Introduction

Oligodendroglial tumors including oligodendrogliomas and oligoastrocytomas commonly exhibit joint allelic losses of chromosomal arms 1p and 19q. Just recently, strong candidates for the putative tumor suppressor genes in these regions have been proposed. Mutations in CIC mapping to chromosome 19q have been found in oligodendrogliomas [3, 14]. Likewise, mutations in FUBP1 mapping to 1p have been detected in oligodendrogliomas, but only in one of these studies [3]. Oligoastrocytomas so far have not been examined thoroughly for CIC and FUBP1 mutations and the sparse data currently available did not reveal alterations in these genes [14].

Human CIC is the homolog of the capicua gene in Drosophila. Capicua acts as downstream transcriptional repressor of receptor tyrosine kinase pathways mediating extracellular signals to gene regulation. Capicua regulation is achieved by phosphorylation through MAP kinases [6]. The role of murine and human CIC has not been investigated in depth but it is involved in the development of granule cells. However, essential functional domains, such as DNA binding domains and critical phosphorylation sites, are conserved and functional. Thus, CIC is mainly expressed in undifferentiated tissue, pointing to a possible role in development [7]. FUBP1 encodes a DNA binding protein shown to be involved in c-myc regulation [4].

Oligodendrogliomas with classical morphological features as well as diffuse astrocytomas are well-defined entities [9]. In contrast, oligoastrocytoma frequently poses diagnostic problems caused by the less stringent definition of sharing both, oligodendroglial and astrocytic morphology and by the problem of representative surgical sampling. This may result in considerable diagnostic variation among different observers. Therefore, the analysis of molecular genetic features of oligoastrocytomas has been hoped for solving classification problems. While combined 1p/19q losses are typical for oligodendrogliomas and TP53 mutations are characteristic for diffuse astrocytomas, oligoastrocytomas unfortunately appear to exhibit both of these alterations [9]. IDH mutations are not helpful in distinguishing these entities due to a very high-mutation rate in all, diffuse astrocytomas, oligodendrogliomas and oligoastrocytomas [1]. Knowledge of the involvement of the recently detected CIC and FUBP1 mutations may help to further clarify the classification of these tumors.

To resolve the current discrepancy regarding CIC mutations in oligodendrogliomas and oligoastrocytomas and to test the potential role of CIC in embryonal human tumors, we analyzed the entire coding sequence of CIC in a series of primary human brain tumors. We went on performing FUBP1 analysis in all oligodendrogliomas, oligoastrocytomas and astrocytomas.

Materials and methods

Tumor specimens average patient age and sex ratio

We analyzed DNA extracted from fresh frozen material of 98 human brain tumors retrieved from the archive of the Department of Neuropathology, Institute of Pathology, Heidelberg. All tumors were diagnosed and classified according to the WHO classification of tumors of the nervous system [9]. The series consisted of 9 oligodendrogliomas WHO grade II (OII), 9 anaplastic oligodendrogliomas WHO grade III (OIII), 16 oligoastrocytomas WHO grade II (OAII), 26 anaplastic oligoastrocytomas (OAIII), 3 diffuse astrocytomas WHO grade II (AII), 7 anaplastic astrocytomas WHO grade III (AIII), 16 glioblastomas WHO grade IV (GBM), including 2 GBM of the giant cell variant as well as 12 medulloblastomas WHO grade IV (MBIV). Oligodendrogliomas were included as a verification set to compare the present study with previously published data. The series of diffuse astrocytomas and GBM was biased for the presence of allelic loss on 19q with some of the tumors also exhibiting additional loss of 1p.

Direct sequencing of CIC FUBP1, IDH1 and IDH2

Fragments spanning all 20 exons of CIC were amplified using 20 ng each of the respective forward and reverse primer. Primer design was based on accession number NM_015125.3. Also, fragments spanning all 20 exons of FUBP1 were amplified using 20 ng each of the respective forward and reverse primer. Primer design was based on accession number NM_003902.3. Primer sequences are given as supplemental data.

For PCR, 100 ng of DNA and HotStar 2× PCR Master Mix (Qiagen, Hilden, Germany) was employed. PCR was performed in a total volume of 30 μl, and included initial denaturation at 95 °C for 180 s, followed by 35 cycles with denaturation at 95 °C for 30 s, annealing at 56 °C for 25 s and extension at 72° C for 40 s. Two microliters of the amplification product were submitted to bidirectional sequencing using the BigDye Terminator v3.1 Sequencing Kit (Applied Biosystems, Foster City, CA, USA). Sequences were determined using a ABI 3500 Genetic Analyzer (Applied Biosystems) and the Sequence Pilot version 3.1 (JSI-Medisys, Kippenheim, Germany) software.

