OUP user menu

Chemically Induced Rat Schwann Cell Neoplasia as a Model for Early-Stage Human Peripheral Nerve Sheath Tumors: Phenotypic Characteristics and Dysregulated Gene Expression

Bernd Koelsch PhD, Linda van den Berg, Dipl Biol, Florian Grabellus MD, Christine Fischer PhD, Andrea Kutritz, Andrea Kindler-Röhrborn MD
DOI: http://dx.doi.org/10.1097/NEN.0b013e31828ea4ac 404-415 First published online: 1 May 2013


Most malignant human tumors display a high degree of intratumoral heterogeneity at the time of diagnosis that contributes to treatment failure. This also applies to malignant peripheral nerve sheath tumors (MPNSTs) and aggressive soft tissue sarcomas that arise sporadically or in the context of neurofibromatosis type 1. On average, MPNSTs measure 10 cm in diameter at diagnosis. To explore molecular changes associated with early malignant progression and that may be present in most, if not all, tumor cells, we generated expression profiles of ethylnitrosourea-induced trigeminal MPNSTs in rats. Because these tumors cause increased intracranial pressure, they become detectable when they are comparatively minuscule. Histologic analyses revealed close resemblance to human MPNSTs. Compared with normal trigeminal nerve tissue, 365 genes were markedly upregulated and 310 genes were consistently downregulated in all MPNST samples. The molecular signature characteristic of early-stage MPNSTs included upregulation of proliferation and tissue remodeling-associated genes, downregulation of genes involved in Schwann cell differentiation, and the absence of transcripts associated with neuronal components. The transforming growth factor-β pathway was consistently upregulated in all tumor samples. These data suggest that the signaling pathways underlying early malignant progression of Schwann cells might be targeted to prevent tumor growth and/or to treat more advanced lesions.

Key Words
  • Ethylnitrosourea
  • Malignant schwannoma
  • Neurooncogenesis
  • Peripheral nervous system
  • Rodent
  • Sarcoma
  • Transforming growth factor-β


At the time of diagnosis, most malignant human tumors display a high degree of intratumoral heterogeneity because of tumor evolution and adaptation that can contribute to treatment failure and drug resistance (1). Large-scale sequencing analyses using multiple samples of a patient primary tumor and metastatic sites have shown that the fraction of mutations found in single biopsies is not present throughout all sampled regions of the same patient's tumor. Furthermore, gene expression profiles indicating a favorable prognosis and molecular signatures pointing to an unfavorable prognosis have been identified in different samples of the same tumor (2). This is particularly discouraging for the search for biomarkers in the context of designing preventive measures, early detection, and/or targeted therapies.

Molecular changes associated with early malignant progression are likely present in every cell or the great majority in small tumors, that is, before diversification has taken place that may mask the early abnormalities. Because they do not manifest clinically before they are large, many human tumors escape early detection. This is often the case for malignant peripheral nerve sheath tumors (MPNSTs), aggressive soft tissue sarcomas that arise either sporadically or in individuals with neurofibromatosis type 1. The prognosis of MPNST is extremely poor. At diagnosis, tumors display an average diameter of 10 cm and exhibit a high degree of intratumoral heterogeneity (3). Therefore, MPNSTs in model organisms that are considerably smaller and more homogeneous can provide useful paradigm for identifying early genetic aberrations that may be targets for prevention and specific therapies.

An animal model of sporadic MPNSTs is provided by ethylnitrosourea (ENU)-induced oncogenesis in the trigeminal nerves of BD rats (4). Because of their anatomic localization, these tumors become evident clinically when they measure less than 0.5 cm in diameter; thus, they fulfill the criteria for early neoplasia. Moreover, the tumor cells have undergone few divisions and, therefore, should be relatively homogeneous. Previously, we were able to establish a large collection of MPNSTs that were induced in different BD strains.

A transversion mutation at nucleotide 2012 in the transmembrane region of the Neu/Erbb2 gene, leading to a constitutional activation of the receptor tyrosine kinase, is a very early event in the development of ENU-induced MPNST in BD rats (5). Dysregulation of neuregulin-1/Erbb signaling has also been observed in human malignant Schwann cell tumors (6, 7). To date, no further genetic alterations in known oncogenes and tumor suppressor genes have been detected in ENU-induced MPNST (8). The Neu/Erbb2 mutation is followed by a loss of heterozygosity on rat chromosome 10, which includes the wild-type allele of the Neu/Erbb2 gene (9). An additional loss of heterozygosity on chromosome 5 could be detected in MPNST of BDIV × BDIX hybrid rats and represents a later step of carcinogenesis (10). In contrast to these structural aberrations, changes in gene expression of trigeminal nerve tissue because of MPNST development have not been investigated.

This study aims at detecting the dysregulation of gene expression in early stages of MPNST as compared with that in normal trigeminal nerve tissue. Because we used tumors that had arisen on different genetic backgrounds, dysregulated expression of distinct genes found in all tumor samples is likely an important, if not necessary, step during the early progression of trigeminal MPNST. Complementary investigation focused on the comprehensive characterization of the morphology of ENU-induced MPNST that should result from the structural and expressional gene alterations observed with regard to the similarity to human tumors.

