Comprehensive genetic and epigenetic analysis of sporadic meningioma for macro-mutations on 22q and micro-mutations within the NF2 locus

Chromosome 22 DNA copy number profiling using array-CGH

We studied 126 sporadic meningioma samples, composing a representative series of tumors operated on at a single hospital in Sweden. All tumor samples studied were analyzed previously using a lower resolution approach [7, 22, 23, 31, 36]. The main rationale in this part of the project was to identify tumors possessing gene dosage aberrations of 22q which did not involve the NF2 locus. We carried out DNA copy number profiling using array-CGH on a genomic microarray covering chromosome 22. This array, containing 480 measurement points across the sequenced part of this autosome, was developed and validated previously [37]. Among the samples used for validation of this array was meningioma T24 DNA, which is a well characterized tumor containing a large terminal deletion of 22q and an additional approximately 150 kb biallelic deletion involving EWSR1, GAR22, RRP22, AP1B1, RFPL1 and NEFH genes [37, 38]. On average, each array-CGH experiment generated ~400 positively scored data points from 22q.

Table 1 summarizes results obtained from DNA copy number profiling of chromosome 22 and other results, as well as relevant clinical information pertaining to the studied patients. The predominant profile detected in 59 tumors was monosomy 22, which accounted for 47% of meningiomas, and this result is consistent with previous studies [7, 22, 23, 31, 36]. The biological consequence of this very common aberration for the affected tumor cells is one copy loss of all genes from 22q. The second category comprised of16 tumors (13%) that were affected by terminal deletions, interstitial deletions and gains, present singly or in combinations (Figure 1A). This number is also consistent with previous analysis using a loss of heterozygosity (LOH) approach [7, 22, 23, 31, 36]. However, due to a much higher density of data points used in the current analysis, we were able to detect additional aberrations as well as exactly define the breakpoints of all rearrangements in samples which were characterized previously using LOH. Figure 1A shows all 16 cases in detail, and Figs. 2 and 3 display array-CGH profiles of 7 tumors providing new information. Tumor M64 (Figs. 1A and 2A) disclosed a previously unknown biallelic deletion affecting the NF2 gene. This was the only locus affected by a homozygous deletion in the studied series of tumors, which confirms the importance of the NF2 gene in meningioma development. This fibroblastic meningioma has an overall 22q profile consistent with a very large (> 30 Mb) terminal deletion of the long arm of chromosome 22. The second deletion affects only one clone (ID 151) [GenBank: AC005529] and specifically targets the NF2 gene, thus inactivating its second allele in tumor cells. Another interesting example is fibroblastic meningioma M11, since it displays a terminal and multiple interstitial deletions, all of which are located outside the NF2 locus (Figs. 1A and 2E). Three of these deletions, located towards the telomere of 22q, were previously characterized [39]. However, current analysis revealed an additional, smaller (~550 kb) deletion, encompassing 5 contiguous clones (ID 85–89) [GenBank: U07000–AP000348], located on the centromeric side of the NF2 locus (Figures 1A, 1B and 2E). This tumor is intriguing since no additional genetic/epigenetic aberration in the NF2 gene could be detected in this sample (see below). The vast majority of copy number aberrations of 22q in meningiomas point to inactivation or a decreased gene dosage. However, in two tumors (M109 and M96, Figs 1A, 2C and 3A) we were able to detect a low copy number gain. For instance, tumor M109 displays a combination of low copy number gain in the centromeric part of 22q and a large terminal deletion (Figure 2A). A minimum overlapping region of gain is approximately 3.8 Mb (clone ID 11–48) [GenBank: AC006285 – AC007957] and is located in 22q11. The importance of these gains for meningioma tumorigenesis is difficult to judge, as they are not frequent events. One possible explanation is that these gains reflect cell-specific chromosomal instability of 22q11, since this region is rich in intra-chromosomal segmental duplications and is prone to rearrangements leading to several genetic disorders.