IDH1 exon 4 encompassing codon 132 and IDH2 exon 4 encompassing codon 172 were subjected to analysis by direct sequencing using an ABI 3500 Genetic Analyzer as previously described [5].

Microsatellite analysis and multiplex ligation-dependent probe amplification (MLPA)

All samples were analyzed for loss of heterozygosity (LOH) of 1p and 19q employing at least three polymorphic microsatellite markers for each chromosomal arm as previously described [2].

MLPA was employed for detection of chromosomal losses on 1p and 19q in all oligodendrogliomas and oligoastrocytomas testing 15 loci on 1p and 8 loci on 19q. Analysis was performed using a commercially available kit (Salsa MLPA, P088, MRC Holland, Amsterdam, The Netherlands). The capability of MLPA to detect partial deletions has been demonstrated previously [13]. PCR products were separated and quantified on an ABI 3500 Genetic Analyzer (Applied Biosystems). Results were analyzed by the MLPA module in the Sequence Pilot 3.1 Software (JSI medical systems). Chromosomal regions were scored as under- or overrepresented if 2 or more probes on 1p or 19q adjacent to each other exhibited a ratio smaller than 0.7 or above 1.3, respectively [13].

Immunofluorescence and immunohistochemistry

Sections from 2 archival cases with known 1p/19q-, IDH- and CIC-status were cut to 3 μm, deparaffinized and incubated in Ventana cell conditioner 1 (Ventana Medical Solutions, Tucson, USA) for 30 min followed by 5 % BSA blocking solution. Primary antibodies against CIC protein, recognizing an epitope spanning residues 1,150–1,200 (Abcam, Cambridge, UK; 1:50), and H09, binding mutant IDH1R132H protein (Dianova, Hamburg, Germany; 1:100) were applied for 2 h at room temperature. Secondary antibodies were anti-rabbit AlexaFluor 488 (Invitrogen, Carlsbad, USA; 1:500) and anti-mouse AlexaFluor 568, incubated for 30 min at room temperature. Slides were mounted with Vectashield Hardening Mounting medium, containing DAPI (Vector Labs, Burlingame, USA).

For immunohistochemistry, sections were processed on a Ventana BenchMark XT® immunostainer (Ventana Medical Systems, Tucson, AZ, USA). The Ventana staining procedure included pretreatment with cell conditioner 1 (pH 8) for 60 min, followed by incubation with anti-CIC (1:150) antibody at 37 °C for 32 min. Incubation was followed by Ventana standard signal amplification, UltraWash, counterstaining with one drop of hematoxylin for 4 min and one drop of bluing reagent for 4 min. For visualization, ultraView™Universal DAB Detection Kit was used.

Results

CIC mutations

For each tumor, data on the status of the IDH genes and on allelic losses on chromosomal arms 1p and 19q are compiled in Table 1.

Table 1 Diagnosis, CIC, FUBP1, 1p/19q and IDH1/2 status
Full size table

CIC mutations were detected in 15/18 oligodendrogliomas encompassing 7/9 in OII and 8/9 OIII and in 14/42 oligoastrocytomas encompassing 6/16 in OAII and 8/26 OAIII. In diffuse astrocytomas preselected for 19q deletion, CIC mutations were detected in 3/10 diffuse astrocytomas encompassing 2/3 in AII and 1/7 in AIII. Of note, all CIC mutant astrocytomas presented with combined LOH 1p/19q. No CIC mutations were detected in the set of 16 glioblastomas and 12 medulloblastomas. The type of CIC mutations is given in Table 2.

Table 2 Characterization of CIC and FUBP1 mutations
Full size table

The frequency of CIC mutations in oligodendrogliomas (83 %) was higher than in oligoastrocytomas (33 %) (p < 0.002; Fisher’s Exact test). While in oligodendrogliomas all patients (100 %) with LOH 1p/19q carried a CIC mutation, only 13/25 (52 %) oligoastrocytoma patients with LOH 1p/19q carried a CIC mutation (p < 0.002; Fisher’s Exact test). A single patient with OAII exhibited a CIC mutation but neither allelic losses on 1p nor 19q.

As previously described [3, 14], CIC mutations occurred throughout the coding sequence with an increased frequency in exon 5 and approximately similar numbers for missense and nonsense mutations.

FUBP1 mutations

DNA for FUBP1 analysis was available from 61 tumors (Table 1). We detected FUBP1 mutations in 7/61 tumors including 2 OII, 1 OIII, 2 OAII, 1 OAIII and in 1 AII. The type of FUBP1 mutations is given in Table 2.

