shown).

Both CI-1033 and erlotinib inhibit EGFR and downstream signaling

We wished to know whether the ERBB inhibitors were effectively blocking signaling through EGFR in neuroblastoma. Both inhibitors decreased EGF-induced phosphorylation of EGFR at a concentration of 0.01 μM (Fig. 4). Near-complete inhibition of EGFR phosphorylation occurred by 1 μM CI-1033, which correlated with a decrease in ERK1/2 and AKT phosphorylation, with no changes in corresponding total protein levels (Fig. 4A). However, despite dramatic inhibition of EGFR and ERK1/2 phosphorylation by 1μM erlotinib, the phosphorylation of AKT only modestly decreased, even at high concentrations of erlotinib, suggesting a role for other ERBB receptors in AKT activation (Fig. 4B).

thumbnail image

Figure 4. Inhibition of targets and downstream proteins with CI-1033 and erlotinib treatment in SK-N-SH cells is depicted. SK-N-SH cells were treated with multiple concentrations of (A) CI-1033 or (B) erlotinib for 24 hours, followed by the addition of EGF (100 ng/mL) for 5 minutes before collection of protein lysate. The phosphorylation of EGFR, AKT, and ERK1/2 was analyzed by western blot analysis. Blots were then stripped and reprobed with total EGFR, AKT, and ERK1/2. Actin was used as the loading control.

Download figure to PowerPoint

In vivo tumor growth with CI-1033 and erlotinib treatment

An established neuroblastoma xenograft model using SK-N-SH cells in immunodeficient NOD-SCID-IL2R gamma knockout mice was used to test the effects of EGFR and pan-ERBB inhibition in vivo. Tumor-bearing mice were treated by daily gavage with erlotinib, CI-1033, or vehicle control. Treatment with CI-1033 for 18 days resulted in a significant reduction in tumor growth (P < .0001). Surprisingly, erlotinib treatment also reduced tumor growth after 18 days of treatment, but to a lesser degree than CI-1033 (P < .005) (Fig. 5A). Similarly, treatment with CI-1033 reduced final tumor weight by an average of 56% (P < .005), compared with a 34% reduction with erlotinib treatment (Fig. 5B).

thumbnail image

Figure 5. In vivo effects of CI-1033 and erlotinib are shown. (A) Subcutaneous SK-N-SH xenografts were either untreated, treated with 10 mg/kg/day of erlotinib, or treated with 30 mg/kg/day of CI-1033. Tumor volume was measured every other day, and all mice were sacrificed after 18 days of treatment. Error bars represent the standard error of the mean (*P < .005; **P < .001). (B) The final weights of each of the subcutaneous tumors are shown (dots), along with the average for each group (bars) (CI-1033, P < .005; erlotinib, P > .05). (C) Representative H&E sections from tumors of mice treated with vehicle control, erlotinib, or CI-1033 are shown. (D) Neuroblastoma tumor sections were stained with anti-EGFR (green) and anti-CD31 (red). Hoechst 33,342 dye (blue) identifies the nuclei. Images are from a control (top) and erlotinib-treated (bottom) tumor.

Download figure to PowerPoint

To evaluate whether there was a role for ERBB signaling in neuroblastoma tumor angiogenesis, we measured CD31 staining in tumor sections. Mean vessel density after treatment with CI-1033 or erlotinib treatment showed a trend toward decreased vascularity in the erlotinib group (data not shown). Interestingly, anti-CD31 staining colocalized with EGFR staining, whereas EGFR staining was also seen in surrounding tumor tissue (Fig. 5D). This suggests that EGFR inhibition on tumor vasculature contributes to ERBB inhibitors' in vivo effect, as has been described in other tumor models,28-31 but further studies are required.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Children with high-risk neuroblastoma have a survival rate of <40%, even with aggressive treatment.1 Neuroblastoma survivors often have serious long-term adverse effects after treatment.32 Children with high-risk neuroblastoma, therefore, need novel therapies to improve overall outcomes and to reduce the incidence and severity of late effects.

This study reports, for the first time, characterization of the ERBB family in neuroblastoma by flow cytometry. We report cell-surface expression of EGFR, HER-3, and HER-4 in a panel of cell lines. Interestingly, comparison with western blot results suggests that much of the expressed EGFR is not displayed at the cell surface, pointing toward regulation of EGFR signaling through receptor recycling. Future studies of ERBB's role in neuroblastoma should include a more detailed evaluation of the intracellular location of each family member, to better define the roles that each molecule plays and how small molecule inhibitors might affect them. The ERBB expression we observe is similar to prior reports of ERBB expression by RT-PCR.5 Primary tumor samples also expressed abundant EGFR and HER-4, with varied HER-2 and HER-3, suggesting that signaling through more than 1 ERBB receptor, particularly HER-4, may play a role in neuroblastoma. It is not clear why HER-2 was detected in some primary samples but no established cell lines. It is possible that the conditioned used to establish neuroblastoma cell cultures select against HER-2 expression or that a broader panel of cell lines might identify samples with HER-2 expression. It is clear, however, that HER-2 expression is not essential in neuroblastoma pathogenesis, to the extent that many cell lines and at least some patient samples had no HER-2 expression.

