If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address correspondence to Roberto Jose Diaz, M.D., Ph.D., F.R.C.S.C., Department of Neurology and Neurosurgery, Montreal Neurological Institute and Hospital, 3801 Rue University, Room 109, Montreal, Quebec H3A 2B4, Canada.
Division of Neurosurgery, The Hospital for Sick Children, Toronto, Ontario, CanadaDepartment of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada
Department of Pediatric Neuro-Oncogenomics, German Cancer Consortium (DKTK) and German Cancer Research Center (DKFZ), Heidelberg, GermanyDepartment of Pediatric Oncology, Hematology, and Clinical Immunology, Medical Faculty, University Hospital Düsseldorf, Düsseldorf, Germany
Department of Pediatric Neuro-Oncogenomics, German Cancer Consortium (DKTK) and German Cancer Research Center (DKFZ), Heidelberg, GermanyDepartment of Pediatric Oncology, Hematology, and Clinical Immunology, Medical Faculty, University Hospital Düsseldorf, Düsseldorf, GermanyDepartment of Neuropathology, Medical Faculty, Heinrich-Heine University Düsseldorf, Düsseldorf, Germany
Division of Neurosurgery, St. Michael's Hospital, Toronto, Ontario, CanadaDivision of Neurosurgery, Department of Surgery, University of Toronto, Toronto, Ontario, Canada
Division of Neurosurgery, The Hospital for Sick Children, Toronto, Ontario, CanadaDepartment of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, CanadaDivision of Neurosurgery, Department of Surgery, University of Toronto, Toronto, Ontario, Canada
Patient-derived xenografts retain the genotype of the parent tumors more readily than tumor cells maintained in culture. The two previously reported clival chordoma xenografts were derived from recurrent tumors after radiation. To study the genetics of clival chordoma in the absence of prior radiation exposure we established a patient-derived xenograft at primary resection of a clival chordoma. Epicranial grafting of clival chordoma collected during surgery was performed. Tumor growth was established in a nonobese diabetic/severe combined immunodeficiency mouse and tumors have been passaged serially for seven generations. Physaliferous cell architecture was shown in the regenerated tumors, which stained positive for Brachyury, cytokeratin, and S100 protein. The tumors showed bone invasion. Single-nucleotide polymorphism analysis of the tumor xenograft was compared with the parental tumor. Copy number gain of the T gene (brachyury) and heterozygous loss of cyclin dependent kinase inhibitor 2A (CDKN2A) was observed. Heterozygous loss of the tumor-suppressor fragile histidine triad (FHIT) gene also was observed, although protein expression was preserved. Accumulation of copy number losses and gains as well as increased growth rate was observed over three generations. The patient-derived xenograft reproduces the phenotype of clival chordoma. This model can be used in the future to study chordoma biology and to assess novel treatments.
Chordoma is a very rare and slowly growing tumor arising in the ventral axial skeleton. Tumor cells express Brachyury, a transcription factor that is restricted topographically to nascent and migrating mesoderm and the notochord during development.
Classic chordoma coexisting with benign notochordal cell rest demonstrating different immunohistological expression patterns of Brachyury and galectin-3.
These tumors are exceptionally resistant to standard DNA-altering chemotherapy, limiting the effectiveness of adjuvant therapies subsequent to surgical resection.
To better understand the tumor response to specific adjuvant therapy, experimental models that allow the assessment of tumor cells in a living microenvironment are needed. Presently, four chordoma xenografts (two clival and two sacral) derived from primary tumor, without cell culture, have been reported in the literature (Table 1).
These tumors were established in the flanks of immunodeficient mice. The clival chordoma xenografts previously described have been derived from tumors with prior radiation exposure and therefore may harbor genetic aberrations that are not representative of the originating oncogenic mutations in the tumor. Therefore, we present the first clival chordoma primary xenograft derived from previously untreated tumor.
The patient-derived chordoma xenograft has been characterized histopathologically and by molecular analysis to determine if phenotype and genotype are maintained in serial generations. We previously identified frequent loss of fragile histidine triad (FHIT) protein expression in chordoma and postulated a potential role in chordoma pathogenesis.
Therefore, FHIT expression and genetic aberration at the FHIT locus were investigated in this patient-derived chordoma xenograft. FHIT is a well-characterized tumor suppressor and a potential target for therapy.
