The role of EXT and growth signalling pathways in osteochondroma
and its progression towards secondary peripheral chondrosarcoma
Hameetman, L.
Citation
Hameetman, L. (2007, April 26). The role of EXT and growth signalling pathways in
osteochondroma and its progression towards secondary peripheral chondrosarcoma.
Department Pathology, Faculty of Medicine / Leiden University Medical Center (LUMC),
Leiden University. Retrieved from https://hdl.handle.net/1887/11865
Version: Corrected Publisher’s Version
License: Licence agreement concerning inclusion of doctoral thesis in the
Institutional Repository of the University of Leiden
Downloaded from: https://hdl.handle.net/1887/11865
Liesbeth Hameetman, Karoly Szuhai,
Ayse Yavas, Jeroen Knijnenburg,
Mark van Duin, Herman van Dekken,
Antonie H.M. Taminiau,
Anne-Marie Cleton-Jansen,
Judith V.M.G. Bovée,
Pancras C.W. Hogendoorn
The role of EXT1 in nonhereditary
osteochondroma: identification of
homozygous deletions 3
Journal of the National Cancer Institute,
(2007); 99: in press
Abstract
Background: Multiple osteochondromas is a hereditary syndrome that is characterized by the formation of cartilage-capped bony neoplasms (osteochondromas), for which exostosis (multiple)-1 (EXT1) has been identified as a causative gene. However, 85% of all osteochondromas present as solitary (nonhereditary) lesions in which somatic mutations in EXT1 are extremely rare, but loss of heterozygosity and clonal rearrangement of 8q24 (the chromosomal locus of EXT1) are common. We examined whether EXT1 might act as a classical tumour suppressor gene for nonhereditary osteochondromas.
Methods: Eight nonhereditary osteochondromas were subjected to high-resolution array- based comparative genomic hybridization (array-CGH) analysis for chromosome 8q. The array-CGH results were validated by subjecting tumour DNA to multiple ligation-dependent probe amplification (MLPA) analysis for EXT1. EXT1 locus-specific fluorescent in situ hybridization (FISH) was performed on nuclei isolated from the three tissue components of osteochondroma (cartilage cap, perichondrium, bony stalk) to examine which parts of the tumour are of clonal origin.
Results: Array-CGH analysis of tumour DNA revealed that all eight osteochondromas had a large deletion of 8q; five tumours had an additional small deletion of the other allele of 8q that contained the EXT1 gene. MLPA analysis of tumour DNA confirmed these findings and identified two additional deletions that were smaller than the limit of resolution of array- CGH. FISH analysis of the cartilage cap, perichondrium, and bony stalk showed that these homozygous EXT1 deletions were present only in the cartilage cap of osteochondroma.
Conclusion: EXT1 functions as a classical tumour suppressor gene in the cartilage cap of nonhereditary osteochondromas.
Context and Caveats
Prior knowledge Mutations in the EXT1 gene cause Multiple osteochondromas, a rare hereditary disorder characterized by multiple benign bone tumours. However, it is not known whether the EXT1 gene is also involved in the more common solitary (nonhereditary) osteochondromas.
Study design Molecular cytogenetic study of eight nonhereditary osteochondromas and the three tissue components (cartilage cap, perichondrium, and bony stalk) of one osteochondroma.
Contribution The authors show that biallelic inactivation of EXT1, which is one of the hallmarks of a classical tumour suppressor gene, occurs in nonhereditary osteochondromas, specifically in the cartilage cap.
Implications All osteochondromas, both hereditary and nonhereditary, are neoplastic and develop as a result of the complete loss one of the EXT genes.
Limitations Due to the rarity of osteochondromas, few tumour samples were examined, and only one yielded sufficient material for analysis of tissue components. LOH due to mitotic recombination without copy number alterations could not be detected by the used techniques.
Osteochondroma is the most common benign bone tumour and represents approximately 50% of all surgically treated benign bone tumours 1. Osteochondromas arise at the external surface of bones that are formed by endochondral ossification, and they consist of a cartilage cap that covers a bony stalk that is continuous with the underlying bone. Chondrocytes located in the cartilage cap have the same spatial organization as those in the epiphyseal growth plate and also undergo endochondral ossification 1. The cartilage cap is covered by a perichondrium that is continuous with the periosteum of the underlying bone. Approximately 15% of osteochondromas occur in the context of Multiple osteochondromas, a hereditary disorder that is characterized by multiple osteochondromas, the number of which can vary substantially between and within families, and is inherited in an autosomal dominant manner2,3, but the vast majority of osteochondromas present as solitary (nonhereditary) lesions.
Multiple osteochondromas is caused by mutations in either of two genes: exostosis (multiple)-1 (EXT1; Online Mendelian Inheritance in Man [OMIM] No. 133700), which is located on chromosome 8q24.11–q24.13, and exostosis (multiple)-2 (EXT2; OMIM No.
133701), which is located on chromosome 11p11–12 4-6. Both genes are ubiquitously expressed4-6. Most of the germline mutations that have been identified in the EXT1 and EXT2 genes lead to premature truncation of the EXT proteins and the loss of protein function [reviewed in 7]. Most hereditary osteochondromas have been reported to be heterozygous for one of the mutations in the EXT genes 8-10. However, the demonstration that some hereditary osteochondromas, in addition to carrying an EXT1 mutation, exhibit loss of the remaining wild-type allele of EXT1 11, is consistent with Knudson’s two-hit model of tumourigenesis 12, and indicates that EXT1 acts as a classical tumour suppressor gene in Multiple osteochondromas.
Somatic mutations in the EXT genes are extremely rare in nonhereditary osteochondromas and have been described in only three cases 13-15, one of which 15 was a nonhereditary secondary peripheral chondrosarcoma, which is a malignant cartilaginous
tumour arising from a preexisting osteochondroma 16. However, the observation that loss of heterozygosity (LOH) and clonal rearrangement at 8q24 (the EXT1 locus) are as frequent in nonhereditary osteochondromas as are EXT1 gene mutations in patients with hereditary osteochondromas suggests that a gene on 8q24—most likely EXT1—is involved in the development of nonhereditary osteochondroma 11,17,18. By contrast, LOH at the EXT2 locus has been reported in only one nonhereditary osteochondroma 18 and, to our knowledge, no somatic mutations in EXT2 have been identified.
In a previous study 19, we examined EXT1 and EXT2 mRNA expression in hereditary and nonhereditary osteochondromas and found that patients with hereditary multiple osteochondromas who had a germline mutation in either of the EXT genes had decreased mRNA expression of the corresponding EXT gene in their tumours compared with the expression found in normal epiphyseal growth plates. By contrast, in nonhereditary tumours, in which EXT1 or EXT2 gene mutations were absent, only EXT1 mRNA expression was decreased.
