| Human Molecular Genetics | Pages |
©1999 Oxford University Press |
Spectrum of hSNF5/INI1 somatic mutations in human cancer and genotype-phenotype correlations
Introduction
Results
The spectrum of somatic hSNF5/INI1 alterations
Genotype-phenotype correlations
Discussion
Types of hSNF5/INI1 mutation observed in human cancers
Mutation of hSNF5/INI1 in tumors with a rhabdoid phenotype
Mutation of hSNF5/INI1 in other tumor types
Materials And Methods
Patient samples
Mutation analysis
Acknowledgements
References
Spectrum of hSNF5/INI1 somatic mutations in human cancer and genotype-phenotype correlations
Received August 20, 1999; Revised and Accepted September 30, 1999
The hSNF5/INI1 gene which encodes a member of the SWI/SNF chromatin ATP-dependent remodeling complex, is a new tumor suppressor gene localized on chromosome 22q11.2 and recently shown to be mutated in malignant rhabdoid tumors. We have searched for hSNF5/INI1 mutations in 229 tumors of various origins using a screening method based on denaturing high-performance liquid chromatography. A total of 31 homozygous deletions and 36 point alterations were identified. Point mutations were scattered along the coding sequence and included 15 nonsense, 15 frameshift, three splice site, two missense and one editing mutations. Mutations were retrieved in most rhabdoid tumors, whatever their sites of occurrence, indicating the common pathogenetic origin of these tumors. Recurrent hSNF5/INI1 alterations were also observed in choroid plexus carcinomas and in a subset of central primitive neuroectodermal tumors (cPNETs) and medulloblastomas. In contrast, hSNF5/INI1 point mutations were not detected in breast cancers, Wilms' tumors, gliomas, ependymomas, sarcomas and other tumor types, even though most analyzed cases harbored loss of heterozygosity at 22q11.2 loci. These results suggest that rhabdoid tumors, choroid plexus carcinomas and a subset of medulloblastomas and cPNETs share common pathways of oncogenesis related to hSNF5/INI1 alteration and that hSNF5/INI1 mutations define a genetically homogeneous family of highly aggressive cancers mainly occurring in young children and frequently, but not always, exhibiting a rhabdoid phenotype.
INTRODUCTION
The malignancy now termed malignant rhabdoid tumor (MRT) was described initially as a `rhabdomyosarcomatoid' variant of Wilms' tumor on the basis of its location in the kidney and of its superficial resemblance to rhabdomyosarcoma in light microscopy (1). Although subsequent studies have clearly shown that cells from this aggressive tumor do not exhibit ultrastructural and immunohistochemical features of skeletal muscle cells, the term rhabdoid has been widely accepted as the name of this tumor. It refers to a phenotype characterized by sheets of polygonal cells displaying vesicular nuclei, prominent nucleoli and hyaline cytoplasmic inclusions consisting of whorls of intermediate filaments (2). Following its initial description in the kidney, authors have reported the observation of the rhabdoid phenotype in tumors from various locations, including soft tissues, liver, lung and central nervous system (CNS), tumors from the latter site being termed atypical and teratoid rhabdoid tumors (ATTRs) (2-6). The absence of precise criteria to define MRTs has led to controversies as to whether the term rhabdoid defines a homogeneous entity or just a phenotypic particularity of various, distinct neoplasias.
Cytogenetic studies of MRT cells have demonstrated minimal karyotypic changes, contrasting with the aggressiveness of the tumor. Recurrent chromosomal abnormalities mainly involve chromosome 22 (7-10). Indeed, translocations of chromosome 22 with a variety of different partners have been described, the breakpoints always involving the 22q11.2 chromosome region. Complete or partial monosomy of chromosome 22 have also been observed recurrently. Altogether, chromosome 22 translocations or deletions were detected in the majority of MRT cases. The observation of similar genetic alterations in renal and extra-renal rhabdoid tumors has provided a strong indication that MRTs might constitute a homogeneous entity.
Alterations of chromosome 22 were confirmed with molecular approaches by the analysis of loss of heterozygosity (LOH) and by fluorescence in situ hybridization (FISH). Indeed, LOH at 22q11.2 loci were observed in >80% of renal cases and defined the smallest region of overlap within 22q11.2. FISH analysis confirmed the chromosome 22 deletions close to the BCR region and demonstrated the occurrence of homozygous deletions (HDs) of this region in both renal and liver MRTs. These molecular data therefore strongly suggested the presence of a tumor suppressor gene (TSG) located within the 22q11.2 region (11-13).
Recently, using a positional cloning approach, we isolated the hSNF5/INI1 gene (14,15) as the target of recurrent, bi-allelic, inactivating alterations in MRTs (16). These alterations, which were either HDs of the gene or truncating mutations of the hSNF5/INI1 coding sequence associated with deletions of the wild-type allele, provided strong evidence that hSNF5/INI1 is the TSG involved in the oncogenesis of MRTs (16). These alterations were observed in renal as well as other abdominal and soft tissue MRTs, and very recently in ATTR (17).
We now report the search for hSNF5/INI1 mutations in various tumor types. In order to evaluate the consistency, the spectrum and the type of hSNF5/INI1 alterations, a large number of MRTs were analyzed. Different malignancies characterized by recurrent LOH at 22q11.2, the region encoding the hSNF5/INI1 gene, were also studied. Finally, tumors belonging to the differential diagnosis of MRTs given their location in the kidney or the CNS, the most frequent sites of renal and extra-renal MRT, were included in the analysis. This study enables us to define the tumors for which hSNF5/INI1 loss of function contributes to oncogenesis and to establish genotype-phenotype correlations.
RESULTS
A total of 229 tumors were screened for the presence of hSNF5/INI1 mutations. This included 72 renal or extra-renal rhabdoid tumors, 107 tumors from the CNS with a diagnosis other than rhabdoid tumors [25 ependymomas, 36 medulloblastomas, 17 central primitive neuroectodermal tumors (cPNETs), 12 choroid plexus tumors and 17 gliomas with LOH at chromosome 22 loci] and 50 tumors with various diagnoses (Table 1).