Patient consent for use of tumor material for research purposes was obtained and tissue collection was conducted as per approval by the St. Michael's Hospital Institutional Review Board. The patient was a 68-year-old man who presented with a 1-year history of neck pain and tongue weakness. Preoperative magnetic resonance imaging and computed tomography scan of the skull base showed a central mass in the lower clivus, extending to the level of the C2 vertebra. No evidence of metastatic disease was found. Heterogenous contrast enhancement, T1-weighted hypointensity, T2-weighted hyperintensity, and bony erosion were observed (Figure 1). An extended image-guided endoscopic transnasal transphenoidal surgical resection was performed. After ensuring sufficient tumor tissue had been obtained for pathologic diagnosis, several pieces of tumor were wrapped in sterile gauze and placed in a partially filled specimen container containing saline to keep the tissue moist. The consulting pathologist confirmed the diagnosis of classic chordoma, not otherwise specified. The tumor showed integrase interactor 1 and p53 expression by immunohistochemistry. The patient had rapid progression of disease after surgical treatment.
Figure 1Imaging features of originating clival chordoma. A: Axial T1-weighted magnetic resonance image (MRI) showing a midline clival mass (T). B: T1-weighted gadolinium-enhanced MRI showing a tumor (T) with heterogeneous enhancement. C: Axial T2-weighted MRI showing a hyperintense microcystic epidural tumor (T) in the region of the lower clivus. D: Axial noncontrast computed tomography showing bony erosion of the clivus (arrowhead).
The specimen container was transported on ice. Under a biological containment hood the tumor sample was divided and a portion was frozen in liquid nitrogen and stored at −80°C. The remainder of the tissue was minced (≤1-mm fragments) and suspended in 750 μL of phosphate-buffered saline (PBS) and kept cold on ice. Two nonobese diabetic/severe combined immunodeficiency γ mice (The Jackson Laboratory, Bar Harbor, ME) were anesthetized by isoflurane gas and a small transverse incision (approximately 4 mm) was made over the posterior parietal skull and suboccipital musculature (n = 2) or over the flank (n = 1). The parietal bone was scraped with the scalpel to thin the outer cortex. The 250 μL of the minced tissue suspension was mixed with 250 μL of Matrigel (Becton Dickinson, Franklin Lakes, NJ) and then implanted by wide-bore pipette into the subcutaneous epicranial space above the posterior parietal bone and suboccipital musculature or into the subcutaneous space over the flank. The incisions were closed with absorbable suture and the mice recovered. Serial inspection of the wound and implantation site was performed on a daily basis for 1 week and then on a weekly basis. Mice received 5 mg/kg ketoprofen analgesic and a bolus of 0.5 mL 0.9% saline subcutaneously in the immediate period after surgery.
Serial Passage of Tumors and Growth Assessment
Animal experiments were approved by the Hospital for Sick Children Animal Care Committee (AUP0204-H) and conducted in accordance with the Ontario Animals for Research Act and the Canadian Council for Animal Care guidelines. Palpable tumors were measured with a digital caliper to estimate the tumor volume (in mm3) using the following formula: tumor volume = length (mm) × width2 (mm2)/2.
Serial transplantation of tumors was performed when the tumor exceeded 1.5 cm in maximum diameter. Mice were anesthetized using isoflurane gas and the tumor was excised and immediately placed into cold PBS for mincing. The tumor bulk was divided into four equal portions: one portion was frozen in liquid nitrogen, one portion was fixed in 4% formaldehyde or cryopreserved in liquid nitrogen isopentane for histologic analysis, and the remaining two portions were resuspended in 1 mL PBS with 1 mL Matrigel. From the PBS:Matrigel mix, 400 μL was implanted into the epicranial space above the posterior parietal bone and suboccipital musculature of anesthetized nonobese diabetic/severe combined immunodeficiency γ mice (n = 4) as per the procedure described in Initial Implantation Procedure. This process was continued for subsequent generations of tumor growth with two modifications. After P1, the implantation ratio was changed to one tumor to three mice for P2 to P4. From P5 and onward, the implantation ratio was one tumor to two mice in order to achieve faster tumor regeneration. The implantation volume was changed to a minced tissue suspension of 75 μL PBS and 75 μL Matrigel to allow for a smaller epicranial pocket. Tumor growth was monitored on a weekly basis and tumor volume measurements were initiated every 15 days when tumors reached a minimum of 0.2 mm in maximum diameter. Mice were euthanized for collection of tumor if any pain, severe disability, or tumor maximum diameter greater than 1.5 cm was reached in accordance with institutional humane animal use protocols. Upon death, the dorsal subcutaneous space was opened and inspected for invasion of the skull by tumor.