Sample ID	Subtype1	CH3	NF2 gene mutations			DNA copy number profiles 4
			Position	Mutation2	Assumed effect3	Chromosome 22	Chromosome 1	Loss of non-22 genes5
M26	A		Ex 2	227_240del	Lys76fs	Monosomy	Monosomy	D1, S1A, S1B
M71	A		Ex8	810+2T>C	Abn. Spl.	Monosomy	1p, 120.8 Mb TD	D1, P2A, S1A, S1B
M38	A		Ex6	573del; 575_581del	Trp191fs	Monosomy	1p 107.9 Mb TD	P1B, P2A, S1A, S1B
M43	A		Ex12	1185del	Lys396fs	Monosomy	1p 120.3 Mb TD, 4.4 Mb A
M46	A		Ex8	810+2T>C	Abn. Spl.	NAD	NAD
M33	A			NAD		30.6 Mb TD [AC009516-AC002055]	1p, 90.2 Mb TD, 45.8 Mb ID
M9	A			NAD		NAD	NAD
M20	A			NAD		NAD	NAD
M13	F		Ex 2	240+2T>C	Abn. Spl.	Monosomy	1p, 142.9 Mb TD	P2A, S1A, S1B
M88	F		Ex4	364-1G>A	Abn. Spl.	Monosomy	1p 84.1 Mb TD; 1q 4.6, 7.1 Mb IDs, 12.5 Mb TD	P2A
M3	F		Ex 2	240+2T>C	Abn. Spl.	Monosomy	1p 119.8 Mb TD
M14	F		Ex3	249del	His84fs	Monosomy	1p 120.2 Mb TD
M15	F		Ex8	683del	Lys228fs	Monosomy	1p 116.0 Mb TD
M39	F		Ex8	810+1G>A	Abn. Spl.	Monosomy	NS
M42	F		Ex10	995del	Lys332fs	Monosomy	NS
M70	F		Ex6	599+1G>A	Abn. Spl.	Monosomy	NS
M91	F		Ex 2	154del	Arg52fs	Monosomy	NS
M102	F		Ex9	865A>T	Lys289X	Monosomy	NS
M115	F		Ex7	634C>T	Gln212X	NAD	1p 119.6 Mb TD
M646	F			NAD		30.4 Mb TD [AC008079-AC002055], NF2 biallelic del [AC005529]	1p 98.8 Mb ID; 1q 24.2 Mb ID	P1B, P2A
M66	F			NAD		Monosomy	1p 118.1 Mb TD	P2A
M117	F			NAD		0.55, 2.8, 4.5 Mb IDs [U07000-AP00034, AC005004-Z69907, AL022311-Z83851]; 5.9 Mb TD [Z93244-AC002055]	1p 116.8 Mb TD, 3.6 Mb A
M32	F			NAD		5.3 Mb ID [AL022329-AL022336], 15.2 Mb TD [AL009049-AC002055]	NS	D1
M4	F			NAD		Monosomy	NAD
M29	F			NAD		Monosomy	NS
M41	F			NAD		Monosomy	NS
M92	F			NAD		Monosomy	NS
M114	F			NAD		18.0 Mb ID [AP000525-Z68324]	NS
M119	F			NAD		NAD	NS
M10	M		Ex13	1341-2A>G	Abn. Spl.	Monosomy	1p 120.3 Mb TD	P2A
M23	M		Ex3	241-2A>C	Abn. Spl.	Monosomy	1p 120.9 Mb TD	P2A
M67	M		Ex4	432C>A	Tyr144X	Monosomy	1p 120.8 Mb TD	P2A
M98	M		Ex14	1519_1520del	Phe507fs	Monosomy	NAD	D1, P2A, S1A, S1B
M57	M		Ex 2	133_144 del	Asp45fs	Monosomy	NS
M72	M		Ex7	634C>T	Gln212X	Monosomy	NS
M40	M			NAD		Monosomy	1p 164.9 Mb TD	P1B, P2A
M76	M			NAD		Monosomy	1p 142.7 Mb TD	P2A
M105	M			NAD		Monosomy	1p 116.3 Mb TD	P1B, P2A
M348	M			NAD		20.0 Mb TD [AC004542-AC002055]	1p 102.8 Mb ID
M44	M			NAD		Monosomy	1p 85.4 Mb TD