Table 1. Types of tumor demonstrating hSNF/INI1 mutations
| Type of tumor | Diagnosis | Particular features | No. of tumors | No. of mutated tumors |
| Renal tumors | Renal rhabdoid tumors | 47 | 32 | |
| Wilms' tumorsa | Anaplastic | 8 | 0 | |
| Non-anaplastic | 10 | 0 | ||
| Mesoblastic nephroma | 1 | 0 | ||
| CNS tumors | ATTRs | 7 | 4 | |
| Medulloblastoma | >36 months | 30 | 2 | |
| <36 months | 6 | 3 | ||
| cPNET | >36 months | 14 | 1 | |
| <36 months | 3 | 1 | ||
| Ependymoma | Adult | 3 | 0 | |
| Child | 22 | 0 | ||
| Plexus choroid | Carcinoma | 6 | 4 | |
| Atypical papilloma | 1 | 1 | ||
| Papilloma | 5 | 0 | ||
| Gliomab | Astrocytoma | 1 | 0 | |
| Anaplastic astrocytoma | 4 | 0 | ||
| Glioblastoma | 12 | 1 | ||
| Others | Extra-renal rhabdoid tumors | 18 | 17 | |
| Rhabdomyosarcoma | 4 | 0 | ||
| Acute lymphoblastic leukemia | 1 | 0 | ||
| Sarcoma | 14 | 0 | ||
| Neuroblastoma | 1 | 0 | ||
| Breast cancerb | 11 | 0 | ||
| Total | 229 | 66 |
bAll cases had chromosome 22 LOH.
Tumor material was available either as frozen tissue (FT) or as paraffin-embedded tissue (PET). Depending on the source, different approaches were used to screen for point mutations. Both DNAs and RNAs were isolated from frozen tumors or cell lines (188 cases). With this material, the nine exons and five overlapping RT-PCR products were submitted to heteroduplex detection by denaturing high-performance liquid chromatography (dHPLC) analysis (18). Optimal parameters for this detection were determined using the software provided by the manufacturer and validated by their ability to retrieve previously described mutations or polymorphisms (16). Representative normal and altered dHPLC profiles for two mutations of exon 2 are shown in Figure 1. PCR fragments exhibiting abnormal dHPLC profiles were sequenced. Only DNA could be isolated from PET (41 cases). Since the use of such material precludes the reliable amplification of DNA fragments >150 bp, the dHPLC procedure described above could not be applied. These tumors were therefore analyzed by direct sequencing of smaller and overlapping PCR products.
Figure 1. dHPLC analysis of hSNF5/INI1 exon 2. (A) dHPLC profile corresponding to a normal exon 2 sequence. (B) Altered dHPLC profile linked to a C->A transversion at codon 47 (case 130). (C) Altered dHPLC profile due to a 19 bp deletion (case 12).
For both types of material, we searched for HDs by competitive co-amplification of a 138 bp hSNF5/INI1 exon 4 fragment with a 104 bp fragment from the GAPDH locus on chromosome 12 (Fig. 2). Finally, given the recent report describing frequent HDs limited to exon 1, this exon was also analyzed by co-amplification with the GAPDH locus in MRT cases for which neither HD for exon 4 nor point mutations were detected (17).
Figure 2. Analysis of homozygous deletion of hSNF5/INI1 by co-amplification of hSNF5/INI1 exon 4 and GAPDH. 1, hSNF5/INI1 exon 4 PCR fragment (138 bp); 2, GAPDH PCR fragment (104 bp). T50, T169, T171, T172 and T179 demonstrate an almost complete absence of amplification of hSNF5/INI1 exon 4 in the tumor DNA as compared with the constitutional DNA which indicates the presence of a homozygous deletion of this exon in the tumor DNA. T186, T187 and T188 show a normal hSNF5/INI1 exon 4 amplification.
The spectrum of somatic hSNF5/INI1 alterations
A total of 31 HD were identified. An almost complete absence of amplification of exon 4 of hSNF5/INI1 as compared with the GAPDH locus was observed in 28 cases (Fig. 2). In two additional cases, amplification of exon 4 was normal but RT-PCR analysis retrieved the skipping of exon 6 from hSNF5/INI1 RNA. In both cases, the analysis of genomic DNA by co-amplification of exon 6 with a control locus confirmed the occurrence of HD limited to this exon. In these two cases, the loss of the exon 6 leads to a frameshift in the coding sequence of hSNF5/INI1. In another case, the genomic DNA could not be PCR amplified for exon 8. Whereas other hSNF5/INI1 exons could be co-amplified efficiently with the control locus, the complete absence of PCR products with this exon was indicative of an HD limited to this exon. Finally, no HD limited to exon 1 was observed.
Thirty-six point mutations were retrieved in 35 tumors (Tables 1 and 2). In 34 tumors, a single mutation could be identified, and in one case two different mutations were retrieved in the same tumor. Most alterations were either frameshift (15 cases) or nonsense (15 cases) mutations. Frameshifts were linked preferentially to microdeletions ranging from 1 to 19 nucleotides and, more rarely, to insertion (one case) or duplication (two cases). The nonsense mutations were due most frequently (nine out of 15 cases) to a C->T transition occurring in the context of a CG dinucleotide. Typical splice site mutations altering the AG consensus of the splice acceptor sites of exons 3 or 6 were identified in two tumors, whereas a more complex mechanism of mutation was observed in one case (no. 8). In this case, a 72 bp insertion containing an in-frame stop codon was identified between exons 1 and 2 at the cDNA level. This insertion was not detected within the genomic sequences analyzed in our screening procedure; however, hybridization and sequence experiments revealed that this 72 bp sequence arose from intron 1 (Fig. 3). Sequence analysis of tumor DNA indicated that an A->G transition created a perfect consensus for a donor splice site within this intron. As a result of this nucleotide change, a pseudo-exon is activated using a cryptic acceptor splice site together with this new donor site. No wild-type isoform was observed on the RT-PCR sequence product, in agreement with the loss of the wild-type allele and with the consistent insertion of this exon in the mRNA. This tumor therefore harbors a complete loss of function of hSNF5/INI1 (Fig. 3). This mutation was also observed in a heterozygous state in constitutional DNA from the same patient. Two missense mutations were detected. One of them (case 170 in Table 2), which changes a conserved proline into a serine at codon 48, was observed on tumor but not constitutional DNA, thereby demonstrating its somatic occurrence. Another mutation (case 185) is a C->T transition leading to the substitution of Ser284 with a leucine. The highly conserved status of this amino acid, together with the absence of detection of this change in other DNAs from the present series, suggests that this DNA variant is a mutation rather than a polymorphism. Constitutional DNA could not be tested to document its acquired appearance. Finally, case 134 displayed at the RNA level an A->G transition leading to the substitution of an arginine by a glycine at codon 127. This change was not detected on the genomic DNA, indicating that it resulted from an editing mechanism. Interestingly, this editing involved all RNA molecules since no wild-type cDNA isoform could be detected by sequence analysis.
