NF2 gene mutations and allelic status of 1p, 14q and 22q in sporadic meningiomas

Abstract

Formation of meningiomas and their progression to malignancy may be a multi-step process, implying accumulation of genetic mutations at specific loci. To determine the relationship between early NF2 gene inactivation and the molecular mechanisms that may contribute to meningioma tumor progression, we have performed deletion mapping analysis at chromosomes 1, 14 and 22 in a series of 81 sporadic meningiomas (54 grade I (typical), 25 grade II (atypical) and two grade III (anaplastic)), which were also studied for NF2 gene mutations. Single-strand conformational polymorphism analysis was used to identify 11 mutations in five of the eight exons of the NF2 gene studied. All 11 tumors displayed loss of heterozygosity (LOH) for chromosome 22 markers; this anomaly was also detected in 33 additional tumors. Twenty-nine and 23 cases were characterized by LOH at 1p and 14q, respectively, mostly corresponding to aggressive tumors that also generally displayed LOH 22. All three alterations were detected in association in seven grade II and two grade III meningiomas, corroborating the hypothesis that the formation of aggressive meningiomas follows a multi-step tumor progression model.

Introduction

Meningiomas are neoplasms that arise from the leptomeningeal covering of the brain and spinal cord, accounting for 15 – 20% of all central nervous system tumors. The current WHO grading system comprises three grades as follows: most meningiomas are slow growing, are generally considered benign tumors, and correspond to grade I (typical meningiomas); about 10% of cases are classified as grade II (atypical) or anaplastic/malignant (grade III), exhibiting more aggressive clinical behaviour, with a higher risk of recurrence when compared to typical grade I meningiomas (Kleihues et al., 1993). These neoplasms are usually sporadic, but a few families have been described with multiple tumors inherited in an autosomal dominant fashion (Memon, 1980; Battersby et al., 1986; Butti et al., 1989; Domenicucci et al., 1989), and they also occur in as many as half of the patients with the dominantly inherited familial neuro-fibromatosis type 2 syndrome (NF2) (Martuza and Eldridge, 1988). Using positional cloning approaches, the candidate gene for NF2 has been isolated from chromosome 22q12 region (Rouleau et al., 1993; Trofatter et al., 1993), and mutations have been observed in both germ-line of NF2 patients and in about 30% of sporadic meningiomas (Ruttledge et al., 1994; Lekanne-Deprez et al., 1994; Merel et al., 1995; Papi et al., 1995; De Vitis et al., 1996; Harada et al., 1996). These findings indicate that inactivation of the NF2 gene is important in the development of a significant number of sporadic cases that also generally display LOH for markers located on chromosome 22. This provides conclusive evidence that NF2 acts as a tumor suppressor gene in at least a subgroup of sporadic tumors. Cytogenetic analysis of meningiomas previously provided evidences of the non-random involvement of chromosome 22 in about 60% of tumors, suggesting that inactivation of tumor suppressor loci (the NF2 gene is now identified as one of them) located in this chromosome might represent an early step in meningioma tumorigenesis (Zang, 1982; Al Saadi et al., 1987; Maltby et al., 1988; Rey et al., 1988; Casalone et al., 1990). Non-random loss of chromosome 14 and structural rearrangements of chromosome 1, generally leading to the loss of short arm regions, were also identified as characteristic cytogenetic features of meningiomas; to a lesser degree, abnormalities of chromosomes 7, 8, 10, 18, 19 and 20 have also been found (Rey et al., 1988; Casalone et al., 1990; Casartelli et al., 1989; Poulsgard et al., 1993; Vagner-Capodano et al., 1993; Lekanne-Deprez et al., 1995Lekanne-Deprez et al., 1995; Lopez Gines et al., 1995; Schneider et al., 1995; Perry et al., 1996). The accumulation of chromosomal abnormalities secondary to chromosome 22 loss, frequently in parallel to the genesis of grade II and III tumors, allowed the proposal that meningiomas might display clonal progression through a pattern, implying the accumulation of several genetic anomalies in which monosomy of chromosome 22 would represent an early step (Rey et al., 1988; Poulsgard et al., 1993). Molecular genetic studies have also recently investigated the association between the allelic loss at genomic regions other than chromosome 22 and tumor progression of meningiomas, suggesting that LOH at 1p, 14q and chromosome 10 appear recurrently associated with grade II and grade III meningiomas (Rempel et al., 1993; Bello et al., 1994; Lindblom et al., 1994; Simon et al., 1995; Menon et al., 1997). A multi-step mechanism similar to that described for other neoplasms (Fearon and Vogelstein, 1990) might thus explain the origin and evolution of aggressive meningiomas. To determine the relationship between NF2 gene inactivation and the genomic regions harboring `progression loci' potentially involved in the neoplastic evolution of meningiomas, we examined NF2 gene mutations and the allelic constitution of 1p, 14q and 22q in a series of 81 sporadic meningiomas. Our data confirmed the association between losses of 1p and/or 14q and the genesis of grade II – III tumors, and provide insights into their relationship with some specific inactivating NF2 gene mutations.