Materials and Methods

Tissue Samples

Ethylnitrosourea-induced MPNST dissected from BDIV, BDIX, (BDIV × BDIX) F1, BDX.BDIV-Mss4a, and BDX. BDIV-Mss4b rats either embedded in paraffin or frozen at − 80°C were used for this study (11). Trigeminal nerves of three 85-day-old untreated female and 3 male rats, respectively, BDIX, BDIV, and BDX.BDIV-Mss4a rats, were used as control tissue for expression analysis using cDNA Microarrays and real-time polymerase chain reaction (RT-PCR). For this purpose, trigeminal nerves were carefully separated from brain tissue. To avoid contamination with ganglion cells, only 1.5 mm of nerve tissue adjacent to the brain nerve junction, not including the smaller branch of the nerve, were immediately frozen in liquid nitrogen and stored at − 80°C until RNA preparation.


Trigeminal tumors of 27 animals were evaluated by light microscopy of hematoxylin and eosin-stained sections. Tumor classification was carried out according to the World Health Organization Classification of Tumours of the Central Nervous System (12). Grading was performed using the 3-grade system of the Fédération Nationale des Centres de Lutte Contre le Cancer, which is based on mitotic count per 10 high-power fields, the occurrence of tumor necrosis, and differentiation of tumor cells/pleomorphism. Tumors were additionally characterized with regard to cellularity, solid or cystic growth architecture, occurrence of extraneural infiltration, mast cell content, and pigmentation (Table 1).

View this table:

Histopathologic Features of Ethylnitrosourea-lnduced Malignant Peripheral Nerve Sheath Tumors

Rat StrainSexAge, daysNecrosisMitotic CountGradeTypePleomorphismCellularityPigmentationMast CellsInfiltration PatternGrowth PatternS100 IRS
BDIX.BDIV-Mss4bF177No01ConventionalWeakLowNoNoIntraneural increased cellularityDiffuse0
BDIX.BDIV-Mss4aF251No41ConventionalWeakIntermediateNoNoIntraneural increased cellularityDiffuse6
BDIX.BDIV-Mss4bF164No41ConventionalModerateIntermediateYesYesNo adjacent tissue availableCystic6
BDIX.BDIV-Mss4bF177No182ConventionalModerateIntermediateYesYesNo adjacent tissue availableCystic8
BDIX.BDIV-Mss4bF189No192ConventionalModerateIntermediateYesYesNo adjacent tissue availableCystic6
BDIX.BDIV-Mss4bF245No572ConventionalModerateHighNoNoNo adjacent tissue availableSolid12
BDIX.BDIV-Mss4bF221No422ConventionalModerateHighNoNoNo adjacent tissue availableCystic8
BDIX.BDIV-Mss4bF184No432ConventionalSevereHighNoNoNo adjacent tissue availableSolid0
BDIX.BDIV-Mss4bF147No892ConventionalSevereHighNoNoNo adjacent tissue availableCystic8
BDIX.BDIV-Mss4bM189No01ConventionalWeakLowYesNoIntraneural increased cellularityDiffuse6
BDIX.BDIV-Mss4bF239No01ConventionalWeakLowNoNoIntraneural increased cellularityDiffuse8
BDIX.BDIV-Mss4bM266No01ConventionalWeakLowNoNoIntraneural increased cellularityDiffuse0
BDIX.BDIV-Mss4bF191No01ConventionalWeakLowNoNoIntraneural increased cellularityDiffusen.d
BDIX.BDIV-Mss4bM182No01ConventionalWeakLowNoNoIntraneural increased cellularityDiffuse0
BDIX.BDIV-Mss4bM171No01ConventionalWeakLowNoNoIntraneural increased cellularityDiffuse12
BDIX.BDIV-Mss4bF180No252ConventionalWeakLowNoNoIntraneural increased cellularityDiffuse8
  • F, female; IRS, Remmele immunoreactive score (13); M, male; n.d., not done.


For immunohistochemistry, 4-µm-thick sections were cut from paraffin-embedded tissue blocks. We used a rabbit polyclonal antibody against S100 (dilution, 1:400; IgG; Dako Cytomation, Glostrup, Denmark) and a mouse monoclonal antibody against cyclin D1 (dilution, 1:2000; IgG1, clone K-2; Zytomed Systems, Berlin, Germany) together with a highly sensitive and specific polymer detection system using horseradish peroxidase (ZytoChem-Plus HRP Polymer-Kit, Zytomed Systems). Development was carried out with a permanent brown chromogen substrate system (Permanent AEC Kit, Zytomed Systems). Finally, nuclei were counterstained with hematoxylin for 5 minutes.

Staining intensity was assessed with the Remmele immunoreactive score by multiplying the level of staining intensity (0–3 points) with the percentage of positive tumor cells (0–4 points) (13). Intensity of staining was judged according to the following scale: negative (0 points); weak (13 points); moderate (48 points); and strong (912 points).

RNA Extraction

For preparation of total RNA, frozen tissue was homogenized in TRIzol Reagent (Invitrogen, Karlsruhe, Germany). RNA was purified according to the manufacturer's instructions and quantified using a Nanodrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA).