Figure 3. Schematic representation of the functional consequences of the mutation detected in case 8. Exon and intron sequences are shown in upper and lower case letters, respectively. (a) Sequence of the hSNF5/INI1 cDNA which reveals an extra 72 bp fragment inserted between exons 1 and 2 and contains an in-frame stop codon. Amino acids are indicated above the coding sequence. (b) Mechanism of the activation of the pseudo-exon. In normal DNA, the 72 bp sequence lacks a functional donor site and therefore is not detected as an exon. The A->G mutation (indicated by an arrow) creates a consensus 5[prime] donor site enabling the correct splicing of this pseudo-exon. The score for a 5[prime] donor site increases from 0.45 for the wild-type to 0.99 for the mutated sequence as determined by the `splice site prediction by neural network' at http://www-hgc.lbl.gov/projects/splice.html . With the same program, the score for the 3[prime] acceptor site of the pseudo-exon is 0.97.
Table 2. Types of hSNF5/INI1 mutation detected in tumors
| Patient no. | Diagnosis | Location | Mutation | Exon | Const. | Material |
| 180 | Rhabdoid | Kidney | 149del 4 bp-rp 2 bp | 2 | N | PET |
| 186 | Rhabdoid | Kidney | 352ins C | 3 | M | PET |
| 173 | Rhabdoid | Kidney | 354del C | 3 | - | PET |
| 191 | Rhabdoid | Kidney | 669del 2 bp | 6 | - | PET |
| 6a | Rhabdoid | Kidney | 772del 13 bp-rp 2 bp | 6 | - | FT |
| 7a | Rhabdoid | Kidney | 950del G | 7 | - | FT |
| 181 | Rhabdoid | Kidney | A631T (K->Stop) | 6 | N | PET |
| 2a | Rhabdoid | Kidney | C141A (Y->Stop) | 2 | N | FT |
| 70 | Rhabdoid | Kidney | C601T (R->Stop) | 5 | - | FT |
| 228 | Rhabdoid | Kidney | G838T (E->Stop) | 7 | - | FT |
| 170 | Rhabdoid | Kidney | C142T (P->S ) | 2 | N | PET |
| 185 | Rhabdoid | Kidney | C851T (S->L) | 7 | - | PET |
| 240 | Rhabdoid | Kidney | C601T(R->Stop)/A->G (SA) | 5/6 | - | CL |
| 74 | Rhabdoid | Kidney | C472T (R->Stop) | 4 | N | FT |
| 244b | Rhabdoid | Kidney | C601T (R->Stop) | 5 | M | PET |
| 73 | Rhabdoid | Kidney | HD | 6 | - | FT |
| 164 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 168 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 169 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 171 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 172 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 175 | Rhabdoid | Kidney | HD | 8 | N | PET |
| 179 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 183 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 184 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 194 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 177 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 165 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 166 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 174 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 196 | Rhabdoid | Kidney | HD | [ge]4 | N | PET |
| 93 | Rhabdoid | Kidney | HD | [ge]4 | - | FT |
| 12a | Rhabdoid | Abdomen | 108del 19 bp | 2 | - | CL |
| 13 | Rhabdoid | Abdomen | HD | [ge]4 | N | FT |
| 10a | Rhabdoid | Abdomen | HD | 4-5 | - | CL |
| 3a | Rhabdoid | Abdomen | HD | 1-9 | - | CL |
| 4a | Rhabdoid | Abdomen | HD | 1-9 | - | CL |
| 9a | Rhabdoid | Abdomen | HD | 1-9 | N | CL |
| 241 | Rhabdoid | Abdomen | HD | [ge]4 | - | CL |
| 64 | Rhabdoid | Liver | HD | [ge]4 | - | FT |
| 11a | Rhabdoid | Liver | HD | 1-9 | - | CL |
| 75 | Rhabdoid | Bladder | HD | 6 | - | FT |
| 14a | Rhabdoid | Spinal | 586dup 17 bp | 5 | - | CL |
| 88 | Rhabdoid | Dorsal | 565dup 7 bp | 5 | - | FT |
| 31 | Rhabdoid | Cervical | C472T (R->Stop) | 4 | M | FT |
| 62 | Rhabdoid | Soft tissue | HD | [ge]4 | - | FT |
| 8a | Rhabdoid | Soft tissue | A->G intron 1 | - | M | FT |
| 1a | Rhabdoid | Lung | HD | 1-9 | N | CL |
| 144 | Rhabdoid | Face | HD | [ge]4 | - | CL |
| 26b | ATTR | CNS | 430del G | 4 | - | FT |
| 68 | ATTR | CNS | C118T (R->Stop) | 2 | - | FT |
| 237b | ATTR | CNS | 591del G | 5 | M | FT |
| 264b | ATTR | CNS | 591del G | 5 | M | FT |
| 24 | cPNET | CNS | G617A (W->Stop) | 5 | - | FT |
| 245b | cPNET | CNS | C601T (R->Stop) | 5 | M | PET |
| 122 | Medulloblastoma | CNS | C472T (R->Stop) | 4 | - | FT |
| 130 | Medulloblastoma | CNS | C141A (Y->Stop) | 2 | - | FT |
| 134 | Medulloblastoma | CNS | A379G (R->G) | 4 | - | FT |
| 138 | Medulloblastoma | CNS | 545del A | 5 | N | FT |
| 224 | Medulloblastoma | CNS | G153A (W->Stop) | 2 | - | CL |
| 90b | PC carcinoma | CNS | 430del G | 4 | N | PET |
| 227 | PC carcinoma | CNS | G->C (SA) | 3 | - | FT |
| 242 | PC carcinoma | CNS | C601T (R->Stop) | 5 | - | FT |
| 266b | PC carcinoma | CNS | 591del G | 5 | - | PET |
| 230 | Atypical PC papilloma | CNS | HD | [ge]4 | N | FT |
| 50 | Glioblastoma | CNS | HD | [ge]4 | N | FT |
aThese cases have been described previously (16).