The gene products of EXT1 and EXT2, the EXT 1 and EXT2 proteins, form a hetero- oligomeric complex in the Golgi apparatus, where they function in heparan sulphate proteoglycan (HSPG) biosynthesis 20. HSPGs are large, multifunctional macroproteins that are involved in several growth signalling pathways in the epiphyseal growth plate 21,22. We previously found that decreased EXT1 or EXT2 mRNA expression in osteochondromas and chondrosarcomas was associated with intracellular accumulation of HSPGs in the Golgi apparatus. By contrast, in growth plates, where expression of HSPG is extracelluler, there is normal expression of the EXT genes 19. It has been shown that a lack of HSPGs at the cell surface affects growth signalling pathways in the growth plate 23 and, possibly, those in osteochondromas 24,25.
The decrease in EXT1 mRNA expression that we observed in nonhereditary osteochondromas 19 suggests that the loss of EXT1 mRNA expression is important for the development of these tumours. However, the fact that we and others 11,14,19,26,27 could find no evidence for somatic mutations or promoter methylation at the EXT1 gene in such tumours implies that other mechanisms may be used to inactivate EXT1 and decrease its mRNA expression. The LOH and clonal rearrangement found at 8q24 in nonhereditary osteochondromas might target the EXT1 gene itself or a regulatory element that affects EXT1 gene transcription. To evaluate this hypothesis, we subjected a series of nonhereditary osteochondromas to high-resolution array-based comparative genomic hybridization (array- CGH) and multiplex ligation-dependent probe amplification (MLPA) analysis. We also examined which part of the osteochondroma (the cartilage cap, bony stalk, or perichondrium) is of clonal origin using EXT1 locus-specific fluorescent in situ hybridization (FISH).
Materials and Methods Patient Material
Frozen and paraffin-embedded tumour tissues were retrieved from the tissue bank of the Leiden University Medical Center. Fresh-frozen material from the cartilage cap was available for eight patients with nonhereditary osteochondromas. Tumour-related data were obtained by review of pathology specimens and reports. We used bone scans and radiographs to confirm that the selected tumour samples were from patients who had only a single
documented osteochondroma lesion. All samples were handled in a coded fashion, following the medical ethical guidelines described in the Code Proper Secondary Use of Human Tissue established by the Dutch Federation of Medical Sciences. Four tumours (those from patients L312, L657, L673, and L1247) were included in previous studies, in which they had been subjected to EXT mutation analysis 11,19. These results were included in this study.
Four patients (L312, L657, L1520, and L1649) provided peripheral blood leukocytes, which were used as the source of normal germline DNA. Normal connective tissue originating from the resection specimen was available for one patient (L1247) and was used as the source of normal germline DNA for that patient. Normal germline DNA was not available for three patients.
DNA and RNA Isolation
Sections (4-µm thick) of frozen tumour tissue were stained with haematoxylin–eosin to identify regions of the tumour sample in which at least 70% of the cells were tumour cells of the cartilage cap; tissue from those regions was used to isolate DNA and RNA. DNA from tumour samples and the connective tissue sample of L1247 was isolated with the use of a Wizard genomic DNA purification kit (Promega, Madison, WI). A salting-out procedure was used to extract normal germline DNA from freshly collected peripheral blood leukocytes 28. Total tumour RNA was isolated as described previously 29.
Mutation Analysis
We screened the entire coding sequences of the EXT1 and EXT2 genes in normal germline DNA (from five patients) and tumour-derived DNA from all eight osteochondromas by direct sequence analysis, as previously described 30.
Array-Based Comparative Genomic Hybridization Analysis of Chromosome 8q
A high-resolution 8q array containing a tiling bacterial artificial chromosome (BAC) clone set was constructed from a previously published clone set 31. This clone set spanned the entire long arm of chromosome 8 and consisted of 618 clones selected from the RPCI-11 library of BAC clones available from the BACPAC Resource Center (Children’s Hospital Oakland Research Institute, Oakland, CA). Control BAC clones (n = 189) representing nonchromosome 8q locations were selected from the same library and used to normalize the signal intensities to the normal chromosome copy number (namely, two) 31. Genomic positions of the BAC clones were retrieved from the University of California Santa Cruz (UCSC) Genome Browser web page (http://genome.cse.ucsc.edu/), which used the Human May 2004 Assembly.
Polymerase chain reaction (PCR) amplification and spotting of the clones, hybridization, and image acquisition procedures were performed as previously described 32 with minor modifications. Briefly, each PCR-amplified clone set was spotted in triplicate per slide.
Genomic DNA (150 ng) was labelled with fluorescent dCTP in an overnight random primer labelling reaction (BioPrime Random Prime Labeling Kit, Invitrogen, Carlsbad, CA). Test DNA (from tumour cells of the cartilage cap or normal leukocytes) and sex-mismatched reference DNA (Promega) were differentially labelled with Cy3-dCTP and Cy5-dCTP, respectively (GE- Healthcare, Amersham, UK). Labelled DNA samples were pooled and precipitated in the presence of Cot1 DNA (Invitrogen). The precipitated DNA was resuspended in a formamide-
based buffer and hybridized to the slides containing the 614 8q-specific and 189 control BAC clones for at least 48 hours, after which the slides were washed and dried.
The slides hybridized with labelled DNAs were scanned at 532 and 635 nm for Cy3 and Cy5, respectively, with the use of a GenePix Personal 4100A scanner (Axon Instruments, Union City, CA) and spot intensities were measured with the use of GenePix Pro 4.1 software.
This software integrates pixel intensities for each spot and excludes spots for which the intensity of the reference DNA was less than five times that of the local background or spots for which more than 3% of the pixels were saturated. The fluorescent intensity ratios of the test (Cy3) and the reference (Cy5) samples of the entire clone set were normalized against the median of the fluorescent intensity ratios of the 189 control BAC clones by using a routine that we developed in Microsoft Excel 2000. For each clone, the average value from triplicate spots and its standard deviation were calculated. Only clones for which at least two of the three spots had fluorescent intensity ratios within the 20% confidence interval (CI) [empirically established and used in previous studies 31,32] of the average intensity were used for further analysis. To display the data, the log2 value of the normalized average ratio was calculated for each clone. The threshold for gains and losses for each target was defined as
±0.33 on a log2 scale.