Complementary DNA synthesis was performed with the SuperScript III Platinum Two-Step qPCR Kit with SYBR Green (Invitrogen) using 0.5 µg RNA per reaction. Kit-included random hexamer and oligo(dT)20 primers were applied for first strand cDNA synthesis according to the following protocol: 10 minutes at 25°C for primer annealing; 50 minutes at 42°C for the reverse transcription step; 5 minutes at 85°C for inactivation of SuperScript III RT; then cooling on ice. Tubes were stored at −20°C until used for subsequent RT-PCR, which was performed with the 7500-Fast-Real-Time-System Instrument in standard mode (Applied Biosystems, Darmstadt, Germany). Hprt was used as a reference gene for relative mRNA quantification of target genes. Two microliters of cDNA samples served as template in a 20-µL RT-PCR reaction. The PCR conditions for mRNA quantification were as follows: 50°C for 2 minutes, 95°C for 2 minutes followed by 40 cycles of 95°C for 15 seconds, and 60°C for 30 seconds. Primer sequences used for mRNA quantification are listed in Table, Supplemental Digital Content 1 (http://www.jnen.oxfordjournals.org/lookup/suppl/doi:10.1097/NEN.0b013e31828ea4ac/-/DC1).

Microarray Hybridization, Scanning, and Data Analysis

RNA samples were processed for hybridization to microarrays on an Illumina Bead Station in the Max-Planck Institute for Molecular Genetics, Berlin, Germany, according to the manufacturer's instructions.

Qualitative integrity tests were carried out on a Bioanalyzer 2100 System (Agilent Technologies, Palo Alto, CA). One hundred nanograms of biotin-labeled cRNA, which was produced using a linear amplification kit (Ambion, Austin, TX), was hybridized to Sentrix Rat-Ref-12-v1 Expression Bead Chips containing gene-specific oligonucleotides (~22.000 genes; Illumina, Inc., San Diego, CA). Hybridization was detected with 1 µg/mL of Cy3-Strepavidin (Amersham Bioscience, Piscataway, NJ). The chips were scanned using an Illumina Bead Reader. All basic expression data analysis was carried out using the BeadStudio software 3.0 (Macrogen, Seoul, Korea). To define genes consistently upregulated or downregulated between tumor and normal nerve tissue, data from each group were combined respectively, and the means of expression values were calculated and compared. Genes with at least 2-fold differential expression between tumor and regular nerve tissue were identified. The resulting gene list was additionally selected for candidates that did not show overlapping expression values between tumor and normal nerve tissue in any of the samples used. Microarray data were deposited at the NCBI Geo repository in accordance with the MIAME standards and can be accessed via http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=vbcldkmaacmkovu&acc=GSE25929.


Histologic Characterization of ENU-Induced Trigeminal MPNST

Median age of the 27 animals carrying trigeminal MPNSTs induced by ENU on postnatal day 1 was 185 days (mean, 202 ± 37.0 [SD] days; range, 147–295 days). The tumor tissue examined originated from MPNSTs induced in BDIX.BDIV-Mss4b (n = 23), BDIX.BDIV-Mss4a (n = 3), and BDIV (n = 1) rats. Regardless of the genetic background, we exclusively found the conventional spindle cell-type histopathologic pattern of MPNSTs in different stages of development (Fig. 1A). Extraneural spread of MPNSTs was mostly observed in older animals (n = 4). In those cases, there was either a predominantly perineural or infiltrative/ destructive phenotype of tumor spread; no adjacent tissue had been dissected in 7 cases. Tumors were exclusively of low- or intermediate-grade malignancy. Necroses were not detected and a severe nuclear pleomorphism was focally seen in only 3 tumors, whereas 10 MPNSTs displayed moderate and 15 displayed weak pleomorphism. Median mitotic count per 10 high-power field was 7 (mean, 18 ± 24 [SD]; range, 0–89). The occurrence of intratumoral mast cells was recognized in all 4 MPNSTs with extraneural spread, in 3 tumors with missing adjacent tissue, and in 3 (21%) of 14 intraneural tumors. Nineteen (70%) tumors expressed the S100 protein, as detected by immunohistochemical investigation (n = 1, weak; n = 14, intermediate; n = 4, strong; Fig. 1B). Results of the morphologic analyses of all tumors are summarized in Table 1.


Ethylnitrosourea-induced malignant peripheral nerve sheath tumor (MPNST) with sweeping fascicles formed by tumor cells that have enlarged and hyperchromatic nuclei. The tumor displays destruction and invasion of adjacent bone (left). (B) Same MPNST with a typical patchy expression of the S-100 protein. Original magnification for both: 400×.

Differential Gene Expression Profiles of MPNSTs and Trigeminal Nerve Tissue

In this study, oligonucleotide microarrays containing probe sets for 22,000 rat genes were used to obtain gene expression profiles of 12 full-blown ENU-induced MPNSTs and 1 microtumor. Most tumors had developed in F1 hybrids of BDIV and BDIX rats (T1, T2, T5, T7, T8, T9, T10); the remaining cases were from BDIV (T11, T12), BDIX (T3, T4), and congenic BDIX.BDIV-Mss4a (T6, MT) rats (11). Trigeminal nerves of 85-day-old BDIV, BDIX, and BDIX.BDIV-Mss4a rats of both sexes were used as reference tissue.

Genes that were at least 2-fold upregulated or down-regulated in tumors compared with normal nerve tissue and showing no overlapping expression values in tumor or normal tissue in any of the samples used were identified. Using these criteria, the expression of 365 genes was significantly upregulated, whereas 310 genes exhibited significantly decreased expression in tumor tissue (Table, Supplemental Digital Content 2, http://www.jnen.oxfordjournals.org/lookup/suppl/doi:10.1097/NEN.0b013e31828ea4ac/-/DC1 and Table, Supplemental Digital Content 3, http://www.jnen.oxfordjournals.org/lookup/suppl/doi:10.1097/NEN.0b013e31828ea4ac/-/DC1). The most pronounced gene expression differences between normal trigeminal nerve tissue and MPNSTs were visualized by performing a hierarchal cluster analysis (Fig. 2). To gain insight into the biologic mechanisms underlying malignant transformation in the peripheral nervous system, these genes were classified regarding their function by using the Gene Ontology database and literature searches (Tables 2, 3).