bFamilial (cases 26 and 90 from one family; cases 237, 264 and 266 from another) and bifocal cases (cases 244 and 245).
Constitutional DNA was available for analysis in 32 cases for which hSNF5/INI1 alterations were detected in the tumor DNA. In 25 cases, the germline DNA displayed a wild-type allele, therefore demonstrating that the mutation was acquired during oncogenesis. In seven cases, the mutation detected in the tumor DNA was also identified at the constitutional level. ATTR nos 237 and 264 were observed in two sibs (Table 2). The MRT no. 244 and the cPNET no. 245 were two successive cancers affecting the same individual. The MRT no. 31 arose in a child with a sib who had died from medulloblastoma. In two cases (nos 8 and 186), neither personal nor familial history of cancer could be documented. In three other cases, although constitutional DNA was not available for study, a germline mutation was highly suspected. Indeed, cases 26 and 90 which harbor the same mutation are from the same sibship, and case 266 belongs to the same family as cases 237 and 264. These familial and bifocal cases are described more thoroughly elsewhere (19). When considering these familial cases, a total of 32 independent mutations were thus identified. Their positions are scattered along the coding sequence of hSNF5/INI1, as indicated in Figure 4.
Figure 4. Distribution of the point mutations of the hSNF5/INI1 gene along the coding sequence. The protein is shown with the different domains encoded by each exon (italic numbers). The gray area represents the SNF5 domain and the hatched area indicates a peptide sequence that can be present or absent in the protein, depending on the use of a cryptic splice donor site in exon 2. The positions of the truncating mutations are shown by arrows above the protein scheme. Missense and editing mutations are depicted by arrows below the protein. Tumors from individuals of the same family are indicated by numbers displayed on the same line.
Besides mutations, a number of polymorphisms were identified (Table 3). Three of them involved the hSNF5/INI1 coding sequence, two being silent and one modifying a methionine into an isoleucine at codon 208. This latter change, which was observed only once, is most probably a polymorphism rather than a mutation since it involves a non-conservative residue and since a somatic, nonsense mutation was identified in the tumor DNA of this individual (case 181). Other polymorphisms were observed within intron sequences (Table 3).
Table 3. Single nucleotide polymorphisms of the hSNF5/INI1 gene
| Localization | Codon | Nucleotide | Nucleotide change | Effect | Allelic frequency (%) |
| Exon 4 | 141 | 423 | C->T | His->His | 1 |
| Exon 5 | 208 | 624 | G->T | Met->Ile | 1 |
| Exon 7 | 299 | 897 | G->A | Ser->Ser | 10 |
| Intron 5 | - | 66 bp 3[prime] to exon 5 DS | C->G | - | 8 |
| Intron 5 | - | 130 bp 5[prime] to exon 6 AS | del 1C | - | 16 |
| Intron 5 | - | 62 bp 5[prime] to exon 6 AS | A->G | - | 8 |
| Intron 5 | - | 58 bp 5[prime] to exon 6 AS | C->A | - | 30 |
| Intron 7 | - | 60 bp 5[prime] to exon 7 DS | ins AA | - | 18 |
Genotype-phenotype correlations
Mutations were observed with a high frequency in rhabdoid tumors whatever their locations since renal rhabdoid tumors demonstrated mutations in 32 of 47 cases, CNS rhabdoid in four of seven cases and other extra-renal MRT in 17 of 18 cases (Table 1).
A total of 18 Wilms' tumors (WTs), including 10 WTs with favorable histology and eight anaplastic WTs, were analyzed. Ten of these cases had been shown previously to harbor chromosome 22 LOH (20). None of these tumors exhibited mutation of hSNF5/INI1.
Among tumors of the CNS, gliomas and ependymomas have been shown to harbor frequent chromosome 22 LOH (21-23). A total of 17 gliomas of various types previously shown to carry chromosome 22 LOH (21) and 25 ependymomas were analyzed for hSNF5/INI1 mutations. Only one of the gliomas demonstrated an hSNF5/INI1 alteration (case 50 in Fig. 2). It was a homozygous deletion detected in a glioblastoma from a 70-year-old woman. None of the ependymomas exhibited hSNF5/INI1 alteration. We also studied other CNS tumors, including medulloblastomas, cPNETs and choroid plexus tumors. hSNF5/INI1 mutations were observed in two of 17 cPNETs and in five of 36 medulloblastomas. Interestingly, mutations were observed more frequently in cPNETs or medulloblastomas of young children. Indeed, one of three cPNETs and three of six medulloblastomas which appeared before 36 months of age demonstrated hSNF5/INI1 alteration, whereas only one of 14 cPNETs and two of 30 medulloblastomas that developed in older patients had hSNF5/INI1 alterations (Table 1). Concerning choroid plexus tumors, which are neoplasms of the cerebral ventricles mainly observed in young children, hSNF5/INI1 mutations were detected with a high frequency (four of six) in carcinoma, the most aggressive form of these tumors. A mutation was also detected in an atypical papilloma, a tumor of intermediate aggressiveness. In contrast, mutations were not observed in papillomas, a benign tumor of the choroid plexus. Finally, hSNF5/INI1 mutations were not observed in a series of 31 tumors of various origins, including four rhabdomyosarcomas, 14 sarcomas and 11 breast cancers previously shown to exhibit LOH on chromosome 22 (24).