Results

LOH studies

Twenty-nine tumors were characterized by allelic losses for loci of chromosome 1 (Figures 1 and 2). Two of these 29 tumors lost alleles at loci spanning the entire short arm of chromosome 1 and the loss extended into the long arm in both cases, suggesting that the entire chromosome 1 had probably been lost in these meningiomas (tumors M-22 and M-60). Interstitial and/or terminal deletions were observed in the remaining 27 cases. Twelve meningiomas displayed terminal deletion of 1p, which in one sample (M-29) was restricted to the most distaly analysed locus D1Z2. Three tumors were characterized by loss of the 1p32-pter region, with break points distal to D1S17 (M-65), D1S57 (M-3) and D1S7 (M-14). Five cases displayed break points at 1p13, distal to D1S440 (M-13, M-15 and M-74) and distal to D1S189 in cases M-72 and M-23. Complex chromosome rearrangements occurred in tumors M-16, M-17 and M-64; LOH was observed for loci at pter-p32 and at p13, with retention of alleles at D1S10, D1S73 and D1S17, respectively, which would define two regions of chromosomal losses on the short arm of chromosome 1. Interstitial 1p deletions were evidenced in 15 meningiomas. In five tumors, locus D1Z2 limited the losses distally, with the proximal border of the deletions defined by break points in the p13-cen region (tumors M-18, M-53, M-76, M-11 and M-75). This was also the proximal deletion border in three samples characterized by retention of heterozygosity at D1S76 (meningiomas M-52, M-39 and M-61). Four additional tumors presented the 1p distal border of the deletion limited by the retention of heterozygosity at locus D1S7, but with a varying length of deletion, as marked by the proximal border according to the loci displaying retention of heterozygosity: NRAS (M-69), D1S9 (M-5 and M-81) and D1S17 (M-7). The break points in this meningioma delimited a small loss involving only locus D1S57. In tumor M-79, LOH was restricted to locus D1S9 and boundaries delimited by retention of heterozygosity at D1S73 and D1S17. Finally, tumors M-31 and M-63 presented LOH at two distinct 1p regions, as evidenced by losses involving D1S77 and ACADM in M-31, and deletion of the D1S7-D1S57 and D1S14 regions in tumor M-63.

Figure 1

Summary of chromosome 1 deletion mapping in sporadic meningioma. The loci analysed and their location is shown to the left (blank: no data available)

Figure 2

Examples of loss of heterozygosity detection with PCR-amplified microsatellites. Data for marker D1S189 of cases M-17 and M-15 are shown. N, T: normal, tumoral DNA

According to these findings, up to three distinct regions of deletion might be proposed. The frequent loss of D1Z2 and D1S76 suggests that 1p36-p35 might be a critical region. A subgroup of tumors with interstitial deletions was defined by the frequent loss of loci D1S57 and D1S17 (located at 1p35-p32 and 1p32-p22, respectively), with case M-7 displaying marker D1S57 as the sole loss. This locus was also frequently deleted in most samples with terminal 1p deletions but, as shown in Figure 3, this group of meningiomas displayed a slightly more telomeric deletion, as indicated by the frequent loss of marker D1S7 instead of D1S17. A third region of interest might be defined by tumors losing marker D1S9 at 1p22-p13. In M-79, the 1p deletion was restricted to this marker, although the loss of this 1p region commonly appeared in both meningioma subgroups, with terminal or interstitial deletions.