View this table:

Genes Significantly Upregulated In Ethylnitorosourea-Induced Malignant Peripheral Nerve Sheath Tumors Compared With Normal Trigeminal Nerve Tissue

Chromosomal Location
SymbolGene NameGenBank IDRNOMbFold Changep
Cell adhesion
CDH17Cadherin 17NM_053977.1526.12.80.000696
COL16A1Procollagen, type XVI, alpha 1XM_345584.35149.13.80.000001
ITGA1Integrin alpha 1NM_030994.12Unknown19.80.000001
ITGA4Integrin alpha 4XM_230033.4362.114.30.000003
NINJ1Ninjurin 1NM_012867.11721.43.10.000001
PCDHB5Protocadherin beta 5XM_001055177.11830.115.10.000020
PCDHGC3Protocadherin gamma subfamily C, 3NM_053943.11830.78.10.000007
Cell cycle
BUB1Budding uninhibited by benzimidazoles 1 homologXM_215849.43115.323.50.000001
BUB1BBudding uninhibited by benzimidazoles 1 homolog, betaXM_342494.33105.115.30.000001
CCNA2Cyclin A2NM_053702.12123.117.50.000002
CCNB2Cyclin B2NM_001009470.1874.923.40.000001
CCND1Cyclin D1NM_171992.21205.410.30.000001
CDC2ACell division cycle 2 homolog ANM_019296.12020.036.90.000001
CDK4Cyclin-dependent kinase 4NM_053593.27Unknown2.90.000001
CDKN3Cyclin-dependent kinase inhibitor 3XM_214152.41522.613.50.000007
MCM3Minichromosome maintenance deficient 3XM_236988.4919.518.50.000001
MCM5Minichromosome maintenance deficient 5XM_001064207.11913.44.40.000001
MCM6Minichromosome maintenance deficient 6XM_001055953.11340.213.00.000001
RGD1562047Cyclin-dependent kinases regulatory subunit 2, similarXM_001054024.11713.323.40.000001
TGFB1Transforming growth factor, beta 1NM_021578.1180.94.80.000001
TGFB2Transforming growth factor, beta 2NM_031131.113102.76.60.000025
TTKTtk protein kinaseXM_001062174.1884.434.90.000006
Chromosomal organization
ASF1BASF1 anti-silencing function 1 homolog BXM_001072446.11923.711.70.000001
ACTG2Actin, gamma 2NM_012893.14117.733.10.003450
ADAM17A disintegrin and metalloproteinase domain 17NM_020306.1641.94.10.000001
ASPMAsp-like, microcephaly associatedXM_213891.41352.927.20.000001
AURKBAurora kinase BNM_053749.11055.88.90.000003
COTL1Coactosin-like 1XM_341700.31950.17.20.000001
DLG7Discs, large homolog 7XM_223937.31523.39.70.000001
KIF11Kinesin family member 11XM_001060913.11241.719.90.000003
KIF15Kinesin family member 15NM_181635.28127.88.30.000005
KIF20AKinesin family member 20AXM_341592.31827.119.90.000001
KIF22Kinesin family member 22NM_001009645.11186.210.90.000001
MYL9Myosin, light polypeptide 9, regulatoryXM_001067182.13144.010.60.000004
NDE1Nuclear distribution gene E homolog 1NM_053347.1100.83.80.000001
PDLIM7PDZ and LIM domain 7NM_173125.11715.25.90.000002
RGD1566336Similar to RIKEN cDNA 4933440J22XM_217737.4130.19.80.000001
RHOCras homolog gene family, member CXM_215659.42200.14.20.000001
TPX2Microtubule-associated protein homolog 2XM_001060351.13140.120.20.000002
Extracellular matrix
ADAMTS1Adam metallopeptidse with thrombospondin type 1 motif 1NM_024400.11125.414.90.000001
COL18A1Procollagen, type XVIII, alpha 1XM_241632.42012.08.20.000001
CSPG2Versican/chondroitin sulfate proteoglycan 2XM_215451.4219.731.70.000001
(Continued on next page)
CTGFConnective tissue growth factorNM_022266.2121.37.10.000022
ECM1Extracellular matrix protein 1NM_053882.12190.57.40.000116
EMILIN1Elastin microfibril interfacer 1XM_001064749.1624.913.40.000092
FBN2Fibrillin 2NM_031826.118Unknown22.80.000003
LGALS3BPLectin, galactoside-binding, soluble, 3 binding proteinNM_139096.110108.43.30.000003
LOXLysyl oxidaseNM_017061.11847.97.70.000761
MMP11Matrix metallopeptidase 11NM_012980.12013.129.70.011806
MMP16Matrix metalloproteinase 16NM_080776.1532.611.10.000099
MMP17Matrix metallopeptidase 17XM_001072688.11228.813.90.000002
POSTNPeriostin, osteoblast specific factorXM_342245.32143.625.80.000195
RGD1560062Similar to Laminin alpha-4 chain precursorXM_001061207.12043.27.90.000001
SPON1Spondin 1XM_579693.11172.014.00.000001
ACPL2Acid phosphatase-like 2NM_001007710.18101.94.80.000001
ANGPT2Angiopoietin 2XM_001065522.11669.019.70.000055
C1QBComplement component 1, q subcomponent, betaNM_019262.15155.64.40.000007
C1QGComplement component 1, q subcomponent, gammaNM_001008524.15155.74.70.000021
GDF15Growth differentiation factor 15NM_019216.11619.311.40.000299