DISCUSSION
We have analyzed a number of tumors in a search for mutations of the recently identified hSNF5/INI1 tumor suppressor. This study enables us to define the types of hSNF5/INI1 mutation observed in human cancers and the spectrum of cancers harboring such alterations.
Types of hSNF5/INI1 mutation observed in human cancers
The most frequent alterations of hSNF5/INI1 are deletions. Indeed, HDs represent close to 50% of the mutations, and microdeletions are the most frequent type of point alteration. Surprisingly, we identified no deletions restricted to exon 1, an observation which contrasts with a recent report indicating that such an alteration accounts for >30% of hSNF5/INI1 mutations (17). This discrepancy could be linked to an overestimation of exon 1 HD in the latter report, possibly related to the low efficiency of the PCR amplification of this exon due to its high GC content. Although we frequently observed a low yield of amplification for this exon, a specific search using co-amplification of a 111 bp fragment overlapping exon 1 and intron 1 sequences and devoid of the GC-rich region, together with a 104 bp GAPDH control fragment, revealed no HD restricted to exon 1.
The other predominant alteration is a C->T transition at CG dinucleotides thought to occur through spontaneous deamination of methylcytosine. Other types of transition, insertion and transversion were observed occasionally. All these mutations, apart from the two missense and the editing changes, are expected to lead to truncated hSNF5/INI1 proteins of various sizes (Fig. 4).
Particular mutations were detected in two cases. In one case, an intronic, constitutional A->G transition activated a splice donor site leading to the insertion of a 72 bp pseudo-exon in the cDNA. The pathogenic role of this mutation is supported by the observation that only RNAs including this exon, which contains an in-frame stop codon, are expressed in the tumor. Interestingly, intronic mutations with similar functional consequences have been described for the ATM gene in ataxia telangiectasia (25). Another case displayed an editing A->G modification resulting in an amino acid substitution at the protein level. Further development of an assay for hSNF5/INI1 function might enable the documentation of the possible oncogenic role of this editing. Interestingly, similar editing mechanisms have been reported in the NF1 and WT1 TSGs and have been suspected to play a significant role in oncogenesis (reviewed in ref. 26).
Mutation of hSNF5/INI1 in tumors with a rhabdoid phenotype
Mutations were observed in most cases diagnosed as MRT whatever their locations, since MRTs from kidney, CNS (ATTR) and other locations demonstrated mutations in 32 of 47, four of seven and 17 of 18 cases, respectively. Mutations were observed in 30 of 34 (88%) MRT cases obtained as FT as compared with 23 of 38 (61%) MRT cases from PET, suggesting that our ability to detect mutation might depend on the starting material. Two factors could contribute to this difference. Firstly, in addition to DNA, RNA was isolated from FT and used to perform RT-PCR which detected exon skipping related to interstitial deletion of hSNF5/INI1. Secondly, as expected, PCR was less efficient with DNA from PET as compared with that from FT. Indeed, ~10% of the genomic sequence of the hSNF5/INI1 gene could not be analyzed reliably in PET. These technical constraints can explain our failure to detect mutations in a number of tumors analyzed from PET which were predominantly renal MRTs. In the ideal situation where the starting material is FT, the frequency of detection of an hSNF5/INI1 mutation in MRTs, whatever their location, is close to 90%. This indicates that the dHPLC approach performed on both DNA and RNA is highly efficient for detecting mutations. It also provides very strong evidence that MRT is a genetically homogeneous entity characterized by hSNF5/INI1 mutations.
Mutation of hSNF5/INI1 in other tumor types
hSNF5/INI1 is localized on chromosome 22q11.2, a region altered recurrently in a number of malignant processes. We were thus interested to test different tumors harboring frequent alteration of this chromosome region for hSNF5/INI1 mutations. Among renal tumors, a recent report has indicated that chromosome 22 LOH is associated with anaplastic forms and unfavorable prognosis in WTs, suggesting that anaplastic WTs could be genetically related to renal MRTs (20). In the present series, hSNF5/INI1 mutations were not detected in any of 18 WTs, even though 10 of these harbored chromosome 22 LOH. This indicates that WTs and MRTs are two genetically distinct entities and that hSNF5/INI1 alterations distinguish MRTs from WTs, the most frequent cancer of the kidney in childhood.
Gliomas, ependymomas and breast cancers, which exhibit recurrent chromosome 22 LOH, were also studied (21-24). Apart from one glioma, none of these tumors harbored hSNF5/INI1 mutations, strongly suggesting that hSNF5/INI1 is not the chromosome 22 tumor gene suspected to be involved in these neoplasias. It will be of interest to delineate further the extent of the HD observed in case 50 in order to define whether it is restricted to the hSNF5/INI1 locus or whether it involves a larger region potentially encoding another suppressor gene.
We also studied other embryonal malignancies of the CNS, including supratentorial cPNET, medulloblastoma and choroid plexus carcinoma (CPC). A possible link between some of these tumors and MRT had been suspected previously on the basis of different arguments. Firstly, these tumours can share clinical characteristics with ATTR, in particular young age of occurrence and high invasiveness and aggressiveness (27). Second, monosomy 22 and 22q rearrangements have been reported as the sole cytogenetic abnormality in 10-20% of cases and have therefore been suggested to identify a particular subgroup of tumors (28-31). Third, associations in the same individual of renal MRTs and embryonal tumors of the CNS, including the aforementioned types, have been reported (3,27,32-34). In the present study, hSNF5/INI1 mutations were retrieved in seven of 53 (13%) tumors diagnosed as cPNET or medulloblastoma and in four of six (67%) tumors diagnosed as CPC. These results strongly support the hypothesis that the hSNF5/INI1 mutation identifies, within the heterogeneous group of embryonal CNS tumors, a subgroup which can be genetically merged with MRTs. A common pathogenetic origin of these tumors and MRTs is strengthened further by our recent observation that, in addition to MRTs, medulloblastoma, cPNET and CPC are observed in the context of a germline hSNF5/INI1 mutation (19). Altogether, these data strongly suggest that a subset of embryonal CNS tumors shares common pathways of oncogenesis with MRT and that hSNF5/INI1 mutation defines a spectrum of highly malignant tumors of early childhood which frequently, but not always, exhibit a rhabdoid phenotype. hSNF5/INI1 alteration now provides a molecular marker to define this spectrum of tumors, and the monitoring of this alteration might thus be of considerable diagnostic interest.