Figure 3

Graphic representation of the LOH frequency for each locus analysed in meningiomas displaying terminal or interstitial 1p deletions. Region 1p36 (locus D1Z2) was lost in the group of tumors with terminal deletion, whereas loss at the 1p22-p13 region (loci D1S9 and D1S189) occurred in a similar frequency in both groups of meningiomas. In region 1p35-p32 (loci D1S7, D1S57 and D1S17), some differences can be observed, as allelic loss is more proximal in the group of meningiomas with interstitial 1p deletion

Of the 23 tumors with LOH for chromosome 14 loci, 11 displayed loss of the entire chromosome, ten were characterized by partial 14q deletions and two samples had complex patterns of allelic loss and retention consistent with complex chromosomal rearrangement. Two deletion regions might be defined, corresponding to 14q22-q24 and 14q32-qter. Tumor M-77 displayed LOH involving only marker D14S13, whereas tumors M-46 and M-62 only showed loss at D14S20. These findings might indicate the existence of two critical distinct subregions within 14q32 (Figure 4).

Figure 4

Summary of chromosome 14q deletion mapping in sporadic meningioma. The loci analysed and their location is shown to the left (blank: no data available)

Unambiguous evidence of LOH at one or more chromosome 22 markers was observed in 44 of 81 (54%) meningiomas (Table 1). In 38 tumors, the results were compatible with the loss of the whole chromosome, because loss of alleles occurred at all informative loci along chromosome 22. In the other six instances, partial chromosome 22 deletions were evidenced with break points located at q11 in five tumors, distal to D22S9 (in two instances), distal to IGLV (in one instance), and distal to D22S20 (in two samples). One tumor presented the break point at q12 distal to D22S32. According to these findings, all but one sample characterized by LOH 22q displayed loss at the NF2 region.

Table 1 Table 1

Screening for NF2 mutations

SSCP analysis revealed abnormal bands in 11 of 81 meningiomas; four mutations were localized in exon 2, two in exon 7 and five in exon 11. None of these mutations was germinal. To characterize the nucleotide sequence alterations detected by mobility shifts in SSCP analysis, we sequenced the abnormal samples after amplification by PCR; the NF2 gene mutations detected are summarized in Table 1.

In exon 2, we identified two different mutations. The first was an A to G transition at position 189 (codon 63), with no apparent change in the protein amino acid sequence (Gly→Gly); this was detected in case M-3. The second mutation was observed in three tumors (M-17, M-36 and M-56) and was a C to T transition at nucleotide 169 of codon 57, creating a stop codon (Figure 5). The first mutation detected in exon 7 was an AT deletion at codons 206 – 207, causing a frameshift mutation (case M-5), whereas the second was a C to A transversion at codon 211 (nucleotide 632) that resulted in a missense Ala→Asp change (case M-66). Four different mutations were identified involving exon 11. C to T transitions were evidenced at codon 341 (nucleotide 1021) (cases M-6 and M-7), codon 337 (nucleotide 999) (tumor M-9) and codon 362 (nucleotide 1084) (tumor M-12) originating nonsense mutations consisting in Arg→Stop or Gln→Stop. The fourth mutation was a T deletion at position 1088 (codon 363), resulting in a frame shift change.

Figure 5

Representative sequencing data. Tumors M-17, M-36 and M-56 displayed an identical NF2 gene mutation: C to T transition at codon 57 of exon 2, which changes an arginine codon (CGA) to a stop codon (TGA)

Association of anomalies

Allelic losses

 Thirty tumors displayed LOH at two or three chromosomal regions (Table 2). Histologically they corresponded to 11 grade I, 17 grade II and two grade III tumors. Both anaplastic grade III tumors accumulated LOH at 1p, 14q and 22q, whereas this situation was evidenced in eight grade II and one grade I tumors. The LOH1p/LOH22q association was more frequently observed (in four grade I and 7 grade II tumors) than the association LOH14q/LOH22q (detected in six grade I and two grade II samples). No case was found displaying LOH1p/LOH14q and a normal chromosome 22 allelic constitution.