FUBP1 mutations appeared to cluster in the FUBP1 DNA binding site spanning exons 5–14. In contrast to CIC, the predominant types of alterations in FUBP1 were frameshift and nonsense mutations. Repetitive Poly-A sequences impaired the selection of optimal primers for exon 6 resulting in lack of data for nucleotide positions 397–412.

Allelic loss of 1p and 19q

Due to the limited number of markers analyzed by conventional microsatellite analysis, we could not safely conclude the extent of the deletions. Therefore, all oligodendrogliomas and oligoastrocytomas were also tested by MLPA. All oligodendrogliomas and oligoastrocytomas with losses exhibited deletion of the entire chromosomal arms.

Age and sex distribution

Mean age for the patient groups with CIC and FUBP1 mutations was: OII 40 years, OIII 45 years, OAII 46 years, OAIII 42 years, AII 39 years, AIII 38 years (one patient). No significant differences were seen between patients with and without mutations, or within the groups of oligodendrogliomas (Student’s t test p > 0.24), oligoastrocytomas (Student’s t test p > 0.4), grade II tumors (Student’s t test p > 0.09), grade III tumors (Student’s t test p > 0.14), OII (Student’s t test p > 0.45) and AII (Student’s t test p > 0.33). Presence of only a single mutant or wild-type case in AIII and OIII, respectively, did not allow for statistical analysis. No significant differences in mutation rates were seen between male and female patients (Student’s t test p > 0.07).

Immunofluorescence and immunohistochemistry

Two tumors with 1p/19q co-deletion and IDH1R132H mutation were analyzed. The OII (case 7) harbored a missense and the OIII (case 10) a truncating nonsense mutation in CIC (Fig. 1). Immunofluorescent double labeling for mutant IDH1R132H and CIC protein yielded a signal for IDH1R132H protein in the tumor cells of both cases. CIC in tumor cells was only detected in the case with missense mutation. Subsequently, we performed immunohistochemistry on both tumors. Staining for CIC was observed in endothelial cells and all tumor cells of the OII with missense mutation (case 7), but only in very few tumor cells in the OIII with truncating mutation (case 10).

Fig. 1
figure 1

Immunofluorescence for CIC (green) and mutant IDH1R132H protein (red) in the tumor infiltration area of two 1p/19q co-deleted, IDH1R132H mutant tumors. CIC nonsense mutation and 19q loss resulted in abrogation of CIC protein and mutually exclusive expression of tumor-specific mutated IDH1R132H (clone H09) and CIC protein. CIC expression is detectable in resident brain tissue (a–d). CIC missense mutation allows the detection of wild type and mutated CIC protein. CIC is seen in both, resident brain tissue and tumor cells characterized by expression of mutant IDH1R132H protein (e–h)

Full size image

Discussion

CIC and FUBP1 mutations in oligodendroglioma have recently been described. The present study investigated a validation set of oligodendrogliomas, a set of oligoastrocytomas, and a set of astrocytomas and glioblastomas biased for allelic losses on chromosomal arms 19q, as well as a set of medulloblastomas.

Validation set of oligodendrogliomas

Data from our validation set match with previous observations. CIC mutations were initially observed by deep sequencing in 18/34 (53 %) [3] and 20/29 (69 %) [14] oligodendrogliomas. By direct sequencing, we detected CIC mutations in 15/18 (83 %), thus confirming a high incidence in oligodendrogliomas and sensitivity of the methods applied. So far, CIC mutations have been tightly linked to the presence of combined allelic losses on 1p and 19q. This holds also true for our oligodendroglioma series with 15/18 (83 %) tumors exhibiting combined loss and 15/15 (100 %) of those tumors with combined loss also carrying a CIC mutation.

FUBP1 mutations were detected only in one of the two previous studies in 5/34 (15 %) [3] cases, while the other study did not find FUBP1 mutations in 16 tumors [14]. We detected FUBP1 mutations in 3/15 oligodendrogliomas with sufficient DNA available. We thus also confirm the incidence of FUBP1 mutations in oligodendrogliomas comparable to that published previously [3].

Oligoastrocytomas

Oligoastrocytomas in our series exhibited CIC mutations in 14/42 (33 %) tumors. This contrasts previous findings not detecting CIC mutations in oligoastrocytomas. However, the previous study [14] analyzed by deep sequencing only 14 oligoastrocytomas without 1p/19q co-deletion, and all but one of our oligoastrocytomas with CIC mutation exhibited 1p/19q co-deletion. That study also examined an extension set containing 11 oligoastrocytomas, 3 of which harbored 1p/19q co-deletion, without detecting CIC mutations in exons 5 and 20 [14].