Multiplex Ligation-Dependent Probe Amplification
All osteochondromas were subjected to a two-colour MLPA assay (using a kit supplied by Service XS, Leiden, The Netherlands) that was to designed to identify deletions in the EXT1 and EXT2 genes that comprise several exons, as previously described 30. We used this assay to verify array-CGH analysis results for the EXT1 locus and to identify small deletions and amplifications of EXT1 that were beyond the limit of resolution of the tiling BAC array clone set (i.e., < 150 kilobases [kb]). Each sample was analyzed in duplicate and the assay was performed according to the supplier’s instructions. Briefly, in a one-tube reaction, combinations of two adjacently annealing oligonucleotide probes were hybridized and ligated. Two probes for unlinked loci were included as a reference. After ligation, the common ends of the probes served as a template for PCR amplification with one primer pair of which the forward primer was labelled with a fluorescent dye. The resulting labelled PCR products were separated according to size using capillary electrophoresis on an ABI 3130 sequencer (Applied Biosystems, Forster City, CA). Data analysis of the different PCR products was performed with the use of GeneScan sequencing software (Applied Biosystems). The height of each exon-specific peak was divided by the sum of the heights of the two reference peaks to produce a ratio. The thresholds for genomic gains and hemizygous losses for each target were defined as ratios of higher than 1.2 and smaller than 0.8, respectively, on a linear scale. Homozygous deletions were defined as ratios smaller than 0.3 (i.e., 0.8-0.5 [the ratio of hemizygous loss]).
Fluorescence In Situ Hybridization of Isolated Nuclei
To investigate which component(s) of the osteochondromas harboured a homozygous EXT1 deletion, we performed FISH on tumour cell nuclei that were isolated, as previously described33, from the cartilage cap and perichondrium components of EDTA-decalcified formalin-fixed, paraffin-embedded tumour tissue from patient L1649. FISH was also performed on paraffin-
Quantitative Reverse Transcription–Polymerase Chain Reaction
For first-strand complementary DNA synthesis, 1 µg of total tumour RNA was reverse transcribed by avian myeloblastosis virus reverse transcriptase (Roche Applied Science) with the use of 100 ng oligo (dT)15 primer (Roche Applied Science) and 50 ng random primers (Invitrogen), according to the manufacturer’s instructions. The sequences of the primers that were used for quantitative reverse transcription–polymerase chain reaction (qPCR) amplification of the RNAs for the EXT1 gene and for the genes to normalize the EXT1 mRNA expression (CPSF6, SRPR, GPR108, and HNRPH1) 19 as well as the expected sizes of the embedded sections of tumour tissue from patient L1649 to investigate the different cell types in the bony stalk component of osteochondromas. After deparaffinization, the sections were pretreated with 0.01 M citrate (pH 6.0) at 80 °C for 80 minutes and predigested with 1 M sodium thiocyanate at 80 °C for 10 minutes, followed by incubation with 100 µg/ml RNase (Roche Applied Science, Penzburg, Germany) at 37 °C for 1 hour, followed by sequential enzymatic digestions with 0.02% collagenase (5 minutes at 37 °C) and 0.4% pepsin ([pH 2.0] at 37 °C for 15 minutes). After these treatments, the slides were rinsed in phosphate- buffered saline, dehydrated in an ethanol series, and dried in air.
We used DNA prepared from BAC clone RP11-357E22 and from fosmid clones G248P86305D6 and G248P8030G11 (obtained from BACPAC Resource Center) as probes for FISH. DNA was isolated from the clones with the use of a High Pure PCR template preparation kit (Roche Applied Science) and labelled with either digoxigenin-11-dUTP or biotin-16-dUTP (both from Roche Applied Science) with the use of a BioPrime Random Prime Labeling Kit. The labelled clones were mixed with an alpha satellite DNA probe 34 specific for the centromere of chromosome 8 that had been directly labelled with Cy5-dUTP by nick translation, as previously described 35. Hybridization of the probe mixture to isolated nuclei and indirect detection of the hapten-conjugated probes were performed according to standard protocols, as previously described 36. Digital fluorescence imaging and image analysis were performed as previously described 37.
Table III-I. Quantitative reverse transcription–polymerase chain reaction primers*
* NCBI = National Center for Biotechnology Information; FW = forward primer; RV = reverse primer.
† CPSF6, GPR108, HNRPH1, and SRPR were used as reference genes to normalize the mRNA expression of EXT1. These genes were selected because their expression remains unchanged in cartilaginous tumors24,38.
amplification products are provided in table III-I 24,38. qPCR was performed with the use of a qPCR Corekit for SybrGreen (Eurogentec, Seraing, Belgium) on an iCycler (BioRad, Hercules, CA). All samples were measured in duplicate. Fluorescent signals generated by PCR were collected in real time and translated to quantitative values using the iCycler software according to the manufacturer’s instructions.
Statistical Analysis
Normalization and data analysis were performed as previously described 38. Briefly, geometric averaging of expression of the four reference genes was performed to acquire reliable normalization of EXT1 gene expression using the geNorm programme 39. This method provides Figure 3.3. Fluorescent in situ hybridization (FISH) analysis of leukocytes and interphase nuclei isolated from paraffin-embedded material from case L1649. (A) Example of a metaphase chromosome spread (upper panel) that was prepared from normal leukocytes and hybridized with two fosmid clones that covered the EXT1 gene (red), Bacterial Artificial Chromosome (BAC) clone RP11-357E22 which contains DNA complementary to a region close to EXT1 (green), and a chromosome 8 centromeric probe (white).
DNA was stained with 4’,6-diamidino-2-phenylindole (DAPI) (blue). The yellow colour indicates overlap of the red and green signal. All signals are present in duplicate and represent normal cell content. Magnifications of the two chromosomes 8 are shown in the middle two panels. Bottom three panels show the three fluorescent signals of the three different probes separately. (B) A haematoxylin–eosin-stained section of osteochondroma from L1649 (left panel) and FISH results of interphase nuclei from the indicated regions of the osteochondroma (right panels). Nuclei from the cartilage cap lacked the red signal, indicating that they had completely lost the DNA complementary to the EXT1 fosmid clones. However, these nuclei continued to display the FISH signals for the BAC clone close to EXT1 locus and the chromosome 8 centromere. Nuclei isolated from all other components of this osteochondroma displayed the expected number of FISH signals.
The images of nuclei from osteoblasts and osteocytes were obtained from a paraffin section that was subjected to FISH analysis. The light blue staining adjacent to the labelled osteocyte nucleus is autofluoresence from bone matrix.
a normalization factor that is representative for the amount of mRNA in each sample. Because osteochondromas histologically resemble the epiphyseal growth plate 1, expression levels in the tumours were related to those of four normal growth plates, for which the EXT1 expression had been determined in a previous study 19. The average of the EXT1 expression level in the growth plates was set to 1. P values were computed by Student’s t test and were considered statistically significant when less than or equal to .05.