Table 2 Table 2

LOH pattern and NF2 gene mutations

 All 11 tumors displaying NF2 gene mutations showed LOH at 22q (Table 1); this was the sole allelic loss in three samples (M-6, M-12, M-56), whereas the remaining eight cases accumulated other LOH. Three tumors presented LOH at 1p, 14q and 22q; they corresponded to two grade II (M-5 and M-7) and one grade III (M-17) meningioma. LOH 14q and LOH 22q, and retention of the allelic constitution of 1p was evidenced in four grade I tumors (M-9, M-36, M-66 and M-77). Finally, one typical meningioma (M-3) displayed LOH1p and LOH22q in association. Identical NF2 inactivant mutations at exon 2 and exon 11 were identified in three and two tumors, respectively. At codon 57 of E2, the same C to T transition was identified in two grade I and one grade III cases. The last (M-17) also displayed LOH at 1p, 14q and 22q, whereas the grade I tumors were characterized by either LOH22q (M-56) or LOH at 14q and 22q (M-36). Similar nonsense mutations due to a C to T transition at codon 341 of exon 11 were detected in cases M-6 and M-7. M-6 was a grade I meningioma which displayed LOH 22q as the sole allelic loss, whereas M-7 was a grade II tumor characterized by LOH at all three chromosomal regions analysed (1p, 14q, 22q). According to these findings, the accumulation of allelic loss at 1p and 14q, secondary to a given inactivating NF2 gene mutation and LOH 22q, seems to be a critical step in the genesis of aggressive forms of meningioma.

Discussion

Previous cytogenetic and molecular genetic studies on meningioma suggested that a genetic model of malignant progression might explain the formation of atypical and anaplastic tumors, as a result of a mutation accumulation process. An early step in this model is the inactivation of a tumor suppressor gene on chromosome 22. At present, up to three genes of this category have been identified in this chromosome: the neurofibromatosis type 2 gene (NF2) (Rouleau et al., 1993; Trofatter et al., 1993), BAM22 gene, a member of the β-adaptin family (Peyrard et al., 1994), and the MN1 gene, disrupted by a balanced translocation in a meningioma cell line (Lekanne-Deprez et al., 1995Lekanne-Deprez et al., 1995).

In our study, we screened 979 of the 1785 nucleotides of the NF2 cDNA corresponding to 54.8% of the NF2 coding sequence. We found 11 cases with mutations representing about 24.5% of mutated cases if the complete coding sequence had been analysed. This frequency is similar to that found in other studies, with the sole exception of Wellenreuther et al. (1995), who found up to 59% of mutations in a series of 70 sporadic meningiomas. C to T transitions were identified in seven cases in our series, confirming that the CpG dinucleotides of the NF2 gene might represent sites prone to point mutations. All 11 mutations in our series occurred at exons 2, 7 and 11 which, in agreement with De Vitis et al. (1996), might be considered exons of preferential involvement, although additional studies are required to verify this.

The inactivation of the NF2 gene does not seem to play a role in the development of malignancy, as most cases displaying this anomaly as the sole alteration are generally classified as grade I meningiomas. Loss of alleles at other genome regions might therefore participate in this process. We have previously reported the association of LOH at 1p and tumor progression of meningiomas (Bello et al., 1994); this alteration was also frequently detected in a series of anaplastic meningiomas (Lindblom et al., 1994). Simon et al. (1995) also found an increased frequency of LOH at 1p, 14q and 10q, parallel to a higher grade of meningioma malignancy. Our results concur with these findings and suggest that up to three distinct 1p regions (1p36, 1p35-p32 and 1p22-p13) might be deleted during meningiomas tumorogenesis, indicating the presence of tumor suppressor genes in these genomic regions. No data on this subject are available, as only a small number of 1p loci had been analysed and no extensive deletion mapping was performed in meningiomas (Bello et al., 1994; Lindblom et al., 1994; Simon et al., 1995). Our findings agree with the data of Mertens et al. (1997), who proposed four 1p regions of preferential involvement in solid tumors. In fact, 1p alterations seem to be associated with a broad spectrum of human malignancies, including both solid tumors and leukemias (Schwab et al., 1996). A close analysis of the 1p alterations in specific tumoral subtypes will probably show the involvement of specific genes.