Immunohistochemical Analysis
Paraffin tissue sections (5 μm) were deparaffinized, rehydrated, and pretreated in citrate buffer, pH 6.0, for 15 minutes. Sections were blocked in 10% goat serum and incubated in primary antibodies for 1 hour at room temperature: Brachyury (1:100, 04-135; Millipore, Burlington, MA), FHIT (1:100, HPA018909; Sigma, St. Louis, MO), pan-cytokeratin (1:100, NBP2-29429; Novus Bio, Littleton, CO), Ki67 (1:50, CRM 325 A; Biocare Medical, Pacheco, CA), INI1 (1:50, 612110; BD Biosciences), and p53 (1:50, 554294; BD Biosciences). Flash-frozen OCT-embedded tumor tissue sections (8 μm) were blocked in 10% goat serum and incubated with anti-nuclei antibody clone 235-1 (1:100, MAB1812; Millipore)
for 1 hour at room temperature. This antibody recognizes human nuclei, but not mouse or rat nuclei. After incubation with primary antibody and washing three times, biotinylated secondary antibody 1:100 (ABC kit; Vector Labs, Burlington, Ontario, Canada) was applied for 30 minutes at room temperature. Sections then were exposed to the avidin-biotin detection system for 40 minutes at room temperature. After three washes, signal was detected using diaminobenzidine, counterstained in hematoxylin, and mounted in Permount (Thermo Fisher Scientific, Waltham, MA). Micrographs were captured with an Infinity1 camera (Lumenera Corporation, Ottawa, Ontario, Canada) with Infinity1 software visualized through an Olympus BX43 light microscope (Olympus, Tokyo, Japan). The Ki67 index was calculated by counting the number of positive nuclei in 10 high-power fields (20× objective).
Genomic DNA was isolated from tumor samples and snap-frozen in liquid nitrogen at the time of passaging. A proteinase K/SDS buffer was used to dissociate the tissue (10 mmol/L Tris-Cl [pH 8.0], 0.1 mol/L EDTA [pH 8.0], 0.5% (w/v) SDS, 20 mg/mL proteinase K) overnight at 55°C. Protein extraction was achieved with phenol-chloroform-isoamyl alcohol using phase-lock centrifuge gels (5-Prime, Gaithersburg, MD) and DNA was precipitated with 100% ethanol and 7.5 mol/L ammonium acetate. The pellet was washed with 75% ethanol and then dried. It was resuspended in Tris-ethanolamine (pH 8.0) buffer and quantified by NanoDrop (NanoDrop, Willmington, DE). Microarray processing was performed in The Centre for Applied Genomics, The Hospital for Sick Children. Briefly, genomic DNA was cleaved with Nsp and Sty nucleases, biotinylated, and hybridized to the Genome-Wide Human Single-Nucleotide Polymorphism (SNP) Array 6.0 (Affymetrix, Santa Clara, CA). Scanning of the microarray was performed on a Affymetrix GeneChip Scanner 3000. Copy number data were analyzed using Partek Genomic Suite 6.6 v6.15.1207 (Partek, St. Louis, MO). Samples were imported, background was normalized using Partek Defaults, and the copy number was calculated from allele-specific intensity. Genomic segmentation was performed using the following parameters: i) 150 minimum genomic markers were required; ii) P value threshold was 0.001; iii) signal-to-noise ratio was 0.3; and iv) diploid status was in the range of 1.7 and 2.3, losses had a copy number <1.7, and gains had a copy number >2.3. Copy number alteration counts were determined by summing the segmentation results <1.7 and >2.3.
Data sets GSE25720 and GSE9023 were downloaded for comparative analysis from GEO at the NCBI website (https://www.ncbi.nlm.nih.gov/geo). GSE25720 samples were analyzed on the Genome-Wide Human SNP Array 6.0 and data analysis was performed as described in the previous paragraph. For GSE9023 samples, which were processed using the klingenkhh BAC 32K Full array, normalized log2 ratios were imported into Partek Genomic Suite to be visualized. Genomic segmentation was performed using the following parameters: i) 10 minimum genomic markers were required; ii) P value threshold was 0.001; iii) signal-to-noise ratio was 0.3; and iv) diploid status was in the range of −0.15 and 0.15, losses had a copy number <−0.15, and gains had a copy number >0.15.