PDGFAPlatelet-derived growth factor, alphaNM_012801.11216.23.50.000001
PNLIPPancreatic lipaseNM_013161.11265.117.20.021419
PTHLHParathyroid hormone-like peptideNM_012636.14Unknown15.20.000003
SERPINE1Serine peptidase inhibitor, clade E, member 1NM_012620.11220.911.20.000131
Golgi apparatus
B4GALT6UDP-Gal:betaGlcNAc beta 1,4-galactosyltransferase 6NM_031740.11812.44.50.000001
GNASGNAS complex locusXM_575296.23165.27.70.003738
HIP1Huntingtin interacting protein 1, transcript variant 2XM_001071106.11223.06.40.000001
LOC311716Similar to Protein KIAA1510 precursorXM_230973.33170.133.20.000007
PTGFRNProstaglandin F2 receptor negative regulatorNM_019243.12196.16.00.000012
ST6GALNAC2Sialyltransferase 7BNM_001031652.110106.823.80.000038
SULF2Sulfatase 2XM_001072989.13153.49.20.000001
Ion binding
MOXD1Monooxygenase, DBH-like 1XM_220095.4121.6161.80.000001
CD68CD68 antigenXM_001079491.11053.55.10.000118
Melanin metabolic process
SILVSilver homologXM_343146.372.037.20.000234
PBKPDZ binding kinaseXM_224300.41545.234.90.000001
Regulation of apoptosis
ALDH1A3Aldehyde dehydrogenase family 1, subfamily A3NM_153300.11120.846.30.000040
CASP7Caspase 7NM_022260.21262.75.60.000001
IGFBP3Insulin-like growth factor binding protein 3NM_012588.11488.011.30.000016
Regulation of T-cell activation
BLMBloom syndrome homologXM_218837.41136.36.50.000063
CASP3Caspase 3, apoptosis related cysteine proteaseNM_012922.21648.93.30.000001
RARARetinoic acid receptor, alphaNM_031528.11087.84.50.000001
Regulation of transcription
CASKCalcium/calmodulin-dependent serine protein kinaseNM_022184.1XUnknown3.00.000034
FOXM1Forkhead box M1NM_031633.14165.47.70.000007
FZD1Frizzled homolog 1NM_021266.2426.07.00.000001
MXD3Max dimerization protein 3NM_145773.11715.329.70.000001
MYCNv-myc myelocytomatosis viral related oncogeneXM_234025.3636.58.80.000193
NKX2-2NK2 transcription factor related, locus 2XM_001056116.13133.515.50.000027
OLIG1Oligodendrocyte transcription factor 1NM_021770.21131.224.80.045772
TCFAP2BTranscription factor AP-2 betaXM_217356.4918.117.60.010974
Tube development
AARDAlanine and arginine rich domain containing proteinNM_145093.1788.439.30.000001
CDCA2Cell division cycle associated 2, transcript variant 1XM_001068286.11541.614.00.000066
DHRS7CDehydrogenase/reductase member 7CXM_001078936.11051.727.90.028976
ECT2Ect2 oncogeneXM_342220.32113.014.40.000001
FBLIM1Filamin binding LIM protein 1NM_001007554.15160.521.70.000001
KCNN4Potassium conductance calcium-activated channel, N 4NM_023021.1179.611.80.000001
KIF23Kinesin family member 23XM_001073723.1861.813.20.000032
KIF4Kinesin family member 4XM_343797.3X88.724.80.000001
PCDH20Protocadherin 20XM_001074783.11563.943.30.000814
PCDHB15Protocadherin beta 15XM_001055818.11830.312.40.000002
PRC1Protein regulator of cytokinesis 1XM_001061201.11126.326.20.000001
TACSTD1Tumor-associated calcium signal transducer 1NM_138541.1611.215.90.000018
TRAF4AF1TRAF4 associated factor 1NM_001004264.13105.329.10.000001
CDCA1Cell division cycle associated 1NM_001012028.11385.312.00.000001
CMTM3CKLF-like MARVEL transmembrane domain containing 3XM_226200.3190.67.70.000001
DUSP6Dual-specificity phosphatase 6NM_053883.2736.95.10.000001
EDG5/S1PR2Sphingosine-1-phosphate receptor 2NM_017192.1820.08.70.000001
EDNRBEndothelin receptor type BNM_017333.115Unknown8.60.000001
EIF4E2Eukaryotic translation initiation factor 4E member 2XM_343616.2986.03.00.000001
EMP1Epithelial membrane protein 1NM_012843.24172.36.30.000064
GPR37G protein-coupled receptor 37NM_057201.1452.25.80.000030
IER5LImmediate early response 5-likeXM_001079588.
MAN2B1Mannosidase 2, alpha B1NM_199404.11924.74.00.000001
MARCH3Membrane-associated ring finger 3NM_001007759.11852.54.30.000001
MGMTO-6-methylguanine-DNA methyltransferaseNM_012861.11196.93.20.008704
MKLN1Muskelin 1NM_031359.1458.63.70.000010
MPZL1Myelin protein zero-like 1NM_001007728.11381.33.30.000001
NNTNicotinamide nucleotide transhydrogenaseNM_001013157.1251.53.20.000001
PLOD2Procollagen lysine, 2-oxoglutarate 5-dioxygenase 2NM_175869.2897.63.10.000039
UBE2TUbiquitin-conjugating enzyme E2TXM_001062580.11346.730.20.000005
  • RNO, rat chromosome.