M101	M			NAD		25.1 Mb ID [AP000525-AL035450], 7.9 Mb TD [AL049758-AC002055]	Monosomy, 1p 2.1 Mb R
M30	M			NAD		Monosomy	NAD	D1, P2A
M112	M			NAD		NAD	1p 114.2 Mb TD
M2	M			NAD		NAD	NAD
M5	M			NAD		NAD	NAD
M77	M			NAD		NAD	NAD
M82	M			NAD		NAD	NAD
M45	M			NAD		NAD	NS
M51	M			NAD		NAD	NS
M54	M			NAD		NAD	NS
M61	M			NAD		NAD	NS
M63	M			NAD		NAD	NS
M65	M			NAD		NAD	NS
M69	M			NAD		NAD	NS
M86	M			NAD		NAD	NS
M95	M			NAD		NAD	NS
M99	M			NAD		NAD	NS
M107	M			NAD		NAD	NS
M108	M			NAD		NAD	NS
M111	M			NAD		NAD	NS
M116	M			NAD		NAD	NS
M120	M			NAD		NAD	NS
M93	P		Ex8	810G>T	Glu270Asp	Monosomy	1p 114.3 Mb ID	P2A
M229	P		Ex13, inter1	1390del; 28309611 C>A	Ala464fs, unknown	Monosomy	NS
M62	P			NAD		Monosomy	NS
M21	P			NAD		NAD	NS
M60	T		Ex14	1519_1528del	Phe507fs	Monosomy	1p 118.2 Mb TD	P2A
M1	T		Ex7	650del	Tyr217X	Monosomy	1p 65.0 Mb TD
M31	T		Ex13	1325del	Ala451fs	Monosomy	1p 53.2, 36.3 Mb IDs
M58	T		Ex 2	115-3G>C	Abn. Spl.	30.1 Mb TD [AC007050-AC002055]	1p 120.2 Mb TD
M90	T		Ex3	241-2A>G	Abn. Spl.	20.1Mb ID [AP000525-AL079295], 13.4 Mb TD [AL049850-AC002055]	1p 83.3 Mb TD
M96	T		Ex12	1175G>T	Glu392X	2.0 Mb ID [AP000525-AC007666], 5.4 Mb A [AC006285-AC000102], 27.5 Mb TD [U07000-AC002055]	1p 43.3 Mb TD, 1q 69.9 Mb TD
M52	T		Ex4	431_432insAA	Tyr144X	Monosomy	NS	P1B
M113	T		Ex4	377del	Leu127X	Monosomy	NAD	D1, P1B, P2A, S1A, S1B
M18	T		Ex12	1198C>T	Gln400X	Monosomy	NS
M37	T		Ex8	745_755del	Arg249fs	Monosomy	NS
M48	T		Ex 1	36_37del	Leu14fs	Monosomy	NS
M75	T		Ex8	750del	Thr251fs	Monosomy	NS
M87	T		Ex4	436G>A	Val146Ile	NAD	NS
M56	T			NAD		Monosomy	1p 116.5 Mb TD	P2A
M85	T			NAD		Monosomy	1p 108.2 Mb TD	P2A
M109	T			NAD		5.7 Mb A [AC006548-AC007957], 30.6 Mb TD [AC009516-AC002055]	1p 120.2 Mb TD
M27	T			NAD		15.5Mb ID [AP000525-AC005005], 14.1 Mb TD [AL049749-AC002055]	NS
M59	T			NAD		Monosomy	NS
M68	T			NAD		Monosomy	NS
M94	T			NAD		Monosomy	NS
M97	T			NAD		Monosomy	NS
M25	T			NAD		NAD	1p 98.7 Mb TD
M117	T			NAD		NAD	1p 36.1 Mb TD
M10410	T	+		NAD		NAD	NS
M8	T			NAD		NAD	NS
M12	T			NAD		NAD	NS
M17	T			NAD		NAD	NS
M35	T			NAD		NAD	NS
M55	T			NAD		NAD	NS
M89	T			NAD		NAD	NS
M103	T			NAD		NAD	NS
M78	U		Ex10	999+1G>T	Abn. Spl.	Monosomy	1p 3.8 Mb TD
M79	U			NAD		NAD	1p 39.1Mb ID, 1q 41.3Mb TD
M83	U			NAD		NAD	NS