Results
Selection of Patient Material
Patients with a nonhereditary osteochondroma were selected based on review of pathologic, clinical, and radiologic data. Fresh-frozen material of nonhereditary osteochondromas was available for eight patients. table III.II summarizes the clinical and tumour related data for these patients.
Table III.II. Characteristics of nonhereditary (solitary) osteochondroma patients*
* The absence of other osteochondromas was based upon review of clinical and radiologic data.
† Normal DNA was obtained from peripheral blood except for patient L1247, for whom DNA was isolated from muscle tissue originating from the resection specimen.
Mutation Analysis of the EXT1 and EXT2 Genes
To confirm that none of patients included in this study was actually a Multiple osteochondromas patient with very mild symptoms (i.e., only one osteochondroma documented) and that none of the tumours had somatic mutations in either the EXT1 or EXT2 gene, we analyzed the entire coding sequences of the EXT1 and EXT2 genes in normal germline DNA (from five patients) and tumour-derived DNA from all eight osteochondromas by direct sequencing.
Apart from known polymorphisms (data not shown), no somatic or germline mutations were detected in either tumour or normal DNA.
High Resolution Array-CGH Analysis of Chromosome 8q in Nonhereditary Osteochondromas We next examined chromosome 8q in the series of eight nonhereditary osteochondromas in more detail with a high-resolution array that contained a tiling BAC clone set. All eight nonhereditary osteochondromas had a large hemizygous deletion of 8q as detected by array- CGH analysis (table III.III). For seven tumours (L657, L673, L1247, L1455, L1520, L1587, and L1649), this deletion extended to the telomere. The smallest region of overlap spanned
approximately 2 megabases (Mb), from 8q24.11 (118.45 Mb position) to 8q24.12 (120.66 Mb position) and comprised six genes, including EXT1. In five tumours (L657, L673, L1455, L1520, and L1649), we identified a homozygous deletion that covered the EXT1 gene, which indicated that a second independent deletion event had occurred in this region. The homozygous deletions were always smaller than the hemizygous deletions, and ranged from
Figure 3.1. Genomic profiles of chromosome 8q of nonhereditary osteochondromas, as determined by high resolution array-based comparative genomic hybridization (array-CGH) analysis. Test DNA, isolated from the cartilage cap of osteochondroma or from normal leukocytes, and reference DNA were differentially labelled and hybridized on an array with a tiling bacterial artificial chromosome (BAC) clone set for chromosome 8q. After normalization of the fluorescent signals, the ratio of test and reference DNAs were log2- transformed. A ratio of 0 indicates that the test sample had two copies of the DNA represented by the BAC clones, similar to the reference DNA.
Ratios between -0.3 and -0.8 indicate a hemizygous deletion in the test DNA, and ratios between -0.9 and -4 indicate homozygous deletions. The data are shown as the genomic distance (in megabases) on the x-axis plotted against the log2-transformed test/reference ratios on the y-axis. Arrows indicate homozygous deletion of BAC clones. (A) Osteo- chondroma L1520. This tumour had a large deletion of the distal part of 8q and a homozy- gous deletion of the DNA represented by one BAC clone (RP11-1061J15), which covered exon 1 of the EXT1 gene. (B) Osteochondroma L657.
This tumour had a large deletion of the distal part of 8q and a homozygous deletion of the DNA represented by four BAC clones, which covered the region where EXT1 is located.
(C) Osteochondroma L1649. This tumour had a large deletion of the distal part of 8q and a homozygous deletion of a region spanning six BAC clones, which included the EXT1 gene.
(D) Osteochondroma L673. This tumour had multiple deletions along 8q and a homozygous deletion of 10 BAC clones, which also covered the EXT1 gene. (E) Normal leukocytes from patient L1587. Note the absence of chromosome gains or losses. At the bottom is shown an ideogram of chromosome 8q in which the location of the EXT1 gene is indicated.
0.16 Mb (covered by clone RP11-1061J15) to 1.8 Mb (covered by 11 clones) (figure 3.1, A–D). None of the samples showed chromosomal gains.
For DNA from tumour samples that have low levels of contamination with normal cells, the typical values of the log2-normalized average ratios for hemizygous and homozygous deletions are -1 and less than -2, respectively, on a log2 scale. However, the observed ratios were between -0.3 and -0.8 for hemizygous deletions and between -0.9 and -4.0 for homozygous deletions. The observed ratios are larger than expected and indicate that most tumour samples indeed contained amounts of normal cell contamination. Normal cells still had 2 copies of chromosome 8q.
All five of the patients for whom we had normal constitutional DNA (i.e., L312, L657, L1247, L1520, and L1649) had a normal genomic profile at 8q (figure 3.1,E). This finding indicated that in these nonhereditary osteochondromas, the 8q deletions, including the homozygous deletions, were of somatic origin.
* The clones covering the beginning and end of the deleted regions and the genomic regions they comprise are given; – = not applicable (no homozygous deletion was identified by array-CGH for these tumor samples); SRO = Smallest region of overlap among all nonhereditary osteochondromas examined.
† Genomic positions were retrieved from the UCSC Genome Browser web page (http://genome.cse.ucsc.edu/), which used the Human May 2004 Assembly.
Table III.III. Summary of 8q chromosomal regions that that were deleted in nonhereditary osteochondromas*
Verification of Array-CGH Results by Multiple Ligation-Dependent Probe Amplification Analysis of EXT1
We next used MPLA analysis to verify the array-CGH results and to facilitate identification of homozygous deletions involving individual exons, which are typically beyond the limit of resolution of the tiling array clone set. MLPA analysis of DNA from the leukocytes of a healthy donor was used to generate a normal profile for the EXT1 MLPA probe set (figure 3.2,A).
MLPA analysis revealed the loss of at least one copy of the EXT1 gene in all eight osteochondromas (figure 3.2). MLPA analysis of tumour DNA also revealed homozygous
deletions of the entire EXT1 gene in osteochondromas from patient L673 (data not shown) and L1649 (figure 3.2,B). These deletions were also identified by the array-CGH analysis.