FUBP1 mutations have not previously been analyzed in these tumors and we detected FUBP1 mutations in 3/13 (23 %) oligoastrocytomas with CIC mutation and 3/37 (8 %) of all oligoastrocytomas. We thus established the presence of FUBP1 mutations in oligoastrocytomas.

The incidence of 1p/19q co-deletions in our oligoastrocytoma series was 25/42 (60 %) and thus lower than that in oligodendrogliomas with 15/18 (83 %). In both sets, CIC mutations with one exception were seen in 1p/19q co-deleted cases. Interestingly, the incidence of CIC mutations in oligodendrogliomas with 1p/19q co-deletion (15/15 cases; 100 %) was significantly higher (p < 0.002, Fisher’s Exact test) than in oligoastrocytomas (13/25 cases; 52 %). Noteworthy, all oligodendrogliomas and oligoastrocytomas have been analyzed by MLPA and there was no difference in extent of deletion. This may point toward heterogeneity among 1p/19q co-deleted oligoastrocytomas and argues against reclassifying all oligoastrocytomas with 1p/19q co-deletions as oligodendrogliomas.

Astrocytomas, glioblastomas and medulloblastomas

In 10 astrocytomas selected for the presence of 19q deletions, we detected 3 CIC mutations among the 5 cases with a 1p/19q co-deletion. On the other hand, none of the 16 GBM including 5 cases with 1p and 19q deletion contained a CIC mutation. This most likely reflects the high-genomic instability of GBM with random occurrence of both, deletions on 1p and 19q. This interpretation finds support in the nearly invariant association of 1p/19q co-deletion with IDH mutations in the tumors with CIC mutations and in the absence of IDH mutations in all GBM samples harboring deletions on 1p and 19q (Table 1). The detection of CIC mutation and 1p/19 co-deletion in 3 diffuse astrocytomas may indicate mis-classification of these tumors and also may hint toward the limitations of routine assessment for diagnosis. We point out that due to the preselected nature of the tumors included, our numbers for CIC mutations are not representative for all diffuse astrocytomas. A single FUBP1 mutation was detected in 1 AII also carrying 1p/19q co-deletion and IDH1 mutation.

The important role of CIC in granule cell development [7] and the report of a single CIC mutation in medulloblastomas [3] prompted us to screen 12 medullobastomas for the presence of CIC mutations. We did not detect mutations and this has also been shown by sequence analyses of all coding genes of medulloblastoma failing to detect CIC mutations [11]. In addition, high expression of CIC protein has been described in medulloblastomas further arguing against a prominent role of inactivating CIC mutations in these tumors [8].

Mutation types of CIC and FUBP1

Among 32 CIC mutations in our series, 16 presumably resulted in punctual protein alterations with 14 being of missense type and 2 representing in frame deletions of 1 amino acid each. The remaining 16 mutations were expected to result in severe alterations on the protein level with 12 constituting frameshift mutations, 2 nonsense mutations, 1 combined missense and frameshift mutation and 1 splice site mutation. The majority of mutations were located at the DNA-binding HMG-boxes, both in previous [3, 14] and our data sets.

All seven FUBP1 mutations represented mutations causing severe protein alterations by producing frameshift mutations in five and nonsense mutations in two instances. The mutational spectrum for CIC [3, 14] and FUBP1 [3] is comparable to that previously described. The majority of FUBP1 mutations localized to the DNA binding domain spanning exons 5–14. The truncating nature of FUBP1 mutations indicates a loss of function and contrasts the potential role of FUBP1 in oligodendroglial tumors to that in liver cancer, where FUBP1 overexpression is linked to tumor growth and migration [10, 12].

The loss of one parental gene copy in combination with the presence of considerable numbers of truncating mutations in both, CIC and FUBP1 raises the possibility to detect complete protein loss with antibody-based detection systems. To this end, we applied double immunofluorescence with antibodies directed against the IDH1R132H mutated protein and against a C-terminal domain of CIC protein to two tumors, one with a truncation and one with a missense mutation. In the tumor with the truncating CIC mutation, mutually exclusive expression of CIC and IDH1R132H protein could be demonstrated while the tumor with missense mutations exhibited co-expression of CIC and IDH1R132H protein (Fig. 1).

In conclusion, we confirm the presence of CIC and FUBP1 mutations in oligodendrogliomas, provide evidence for such mutations in oligoastrocytomas and a small fraction of astrocytomas and point out the heterogeneous mutation pattern within the group of 1p/19q co-deleted oligoastrocytomas.