1 Tumors are grouped according to histopathological subtypes. Abbreviations: A, Anaplastic; F, Fibroblastic; M, Meningothelial; P, Psammomatous; T, Transitional and U, Unclassified.

2 Position of the mutations according to numbering of the cDNA sequence, nucleotide number one is the A in the ATG translation initiation codon [GenBank:AF369658]. Abbreviations: substitution, >; deletion, del; insertion, ins; No Aberration Detected, NAD. Donor splice site mutations indicated by the last nucleotide of the preceding exon, a plus sign and position in the intron (e.g. 240+2T>C). Acceptor splice site mutations denoted by the first nucleotide of the following exon, a minus sign and position in the intron (e.g. 241-2A>C).

3 Amino acid number one is the translation initiator Methionine. Abbreviations: translation termination codon, X; frame shift, fs; abnormal splicing, Abn. Spl.

4 Data from copy number imbalances on chromosome 1 were previously published [34]. Abbreviations: Terminal Deletion, TD; Interstitial Deletion, ID; Amplification/gain, A; Not Studied, NS; No Aberration Detected, NAD; Retained segment R. Genbank accession numbers in squared brackets.

5 Loss of non-chromosome 22 clones spanning the DAL1, D1; PTCH2A, P2A; PTCH1B, P1B; SUFU1A, S1A; and SUFU1B, S1B genes.

6 M64 displays a homozygous deletion of the NF2 gene.

7 M11 demonstrates multiple deletions not affecting the NF2 gene.

8 M34 displays a terminal deletion not affecting the NF2 locus.

9 M22 presents two mutations in the NF2 locus, in exon 13 and in the CNG inter1. The position of the inter1 mutation is according to numbering on chromosome 22 (UCSC Homo sapiens Genome Browser, assembly July 2003). Ten bases on each side of the mutation: agcaaagagaCaaattaccca.

10 Methylation of a single CpG site in the intron 1 of the NF2 gene in ~40% of DNA molecules.

All samples displaying aberrations other than monosomy 22 are graphically summarized in Fig 1A. This analysis enables the discrimination of several minimal overlapping deleted regions which should be studied further. Two of these regions, named as R1 (~550 kb) and R2 (~250 kb) are small enough to allow for detailed analysis of a limited number of candidate genes (Figs. 1B and 1C). The breakpoint cluster region (BCR) gene is one interesting candidate located in R1. This gene is located at the site of the chromosome 22 breakpoint generating the Philadelphia chromosome (9;22) translocation that creates the BCR/ABL fusion gene found in chronic myeloid leukemia and acute myeloid leukemia. The normal BCR gene was reported to have a serine/threonine kinase activity [40]. Another candidate gene located within R1 is the RGR oncogene. This gene was originally identified as a fusion protein which can promote tumor development in nude mice, but has also displayed oncogenic properties when not in the fusion form [41]. The RGR gene has frequently been found to be altered in human T-cell malignancies [42]. The zinc finger protein encoding gene ZCWCC1, located within R2, may also be a candidate for further investigation. The relative weakness of the R1 and R2 regions is that they are defined by rare cases with small deletions. There is therefore a possibility that the latter chromosome 22 aberrations are neutral "passenger mutations", with little consequence to tumorigenesis. Yet another apparent, large candidate region is located towards the telomere of 22q, starting from clone ID 390. This segment is affected by hemizygous deletion in all but one (M114) of the studied tumors which disclosed gene copy number imbalances of chromosome 22.

The third major category of tumors (51 cases, 40%) was composed of samples that did not display detectable changes in gene dosage on 22q (Table 1, "N o A berration D etected"; NAD) despite the high average resolution (~75 kb per data point). There may be several possible explanations for these results. One possibility is that additional genes on 22q (other than NF2 tumor suppressor) are important for meningioma development and these are mutated by alternative mechanisms, such as point mutation or epigenetic silencing. Another explanation might be that this subset of samples involves an alternative tumorigenic pthway which does not involve chromosome 22-located genes.

The current version of the chromosome 22 array has been supplemented with additional non- 22q-derived genomic clones from other chromosomes, which encompass genes (e.g. PTCH2A, PTCH1B, SUFU and DAL1 genes) of potential interest in the fields of meningiomas and brain tumors in general. The complete list of these additional clones is provided in additional file 1. Table 1 shows the samples in which loss of DNA copy number was detected for some of these loci. The highest rate of loss detected was at the PTCH2A locus on 1p34.1, which displayed a rate of loss of 18%. The other loci, namely DAL-1, PTCH1B, SUFU1A and SUFU1B, displayed a level of DNA copy number loss of 6%, 5%, 6% and 6%, respectively. We also detected DNA copy number losses in a number of non-chromosome 22-derived control clones. The most frequently deleted control clones were those derived from chromosomes 1 and 3, suggesting the likely importance of other regions of the genome in meningioma development/progression.