MPLA analysis also identified a homozygous deletion of exon 1 in osteochondroma from patient L1520 (figure 3.2,C), a result that was consistent with array-CGH results that showed homozygous deletion of the region covered by a single BAC clone, RP11-1061J15, which only included exon 1 of the EXT1 gene. In osteochondroma from patient L312, MLPA analysis revealed the complete loss of exons 2, 3, and 4 (figure 3.2,D), identifying a homozygous deletion of at least 7.3 kb. Osteochondroma L1587 showed complete loss of exon 6 Figure 3.2. Multiple ligation-dependent probe amplification (MLPA) analysis of the EXT1 gene in nonhereditary osteochondromas. DNA isolated from the cartilage cap of the osteochondromas was subjected to the MLPA assay for EXT1. Pairs of oligonucleotide probes of every exon (e.g., for exon 1, two probe pairs 1-1 and 1-2) of EXT1, complemented with two probe pairs for unlinked loci (labelled C), were hybridized, ligated, and polymerase chain reaction (PCR) amplified with fluorescent primers. PCR products were separated using capillary electrophoresis and an ABI 3130 genetic analyzer (Applied Biosystems), and data on the PCR products were analyzed in GeneScan. The peak height was measured in arbitrary units (AU) and displayed in A-D. (A) Peak pattern of the EXT1 MLPA probe set and two references peak of DNA isolated from leukocytes of a healthy donor, demonstrating normal peak heights of the EXT1 exon products, referred to as normal profile. All exons are numbered. (B) Osteochondro-ma L1520. The ratio between the exon peak and the two control peaks for all EXT1 exon peaks was reduced by approximately 50% compared with the normal profile, and the two peaks for exon 1 (arrows) were lost completely, indicating homozygous deletion of exon 1. (C) Osteochondroma L1649. All EXT1 exon peaks were lost completely. (D) Osteochondroma L312. The ratio between the exon peaks and the two control peaks for all EXT1 exon peaks was reduced by approximately 50% compared with the normal profile compared with the height of the corresponding peaks in the normal profile and the peaks for exons 2, 3, and 4 (arrows) were almost completely lost, suggesting a homozygous deletion of these exons. (E) Schematic representation of all MLPA results. The peak heights of each EXT1 exon–
specific peak were normalized against the two reference peaks. The normalized peak height thresholds for gains and hemizygous losses for each target were defined as more than 1.2 and less than 0.8, respectively, on a linear scale. Homozygous deletions were defined as less than 0.3. The control sample is the leukocytes sample shown in (A). MLPA analysis of L1247 and L1455 failed due to technical difficulties. White indicates no loss of alleles; light gray indicates the loss of one allele; and dark gray indicates the loss of both alleles.
(figure 3.2,E), demonstrating a homozygous deletion of at least 119 bp (the size of this exon). Unlike the array-CGH analysis, MPLA analysis of osteochondroma from patient L657 did not show a homozygous deletion of the EXT1 gene, but only a hemizygous deletion. This result probably reflects contamination of the tumour DNA sample with normal DNA, as was suggested by the log2 ratios found for hemizyous (-0.6) and homozygous deletion (-1.2) in the array-CGH experiments (figure 3.1,B). The PCR amplification for MLPA analysis of osteochondromas from patients L1247 and L1455 failed twice; because of the limited amount of material available, it was impossible to obtain MLPA data for these cases.
EXT1 Locus-Specific Fluorescent In Situ Hybridization Analysis of Osteochondroma Tissue Components
Array-CGH analysis was performed with DNA isolated from cells of the cartilage cap, which is known to be of clonal origin 11,18. However, osteochondromas consist of three components: a cartilage cap, a bony stalk, and a perichondrium 1. Of the latter two components, it is still debated whether they are also of clonal or reactive/remodelled host bone origin. To examine whether the bony stalk and perichondrium of osteochondroma also harboured the homozygous deletion of EXT1, we performed FISH analysis on formalin-fixed paraffin-embedded material of patient L1649. It was not possible to perform EXT1 locus–specific FISH of the three tissue components in a single paraffin section because they behaved differently during pretreatment of the tissue section. Instead, we performed FISH on paraffin sections of the bony stalk only and on isolated nuclei of formalin-fixed, paraffin-embedded microdissected tissues containing the either cartilage cap or the perichondrium.
For FISH analysis, two fosmid clones covering the EXT1 gene were selected as EXT1 locus–specific probes. We used BAC clone RP11-357E22, which contains DNA located close to the EXT1 locus that was not altered in tumour DNA from patient L1649, and a centromeric probe for chromosome 8 as controls, allowing us to detect chromosome 8 and alterations in both EXT1 loci. FISH performed on normal metaphase spreads showed fluorescent signals for each of the three probes on two chromosomes, both of which were identified as chromosome 8 (figure 3.3,A, see page 56). FISH signals for the fosmid clones were not observed in the nuclei from the cartilage cap, confirming the homozygous deletion of EXT1 (figure 3.3,B). By contrast, we observed two FISH signals of the EXT1 fosmid clones in nuclei isolated from the perichondrium of case L1649, indicating that both copies of the EXT1 gene were still present (figure 3.3,B). We also observed two FISH signals for the EXT1 fosmid clones in osteoblasts, osteocytes, bone marrow, and epithelial cells in the bony stalk (figure 3.3,B). The FISH results showed that the homozygous deletion of EXT1 was found only in nuclei of the cartilage cap and not in nuclei of the perichondrium and the bony stalk. Thus, in osteochondroma only the cartilage cap appears to be of clonal origin.
EXT1 mRNA Expression in Osteochondromas
Finally, to evaluate whether the homozygous deletion of EXT1 indeed abolished its mRNA expression, we examined EXT1 mRNA expression in the eight nonhereditary osteochondromas by qPCR and compared them with expression of four normal growth plates. For seven of the osteochondromas, sufficient amounts of RNA were available for determining EXT1 mRNA expression by qPCR. For each of these tumours, the EXT1 mRNA expression level is represented
Table III.IV. EXT1 mRNA expression in nonhereditary osteochondromas*
* EXT1 expression is presented as fraction of the average EXT1 mRNA expression of four growth plates;
ND = not determined, because the RNA isolation failed.
as the fraction of the average EXT1 mRNA expression in the growth plate (table III.IV). The average EXT1 mRNA level was statistically significantly lower in osteochondromas than in four normal epiphyseal growth plates (0.12 versus 1.00; difference = 0.88, 95% CI = 0.51 to 1.15, P = .001 [Student’s t-test]). Thus, the decreased EXT1 mRNA expression in osteochondroma is consistent with the homozygous deletions of EXT1 that we identified with array-CGH and MLPA analysis.
Discussion
In this study, we identified homozygous deletions of the EXT1 gene in seven of eight nonhereditary osteochondromas by two types of analyses, array-CGH and MLPA. The smallest homozygous deletion identified by array-CGH analysis covered the EXT1 promoter region and exon 1, the largest exon of the gene 4; however, the higher resolution of the MLPA technique allowed us to identify homozygous deletions as small as single exons. All of the homozygous deletions we identified resulted from a large deletion comprising a large part of the chromosome arm on one allele and a small deletion of only the EXT1 gene (or part of it) on the other allele. Frei et al. have suggested that the small deletions were more likely to be the consequence of erroneous recombination events than to represent a somatic point mutation or replication errors that commonly occur in stretches of similar nucleotides 40.These results indicate that two distinct events (either simultaneous or consecutive) involving both alleles of chromosome 8 occurred in these tumours.