Simon et al. (1995) proposed 14q24.3-q32.33 as the critical region involved in losses in meningiomas. This fact is also shown by our data which suggest that two minimal regions of deletion would consist of 14q22-24 and 14q32, perhaps excluding the putative target of chromosome 14 deletions in colon cancer as the gene involved in atypical and anaplastic meningioma, as previously suggested by Simon et al. (1995). Construction of detailed partial deletion maps in meningioma, together with data from reports on other neoplasms displaying 14q losses (Takayama et al., 1992; Young et al., 1993; Fujino et al., 1994; Chang et al., 1995; Kovacs, 1993; Bandera et al., 1997) would support with the hypothesis that two or more critical chromosome 14 regions may harbor tumor suppressor genes, located primarily at 14q12-q21, 14q22-q24 and 14q32 (Menon et al., 1997; Bandera et al., 1997).

Indirect evidence of the non-random role of the inactivation of loci from 1p and 14q in meningioma progression has been shown by the observations obtained in several tumors in our series which displayed identical NF2 mutations. C to T transitions at codon 57 of exon 2 were detected in cases M-56, M-36 and M-17, and at codon 341 of exon 11 in tumors M-6 and M-7. Only those cases which, in addition to the NF2 gene inactivation, accumulated LOH at both 1p and 14q, corresponded to grade II or grade III meningiomas. Tumors M-9, M-36, M-66 and M-77 were classified as grade I, and displayed NF2 gene inactivation, LOH at 14q and retention of alleles at 1p. The alteration at 14q probably produces grade I tumors, and LOH at 1p thus seems to be linked to the development of aggressive meningiomas, as suggested by the findings in cases M-3, M-5, M-7 and M-17. We should consider that meningiomas escaping from this progression model do exist (tumors M-37, M-38, M-65, M-75, in our series), and perhaps they follow a distinct molecular pathway, i.e. inactivation of other 14q/1p progression loci, or at 10q loci (Rempel et al., 1993; Simon et al., 1995).

Simultaneous loss at 1p and 14q is not exclusive to aggressive meningioma, as a similar association has been detected in neuroblastoma (Takayama et al., 1992). Cooperative inactivation of tumor progression genes located here might be a relevant event in the pathogenesis of certain tumoral subtypes. Simultaneous loss of 1p and 19q occurs in up to 75% of brain tumors with a major oligodendroglial component (Reifenberger et al., 1994; Bello et al., 1995; Kraus et al., 1995), thus providing evidence for the location at 1p of a tumor-related gene that frequently appears altered in association to locus anomalies from other genomic regions. Bieche et al. (1994) proposed the location of a tumor suppressor gene on chromosome 1p32-pter that controls the amplification of MYC family genes in breast tumors and, recently, the RAD54 gene (Rasio et al., 1997) and p73 (a TP53 related gene) (Jost et al., 1997; Kaghad et al., 1997), both located at 1p, have been proposed as candidate loci for the alterations of the short arm of chromosome 1.

According to the data presented here, some atypical and malignant meningiomas may result from progression of grade I tumors and should not necessarily be considered as a separate, distinct class of meningiomas. If the existence of more than one tumor-related locus at 1p and 14q is confirmed, the involvement of different cooperative genes in these regions might suggest the existence of distinct biological forms of meningioma.

Materials and methods

Human tissue samples

Paired blood and tumor samples were obtained from 81 patients with meningiomas. All tumors were classified by histologic examination and graded according to World Health Organization (WHO) guidelines (Kleihues et al., 1993), as follows: 54 grade I (typical), 25 grade II (atypical), and two grade III (anaplastic) meningiomas. The tumoral cell content in all samples was estimated at approximately 75% by histologic analysis. Available information on clinico-pathological data is shown in Table 1.