Mutation analysis of NF2 exons and evolutionarily conserved non-genic sequences (CNGs)

Inactivating micro-mutations in the NF2 tumor suppressor gene are found in meningiomas of all histopathological subtypes and malignancy grades and are thought to be an early event in tumorigenesis [43]. However, large differences (20–60%) of mutation frequency in sporadic meningiomas were previously reported [7, 13–18]. Our rationale was to reassess the mutation frequency of NF2, especially when combined with very high resolution array-CGH analysis of these tumors. The rationale was also to test the hypothesis that mutations might be located in evolutionarily well conserved non-genic sequences within introns or in the 5' promoter region of the NF2 gene, in addition to mutations in the exons. Comparisons of the orthologous genomic sequence from the NF2 locus derived from five species (human, baboon, mouse, rat and pufferfish) were previously performed in order to identify CNGs that might be of functional importance for the NF2 gene [35]. The exonic sequence of the NF2 gene constitutes ~2.7% of this locus. When applying a threshold of 70% sequence identity over >100 bp of gap-free alignment in comparisons between human and mouse, the resulting level of extra-exonic conservation was 3.8%. We selected CNGs for mutational analysis that displayed a similar level of evolutionarily conservation as the protein coding sequence. In total, 17.5 kb was sequenced in 100 tumors, which included all 17 exons, the highly conserved exon flanking sequences and seven CNGs not contiguous with exons (Figure 4A). Four of the CNGs were positioned upstream of exon 1 and the remaining three CNGs were located in introns 1, 4 and 10.

In case M22 we identified two tumor-specific micro-mutations, one mutation causing a frame shift in exon 13 introducing a premature stop codon and the second being a mutation in the inter1 CNG element. Inter1 is situated approximately 15 kb upstream of the NF2 gene (Figure 4A). It displays sequence conservation between human and pufferfish, which are evolutionarily very distant species [35], suggesting that inter1 may possess functional properties. Details on the exact position of the inter1 mutation on chromosome the 22 sequence are shown in Table 1. The tumor-specific mutation in inter1 CNG is one of three changes present in M22, as this tumor also contains monosomy 22. Thus, we may speculate that three hits are affecting the NF2 gene. An alternative explanation is that one of these mutations has little consequence to the normal function of the NF2 gene.

Methylation profiling of the NF2 promoter region by bisulfite modification followed by pyrosequencing

Methylation of promoter associated CpG-islands is known as a possible mechanism for inactivation of tumor suppressor genes in a variety of human cancers. It has also been suggested as a mechanism of repressing transcription of the NF2 gene in schwannomas [50] and in meningiomas [21]. Methylation of other cancer related genes (GSTP1, THBS1, CDKN2A, CDH1, BRCA1, p14 ^ARF and RB1) has also been reported to be associated with malignant progression of meningiomas [20, 51, 52]. We used an established method of bisulfite treatment of tumor DNA followed by Pyrosequencing (Biotage, Uppsala, Sweden) which allows detection and semi-quantitative estimation of the degree of methylation at the selected CpGs [53, 54]. Two overlapping PCR products of 339 bp and 550 bp covered ~750 bp of the upstream promoter region, the first exon and approximately 70 bp of intron 1, immediately downstream of the first exon (Figure 4B). These two products were Pyrosequenced with 8 sequencing primers. In total, 40 CpG sites were sequenced, between positions -591 and +166 bp (position +1 refers to transcriptional start site). This analysis allowed interrogation of up to 11 CpG sites per assay, distributed over 79 bp and with a maximum of 60 dispensations of nucleotides (additional file 2). Methylation was identified in only one tumor sample (M104), at a single CpG site at position +120, in ~40% of DNA molecules in this tumor (Figure 4B). Kino et al. reported a 70 bp region (-591 to -522) essential for NF2 expression in schwannomas [50]. Three CpG sites within this region (at positions -591, -586 and -581) appeared to be of particular importance for silencing of the NF2 gene upon methylation in schwannomas. We paid particular attention to these CpG sites but no methylation was detected. In summary, upon comprehensive analysis of 40 CpG sites, we conclude that methylation does not play a major role in the silencing of the NF2 gene in meningiomas. These results are in accordance with a report by van Tilborg et al [20] although contradicting the results from Lomas et al. [21].