RNAscope in Situ Hybridization
The RNAscope in situ hybridization (ACDBio, Newark, CA) assay was performed on formalin-fixed, paraffin-embedded sections from the originating patient-derived xenograft (PDX0) and third-generation (PDX3) tumors. Briefly, tissue sections were deparaffinized and treated with hydrogen peroxide, followed by RNA target retrieval and protease treatment. After the sections were prepared, a commercial probe against human T gene mRNA was used to hybridize to the target, followed by staining of the targets with RNAscope 2.5 HD Red Detection reagent (ACDBio). Slides were counterstained with hematoxylin and mounted using EcoMount (BioCare Medical, Pacheco, CA). Whole slide scans were acquired using an Aperio AT2 Scanner (Leica Biosystems, Wetzlar, Germany) at ×40 magnification. Images then were analyzed using Definiens Developer XD software version 2.6 (Definiens, Munich, Germany), using a custom-programed algorithm to count the number of spots per nucleus (indicating single RNA molecules) in the entire tumor region.
Statistical Analysis
Prism 7 (GraphPad Software, Inc., La Jolla, CA) was used for statistical analysis. All measures are reported as means ± SEM. Means for survival time, tumor diameter, and Ki67 index were compared by an unpaired Welch t-test with a P value <0.05 set as significant. RNAscope data were compared using negative binomial regression.
Results
Human Clival Chordoma Can Be Regenerated in Nonobese Diabetic/Severe Combined Immunodeficiency Mouse Epicranial Space
Clival chordoma tissue obtained at the time of surgical resection (Figure 1) was implanted in a Matrigel (Corning, Corning, NY) slurry within the epicranial space over the posterior parietal bone and in the subcutaneous space of the flank of nonobese diabetic/severe combined immunodeficiency mice. After 4.5 months a palpable and visible tumor formed in the epicranial space in 2 of 2 mice (Figure 2). No tumor formed in the subcutaneous space at the flank of the third host mouse by 4.5 months. The tumor from the epicranial space from one of the mice was passaged serially and is now in the 7th generation (Table 2). The patient-derived xenograft has been made available to the research community through the Chordoma Foundation (https://www.chordomafoundation.org, last accessed September 4, 2017). On gross pathology, bony invasion was observed in each generation from PDX3 to PDX7; however, it was not apparent for every tumor within each generation (Table 2). Microscopic bony invasion consisted of nests of tumor cells surrounded by mesenchymal cells (Figure 2). A pattern of successive growth rate increase over successive generations was observed as the tumor to host transplantation ratio was changed from 1:4 to 1:2. The median time to the appearance of the first tumor measuring >0.2 mm in maximum diameter was 104 ± 12 days (n = 15) in PDX4 (1:3 transplant ratio) compared with 44 ± 11 days (n = 10) in PDX7 (1:2 transplant ratio) (P = 0.001). The estimated volumetric tumor growth rates showed a wide range of variability in the PDX4 compared with the PDX7 generation (Figure 3A). The Ki67 labeling index was significantly higher in the PDX7 generation (10% ± 2%) compared with parental tumor (3% ± 1%; P = 0.002) (Figure 3B). Along with the faster tumor growth a decrease in mouse survival was observed over generations, with mean survival times of 35 ± 3 weeks (n = 15) in PDX4 compared with 27 ± 1 week (n = 10) in PDX7 (P = 0.026).
Figure 2A: Vivisection of xenograft tumor (arrowheads) overlying the calvarium and extension to subcutaneous tissue overlying the thoracic spine. B: Hematoxylin and eosin–stained chordoma xenograft tumor (arrowheads) invading into mouse skull bone. Scale bar = 50 μm.
Figure 3A: Tumor growth curves for individual mice over time since implantation of tumor xenograft. Growth in the mice in the fourth generation (PDX4) and seventh generation (PDX7) is compared showing earlier tumor formation in PDX7. B: Histologic sections of original tumor, PDX4, and PDX7 showing Ki67 immunohistochemistry. Graph quantifying the Ki67 index in original tumor, PDX4, and PDX7. The Ki67 index is higher in PDX7 tumor compared with original tumor. **P < 0.01. Scale bars = 50 μm.