Genes stimulating proliferation through their role in cell cycle regulation would be expected to be upregulated also in early-stage tumor tissue. Indeed, 15 genes that have various roles during cell proliferation displayed highly elevated expression in ENU-induced rat MPNSTs (Table 2). Among them, cyclin D1 (Ccnd1), which mediates the proliferative response to cAMP in Schwann cells, showed a 200-fold up-regulation compared with trigeminal nerve tissue and was strongly expressed in all ENU-induced MPNST on the protein level (Fig. 4) (16). Interestingly, a marked upregulation of the Tgfb1 gene was also detected. This was accompanied by elevated expression of a group of genes, namely Cspg2/Vcan (17), Postn/Osf2 (18), or Tgfbi, a paralog of Postn (19), all of which have been described as induced by transforming growth factor-β1 (Tgfb1), indicating an activation of the Tgfb pathway. Apart from its role in cell proliferation, the TGFs (Tgfb1, Tgfb2, and Tgfb3) have also been described to be present in the cyst fluid of transplanted ENU-induced rat MPNST and may play an immunosuppressive role by inhibiting lymphocyte proliferation so that tumor cells escape immuno-surveillance (20, 21). Because we previously showed that T lymphocytes invade trigeminal nerves as a consequence of tumor initiation and seemed to be involved in the resistance mechanism against ENU-induced MPNST development in BDIV rats, upregulation of Tgfb1 in MPNST seems plausible (22, 23).

View this table:

Genes Significantly Downregulated in Ethylnitrosourea-Induced Malignant Peripheral Nerve Sheath Tumors Compared With Normal Trigeminal Nerve Tissue

Chromosomal Location
SymbolGene NameGenBank IDRNOMbFold Changep-Value
CNTFCiliary neurotrophic factorNM_013166.11215.820.80.000136
MBPMyelin basic proteinNM_001025293.11879.035.90.014727
MAP1BMicrotubule-associated protein 1bXM_215469.4230.410.70.000195
SNCASynuclein, alphaNM_019169.2489.632.80.000493
TAC1Tachykinin 1NM_012666.1432.736.80.001130
AOX1Aldehyde oxidase 1NM_019363.2956.817.60.002535
Extracellular region
TGFATransforming growth factor alphaNM_012671.14120.46.30.000420
CHGBChromogranin BNM_012526.13120.65.90.008129
TUBA4Tubulin, alpha 4NM_001007004.1974.525.90.013571
LECT1Leukocyte cell-derived chemotaxin 1NM_030854.11560.935.60.011099
Glutathione metabolism
GSTM3Glutathione S-transferase, mu type 3NM_031154.12203.516.80.000387
Golgi membrane
NTRK1Neurotrophic tyrosine kinase, receptor, type 1NM_021589.12179.841.60.000691
FUT8Fucosyltransferase 8 fucosyltransferase)NM_001002289.16100.07.10.006179
Heterotrimeric G-protein complex
RGS7Regulator of G-protein signaling 7NM_019343.11390.715.00.006259
Ion channel complex
KCNS3Potassium voltage-gated channel S member 3NM_031778.2634.521.80.007466
SCN10ASodium channel, voltage-gated, type 10, alphaNM_017247.18124.621.20.000576
SCN11ASodium channel, voltage-gated, type XI, alphaNM_019265.28124.721.10.002923
SCN4BSodium channel, voltage-gated, type IV, betaNM_001008880.1848.136.80.005867
Membrane fraction
SLC17A6Solute carrier family 17, member 6NM_053427.11101.521.30.000322
FDFT1Farnesyl diphosphate farnesyl transferase 1NM_019238.21542.45.00.000759
CD24CD24 antigenNM_012752.22047.524.20.005478
MALmal, T-cell differentiation proteinNM_012798.13115.243.90.000865
GNAOGuanine nucleotide binding protein, alpha oNM_017327.11911.515.80.006533
NBL1Neuroblastoma, suppression of tumorigenicity 1NM_031609.15Unknown12.00.000001
Neuron projection
FBXO2F-box only protein 2NM_053511.15165.222.30.006879
SNAP25Synaptosomal-associated protein 25NM_030991.13124.939.30.000268
ANK1Ankyrin 1, erythroidXM_240464.31673.325.50.000317
PIB5PAPhosphatidylinositol bisphosphate 5-phosphatase, ANM_133562.11484.120.90.008625
SNCGSynuclein, gammaNM_031688.11610.020.60.000395
CABP1Calcium-binding protein 1, transcript variant 3NM_001033676.11242.722.20.002197
STMN2Stathmin-like 2NM_053440.2295.327.60.003760
GABBR2Gamma-aminobutyric acid B receptor 2NM_031802.1563.236.30.001984
P2RXL1Purinergic receptor P2X-like 1, orphan receptorNM_012721.11185.420.10.009438
NEF3Neurofilament 3, mediumNM_017029.11547.720.00.001899
CHRNA3Cholinergic receptor, nicotinic, alpha polypeptide 3XM_001072823.1854.940.20.003179
MGLLMonoglyceride lipaseNM_138502.24122.924.60.000276
HTR3A5-Hydroxytryptamine receptor 3aNM_024394.18Unknown38.50.001119
RGD1559440Ca2+-dependent activator for secretion protein 2XM_231528.4449.723.10.000367
Tight junction
CLDN19Claudin 19NM_001008514.15139.840.80.001479
EEF1A2Eukaryotic translation elongation factor 1 alpha 2NM_012660.23170.316.00.001032
  • RNO, rat chromosome.