All of the nonhereditary osteochondromas in our study demonstrated physical loss of 8q24 (and thus the EXT1 gene). By contrast, complete loss of the wild-type EXT1 gene by homologous recombination of the mutated allele has been described in hereditary osteochondromas 11 (figure 4.4). We speculate that, because small single–base pair mutations are rare in nonhereditary osteochondromas 13-15, physical loss of 8q24 is necessary to inactivate both copies of the EXT1 gene (figure 4.4).
To our knowledge, this is the first report to identify multiple cases of homozygous deletion of EXT1 in osteochondromas. Our results resolve questions about the role of EXT1 as potential tumour suppressor gene in the development of osteochondromas. An ongoing debate in the literature has been whether EXT1 fits the classical two-hit model for tumour suppressor genes 11,12 or whether haploinsufficiency for EXT1 via mutational inactivation (i.e., the loss of one allele) is sufficient for the formation of osteochondromas 8. This so- called haploinsufficiency model was based on studies in which 90% of the hereditary osteochondromas demonstrated heterozygous mutations in EXT1 or EXT2 8-10. However, we have clearly demonstrated that biallelic inactivation of EXT1, which is one of the hallmarks of a classical tumour suppressor gene 41, also occurs in nonhereditary osteochondromas.
Previous studies 11,17,18 have demonstrated that the cartilage cap of osteochondroma is of clonal origin and thus is neoplastic. The clonal origin of the cartilage cap is confirmed by our identification of homozygous deletions in EXT1 in DNA isolated from the cartilage cap.
However, it has remained unclear whether cells that form the bony stalk of an osteochondroma and the overlying perichondrium are also of clonal origin. By performing FISH analysis on the different components of an osteochondroma, we demonstrated that nuclei isolated from cells comprising the cartilage cap harboured a homozygous deletion of EXT1, whereas nuclei isolated from cells comprising the perichondrium and bony stalk had no such deletion. Thus, these results demonstrate that the perichondrium and the bony stalk of osteochondromas are not neoplastic.
Our finding that the cartilage cap is the only neoplastic component of osteochondroma revives a longstanding debate about the cell of origin of osteochondromas. Several theories have been suggested over the years 42,43, but so far, to our knowledge, no compelling evidence supporting any one of them has been reported in the literature. Recently, it was suggested that specific cells in the perichondrium may give rise to chondrocytes that are necessary for the development and continued growth of osteochondromas 44. However, this suggestion is inconsistent with our FISH results for the different components of osteochondroma, which suggest that the cell of origin most likely resides in the growth plate.
Figure 3.4. Schematic representation of the genomic alterations on chromosome 8q in hereditary and nonhereditary osteochondro- mas. According to Knudson’s two-hit model for tumour suppressor genes 12, both alleles of EXT1 must be inactivated for hereditary (Multiple osteochondromas) or nonhere- ditary (solitary) osteochondroma formation.
For hereditary osteochondromas, after inactivation of the first allele (∆EXT1), inactivation of the second allele can be achieved either by physical loss of the remaining wild-type (WT) allele or by homologous recombination of the mutated allele. We show that in nonhereditary osteochondromas both WT alleles are lost, usually by loss of 8q and a small EXT1 deletion, resulting in homozygous EXT1 deletion.
Conventional LOH analysis, which has been used to analyze osteochondromas in the past, usually does not detect homozygous deletions 11,17,18. To our knowledge, only one other study 14 has identified a homozygous deletion of EXT1 in a single nonhereditary osteochondroma. Even tumours that have little contamination by normal cells show retention of heterozygosity by conventional LOH analysis, which severely hampers the distinction
between homozygous deletion and retention of heterozygosity. Homozygous deletions are typically detected only in cell lines that completely lack contamination 45. Our results show the benefit of using high-resolution array-CGH analysis with a tiling clone set to identify homozygous deletions.
Recently we demonstrated that the loss of EXT expression in osteochondromas results in the disordered distribution of HSPGs 19. In these osteochondromas, HSPGs were no longer present at the cell surface but accumulated in the cytoplasm, where they were concentrated in the Golgi apparatus. HSPGs are known to be involved in several signalling pathways in the growth plate 21. It is possible that the lack of HSPGs at the cell surface in osteochondromas might have a functional effect on HSPG-dependent signalling pathways, e.g., Indian hedgehog (IHH) signalling. In the growth plate, IHH requires interaction HSPGs to diffuse through the extracellular matrix to its receptor 23. However, two recent studies 24,25 showed that IHH signalling was still present in osteochondromas despite the absence of HSPGs at the cell surface. Benoist-Lasselin et al. 25 also demonstrated that IHH was expressed in all cells of the cartilage cap, whereas IHH expression in the growth plate is restricted to the transition zone46. We speculate that osteochondroma cells circumvent the impaired diffusion capacities that result from diminished amounts of HSPGs at the cell surface by producing IHH in every cell of the cartilage cap, resulting in cell-autonomous (i.e., autocrine) IHH signalling. Other HSPG-dependent growth signalling pathways that are affected in osteochondroma are the parathyroid hormone–like hormone signalling pathway, which is a downstream target of IHH46, and the fibroblast growth factor signalling pathway 47.
Our study has several limitations. One is the infrequent occurrence of osteochondromas, which inhibited the collection of a larger series of fresh frozen tumours. Furthermore, the low cellularity and excess of extracellular matrix hampered DNA and RNA analyses. Therefore, we were unable to extend the series reported here. In general, array-CGH and MLPA are perfectly suitable for detecting physical loss of genomic regions 48. However, LOH due to mitotic recombination without copy number alterations cannot be detected by these techniques and will require complementary LOH analysis.
In conclusion, our identification of the homozygous deletion of EXT1 in the cartilaginous cells of nonhereditary (solitary) osteochondromas proves that EXT1 acts as a classical tumour suppressor gene in osteochondroma formation and supports the two-hit model for osteochondroma development. The absence of this deletion in the perichondrium and the bony stalk indicates that the cartilage cap constitutes the neoplastic lesion and that both hereditary and nonhereditary osteochondromas arise from cells of the growth plate rather than from the perichondrium, as previously suggested 44. Our results indicate that all osteochondromas develop as result of loss of EXT1 and subsequent abrogation of HSPG- dependent signalling pathways.