Expression of Chordoma Markers over Xenograft Generations
The xenograft tumor showed similar cytology as the original tumor with lobules of physaliferous cells (Figure 4). Nuclear Brachyury expression and cytokeratin expression was conserved over generations from original tumor to PDX3 (Figure 4). The original and PDX3 tumor did not express epithelial membrane antigen, as expected for a chordoma (Figure 4). The identity of the chordoma xenograft was confirmed by immunohistochemistry independently by Valasciences, Inc. (San Diego, CA). The physaliferous cells of the patient-derived xenograft expressed human FHIT in a cytoplasmic and perinuclear location, although cytoplasmic expression was prominent (Figure 5). FHIT expression was conserved over the first three generations. Absent anti-human FHIT immunolabeling of connective tissue components in the xenografts was observed, as would be expected owing to the antibody specificity (Figure 5). Concordantly, anti-human nucleus antibody staining of physaliferous cells, but not the connective tissue component of the xenograft tumors, was observed (Figure 5).
Figure 4Histologic sections of original tumor and third-generation xenograft (PDX3) showing immunohistochemistry positive for Pan-cytokeratin (Pan-CK) and nuclear Brachyury, and absence of epithelial membrane antigen (EMA). Tumor cytology showing lobules of physalipherous cells surrounded by a thin layer of spindle-like cells is preserved in the xenograft on hematoxylin and eosin (H&E) staining. Scale bars = 40 μm.
Figure 5A: Histologic section of parental chordoma immunolabeled with anti-human FHIT antibody. B: Histologic section of third-generation xenograft immunolabeled with anti-human FHIT antibody. Prominent immunolabeling with anti-human FHIT antibody is seen in physaliferous tumor cells (arrow), but not in adjacent mouse connective tissue (arrowheads). C: Histologic section of U87 human glioblastoma xenograft (negative control) immunolabeled with anti-human FHIT. D: Histologic section of human kidney (positive control) immunolabeled with anti-human FHIT. E: Histologic section of eighth-generation xenograft immunolabeled with anti-human nuclear antibody identifying physaliferous tumor cells as human and associated connective tissue (arrowheads) as originating from mouse. F: Histologic section of mouse muscle (negative control) immunolabeled with anti-human nuclear antibody. Scale bars = 20 μm.
Chordoma Genome Accumulates Copy Number Gains and Losses over Generations
To show the resemblance of the original tumor to other clival chordomas a karyogram was generated from the SNP data collected (Supplemental Figure S1). Copy number analysis shows gains and losses throughout the tumor genome (Figure 6A). The tumor established in the immunocompromised mouse at PDX0 and PDX3 carries the same key copy number aberrations as the original patient tumor, namely heterozygous loss of chromosome 9 (which includes the CDKN2A locus) and copy number gain of a region of chromosome 6 encompassing the T gene (brachyury) (Supplemental Figure S2 and Figure 6B). The PDX3 tumor showed significantly higher levels of T gene labeling per nucleus as compared with PDX0 (P < 0.001) (Figure 6C). Heterozygous loss of FHIT was observed over the first three generations (Supplemental Figure S3). The major genomic difference between the original tumor and the xenograft tumor are chromosome 19 losses (Supplemental Figure S4). Furthermore, copy number gains in chromosome 21 are lost in the transition from PDX0 to PDX3 (Supplemental Figure S5). The presence of gains or losses in 150 probe segments across the genome increased as follows: original tumor (n = 323), PDX0 (n = 937), and PDX3 (n = 1407). Unlike the MUG-Chor1 cell line
parental tumor, which does not show T gene copy number gain, the parental tumor in this study showed T gene copy number gain and this persisted in PDX1 and PDX3 (Supplemental Figure S6). MUG-Chor1 cells in passages 5 and 13 showed acquisition of T gene copy number gain. Among 11 chordoma tumor samples studied in GSE9023, only two samples showed T gene copy number gain. No highly recurrent copy number change was observed across tumor, cell line, and patient-derived xenograft samples. There was significant variation in copy number gains and losses across samples; however, chromosomal alterations in 1p, 3, 4, 7, 9p, 10, 12, 14, 19, and 22 were frequent. SNP data from the present study are available at (https://www.ncbi.nlm.nih.gov/geo; accession number GSE103342).