DNA extraction

High molecular weight DNA was extracted from the tumor tissue and peripheral blood leukocytes by standard methods, as described previously (Rey et al., 1992).

RFLP analysis

DNA samples were digested to completion with restriction enzymes, electrophoresed through 0.8 – 1% agarose gels, transferred to nylon membranes (Zeta-Probe, Bio-Rad, Richmond, VA, USA), and hybridized to probes radiolabeled by nick-translation or random oligonucleotide priming procedures (Rigby et al., 1977; Feinberg and Vogelstein, 1984). A panel of DNA probes was used to detect LOH at 29 loci on chromosomes 1, 14 and 22 as follows: D1Z2 (p1.79); D1S76 (pCMM12.1); D1S77 (pMCT58); D1S7 (pMS1); D1S57 (pYNZ2); D1S17 (p3.18), ACADM (pMCAD); F3 (pHTF8); D1S9 (p1.04); D1S10 (p1.08); D1S14 (p6.02); D1S73 (pEFD53.2); NRAS (pCN2); D1S61 (pMLAJ1); D1S66 (pHBl40); D14S13 (pCMM101); D14S18 (pHMZ9); D14S16(pTHHH37.1); D14S17 (pEFZ18.2); D14S20 (pMCOC12); D14S23 (cKKA39); D22S9 (p22/34); IGLV (pV3.3); D22S20 (pFZVIA2); D22S32 (pEFZ31); MB (pHM27.B2.9-MB); PDGFB (pPDGF-B17); D22S80 (pFZVD11); D22S171 (pFZIXC11).

Microsatellite analysis

LOH studies were also performed by analysis of CA repeat polymorphisms. Oligonucelotide primer pairs were obtained from GENSET SA (France) for the amplification of microsatellites from the following loci: D1S189, D1S440, D14S66, D14S70, D14S72 and D14S77. Each PCR was performed in a 20 μl mixture containing 100 ng of genomic DNA, 0.4 μM of each primer, 0.4 μM each of dATP, dCTP, dTTP and dGTP and 0.5 units of Taq polymerase in GeneAmp PCR buffer, on a PTC-100 Programmable Thermal Controller (MJ Research, Inc, Watertown, MA, USA). For all primer pairs, the mixtures were first heated to 94°C for 5 min, then subjected to 35 PCR cycles, of 94°C for 30 s, 57°C for 30 s and 72°C for 30 s, with a final 5 min extension at 72°C. To improve the results obtained with some primer pairs, the annealing temperature was varied from 52 – 63°C, and the extension time was increased to 1 min. The alleles were resolved on 6% polyacrylamide sequencing gels, then silver-stained as described by Bender et al. (1994). Scanning densitometry to determine allelic dosage was performed as described (Rey et al., 1992).

NF2 gene mutation analysis

Eight of the 17 exons of the NF2 gene (exons 2, 5, 7, 8, 9, 11, 12 and 15, corresponding to approximately 55% of the coding region) were examined for point mutations by PCR/SSCP analysis, using the primers and PCR conditions specified by Rouleau et al. (1993). SSCP analysis was performed as follows: 6 μl of the radioactive PCR product were mixed with 4 μl of stop solution (95% formamide, 20 mM EDTA, 0.05% bromophenol blue, 0.05% xylene cyanol). Samples were heated to 95°C for 5 min and then loaded onto a 6% non-denaturing polyacrylamide gel (with or without 5% glycerol). A non-denaturing control was also included so that single-stranded fragments could be easily identified. The samples were electrophoresed at 10 W for 5 – 18 h at room temperature. Abnormal bands detected by SSCP analysis were reamplified by PCR, the DNA purified on a gel and used as a template for the sequencing reactions with the same primers used for PCR. The corresponding region of peripheral blood leucocytes derived DNA was also sequenced. Sequencing was performed using the Sequenase PCR Product Sequencing kit (USB-Amersham, UK), according to the manufacturer's indications.