Within this spectrum of tumors, the position of the hSNF5/INI1 point mutations does not appear to be correlated with particular tumor types or body localization. However, it is noteworthy that HD was observed preferentially in tumors localized outside the CNS whereas CNS tumors exhibited almost exclusively point mutations (Table 2), suggesting that the chromosome mechanisms leading to hSNF5/INI1 inactivation could depend on the tissue of origin (35).
Schematically, TSGs can be divided into two groups depending on the spectrum of tumors which demonstrate alterations (36). A subset of TSGs, such as p53, Rb or p16, can be somatically mutated in a broad range of neoplasias, whereas their mutation at the constitutional level only predisposes to a limited spectrum of cancer. In contrast, for other TSGs, such as NF2, VHL or APC, acquired mutations are identified in a restricted range of tumors similar to that associated with germline mutation. hSNF5/INI1 behaves more like a TSG from the latter group than one from the former since hSNF5/INI1 acquired mutations are restricted to MRTs and a subset of aggressive CNS tumors, a spectrum of tumors which fully overlaps with that observed in the context of a germline mutation (19).
hSNF5/INI1 encodes a member of the ATP-dependent, chromatin remodeling SWI/SNF complex which is thought to facilitate the access of transcription factors to DNA (37). In yeast, mutations in different members of the complex result in similar swi and snf phenotypes, in particular a deficiency of induction of the HO gene necessary for the mating type switching and of the SUC2 gene required for growth on sucrose and raffinose. This report clearly establishes that hSNF5/INI1 mutation accounts for most, if not all, MRTs, suggesting that mutations in other SWI/SNF members might not lead to the same phenotype. The observation that the hSNF5/INI1 protein, in contrast to other components such as BRG1 or hBRM, is an integral component of the diverse SWI/SNF complexes present in the cell could indicate that the loss of wild-type hSNF5/INI1 hampers the function of a wide range of complexes (38). Alternatively, it may indicate that, in addition to its role as a member of the SWI/SNF complex, the hSNF5/INI1 protein has other cellular functions linked to its tumor suppressor role.
MATERIALS AND METHODS
Patient samples
A total of 229 tumor samples including 41 paraffin-embedded fragments and 188 snap-frozen tissues were studied for hSNF5/INI1 mutations. Constitutional material was obtained from blood or non-tumor tissues.
Mutation analysis
DNA from blood and frozen tumor samples was isolated using standard procedures with phenol/chloroform extraction and ammonium acetate/ethanol precipitation. Paraffin-embedded fragments were deparaffinized with xylene and rinsed in ethanol prior to DNA extraction using a QiaAmp Tissue kit (Qiagen, Germany). RNA was isolated using the Trizol reagent (Gibco BRL, Life Technologies, Rockville, MD).
PCR was performed using the GeneAmp PCR kit with AmpliTaq Gold (Perkin Elmer, Norwalk, CT). Annealing temperatures were 60°C for all the PCR primers except that for exon 1 (64°C/5% DMSO). DNA from PET was amplified and sequenced using short PCR fragments (16 fragments were needed for the screen of the entire gene). The sequences of the primers are available on the web site of the Institut Curie (http://www.curie.fr/sr/unites/u509 ).
HDs were screened by co-amplification of a 138 bp fragment from exon 4 of the hSNF5/INI1 gene together with a 104 bp fragment of GAPDH.
For mutation detection, dHPLC analysis using Wave technology (Transgenomic, San Jose, CA) and direct sequencing were used. dHPLC was carried out on automated HPLC instrumentation. A total of 14 primer pairs were used for amplifying hSNF5/INI1 exons, and the cDNA was divided into five parts after RT-PCR. For the detection of heteroduplexes, equal amounts of tumoral and normal DNA were PCR amplified then submitted to a final denaturation of 10 min at 98°C followed by a gradual reannealing from 98 to 25°C over a period of 40 min. Wavemaker 1.2.2 software (Transgenomic) was used to predict the mean melting temperature of each PCR fragment and the appropriate linear acetonitrile gradient necessary to distinguish hetero- and homoduplexes. These conditions were also evaluated experimentally by analyzing the elution time of PCR products from known mutants (16). The start and end points of the gradient were adjusted according to the size and the percentage of GC nucleotides of each PCR fragment. For sequencing, PCR products were purified on solid-phase columns using the Qiaquick PCR purification kit (Qiagen) then sequenced using the Big Dye Terminator kit (Perkin Elmer) according the manufacturers' specifications. Samples were loaded on an Applied Biosystems 373A or 377 sequencer. The sequences were submitted to Factura and Sequence Navigator softwares (Perkin Elmer) for mutation detection.
ACKNOWLEDGEMENTS
We thank the following colleagues for technical help and clinicians for providing samples used in this study: A. Abel, P. Ambros, D. Amram, M.F. Belin, J. Bénard, N. Blot, A. Bornemann, P. Bret, L. Brugières, V. Caux, A. Chompret, V. Costes, P. Déchelotte, C. Depardieu, F. Doz, M. Fabre, J.C. Fournet, J.L. Gala, E. Gilbert, R. Hamelin, R. Handgretinger, F. Jaubert, C. Kalifa, T. Klingelbiel, N. Kopp, E. Koscielniak, H. Kovar, J. Lange, A. Laquerrière, A. Lauge, F. Lemoine, G. Marguerite, R. Meyermann, J. F. Mosnier, C. Mottolese, S. Pagès, M. Peter, D. Rancherre-Vince, M. Robinson, P. Ruck, G. Saint-Pierre, E. Salame, I. Salmon, N. Saran, N. Schneider, E. Sheridan, D. Stoppa-Lyonnet, P. Validire, J.P. Vannier, G. Vassal, I. Versteege and P. Vielh. This work was supported by grants from the Association pour la Recherche contre le Cancer, the Institut Curie, the Institut National de la Santé et de la Recherche Médicale and the Programme Hospitalier de Recherche Clinique. N.S. is the recipient of a fellowship from the Ministère de l'Education Nationale, de la Recherche et de la Technologie.