Notes
The Dutch Cancer Society (grant number: RUL 2002-2738), financially supported the experiments performed for the study. The departments of Pathology of the Leiden University Medical Center and Erasmus Medical Center are partners of the EuroBoNeT consortium, a European Commission granted Network of Excellence for studying the pathology and genetics of bone tumours, which financed the collaboration between the different research groups
who participated in this study. The funding agencies had no role in the design of the study;
the collection, analysis, or interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication.
We thank Marja van den Burg for expert assistance with the array experiments.
The chromosome 8 idiograms shown in figure4.4 were obtained from
http://www.pathology.washington.edu/research/cytopages. This study was presented at Human Genome Organisation’s 11th Human Genome Meeting.
References
1. Khurana J, Abdul-Karim F, Bovée JVMG. (2002) Osteochondroma. In World Health Organization classification of tumours. Pathology and genetics of tumours of soft tissue and bone, Fletcher CDM, Unni KK, Mertens F (eds). IARC Press: Lyon. pp. 234-236
2. Wicklund LC, Pauli RM, Johnston D, Hecht JT. (1995) Natural history study of hereditary multiple exostoses. Am J Med Genet 55:43-46
3. Bovée JVMG, Hogendoorn PCW. (2002) Multiple osteochondromas. In World Health Organization classification of tumours. Pathology and genetics of tumours of soft tissue and bone, Fletcher CDM, Unni KK, Mertens F (eds). IARC Press: Lyon. pp. 360-362
4. Ahn J, Ludecke H-J, Lindow S, Horton WA, Lee B, Wagner MJ, Horsthemke B, Wells DE. (1995) Cloning of the putative tumour suppressor gene for hereditary multiple exostoses (EXT1). Nature Genet 11:137-143
5. Wuyts W, Van Hul W, Wauters J, Nemtsova M, Reyniers E, Van Hul E, De Boulle K, De Vries BBA, Hendrickx J, Herrygers I, et al. (1996) Positional cloning of a gene involved in hereditary multiple exostoses. Hum Mol Genet 5:1547-1557
6. Stickens D, Clines G, Burbee D, Ramos P, Thomas S, Hogue D, Hecht JT, Lovett M, Evans GA.
(1996) The EXT2 multiple exostoses gene defines a family of putative tumour suppressor genes.
Nature Genet 14:25-32
7. Zak BM, Crawford BE, Esko JD. (2002) Hereditary multiple exostoses and heparan sulfate polymerization. Biochim Biophys Acta 1573:346-355
8. Hall CR, Cole WG, Haynes R, Hecht JT. (2002) Reevaluation of a genetic model for the development of exostosis in hereditary multiple exostosis. Am J Med Genet 112:1-5
9. Bernard MA, Hogue DA, Cole WG, Sanford T, Snuggs MB, Montufar-Solis D, Duke PJ, Carson DD, Scott A, Van Winkle WB, et al. (2000) Cytoskeletal abnormalities in chondrocytes with EXT1 and EXT2 mutations. J Bone Miner Res 15:442-450
10. Legeai-Mallet L, Rossi A, Benoist-Lasselin C, Piazza R, Mallet JF, Delezoide AL, Munnich A, Bonaventure J, Zylberberg L. (2000) EXT 1 gene mutation induces chondrocyte cytoskeletal abnormalities and defective collagen expression in the exostoses. J Bone Miner Res 15:1489-1500
11. Bovée JVMG, Cleton-Jansen AM, Wuyts W, Caethoven G, Taminiau AHM, Bakker E, Van Hul W, Cornelisse CJ, Hogendoorn PCW. (1999) EXT-mutation analysis and loss of heterozygosity in sporadic and hereditary osteochondromas and secondary chondrosarcomas. Am J Hum Genet 65:689-698 12. Knudson AG, Jr. (1971) Mutation and cancer: statistical study of retinoblastoma. Proc Natl Acad Sci
USA 68:820-823
13. Hecht JT, Hall CR, Snuggs M, Hayes E, Haynes R, Cole WG. (2002) Heparan sulfate abnormalities in exostosis growth plates. Bone 31:199-204
14. Bernard MA, Hall CE, Hogue DA, Cole WG, Scott A, Snuggs MB, Clines GA, Ludecke HJ, Lovett M, Van Winkle WB, et al. (2001) Diminished levels of the putative tumor suppressor proteins EXT1 and EXT2 in exostosis chondrocytes. Cell Motil Cytoskeleton 48:149-162
15. Hecht JT, Hogue D, Wang Y, Blanton SH, Wagner M, Strong LC, Raskind W, Hansen MF, Wells D.
(1997) Hereditary multiple exostoses (EXT): mutational studies of familial EXT1 cases and EXT- associated malignancies. Am J Hum Genet 60:80-86
16. Bertoni F, Bacchini P, Hogendoorn PCW. (2002) Chondrosarcoma. In World Health Organisation classification of tumours. Pathology and genetics of tumours of soft tissue and bone, Fletcher CDM, Unni KK, Mertens F (eds). IARC Press: Lyon. pp. 247-251
17. Mertens F, Rydholm A, Kreicbergs A, Willen H, Jonsson K, Heim S, Mitelman F, Mandahl N. (1994) Loss of chromosome band 8q24 in sporadic osteocartilaginous exostoses. Genes Chromosomes Cancer 9:8-12
18. Bridge JA, Nelson M, Orndal C, Bhatia P, Neff JR. (1998) Clonal karyotypic abnormalities of the hereditary multiple exostoses chromosomal loci 8q24.1 (EXT1) and 11p11-12 (EXT2) in patients with sporadic and hereditary osteochondromas. Cancer 82:1657-1663
19. Hameetman L, David G, Yavas A, White SJ, Taminiau AHM, Cleton-Jansen AM, Hogendoorn PCW, Bovée JVMG. (2007) Decreased EXT expression and intracellular accumulation of HSPG in osteochondromas and peripheral chondrosarcomas. J Pathol 211:399-409