Hierarchical cluster analysis of gene expression comparing ethylnitrosourea-induced malignant peripheral nerve sheath tumors (T1–T12), microtumor (MT), and normal trigem-inal nerve (WT) tissue. See dendrogram above the heat map. RNAs used for hybridizations are arranged in columns; individual genes (91 unique probe sets) are in rows. Expression signal strength is indicated by color (red, high expression; green, low expression). Color scale of arbitrary signal strength expression is shown on the right. Red asterisks indicate gene expression data that were confirmed by real-time polymerase chain reaction; green asterisks indicate additional immunohistochemistry.

Validation of Microarray Data by RT-PCR

To confirm the gene expression data obtained by microarray analysis, we conducted quantitative RT-PCR assays for 20 genes displaying the strongest expression differences between MPNST and normal trigeminal nerve tissue (Fig. 2). According to the expression array data, Aard, Cspg2, Ednrb, Itga1, Mlph, Moxd1, Postn, Prc1, Silv, Spon1 were significantly upregulated in MPNST compared with trigeminal nerve tissue, whereas Aox1, CD24, Cldn19, Cntf, Fxyd2, Gstm3, Lect1, Mal, Mbp, Nbl1 exhibited higher expression in control nerve tissue than in the tumors. These data could be reproduced by RT-PCR (Fig. 3).


Validation of microarray data by real-time polymerase chain reaction. The mRNA-expression of 10 genes was examined in ethylnitrosourea-induced malignant peripheral nerve sheath tumors (MPNSTs) (T1–T9) versus untreated trigeminal nerve tissue (WT). Bar plots show expression in each MPNST and normal trigeminal nerve, box plots show median values (horizontal bars in red boxes: MPNST; black boxes represent normal nerve tissue). (A) Genes upregulated in MPNST. (B) Genes downregulated in MPNSTs. According to the Wilcoxon signed-rank test, the expression of these genes was significantly different between MPNST and WT, p < 0.01.

Immunolocalization of the Cyclin D1 Protein in MPNST Versus Trigeminal Nerve Tissue

Abundant amounts of Ccnd1 mRNA in all MPNSTs investigated in contrast to low levels in trigeminal nerve tissue point to an important role this gene may play in PNS onco-genesis. Therefore, we investigated whether this difference is reflected by a corresponding protein expression pattern. Immunohistochemical investigations on MPNST induced in BDIX.BDIV-Mss4b using an anti-cyclin D antibody showed strong, mainly nuclear staining in every tumor (n = 8); this staining was almost totally absent in 85-day-old trigeminal nerve tissue (n = 7) (Fig. 4 A, B).


(A) Ethylnitrosourea-induced malignant peripheral nerve sheath tumors strongly express cyclin D1, as predicted by elevated levels of mRNA detected by expression profiling and real-time polymerase chain reaction. (B) A trigeminal nerve from an 85-day-old rat lacks cyclin D1 expression. Original magnification for both: 400×.


Preventive measures and curative cancer therapies counteracting the abnormal biology of cancer cells would ideally require target signal transduction pathways aberrant in every preneoplastic and/or full-blown tumor cell. However, it is evident that most somatic genetic mutations are not present ubiquitously within a tumor, and molecular signatures that are meant to predict a patient's outcome vary within the same tumor (2).

With this study, we aimed at detecting molecular alterations at the level of gene transcription tightly associated with early stages of tumor progression in MPNST. Ethylnitrosourea-induced MPNSTs in rats of the BD strains predominantly arise in the trigeminal nerves close to the brain-nerve junction. They invite analysis of early genetic and epigenetic alterations propelling tumor progression. Because of their anatomic localization in the skull, they become clinically evident early when they measure about a few millimeters in diameter and when the tumor cells still are comparatively homogeneous. This surpasses the possibilities offered by human MPNSTs, which usually display a several thousandfold larger volume at detection and have a high degree of intratumoral heterogeneity as a result of secondary cellular diversification caused by tumor evolution and adaption (3).

Before global gene expression profiling, we performed a detailed histologic analysis of ENU-induced MPNST to investigate the morphologic correlates of dysregulated gene expression and the similarity to that in human MPNST. The MPNSTs investigated histologically had arisen in different BDIX-derived rat strains and in a BDIV rat (11). The light microscopic findings differed regarding tumor stage but did not vary depending on the genetic background. In the great majority of tumors, there was a conventional spindle cell phenotype with variable degrees of cellularity, pleomorphism, and mitotic activity. This is in accordance with findings of other authors (5, 14). Most tumors expressed the S100 protein focally, as is seen in human MPNST (15). As expected, only the histology of the few most advanced ENU-induced rat MPNSTs resembled the average human tumor; all other ENU-induced tumors of the trigeminal nerves represented early tumor stages.