REFERENCES
+To whom correspondence should be addressed. Tel: +33 1 42 34 66 81; Fax: +33 1 42 34 66 30; Email: delattre{at}curie.fr
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
This article has been cited by other articles:
![]() |
B. Weissman and K. E. Knudsen Hijacking the Chromatin Remodeling Machinery: Impact of SWI/SNF Perturbations in Cancer Cancer Res., November 1, 2009; 69(21): 8223 - 8230. [Abstract] [Full Text] [PDF] |
||||
![]() |
H. Lin, R. P.C. Wong, M. Martinka, and G. Li Loss of SNF5 Expression Correlates with Poor Patient Survival in Melanoma Clin. Cancer Res., October 15, 2009; 15(20): 6404 - 6411. [Abstract] [Full Text] [PDF] |
||||
![]() |
Y. Liu, G. L. Borchert, S. P. Donald, B. A. Diwan, M. Anver, and J. M. Phang Proline Oxidase Functions as a Mitochondrial Tumor Suppressor in Human Cancers Cancer Res., August 15, 2009; 69(16): 6414 - 6422. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. Das, J. Cano, and G. V. Kalpana Multimerization and DNA Binding Properties of INI1/hSNF5 and Its Functional Significance J. Biol. Chem., July 24, 2009; 284(30): 19903 - 19914. [Abstract] [Full Text] [PDF] |
||||
![]() |
J.-P. Kilday, R. Rahman, S. Dyer, L. Ridley, J. Lowe, B. Coyle, and R. Grundy Pediatric Ependymoma: Biological Perspectives Mol. Cancer Res., June 1, 2009; 7(6): 765 - 786. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. E. Elsheikh, A. R. Green, E. A. Rakha, D. G. Powe, R. A. Ahmed, H. M. Collins, D. Soria, J. M. Garibaldi, C. E. Paish, A. A. Ammar, et al. Global Histone Modifications in Breast Cancer Correlate with Tumor Phenotypes, Prognostic Factors, and Patient Outcome Cancer Res., May 1, 2009; 69(9): 3802 - 3809. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. Caramel, F. Quignon, and O. Delattre RhoA-Dependent Regulation of Cell Migration by the Tumor Suppressor hSNF5/INI1 Cancer Res., August 1, 2008; 68(15): 6154 - 6161. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. Glaros, G. M. Cirrincione, A. Palanca, D. Metzger, and D. Reisman Targeted Knockout of BRG1 Potentiates Lung Cancer Development Cancer Res., May 15, 2008; 68(10): 3689 - 3696. [Abstract] [Full Text] [PDF] |
||||
![]() |
B. A. McKinley and M. V. Sukhodolets Escherichia coli RNA polymerase-associated SWI/SNF protein RapA: evidence for RNA-directed binding and remodeling activity Nucleic Acids Res., December 18, 2007; 35(21): 7044 - 7060. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. Chai, X. Lu, V. Godfrey, C. Fletcher, C. W.M. Roberts, T. Van Dyke, and B. E. Weissman Tumor-Specific Cooperation of Retinoblastoma Protein Family and Snf5 Inactivation Cancer Res., April 1, 2007; 67(7): 3002 - 3009. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. Klochendler-Yeivin, E. Picarsky, and M. Yaniv Increased DNA Damage Sensitivity and Apoptosis in Cells Lacking the Snf5/Ini1 Subunit of the SWI/SNF Chromatin Remodeling Complex. Mol. Cell. Biol., April 1, 2006; 26(7): 2661 - 2674. [Abstract] [Full Text] [PDF] |
||||
![]() |
P. Modena, E. Lualdi, F. Facchinetti, L. Galli, M. R. Teixeira, S. Pilotti, and G. Sozzi SMARCB1/INI1 Tumor Suppressor Gene Is Frequently Inactivated in Epithelioid Sarcomas Cancer Res., May 15, 2005; 65(10): 4012 - 4019. [Abstract] [Full Text] [PDF] |
||||
![]() |
R. G.J. Vries, V. Bezrookove, L. M.P. Zuijderduijn, S. K. Kia, A. Houweling, I. Oruetxebarria, A. K. Raap, and C. P. Verrijzer Cancer-associated mutations in chromatin remodeler hSNF5 promote chromosomal instability by compromising the mitotic checkpoint Genes & Dev., March 15, 2005; 19(6): 665 - 670. [Abstract] [Full Text] [PDF] |
||||
![]() |
B. Suarez-Merino, M. Hubank, T. Revesz, W. Harkness, R. Hayward, D. Thompson, J. L. Darling, D. G.T. Thomas, and T. J. Warr Microarray analysis of pediatric ependymoma identifies a cluster of 112 candidate genes including four transcripts at 22q12.1-q13.3 Neuro-oncol, January 1, 2005; 7(1): 20 - 31. [Abstract] [PDF] |
||||
![]() |
C. B. Zraly, D. R. Marenda, and A. K. Dingwall SNR1 (INI1/SNF5) Mediates Important Cell Growth Functions of the Drosophila Brahma (SWI/SNF) Chromatin Remodeling Complex Genetics, September 1, 2004; 168(1): 199 - 214. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. Medjkane, E. Novikov, I. Versteege, and O. Delattre The Tumor Suppressor hSNF5/INI1 Modulates Cell Growth and Actin Cytoskeleton Organization Cancer Res., May 15, 2004; 64(10): 3406 - 3413. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. J. Guidi, T. M. Veal, S. N. Jones, and A. N. Imbalzano Transcriptional Compensation for Loss of an Allele of the Ini1 Tumor Suppressor J. Biol. Chem., February 6, 2004; 279(6): 4180 - 4185. [Abstract] [Full Text] [PDF] |