Figure 6A: Copy number losses (blue) and gains (red) across chromosomes in parental tumor (O), originating xenograft (PDX0), and third-generation xenograft (PDX3). B: Chromosomal map showing copy number gain of a segment of chromosome 6, which includes the T gene. C: RNAscope analysis using a probe targeting the T gene. Each spot indicates a single RNA molecule. A custom algorithm was used to count the number of spots per nucleus. Nuclei and spots were segmented by Definiens XD version 2.6 and delineated. PDX3 showed significantly higher (P < 0.001) levels of T gene labeling compared with PDX0, thus indicating copy gain of the gene. Scale bar = 10 μm. Chr6, chromosome 6.
There is growing interest in establishing validated models for investigation of novel therapeutics for chordoma, a tumor that is highly resistant to standard chemotherapy.
Here, we report the growth and serial passage in a nonobese diabetic/severe combined immunodeficiency mouse of a patient-derived clival chordoma that had not been irradiated previously. This is the third patient-derived clival chordoma model reported in the literature; however, it is the first derived from a previously nonirradiated tumor.
The advantage of this model is that it harbors the initiating mutations required for tumor formation. Tumor formation in the mouse was initially slow, taking longer than 3 months. However, with higher starting cell numbers, tumors now are palpable as early as 15 days after implantation. The slow growth of the tumor followed by a period of rapid expansion and subsequent death of the mice in all generations follows the same clinical course as observed in humans. The xenografts have the same cytology as the original tumor, namely physaliferous cells with round nuclei in lobules surrounded by connective tissue stroma (Figure 5E). The connective tissue within the xenografts is derived from the host mouse cells. Thus, physaliferous cells are able to direct the formation of supportive stroma. This is an important finding because this model would allow investigation of changes in the tumor microenvironment as a way of suppressing tumor growth. The xenograft tumor cells invade bone as is seen with human chordoma. The preoperative imaging of the parental tumor showed destruction of the clivus in an endophytic (type 1) pattern (Figure 1). Clival chordomas with this pattern of growth recently have been associated with differential expression of cell motility and invasion proteins compared with exophytic (type 2) clival chordomas.
The present xenograft clival chordoma model could be used to further study the cell interactions that result in the observed tumor cytology, signaling processes in the tumor microenvironment, and mechanisms of bony invasion. Studying tumor in its native environment avoids alterations in matrix protein expression and cellular behavior that may be seen in 2-dimensional cell culture systems.
The established tumor xenograft, which had not been radiated previously, also will be useful to assess genomic and proteomic changes that occur upon radiation treatment and to study the effect of radiation sensitizers.
At a genetic level the initial xenograft (PDX0) and its third generation (PDX3) show copy number changes compared with the parental tumor, indicating instability of the genome in this clival chordoma model. An associated increase in tumor growth was noticed over the PDX generations with an increased Ki67 index in PDX4 compared with the parental tumor. The preservation of copy number gains of a region of chromosome 6 involving the T gene locus in the clival chordoma PDX model reported here and in a sacral chordoma cell line (MUG-Chor1)
is striking because it was preserved through three generations in our model and through multiple passages in the MUG-Chor1 cell line. By using RNAscope analysis a propensity to acquire increased T gene copy number from the PDX0 to the PDX3 generation was identified. These observations support the importance of T gene expression in chordoma tumor maintenance. Brachyury expression has been shown to be important in the maintenance of chordoma cell lines.
In addition, the increased T gene copy number may contribute to the enhanced proliferation of later generations in the xenograft model because increased copy number has been associated with phosphatidylinositol 3-kinase/Akt pathway activation.
Brachyury gene copy number gain and activation of the PI3K/Akt pathway: association with upregulation of oncogenic Brachyury expression in skull base chordoma.
The SNP analysis performed on the patient-derived chordoma xenograft described here showed similar patterns of copy number gains and losses as that observed in SNP data from both chordoma cell lines and tumors. Previously, heterozygous loss of CDKN2A was reported in 58% and homozygous loss was reported in 12% of a set of spinal and sacral chordomas.
In this model, it was shown that CDKN2A is lost heterozygously, but as part of widespread heterozygous chromosome 9 loss. The 9p deletions, including the CDKN2A locus, occurred in 4 of 18 (22%) clival chordomas.
Similarly, heterozygous loss of FHIT is seen as part of the loss of the short arm of chromosome 3. FHIT encodes a tumor-suppressor protein that is lost in a number of cancers.