20. Esko JD, Selleck SB. (2002) Order out of chaos: assembly of ligand binding sites in heparan sulfate.
Annu Rev Biochem 71:435-471
21. Knudson CB, Knudson W. (2001) Cartilage proteoglycans. Semin Cell Dev Biol 12:69-78 22. Selleck SB. (2000) Proteoglycans and pattern formation. Sugar biochemistry meets developmental
genetics. T I G 16:206-212
23. Koziel L, Kunath M, Kelly OG, Vortkamp A. (2004) Ext1-dependent heparan sulfate regulates the range of Ihh signaling during endochondral ossification. Dev Cell 6:801-813
24. Hameetman L, Rozeman LB, Lombaerts M, Oosting J, Taminiau AHM, Cleton-Jansen AM, Bovée JVMG, Hogendoorn PCW. (2006) Peripheral chondrosarcoma progression is accompanied by decreased Indian Hedgehog (IHH) signalling. J Pathol 209:501-511
25. Benoist-Lasselin C, de Margerie E, Gibbs L, Cormier S, Silve C, Nicolas G, Lemerrer M, Mallet JF, Munnich A, Bonaventure J, et al. (2006) Defective chondrocyte proliferation and differentiation in osteochondromas of MHE patients. Bone 39:17-26
26. Ropero S, Setien F, Espada J, Fraga MF, Herranz M, Asp J, Benassi MS, Franchi A, Patino A, Ward LS, et al. (2004) Epigenetic loss of the familial tumor-suppressor gene exostosin-1 (EXT1) disrupts heparan sulfate synthesis in cancer cells. Hum Mol Genet 13:2753-2765
27. Tsuchiya T, Osanai T, Ogose A, Tamura G, Chano T, Kaneko Y, Ishikawa A, Orui H, Wada T, Ikeda T, et al. (2005) Methylation status of EXT1 and EXT2 promoters and two mutations of EXT2 in chondrosarcoma. Cancer Genet Cytogenet 158:148-155
28. Miller SA, Polesky HF. (1988) A simple salting out procedure for extracting DNA from human nucleated cells. Nucleic Acids Res 16:1215
29. Baelde HJ, Cleton-Jansen AM, van Beerendonk H, Namba M, Bovée JVMG, Hogendoorn PCW. (2001) High quality RNA isolation from tumours with low cellularity and high extracellular matrix component for cDNA microarrays: application to chondrosarcoma. J Clin Pathol 54:778-782
30. Vink GR, White SJ, Gabelic S, Hogendoorn PC, Breuning MH, Bakker E. (2004) Mutation screening of EXT1 and EXT2 by direct sequence analysis and MLPA in patients with multiple osteochondromas:
splice site mutations and exonic deletions account for more than half of the mutations. Eur J Hum Genet 13:470-474
31. van Duin M, van Marion R, Watson JE, Paris PL, Lapuk A, Brown N, Oseroff VV, Albertson DG, Pinkel D, De Jong P, et al. (2005) Construction and application of a full-coverage, high-resolution, human chromosome 8q genomic microarray for comparative genomic hybridization. Cytometry A 63:10- 19
32. Knijnenburg J, Szuhai K, Giltay J, Molenaal L, Sloos W, Poot M, Tanke HJ, Rosenberg C. (2005) Insights from genomic microarrays into structural chromosome rearrangements. Am J Med Genet 132A:36-40
33. Jordanova ES, Riemersma SA, Philippo K, Giphart-Gassler M, Schuuring E, Kluin PM. (2002) Hemizygous deletions in the HLA region account for loss of heterozygosity in the majority of diffuse large B-cell lymphomas of the testis and the central nervous system. Genes Chromosomes Cancer 35:38-48
34. Donlon T, Wyman AR, Mulholland J, Barker D, Bruns G, Latt S et al. Alpha satellite-like sequences at the centromere of chromosome #8. Am.J.Hum.Genet. 39, A196. 1986.
35. Wiegant J, Raap AK. (2001) Probe Labeling and Fluorescence In Situ Hybridization. In Current Protocols in Cytometry, Robinson JP, Darzynkiewicz Z, Dean PN, Orfao A, Rabinovitch PS, Stewart CC et al. (eds). John Wiley & Sons, Inc.: New York, NY. p. 8.3.1-8.3.21
36. Wiegant J. (2001) Immunocytochemical Detection. In Current Protocols in cytometry, Robinson JP, Darzynkiewicz Z, Dean PN, Orfao A, Rabinovitch PS, Stewart CC et al. (eds). John Wliey & Sons, Inc.: New York, NY. p. 8.4.1-8.4.18
37. Szuhai K, Knijnenburg J, Ijszenga M, Tanke HJ, Baatenburg De Jong RJ, Bas Douwes DP, Rosenberg C, Hogendoorn PC. (2004) Multicolor fluorescence in situ hybridization analysis of a synovial sarcoma of the larynx with a t(X;18)(p11.2;q11.2) and trisomies 2 and 8. Cancer Genet Cytogenet 153:48- 52
38. Rozeman LB, Hameetman L, Cleton-Jansen AM, Taminiau AHM, Hogendoorn PCW, Bovée JVMG.
(2005) Absence of IHH and retention of PTHrP signalling in enchondromas and central chondrosarcomas. J Pathol 205:476-482
39. Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A, Speleman F. (2002) Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol 3:research0034.1-0034.11
40. Frei K, Szuhai K, Lucas T, Weipoltshammer K, Schofer C, Ramsebner R, Baumgartner WD, Raap AK, Bittner R, Wachtler FJ, et al. (2002) Connexin 26 mutations in cases of sensorineural deafness in eastern Austria. Eur J Hum Genet 10:427-432
41. Sherr CJ. (2004) Principles of tumor suppression. Cell 116:235-246
42. Porter DE, Simpson AHRW. (1999) The neoplastic pathogenesis of solitary and multiple osteochondromas. J Pathol 188:119-125
43. Bovée JVMG, Hogendoorn PCW. (2000) Letter to the editor Re. Review Article entitled “The neoplastic pathogenesis of solitary and multiple osteochondromas”. J Pathol 190:516-517
44. Hecht JT, Hayes E, Haynes R, Cole WG, Long RJ, Farach-Carson MC, Carson DD. (2005) Differentiation-induced loss of heparan sulfate in human exostosis derived chondrocytes.
Differentiation 73:212-221
45. Kamb A, Gruis NA, Weaver-Feldhaus J, Liu Q, Harshman K, Tavtigian SV, Stockert E, Day III RS, Johnson BE, Skolnick MH. (1994) A cell cycle regulator potentially involved in genesis of many tumor types. Science 264:436-440
46. Vortkamp A, Lee K, Lanske B, Segre GV, Kronenberg HM, Tabin CJ. (1996) Regulation of rate of cartilage differentiation by indian hedgehog and PTH-related protein. Science 273:613-622 47. Bovée JVMG, Van den Broek LJCM, Cleton-Jansen AM, Hogendoorn PCW. (2000) Up-regulation of
PTHrP and Bcl-2 expression characterizes the progression of osteochondroma towards peripheral chondrosarcoma and is a late event in central chondrosarcoma. Lab Invest 80:1925-1933 48. Speicher MR, Carter NP. (2005) The new cytogenetics: blurring the boundaries with molecular
biology. Nat Rev Genet 6:782-792