||||
![]() |
H. Kang, K. Cui, and K. Zhao BRG1 Controls the Activity of the Retinoblastoma Protein via Regulation of p21CIP1/WAF1/SDI Mol. Cell. Biol., February 1, 2004; 24(3): 1188 - 1199. [Abstract] [Full Text] [PDF] |
||||
![]() |
I. Oruetxebarria, F. Venturini, T. Kekarainen, A. Houweling, L. M. P. Zuijderduijn, A. Mohd-Sarip, R. G. J. Vries, R. C. Hoeben, and C. P. Verrijzer p16INK4a Is Required for hSNF5 Chromatin Remodeler-induced Cellular Senescence in Malignant Rhabdoid Tumor Cells J. Biol. Chem., January 30, 2004; 279(5): 3807 - 3816. [Abstract] [Full Text] [PDF] |
||||
![]() |
T. J. MacDonald, B. R. Rood, M. R. Santi, G. Vezina, K. Bingaman, P. H. Cogen, and R. J. Packer Advances in the Diagnosis, Molecular Genetics, and Treatment of Pediatric Embryonal CNS Tumors Oncologist, April 1, 2003; 8(2): 174 - 186. [Abstract] [Full Text] [PDF] |
||||
![]() |
S. Hwang, D. Lee, Y. Gwack, H. Min, and J. Choe Kaposi's sarcoma-associated herpesvirus K8 protein interacts with hSNF5 J. Gen. Virol., March 1, 2003; 84(3): 665 - 676. [Abstract] [Full Text] [PDF] |
||||
![]() |
D. R. Marenda, C. B. Zraly, Y. Feng, S. Egan, and A. K. Dingwall The Drosophila SNR1 (SNF5/INI1) Subunit Directs Essential Developmental Functions of the Brahma Chromatin Remodeling Complex Mol. Cell. Biol., January 1, 2003; 23(1): 289 - 305. [Abstract] [Full Text] |
||||
![]() |
K. Roy, I. L. de la Serna, and A. N. Imbalzano The Myogenic Basic Helix-Loop-Helix Family of Transcription Factors Shows Similar Requirements for SWI/SNF Chromatin Remodeling Enzymes during Muscle Differentiation in Culture J. Biol. Chem., September 6, 2002; 277(37): 33818 - 33824. [Abstract] [Full Text] [PDF] |
||||
![]() |
Z.-K. Zhang, K. P. Davies, J. Allen, L. Zhu, R. G. Pestell, D. Zagzag, and G. V. Kalpana Cell Cycle Arrest and Repression of Cyclin D1 Transcription by INI1/hSNF5 Mol. Cell. Biol., August 15, 2002; 22(16): 5975 - 5988. [Abstract] [Full Text] [PDF] |
||||
![]() |
D. Y. Wu, D. C. Tkachuck, R. S. Roberson, and W. H. Schubach The Human SNF5/INI1 Protein Facilitates the Function of the Growth Arrest and DNA Damage-inducible Protein (GADD34) and Modulates GADD34-bound Protein Phosphatase-1 Activity J. Biol. Chem., July 26, 2002; 277(31): 27706 - 27715. [Abstract] [Full Text] [PDF] |
||||
![]() |
K. E. Neely, A. H. Hassan, C. E. Brown, L. Howe, and J. L. Workman Transcription Activator Interactions with Multiple SWI/SNF Subunits Mol. Cell. Biol., March 15, 2002; 22(6): 1615 - 1625. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. B. Zraly, Y. Feng, and A. K. Dingwall Genetic and Molecular Analysis of Region 88E9;88F2 in Drosophila melanogaster, Including the ear Gene Related to Human Factors Involved in Lineage-Specific Leukemias Genetics, March 1, 2002; 160(3): 1051 - 1065. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. Wild, P. Langer, A. Ramaswamy, B. Chaloupka, and D. K. Bartsch A Novel Insulinoma Tumor Suppressor Gene Locus on Chromosome 22q with Potential Prognostic Implications J. Clin. Endocrinol. Metab., December 1, 2001; 86(12): 5782 - 5787. [Abstract] [Full Text] [PDF] |
||||
![]() |
E. Kolomietz, J. Al-Maghrabi, S. Brennan, J. Karaskova, S. Minkin, J. Lipton, and J. A. Squire Primary chromosomal rearrangements of leukemia are frequently accompanied by extensive submicroscopic deletions and may lead to altered prognosis Blood, June 1, 2001; 97(11): 3581 - 3588. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. J. Guidi, A. T. Sands, B. P. Zambrowicz, T. K. Turner, D. A. Demers, W. Webster, T. W. Smith, A. N. Imbalzano, and S. N. Jones Disruption of Ini1 Leads to Peri-Implantation Lethality and Tumorigenesis in Mice Mol. Cell. Biol., May 15, 2001; 21(10): 3598 - 3603. [Abstract] [Full Text] |
||||
![]() |
T. Takeuchi, S. Nicole, A. Misaki, M. Furihata, J. Iwata, H. Sonobe, and Y. Ohtsuki Expression of SMARCF1, a Truncated Form of SWI1, in Neuroblastoma Am. J. Pathol., February 1, 2001; 158(2): 663 - 672. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. A. Biegel, B. Fogelgren, J.-Y. Zhou, C. D. James, A. J. Janss, J. C. Allen, D. Zagzag, C. Raffel, and L. B. Rorke Mutations of the INI1 Rhabdoid Tumor Suppressor Gene in Medulloblastomas and Primitive Neuroectodermal Tumors of the Central Nervous System Clin. Cancer Res., July 1, 2000; 6(7): 2759 - 2763. [Abstract] [Full Text] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||

