Studies of global gene expression in early-stage MPNSTs that arose on various genetic backgrounds compared with genetically matched normal trigeminal nerve tissue of adult rats should allow identification of genes that are crucial for early malignant progression of Schwann cells. These genes should be expressed to a similar extent in all MPNSTs independent of the genetic background, and the strength of expression should differ significantly from that of wild-type nerve tissue. Accordingly, we identified 2 sets of genes that were either markedly upregulated or downregulated in all MPNST specimens. These differentially expressed genes were functionally classified into different biologic categories. Several of them are involved in signaling pathways playing important roles in malignant transformation.

Tissue remodeling in ENU-induced MPNST first indicated by a marked softening of tumor tissue compared with normal nerve and later by complete destruction of normal nerve structure was reflected by increased expression of genes belonging to the functional classes of extracellular matrix molecules, cell adhesion proteins, and proteases in all ENU-induced MPNSTs investigated. Many of these are known to play an important role in human cancer, including Col16a1, Col18a1, Itga1, Itga4, Ecm1 fibrillin/Fbn2, Spon1, Mmp11, Mmp16, and Mmp17 (Table 2).

In addition, there were genes with unexplained functions in carcinogenesis that were upregulated in ENU-induced MPNSTs, such as monooxygenase1 (Moxd1), which was more than 160-fold overexpressed, and Aard, a gene of unknown function that is highly expressed in mouse testis that has never previously been associated with the malignant phenotype (24).

Most of the genes that were downregulated in ENU-induced MPNSTs are associated with the differentiated state of normal nerve tissue, that is, myelinating and nonmyelinating Schwann cells, the precursor cells which are thought to represent the cells of origin for MPNSTs (25). Another group of genes that were downregulated are normally expressed in nerve tissue components but not represented in the tumors, for example, ganglion cells with their axons and synapses (Table 3). Transcripts of genes such as Mbp, which encodes a structural myelin protein, and Scn4b, which is responsible for action potential initiation and propagation in excitable cells, were essentially not present in MPNST tissue; those involved in syn-aptic processes, such as transcripts of Synpr, Snap25, and Htr3a, were also not present.

In addition, the telomeric region on chromosome 10 (107–110, 7 Mb), which is frequently deleted in ENU-induced rat MPNSTs (10), harbors 2 genes that are significantly down-regulated: neuronal pentraxin 1 (LOC497675/NPTX1) and Fructosamine 3 kinase (Fnsk/Fn3k).

In summary, the molecular signature characteristic of very early stage MPNST includes upregulation of proliferation-associated genes, genes involved in tissue remodeling and of different partially unknown function as well as the down-regulation of genes maintaining the differentiated Schwann cell phenotype and the absence of transcripts being associated with nerve tissue components not represented in MPNST-like ganglion cells, axons, and synapses. In comparison with the gene expression profiles of human MPNSTs, it is not surprising that the expression of only a few genes in the rat MPNSTs were dysregulated in a similar way.

Sporadic and Nf1-associated human MPNSTs could not be distinguished by gene expression profiling (26). Thus, the different etiology of ENU-induced rat MPNSTs versus human tumors might be more indicative of the different tumor stages. We performed the present study because we intended to detect pathways that are deregulated in early malignant progression and as a consequence would be detectable in most if not all tumor cells. Therefore, the molecular signature of ENU-induced rat MPNSTs at the mRNA level might correspond to the expression profile of a small subset of human MPNSTs that might possibly be detected in the early stages only by accident.

Moreover, in most studies, gene expression of MPNSTs or MPNST cell lines is compared with profiles of Schwann cell lines or primary cultured Schwann cells (25, 27, 28). In contrast, gene expression profiles obtained for primary ENU-induced trigeminal MPNST tissue in this study were compared with those of regular trigeminal nerve tissue. Because the 1.5 mm of nerve tissue adjacent to the brain-nerve junction that was obtained was devoid of ganglion cells and blood vessels, these pieces of tissue seemed to constitute an appropriate control.

With the strategy used in this study, we meant to rule out expression artifacts caused by culture conditions. It is known that cells that are part of a 3-dimensional tissue network show different expression patterns than they do in monolayer tissue culture. Although it was previously shown that the expression of many genes in MPNSTs and MPNST cell lines corresponded to each other, the expression of stem cell and mature cell markers varied considerably in the MPNST cell line S462, depending on whether it was growing adherently or in spheres (27, 29). Whereas spheres seemed to be enriched in stemlike cells and exhibited a neural crest signature, adherent cells seemed to be more differentiated. Nevertheless, some genes found upregulated in a fraction of human MPNSTs, such as BUB 1, CASP3, CCNB2, CCND1, CDC6, CDC20, CTGF, FOXM1, KIF4, MCM6, PBK, POLE, POSTN, or TTK, were also identified in the present study.

Having gained insight into critical pathways underlying early malignant progression of rat Schwann cells will enable us to target, for example, Tgfb signaling in the same rat model to prevent ENU-induced MPNST growth and/or to treat more advanced lesions. These experiments will determine whether targets defined in early lesions represent effective therapies.


The authors thank Aydah Sabah at the Max Planck Institute of Molecular Genetics, Berlin, Germany, for hybridization of expression arrays.


  • This work was supported by the Wilhelm Sander Stiftung, Munich, Germany (Grant No. 2005.093.1 to Andrea Kindler-Röhrborn).

  • Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Web site (http://jnen.oxfordjournals.org/).


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
View Abstract