Supporting a role for ERBB signaling in neuroblastoma, we observed different effects from the pan-ERBB inhibitor CI-103326, 27 compared with selective EGFR inhibition by erlotinib.33, 34 Whereas both inhibitors blocked EGF-induced EGFR and ERK1/2 phosphorylation at similar concentrations, CI-1033 was more effective at inhibiting proliferation, inducing PARP-cleavage and apoptosis and inhibiting AKT phosphorylation. This effect correlated with more potent in vivo growth inhibition with CI-1033 than erlotinib. These results suggest that ERBB receptors other than EGFR may contribute to neuroblastoma growth and that differential effects on downstream targets, such as AKT, are an essential feature. Given its expression in nearly all neuroblastoma cell lines and patient samples, and particularly its consistent surface expression by flow cytometry, HER-4 is the most likely ERBB receptor to explore in neuroblastoma. As mentioned before, little is known about the role of HER-4 in tumorigenesis; however, its multiple isoforms, gamma-secretase–mediated cleavage and the transcriptional effects of its liberated intracellular domain, provide for multiple mechanisms of dysregulation to be explored.35-39

Activating mutations in EGFR of nonsmall-cell lung cancer (NSCLC) can predict a response to erlotinib treatment.40, 41 An analysis of 36 neuroblastoma cell lines revealed no EGFR mutations. Furthermore, no mutations of K-Ras exon 2, which can cause resistance to cetuximab in colorectal cancer,42 were found in 36 neuroblastoma cell lines. Therefore, the lack of neuroblastoma response to erlotinib was not due to the presence of any of the known resistance-causing mutations.

Because there was no effect on neuroblastoma cell growth in vitro with erlotinib treatment, we did not expect suppression of tumor-cell growth in vivo with erlotinib. We were surprised, therefore, to find that erlotinib-treated xenograft tumors grew more slowly than the untreated tumors. Given the limited effects of erlotinib in vitro, the in vivo growth inhibition likely arose because of modulation of the tumor microenvironment. We saw strong colocalization of CD31 and EGFR by immunohistochemistry and, with erlotinib, found a trend toward reduced mean vessel density. Given that erlotinib-treated tumors were smaller than the untreated tumors and also tended to have fewer vessels per high-powered field, it seems likely that the total vessel content of erlotinib-treated tumors was truly less than that of untreated tumors. However, our in vivo experiments were not powered nor designed to assess this difference. Although several reports have shown EGFR expression in endothelial cells of tumor models and anti-EGFR treatment reducing the endothelial cell number,28-31 others have not confirmed these findings. For example, Amin et al have reported a decreased in vivo tumor volume in a melanoma xenograft with gefitinib treatment, presumably through targeting of blood vessels, but they did not find a decrease in vessel density.31 Further analysis is required to determine the contribution of erlotinib and CI-1033 treatment to angiogenesis in vivo. Given that interactions between tumor cells and stromal elements appear to contribute to ERBB's impact on neuroblastoma growth, orthotopic models in which neuroblastoma cells are placed under the adrenal capsule may be the best models for addressing these questions in vivo.

CI-1033, through its pan-ERBB inhibition, is likely having an antiangiogenic effect through blockade of EGFR and also a direct antitumor effect as seen in vitro. This would account for smaller final tumor volume with CI-1033 treatment, compared with erlotinib treatment. However, the different targeting profiles, the irreversible binding of CI-1033, and the bioavailability of erlotinib and CI-1033 may also contribute to differences observed in vivo. CI-1033 has been reported to reach a plasma level of 1-3 μM in a mouse model with an oral dose of 20 mg/kg,23 so the 30 mg/kg/day that we administered was expected to reach concentrations where effects were observed in vitro. It also is still expected to be specific for all ERBB kinases at this concentration.27 The dose given of erlotinib, 10 mg/kg/day, is equivalent to the dose currently given to lung cancer patients (150 mg). This dose was also shown to have little toxicity in mice and to reach a peak plasma level of about 13 μM after 0.5 hours,43 higher than the levels we found to be ineffective in vitro. We chose this dose also because we wished to evaluate the minimum of a biologically effective dose24 and to be certain that we were within the dose range in which the drug remains specific for EGFR, because erlotinib can inhibit HER-2 signaling at higher concentrations.44 At a dose of 10 mg/kg/day, serum concentrations of erlotinib are identical with either the intravenous or oral routes,24 and oral administration at this dose has been shown to inhibit EGFR autophosphorylation and EGFR-mediated growth in xenograft models.

In summary, we report the cell surface protein expression of multiple ERBB receptors in neuroblastoma, notably EGFR and HER-4. We demonstrate multiple significant differences in the effects of an EGFR selective inhibitor, erlotinib, compared with a pan-ERBB inhibitor, CI-1033, particularly their proapoptotic and antitumor effects. Although further work to explore the direct effects of non-EGFR ERBB signaling in neuroblastoma is needed, we present data that support the use of pan-ERBB inhibitors such as the successor compound to CI-1033, PF-00299804, in neuroblastoma over the use of EGFR-specific agents.

CONFLICT OF INTEREST DISCLOSURES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

This work was supported by the Lorrie Olivier Family Neuroblastoma Research Fund and by NIH grant 5K08CA118730 to DPMH.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES