Abstract
Germ-line and acquired mutations of the hSNF5/INI1 tumor suppressor gene have been reported in central nervous system (CNS), renal, and soft-tissue rhabdoid tumors. The present study was designed to compare the types of INI1 alterations among tumors from diverse anatomical sites and identify mutation hot spots. Fluorescence in situ hybridization and PCR-based microsatellite, heteroduplex, and sequence analysis were used to characterize chromosome 22 deletions and INI1 mutations among 100 primary rhabdoid tumors. Deletions and/or mutations of INI1 were detected in 75 patients, including 42 children with atypical teratoid/rhabdoid tumors of the brain or spinal cord and 6 children with a brain and a renal or soft-tissue tumor. Nineteen tumors arose in the kidney (in one child, bilaterally) and eight tumors were extra-renal. Homozygous deletions detected by fluorescence in situ hybridization were most often seen in CNS and extra-renal rhabdoid tumors, whereas truncating mutations were detected in a high percentage of CNS and kidney tumors. The highest frequencies of INI1 mutations for kidney tumors were seen in exons 2, 6, and 7, compared with exons 5 and 9 for CNS tumors. Two potential hot-spot mutations for CNS atypical teratoid/rhabdoid tumors were noted, including a C-to-T transition in codon 201 in exon 5 and a cytosine deletion in exon 9. Germ-line mutations were noted in 10 children, including 4 patients with two primary tumors. The majority of rhabdoid tumors from all sites contained deletions and/or mutations of the INI1 gene. Specific mutations were nonrandomly associated with anatomical site.
INTRODUCTION
Tumors composed partly or totally of rhabdoid cells are clinically aggressive malignancies that predominantly arise in infants and young children. The most common locations are in the CNS3 and kidney, although such tumors may arise in almost any site. The differential diagnosis is often challenging because of the presence of overlapping histological features and complex immunophenotypes among rhabdoid and other embryonal tumors. In the brain, these tumors may be composed only of rhabdoid cells or, more commonly, may contain areas of rhabdoid cells juxtaposed to areas of primitive neuroepithelial cells and/or mesenchymal tissue and/or epithelial tissue. Such lesions have been called AT/RT (1) . These tumors are often misdiagnosed as medulloblastoma or primitive neuroectodermal tumor and, to a lesser extent, choroid plexus carcinoma or germ-cell tumor (1 , 2) .
Whereas malignant rhabdoid tumor of the kidney and CNS AT/RT are now commonly recognized entities, the classification of extra-renal rhabdoid tumor is still subject to debate. The presence of rhabdoid cells has been documented in a variety of malignancies in both adults and children. However, this histological finding alone is not sufficient to classify a tumor as rhabdoid. Because the prognosis is extremely poor for children with rhabdoid tumors, an accurate diagnosis is essential to ensure appropriate treatment. Molecular genetic analysis of the INI1 locus may thus have clinical utility in a diagnostic setting.
Deletions and mutations of the hSNF5/INI1/SMARCB1 locus in chromosome band 22q11.2 have been demonstrated in rhabdoid tumors of the kidney, CNS, and extra-renal sites (3 , 4) . Identification of predisposing germ-line mutations, with subsequent loss of the wild-type allele in tumors (4, 5, 6, 7, 8) , demonstration of acquired compound heterozygous mutations in primary tumors, and generation of several murine Ini1 knockout models in which heterozygous mice develop soft-tissue rhabdoid tumors (9, 10, 11) all support the classification of INI1 as a tumor suppressor gene.
INI1 is a member of the ATP-dependent SWI/SNF chromatin-remodeling complex (12 , 13) . The role of the INI1 gene and SWI/SNF complex in rhabdoid tumors has recently been reviewed (14) . Although the specific function of INI1 in rhabdoid tumor development is unknown, it is hypothesized that INI1 affects its tumor-suppressor function by modulating the transcription of cellular genes. It appears that only a subset of genes use the SWI/SNF complex for transcriptional regulation, which may be related to the strength of the promoter and activator and the state of chromatin condensation at the regulatory sites of genes (14) .
In the present study, the nature and frequencies of INI1 alterations in a large series of rhabdoid tumors from different anatomical sites is described. The types of deletions and the location of mutations within the coding sequence for renal, extra-renal, and CNS rhabdoid tumors may ultimately yield clues regarding the function of INI1 in the development of rhabdoid tumors in young children.
PATIENTS AND METHODS
Clinical Cases.
Patients were referred through physicians at The Children’s Hospital of Philadelphia and from member institutions of the Children’s Oncology Group. DNA samples from patients 50120 to E452 were obtained from the National Wilms Tumor Study Group Tumor Bank. Cases from outside institutions were reviewed to confirm a diagnosis of rhabdoid tumor or AT/RT, and all other histological types were excluded. Fifteen cases have been reported previously (4 , 8) . Matched peripheral blood samples from patients and parents were used as a source of normal DNA.
FISH.
Cytogenetic pellets from freshly biopsied tumors, touch imprints prepared from frozen tissue, or nuclei isolated from formalin fixed tissue were analyzed by interphase FISH as previously described (4) . Cosmid clones for INI1 were hybridized simultaneously with a probe for the Ewing’s sarcoma locus in 22q12 as an internal control.
DNA and RNA Isolation.
Total RNA was isolated from tumor tissue using TRIzol (Invitrogen Life Technologies, Inc., Carlsbad, CA), and DNA from normal and tumor tissue or lymphocytes was extracted using a DNA Isolation Kit (Gentra Systems, Minneapolis, MN).
Microsatellite Analysis.
Paired tumor DNA and lymphoblast DNA samples (patients 88-128 through 98-292) were analyzed for LOH by radioactive end labeling, PCR-based microsatellite and gel analysis using standard conditions, as reported previously (4) . A fluorescent end-labeling PCR-based method for LOH was used for the other specimens. PCR primers for di- or trinucleotide microsatellite markers were synthesized (Integrated DNA Technologies, Inc., Coralville, IA) with either a 6-carboxyfluorescein or 4,7,2′,7′-tetrachloro-6-carboxyfluorescein fluorescent-dye tag for the following loci: D22S303, D22S257, D22S1685, D22S345, and TOP1P2. Amplification was performed with 50 ng of genomic DNA from normal and tumor tissue in 25 μl reactions. The PCR conditions were as follows: initial denaturation at 94°C for 4 min, followed by 30 cycles at 94°C for 30 s, 55–58°C for 30 s, and 72°C for 1 min and then a final extension at 72°C for 8 min. The PCR products were analyzed by electrophoresis with 2% agarose gels. Detection of labeled PCR products was performed by the Nucleic Acid/Protein Core facility of The Children’s Hospital of Philadelphia on a 3700 Genetic Analyzer (Applied Biosystems, Foster City, CA). Data were collected and analyzed using GeneScan and GeneTyper (Applied Biosystems). LOH was determined by densitometric analysis (15) . Allelic ratios were calculated by determining the percentage loss of intensity for the tumor alleles compared with the normal alleles. LOH was considered present when the calculated loss ratio for a tumor allele was greater than 60%.
RT-PCR.
Preparation of cDNA for RT-PCR was performed according to the manufacturer’s protocol (Invitrogen). Two pairs of primers were designed to amplify the INI1 cDNA: INICD1–4 forward 5′-CTG AGC AAG ACC TTC GGG CAG-3′, and INICD1–4 reverse 5′-GAT GGC TGG CAC AAA CGT CAG-3′; and INICD4–9 forward 5′-ACC CTG TCC AAC AGC TCC CA-3′ and INICD4–9 reverse 5′GGC CCA ATC TTC TGA GAT GC-3′. The PCR conditions were as follows: initial denaturation at 94°C for 4 min, followed by 30 cycles at 94°C for 30 s, 64°C for 30 s, and 72°C for 1 min and then a final extension at 72°C for 10 min. The PCR products were analyzed by electrophoresis using 1% agarose gels. Individual bands were isolated and purified with the QIAquick gel extraction kit (Qiagen, Inc., Valencia, CA) for sequencing.
Mutation Analysis.
Oligonucleotide primers for exons 1–9 were designed from the intron/exon boundary sequences for the INI1 gene (GenBank accession nos. AP000349-350). Primer sequences are available on request. PCR products for individual exons were analyzed by the heteroduplex method and/or direct sequencing. Heteroduplex analysis was performed based on published methods (4 , 16) . Briefly, genomic DNA was amplified in a 50-μl reaction mixture containing 50 mm KCl, 1.5 mm MgCL2, 10 mm Tris-HCl (pH 8.3), 125 μm each deoxynucleotide triphosphate, 1 unit of TaqDNA Polymerase (PGC Scientific Corp., Gaithersburg, MD), and 25 pmol of each primer. The PCR conditions were as follows: initial denaturation at 95°C for 5 min, 30 cycles at 94°C for 30 s, 60–64°C for 30 s, and 72°C for 1 min and then a final extension at 72°C for 10 min. Heteroduplex formation was enhanced by mixing equal volumes of sample and wild-type PCR product, denaturing at 95°C for 5 min and annealing at 68°C for 30 min. Five microliters of this reaction mixture were electrophoresed through a vertical, 1-mm-thick 15% nondenaturing acrylamide gel. Gels were run at room temperature in 0.5× glycerol tolerant gel buffer (USB, Cleveland, OH) for 6.5 h at 40 W. Gels were stained with 2 μg/ml ethidium bromide, and bands were visualized using a Bio Rad Gel Doc 1000 imaging system (Bio Rad, Hercules, CA). To confirm any base changes observed by heteroduplex screening, the PCR products were sequenced directly.
Sequence Analysis.
Sequencing of the RT-PCR or PCR products was performed by the Nucleic Acid/Protein Core facility of The Children’s Hospital of Philadelphia. The cDNA sequence was compared with the GenBank database (accession no. U04847) using the BLAST program through the network server at the National Center for Biotechnology Information. Nucleotide 406 in codon 136 is a cytosine instead of a thymidine, and nucleotide 1145 in codon 382 is a cytosine instead of a guanine. The sequence of individual exons was compared with the reported sequences in GenBank (accession nos. AP000349-350).
RESULTS
One hundred rhabdoid tumors were analyzed for deletions and mutations of the INI1 gene. Twenty-five tumors (12 CNS, 10 renal, 3 extra-renal) did not demonstrate mutations. The deletion status of chromosome 22, as determined by standard cytogenetic, FISH, or LOH analysis for the remaining 75 tumors is shown in Table 1⇓ . Forty-two patients had primary tumors of the brain or spinal cord. Six children presented with a brain tumor and either a renal tumor or soft-tissue tumor. Nineteen patients had a renal tumor (in 1 patient tumors were bilateral) and 8 tumors were extra-renal.
INI1 deletions and mutations in atypical teratoid and rhabdoid tumors
Deletion analysis was performed for 57 of 75 tumors. Four tumors did not demonstrate loss of 22q11.2. A total of 19 tumors had homozygous deletions of the INI1 gene, as determined by interphase FISH analysis. Included among the 19 tumors were 10 brain tumors, 3 kidney tumors, and 6 extra-renal soft-tissue tumors. Thirty-four tumors had evidence for a heterozygous cytogenetic deletion of INI1 or LOH for markers on 22q11.2. This included four tumors (patients 98-156, 98-240, 99-192, and 50495) for which FISH results were normal, but the microsatellite analysis revealed LOH. These results are consistent with loss and duplication or mitotic recombination of chromosome 22.
As shown in Table 1⇓ , 58 mutations were identified in 56 rhabdoid tumors. Two of the four tumors without evidence for a deletion of chromosome 22q11.2 (patients 91-114 and 00-133) had two different mutations. Blood samples were not available from either of these AT/RT patients to determine whether the mutations were present in the germ line.
Two tumors had intragenic deletions detected by RT-PCR analysis, one involving exons 4 and 5 (patient 00-214), and the other resulting in loss of exon 6 (patient 98-156). The genomic DNA sequence at the intron-exon boundaries was normal in both cases, eliminating a simple splice site mutation as the mechanism for the deletion.
Twenty-nine of the 58 mutations were single-bp point mutations, all of which resulted in the production of a novel stop codon, and thus predicted premature truncation of the protein. Single-bp mutations were observed in 15 CNS AT/RTs, 9 kidney tumors, 1 lung tumor, and 4 patients with a brain and kidney or a brain and soft-tissue tumor. Twenty-seven mutations were deletions or insertions that resulted in a frameshift, and therefore introduced a novel stop codon. These mutations were identified in 16 brain AT/RTs, 2 spinal cord tumors, 7 kidney tumors, and 1 patient with a brain and kidney tumor.
The mutations were distributed among most of the coding exons of the INI1 gene, although several exons were clearly under-represented. There were no exon 1 coding sequence mutations in this series of rhabdoid tumors, although we have analyzed one cell line that had a 1-bp deletion causing a frameshift (data not shown). No mutations were identified in exon 8, and there were only three mutations in exon 3. The highest frequency of mutations was observed for exons 2 and 9 (n = 12 mutations each), followed by exon 5 (n = 10), 6 (n = 9), 4 and 7 (n = 5).
Several hot spots were observed in this series of tumors. Four of five mutations in exon 4 were identical. A C472T transition at codon 158 that resulted in an arginine to a stop codon change was identified in two brain tumors (patients 88-128 and 00-133) and two kidney tumors (patients 50213 and E411). Patient E411 demonstrated this mutation in the germ line.
Six tumors contained a C601T transition at codon 201 in exon 5, also resulting in an arginine to stop codon change. Interestingly, one patient (patient 01-225) with a brain and lung tumor had the identical mutation present in both tumors, but the mutation was not found in the blood DNA. Although the histological features were distinct, one tumor may have been a metastatic lesion from the other. The remaining five tumors were all children with CNS AT/RTs. Two identical mutations in exon 6 (C727T in codon 243) were both observed in rhabdoid tumors of the kidney. The same mutation in exon 2 (C118T) was observed in a lung tumor (patient 00-272) and a brain tumor (patient 01-323).
The most striking hot spot for mutations in the INI1 gene was observed for exon 9. An identical single cytosine deletion at the 3′ end of the coding sequence was detected in 10 rhabdoid tumors, all of which were somatic in origin and all of which were found in CNS AT/RTs. In addition, one tumor (01-90) contained a deletion of a guanosine in the adjacent codon. This common single-bp deletion removes one of four cytosines causing a frameshift. The stop codon is eliminated and is predicted to add an additional 100 amino acids to the end of the protein. Western blot analysis of two cases with this genomic alteration was performed, but no protein was detected (data not shown). This suggests that the message or protein may be unstable.
Germ-Line Mutations.
Matched normal tissue was available for 23 of the 75 patients. Exons that harbored INI1 mutations in tumors were PCR amplified from the normal DNA and directly sequenced. Ten of the 23 cases were shown to have the identical mutation in normal and tumor DNA, consistent with a germ-line mutation. Germ-line mutations were identified in two children with CNS AT/RT (98-74, 01-08), four infants with primary tumors of the brain and kidney (patients 97-103, 99-202, 50796, and 51510), one child with bilateral kidney tumors (patient 50340), and three patients with kidney tumors (patients 50462, 51529, and E411). Among the 10 germ-line mutations identified, 7 were single-bp point mutations that introduced a novel stop codon, and 3 were intragenic deletions or insertions that resulted in a frameshift.
Maternal and paternal blood DNA samples were available for 4 of the 10 children. None of the parents carried a mutation, suggesting that these were de novo germ-line mutations.
DISCUSSION
The aim of the present study was to define the spectrum of mutations in a large series of rhabdoid tumors from different anatomical sites. Seventy-five of 100 patients with histologically confirmed rhabdoid tumors originating in the brain, spinal cord, kidney, and soft tissues were shown to contain homozygous deletions or mutations of the INI1 gene in chromosome band 22q11.2. On the basis of these data, we estimate that an alteration of the INI1 gene should be detected in ∼75% of rhabdoid tumors for which high-quality nucleic acids can be isolated. Determining the sensitivity of the mutation analysis, however, is hampered by the large number of tumors for which only paraffin-embedded tissue is available, as well as the difficulty inherent in making a diagnosis of rhabdoid tumor, particularly for patients with tumors located in the CNS and soft tissues. Alternative mechanisms to coding sequence mutations or deletions that have the potential to inactivate INI1 include promoter alterations that decrease or inhibit transcription, posttranscriptional modification, defects in protein translation, and posttranslational modification. As reported previously (17) , hypermethylation of the INI1 promoter does not appear to be the mechanism for decreased expression of INI1 seen in tumors without coding-sequence mutations.
As shown in Table 1⇓ , the majority of tumors had evidence for inactivation of both INI1 alleles, consistent with its function as a tumor suppressor gene. Homozygous deletion, cytogenetic deletion, or LOH with a coding-sequence mutation in the remaining allele, or two independent mutations were present in all but two cases (patients 01-310 and E411). For the majority of cases in which FISH could not be performed, or blood was not available for LOH studies, analysis of the raw sequence data were consistent with loss of the wild-type allele.
As shown in Table 2⇓ , there appears to be a correlation between the type of INI1 alteration detected and anatomical location. The highest frequency of homozygous deletions of INI1 was observed among the soft-tissue rhabdoid tumors, such as those arising in the lung, mediastinum, thigh, and so forth. Standard cytogenetic studies have demonstrated that a majority of soft-tissue tumors contain cytogenetically balanced or unbalanced translocations or deletions of chromosome 22 (Ref. 18 ; data not shown). In contrast, many of the CNS tumors for which karyotypes are prepared have no detectable cytogenetic aberrations (4) . FISH analysis for these tumors is a sensitive means of detecting INI1 deletions because the presence of normal contaminating cells can be quantified. FISH analysis alone, however, will not detect loss and duplication of chromosome 22 or mitotic recombination, which is best determined by microsatellite analysis. Thus a combination of cytogenetic, FISH, and molecular analysis is required to cover the range of abnormalities of INI1 that might be present.
Frequency of INI1 deletions and mutations in pediatric rhabdoid tumors
The frequency of mutations was similar for AT/RTs of the CNS (76%) and renal rhabdoid tumors (84%). As might be expected, the highest frequency of both somatic (six of six) and germ-line mutations (four of six) was found for patients with primary tumors of the brain and either the kidney or soft tissues. One patient presented in the newborn period with a brain and kidney tumor. An exon 4–5 deletion was detected in the kidney tumor; however, normal tissue was not available to rule out a germ-line mutation. In contrast, one patient with an AT/RT and rhabdoid tumor of the lung had identical mutations in the two tumors, and yet the mutation was not detected in the DNA extracted from the blood. A sample mix-up is unlikely because the microsatellite analysis demonstrated the same alleles in the normal and tumor DNA. This suggests that one of the lesions may have been a metastasis from the primary tumor. Alternatively, an early somatic event may have led to the presence of the mutation in specific tissues, including the brain and lung, but not the blood.
A total of 10 blood samples demonstrated mutations within the coding sequence of the INI1 gene. However, a true estimate of the frequency of germ-line mutations in this series of patients cannot be made because blood samples were available for only 23 of the 75 children. We would predict that the frequency of germ-line mutations is high, emphasizing the importance of germ-line screening and appropriate genetic counseling for families of these children. The fact that we did not identify any inherited mutations is not surprising, given the very low frequency of families in which multiple first-degree relatives have presented with rhabdoid tumors (5 , 6) .
There were no coding-sequence mutations within exons 1 or 8 in this series of tumors. The locations of the 58 mutations (not including the exon 4–5 and exon 6 deletion) in the remaining exons of INI1 are shown in Fig. 1⇓ . Only three mutations in exon 3 were identified. The number of mutations among the AT/RTs and renal rhabdoid tumors was the same for exons 2 and 6. However, as shown in Table 3⇓ , the percentage of mutations in both exons was 2-fold higher for the patients with kidney tumors as compared with the CNS tumors. The percentage of tumors with mutations in exon 7 was also higher for the renal rhabdoid tumor patients than the AT/RT patients, although the total numbers were small for both groups. In contrast, both the number and percentage of tumors with mutations in exons 5 and 9 were much greater for the AT/RTs than any other group.
INI1 mutations in exons 2–9 among 56 patients with rhabdoid tumors.
Distribution of mutations among INI1 exons 1 through 9 in rhabdoid tumors of the CNS and kidney
The most significant finding in this study was the recurring cytosine deletion in exon 9. The identical mutation was detected in 10 tumors, all of which were located in the brain. It is intriguing to note that given the frequency of the mutation, it has not yet been detected in the germ line. This is one of the only mutations not predicted to result in a truncated protein. In contrast, it is expected to result in a protein with an extra 100 amino acids. Western blot analysis (data not shown) did not demonstrate the presence of a larger protein. In fact, no protein for tumors with the exon 9 mutation was observed, suggesting that the message and/or protein were unstable. It also raises the possibility that this alteration could function as a dominant negative and interfere with normal development. This may account for why we have not yet observed this as a germ-line mutation.
The identical mutation has been reported by Weber et al. (19) in a tumor that was characterized as a choroid plexus carcinoma. We have reported two patients who were initially thought to have primitive neuroectodermal tumors, but whose tumors contained this exon 9 mutation. Subsequently, review of the histological and immunohistochemical features indicated that they were AT/RT. Choroid plexus carcinomas and AT/RTs have numerous overlapping histological features, immunophenotypic staining patterns, anatomical location, and characteristic early age at diagnosis. Rorke et al. (1) and Burger et al. (2) have reported that CNS AT/RT may be misdiagnosed as choroid plexus carcinoma. Given the frequency of the exon 9 mutation in CNS AT/RT, identification of this abnormality should raise the possibility that the tumor in question may be an AT/RT. Regardless of the pathologic diagnosis, it is likely that the clinical course of these patients will be similar.
We reported recently that the gene-expression profiles of AT/RT might be used to distinguish rhabdoid tumor from a variety of other embryonal tumors of the CNS (20) . It was of interest to note that regardless of anatomical site, the expression profiles were similar between the CNS, renal, and extra-renal tumors. Because the total numbers were small with respect to anatomical location, further studies need to be performed to confirm these preliminary findings.
Clinical follow-up data are not yet available for the majority of patients in the present study, thus we did not correlate specific mutations or deletions with outcome. The treatment protocols for AT/RT and renal and extra-renal rhabdoid tumor patients are quite varied. Prospective clinical trials designed to allow analysis of genotype with outcome are in development and should allow us to address these questions.
Acknowledgments
The support of the pediatric branch of the Cooperative Human Tissue Network, the National Wilms Tumor Study Group Tumor Bank, and the Children’s Oncology Group is gratefully acknowledged. We thank the following clinicians for their contributions: Minnie Abromowitch, Jeffrey Auletta, Jean Belasco, Deborah A. Belchis, Anne Bendle, Peter Burger, Herbert Bevan, David Biddle, Andrew Bolen, Adrian Charles, B. A. Chrenka, Katrina Conard, Albert S. Cornelius, Mary Davis, Najat Daw, Dennis Drehner, Anne Christine Duhaime, Jeffrey Golden, Paul Grundy, Sri Gururangan, Joanne Hilden, Marianna Horne, Anna Janss, Amy Lowichik, Mark Matthews, Thomas Milligan, James Olson, Mahendra Patel, Elizabeth Perlman, Peter Phillips, Scott Pomeroy, James M. Powers, Harker Rhodes, Carolyn Russo, Deborah Schofield, Violet Shen, Ana Sotrel, Leslie Sutton, Beth Trost, Hannes Vogel, Andrew Walter, and David Zagzag. The technical assistance of Benjamin Fogelgren, Jun-Ying Zhou, David Kim, and Jacquelyn Roth is also appreciated.
Footnotes
-
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
-
↵1 This study was supported by grants (to J. A. B.) from the NIH (CA 46274) W. W. Smith Charitable Trust and the Pediatric Brain Tumor Foundation.
-
↵2 To whom requests for reprints should be addressed, at Room 1002 Abramson Research Building, The Children’s Hospital of Philadelphia, 3615 Civic Center Boulevard, Philadelphia, PA 19104. Phone: (215) 590-3856; Fax: (215) 590-3764; E-mail: biegel{at}mail.med.upenn.edu
-
↵3 Abbreviations used: CNS, central nervous system; AT/RT, atypical teratoid/rhabdoid tumor; FISH, fluorescence in situ hybridization; LOH, loss of heterozygosity; RT-PCR, reverse-transcription PCR.
- Received April 10, 2002.
- Revision received July 22, 2002.
- Accepted July 23, 2002.
References
Volume 8, Issue 11