Summary of macro- and micro-mutations on chromosomes 22 and 1 in meningiomas

We observed pronounced differences of the NF2 gene mutation frequency between different histopathological subtypes of meningiomas (Table 3). There was a clear trend towards an increased frequency of the biallelic NF2 gene inactivation in meningothelial, transitional and fibroblastic subtypes which were 6 out of 33 tumors (18%), 12 out of 31 tumors (39%) and 11 out of 21 tumors (52%), respectively. Based on these results, we may speculate that the biallelic inactivation of the NF2 gene is not as important for the development of the meningothelial subtype of meningioma, when compared to the transitional and fibroblastic tumors. A related trend was even more pronounced, when the frequencies of large chromosome 22 deletions were correlated to the three above mentioned histopathological subtypes of meningioma. The presence of large hemizygous deletions showed marked differences between the meningothelial, transitional and fibroblastic subtypes, comprising 13 out of 33 cases (39%), 20 out of 31 cases (65%) and 19 out of 21 cases (90%), respectively. It should be stressed here that 15 out of 19 fibroblastic meningiomas, which displayed gene dosage imbalances on 22q, were monosomy 22 cases. The three remaining tumors (M64, M32, and M114) all showed a minimum loss of 18 Mb of 22q. It may thus be hypothesized that one copy loss of genes located within the 18 Mb on 22q is important for generation of the fibroblastic phenotype of cells.

More than half (61%) of the meningothelial cases were among the category of tumors with both chromosome 22 copies retained. One possible explanation for this finding is that meningothelial tumors more frequently harbor mutations in additional genes that are located outside chromosome 22. Our results correlate well with previous studies by Wellenreuther et al. [15] as well as with Evans and colleagues [17] that identified NF2 gene mutations in 4/26 (15%) and 1/20 (5%) of meningothelial tumors. The above results from DNA level analysis are supported by analysis of merlin/schwannomin protein expression, which show normal expression in the greater part of meningothelial tumors, but severely reduced expression levels in the majority of fibrous and transitional meningiomas [55–57]. Furthermore, correlation between tumor localization and histological subtype was previously reported. The majority of meningiomas with no chromosome 22 aberrations were localized in the anterior skull base and were predominantly meningothelial [19]. Taken together, these results might suggest that cellular origin plays a role in the genetic pathways which are operative during tumor formation of meningiomas. In the context of our results from chromosome 22, it is interesting to correlate this to the data obtained from chromosome 1, as genetic aberrations affecting these two autosomes are the most frequent genetic abnormalities in meningiomas [31, 36]. Furthermore, aberrations of chromosome 1 are also related to recurrence-free survival from the disease [31]. We have recently completed a high resolution analysis of chromosome 1 using the same tumor series [34] and 49 of the samples were studied with similar resolution on both chromosomes (Table 1). There is a strong relationship between changes on both chromosomes, with 67% of samples showing concurrent aberrations. These results clearly indicate the need for a genome-wide and high-resolution array-based characterization of chromosomal aberrations in meningiomas.

Patient samples

Tumor tissue and matched samples of peripheral blood were collected and DNA was isolated using standard methods [59]. Samples were obtained from the Dept. of Neurosurgery, Karolinska Hospital, Stockholm, Sweden. All analyzed tumor samples were reported previously in various studies [7, 22, 23, 31, 34, 36]. Ethical approval was granted from the Research Ethics Committee, Faculty of Medicine, Uppsala University. Copy number profiling of chromosome 22 was performed on 126 sporadic meningiomas and this sample set comprised 39 meningothelial, 38 transitional, 28 fibroblastic, 9 anaplastic, 6 psammomatous, 1 angiomatous and 5 tumors of unclassified histopathological subtype. For the analysis for point mutations and methylation in the NF2 gene promoter, we used a subset of 100 tumor DNA samples (Table 1 and additional file 3).