In mice lacking one allele of the FHIT gene, 78% developed forestomach tumors after a single dose of the carcinogen N-nitrosylbenzylamine and the spontaneous tumor formation occurred at an average of 0.94 tumors per mouse after 21 months.
Thus, a single-hit event such as mono-allelic FHIT loss may be sufficient to trigger chordomagenesis. Despite expression of FHIT protein as observed in the established xenograft, protein function may be altered because altered transcripts have been reported in the FHIT gene in cervix cancer.
The marked expression of FHIT in tumor cells suggests an important role of this protein in the cell biology of these cells, similar to expression seen in renal collecting tubule cells. It is interesting to note that both notochord cells, as well as the embryonic cells that give rise to intermediate mesoderm and subsequently renal collecting tubules, both showed expression of Brachyury.
Alterations of the mRNA transcript from the FHIT allele were not explored in this study; thus, the question of whether functional FHIT protein is present in this model remains unanswered at this time.
Large- and small-scale genomic alterations including chromosomal imbalance and increasing copy number gains/losses through the whole genome across the tumor generations from parental to the third generation were observed. This study is the first to show chromosomal genomic instability in chordoma as the xenograft tumor grows through multiple generations. It is uncertain if the accumulation of copy number changes is an inherent characteristic of all chordoma cells or the result of selection bias toward tumor cells that are more likely to adapt to a nonhuman extracellular environment owing to loss of DNA repair mechanisms. However, the finding that at least a subpopulation of chordoma cells show accumulation of copy number changes may explain the resistance to radiation and chemotherapy shown by this tumor. It is known that genomic instability imparts resistance to chemotherapy in other tumors such as melanoma and colorectal cancer.
We have established a patient-derived xenograft in immunocompromised mice from a previously nonirradiated clival chordoma. The tumor can be passaged and reproducibly generates a mass with cytology and immunohistology consistent with chordoma. Genomic gains/losses accumulate over generations, suggesting that chromosomal instability may play an important biological role in chordoma progression and chemotherapeutic resistance.
Acknowledgment
We thank the Chordoma Foundation for providing liaison support for independent immunohistochemical characterization of the xenograft.
Heterozygous loss of CDKN2A is maintained from parental to PDX0 and PDX3 tumor. Chr9, chromosome 9; hg, human genome; MTAP, methylthioadenosine phosphorylase.
Heterozygous loss of FHIT as part of a large 3p isochromosome deletion is maintained from parental to PDX0 and PDX3 tumor. Chr3, chromosome 3; hg, human genome; PTPRG, protein tyrosine phosphatase receptor type G.
Chromosome 19 (Chr19) single-nucleotide polymorphism map shows acquisition of large isochromosome deletions (blue) and copy number gain (red) in small regions near the centromere in the transition from parental tumor to the first generation.
Chromosome 21 (Chr21) single-nucleotide polymorphism map shows acquisition of isochromosome deletions (blue) in the q arm and loss of large copy gain regions (red) in the transition from PDX0 to PDX3.
Comparison of chordoma samples across multiple data sets. Top: Data from this study showing segmentation across the entire genome of clival tumor and two PDX generations of 0 and 3. Middle: Data from a study looking at copy number of the original sacral tumor and the genomic stability in the derived cell line over several passages. Bottom: Patient samples analyzed using a chromosome tiling array showing recurrent copy number alterations in chordoma tumor samples. Cases 1, 2, 3, 4, 8, 10, and 11 are sacral and cases 5, 6, 7, and 9 are clival.
Classic chordoma coexisting with benign notochordal cell rest demonstrating different immunohistological expression patterns of Brachyury and galectin-3.
Brachyury gene copy number gain and activation of the PI3K/Akt pathway: association with upregulation of oncogenic Brachyury expression in skull base chordoma.
Supported by Brainchild, Megan's Walk, the Laurie Berman and Wiley Family Funds for Brain Tumor Research (J.T.R.), and German Research Foundation/Deutsche Forschungsgemeinschaft grant RE 2857/2-1 (D.P. and M.R.).
30869-6&id=gr1.jpg){kind=link}
30869-6&id=gr2.jpg){kind=link}
30869-6&id=gr3.jpg){kind=link}
30869-6&id=gr4.jpg){kind=link}
30869-6&id=gr5.jpg){kind=link}
30869-6&id=gr6.jpg){kind=link}