DNA copy number profiling using the chromosome 22 genomic microarray

The genomic microarray used to assess DNA copy number status of chromosome 22 has been described previously [37]. In brief, 480 genomic clones derived from the minimal tiling path of chromosome 22, 8 control clones from chromosome X and 30 non-chromosome 22 autosomal control clones were isolated and grown in bacterial cultures overnight. The clone DNA (cosmid, fosmid, BAC and PAC) was purified using Qiagen kits and quantified using the picoGreen assay (Molecular Probes, Inc., Eugene, OR). All clone DNA were verified using STS PCR. The prepared clone DNA stock was amplified by DOP-PCR [60]. A second round of PCR on the DOP-PCR products using an amino linked universal primer was performed and amplification products were printed in quadruplicate on CodeLink™ slides (GE Healthcare, UK). Protocols used for the hybridization and washing procedures have been previously described [37, 61]. Image capture and analysis was performed using the axon 4000B Scanner and GenePix 6 image analysis software. For all clones on the array the average, standard deviation and coefficient of variation (CV) for each of the 4 replicas was calculated. Clones displaying a CV above 5% between a minimum of 3 replica spots were discarded from further analysis. Array-CGH profiles are plotted in Microsoft Excel with the clone IDs on the X-axis and the normalized fluorescent ratio intensities on the y-axis. The average fluorescent ratio value for a given set of continuous clones on the array is referred to as the A verage N ormalized I nter L ocus F luorescence R atio or ANILFR value.

PCR and sequencing

Exons and conserved non-genic sequences (CNGs) in the NF2 locus were PCR amplified from genomic DNA derived from one hundred sporadic meningiomas, with customized primers using Ampli Taq Gold (Applied Biosystems) (primer list in Additional file 4). Peripheral blood DNA from phenotypically normal control persons was also PCR amplified and sequenced for CNGs in which mutations were observed. PCR fragments were purified by Exosapit (GE Healthcare, UK) digestion prior to sequencing. Sequencing of amplicons was performed with customized nested primers using ET Terminator (GE Healthcare, UK) and Big Dye v.3.1 (Applied Biosystems) sequencing chemistries, and ethanol precipitated according to standard protocol [59]. Electrophoresis and fluorescent detection was carried out on Megabace 1000 (GE Healthcare, UK) and ABI 3700 (Applied Biosystems) capillary sequencing machines. Processing and analysis of sequencing data was performed using the Staden package [62].

Methylation profiling by pyrosequencing of bisulfite treated DNA

Pyrosequencing (Biotage, Uppsala, Sweden) was used as method for detection of methylation of cytosines situated 5'of a guanine [53]. Briefly, the DNA was bisulfite treated using EZ-96 DNA methylation kit (Zymo research, Orange, CA, USA) for conversion of unmethylated cytosines into uracils. In the subsequent PCR uracil amplifies as thymidine. Methylated cytosines remain unchanged after treatment. Cytosines not followed by a guanine are not methylated and will therefore be converted to a thymidine after bisulfite treatment and PCR. These thymidines were used as internal controls for the completeness of bisulfite treatment. Human genomic DNA was methylated in vitro by the CpG methylase M. Sss I and used as positive controls for detection of methylation. Bisulfite treated DNA were amplified in a 50 μl reaction using Platinum Taq DNA polymerase (Invitrogen Life Technologies, Carlsbad, CA, USA) according to manufacturers protocol. Primers for amplification and sequencing were custom designed using PSQ assay design software (Biotage, Uppsala, Sweden). Data of primers used for amplification and sequencing, as well as the nucleotide sequence analyzed and the dispensation order of nucleotides during sequencing is shown in Additional file 2. One PCR primer in each pair was biotinylated for conversion of the PCR product to a single stranded template. The cycling condition used were: 94°C for 3 minutes, followed by 20 cycles of 94°C for 20 seconds, 65°C minus 0.5°C/cycle for 20 seconds, 72°C for 1 minute, followed by 40 cycles of 94°C for 20 seconds, 55°C for 20 seconds, 72°C for 1 minute and finally 72°C for 10 minutes. 45 μl of PCR product was used for the Pyrosequencing reaction. Sequencing was performed on a PSQ 96MA system with PSQ Gold reagent kit (Biotage, Uppsala, Sweden) according to manufacturer's protocol. Pyro Q-CpG software (Biotage, Uppsala, Sweden) was used for data analysis.