Fibroblast Growth Factor Receptor-1 α-Exon Exclusion and Polypyrimidine Tract-binding Protein in Glioblastoma Multiforme Tumors

Introduction

Alternative recognition of the α exon during processing of FGFR-1³ RNA produces receptor forms that vary in their affinity for fibroblast growth factor (1 , 2) . Because FGFR-1 plays a primary role in many cell growth and differentiation pathways, precise regulation of its RNA splicing is critical. However, normal recognition of the α exon is altered during the malignant progression of glial cells, producing a receptor form lacking the α exon and with enhanced affinity for fibroblast growth factor (3 , 4) . Expression of this form of FGFR-1 in glial cells is believed to provide a cell-growth advantage and possibly to contribute to glial cell malignancy (3) . Using a cell culture model, we previously identified two intronic RNA sequences flanking the α exon, ISS-1 and ISS-2, that are required for glioblastoma cell-specific FGFR-1 RNA splicing (5 , 6) . Deletion or mutation of either of these elements reverses the splicing phenotype observed in glioblastoma cells so that the FGFR-1 mRNA includes the α exon. In this study, we found that the trans-acting factor PTB specifically bound to the upstream element ISS-1 and was overexpressed in patient glioblastomas, suggesting that PTB may regulate glioblastoma-specific FGFR-1 RNA splicing.

Materials and Methods

RNA Isolation and RT-PCR.

The transfection of cell lines, RNA isolation, and RT-PCR analysis were performed as described previously (7) . Total RNA was isolated from ∼100 mg of normal and tumor tissue by sonication in Catrimox-14 (Qiagen, Chatworth, CA) as described previously (8) . Because of the presence of nonspecific amplification bands, the RT-PCR protocol used to amplify tissue-derived RNA was modified from a previously described procedure to include two amplification steps (9) . Briefly, reverse transcription was performed with the FGFR-1-specific primer Endo-R using 5 μg of total RNA in a 20-μl reaction volume. A first round of 11 cycles of PCR (1 min at 94°C, 1 min at 55°C, and 2 min at 72°C) was performed with 10 μl of the cDNA and the primers Endo-F and Endo-R (9) in a final volume of 50 μl. This was followed by a second round of 17 cycles of PCR (1 min at 94°C, 1 min at 66°C, and 2 min at 72°C) with 0.1 μl of the first-round PCR mixture and primers FP183 (5′-CTTCTGGGCTGTGCTGGTCA-3′) and a mixture of unlabeled plus 0.08 pmol of ³²P end-labeled FP184 primer (5′-TCTTTTCTGGGGATGTCCAA-3′). A single pair of RNA samples (Fig. 3 ⇓ , Lanes 5 and 6) failed to amplify under these conditions and required 1 μl of the first-round PCR mixture. These RT-PCR conditions were found to be within the linear amplification range for RNA isolated from SNB-19 and JEG-3 cell lines (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Western analysis of PTB expression in human brain samples (top two panels). Total cellular protein was prepared from tumor (T) and adjacent normal (N) tissue obtained from patients diagnosed with low-grade astrocytoma (LGA), anaplastic astrocytoma (AA), or glioblastoma multiforme (GBM). One patient with a prior history of oligodendroglioma (Rec), had radionecrosis with a lesion mimicking a tumor (R). Western analysis was performed as described in “Materials and Methods” using ∼50μ g of protein in each lane. Actin antibody was included with the PTB antibody during the primary incubation to provide an additional control for protein sample integrity and loading. RT-PCR analysis of FGFR-1 transcripts was performed on total RNA isolated from the tissue pairs in the middle panel as described in “Materials and Methods” (bottom panel). Arrows indicate the α-exon inclusion (I) and exclusion (E) products.

Results and Discussion

Failure to include the α exon during processing of the FGFR-1 primary transcript has been correlated with glial cell malignancy (3) . Our DNA sequence analysis of the FGFR-1 gene and studies of cell-culture models of α exon splicing suggested that changes in the expression of trans-acting factors cause this aberrant RNA splicing (9) . We previously identified a series of RNA sequence elements that are required to maintain regulated FGFR-1 splicing in a cell culture model system that mimics the patterns observed in normal brain glial cells and glioblastomas (9) . The reporter construct (pFGFR-17) contains a 3.8-kb fragment of the FGFR-1 genomic sequence inserted into intron 1 of the human metallothionein IIA gene (Fig. 1A) ⇓ . In RNA transcripts derived from pFGFR-17, the α exon was included in a cell-specific manner: the exon was predominantly excluded (71%) in the glioblastoma cell line SNB-19, but predominantly included in JEG-3 cells (83%; Fig. 1B ⇓ ). Two intronic sequences flanking the α exon, ISS-1 and ISS-2, inhibit α-exon inclusion in SNB-19 cells (5 , 6) . The ISS-1 element was mapped by the deletion construct, pFGFR-D1, to a 40-nt sequence 241 nt upstream of the α exon. The deletion of this short sequence had a dramatic effect on exon inclusion in SNB-19 cells, increasing it from 29% to 70%, with little effect on exon inclusion in JEG-3 cells (83% versus 86%; Fig. 1B ⇓ ). Within this 40-nt region are two YUGCYYYY elements that resemble the regulatory sequences found to inhibit exon recognition in the α-tropomyosin gene (15 , 16) . These sequences were mutated to determine whether they played a similar role in α-exon recognition. Mutation of the first element to create pFGFR-M2 increased α-exon inclusion to a level similar to that of pFGFR-D1, whereas mutation of the second element (pFGFR-M4) only partially increased inclusion. Both mutations had only a small effect on splicing in JEG-3 cells (Fig. 1B) ⇓ .

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Inhibition of α-exon inclusion in SNB-19 cells was mediated through the ISS-1 element. A, the pFGFR-17 splicing construct and the cell-specific RNA splicing pathways. The FGFR-1 genomic sequence was inserted into intron 1 of the human metallothionein IIA gene. The numbers indicate the lengths of the exons and introns in bp. The positions of the oligonucleotide primers and the sizes of the RT-PCR products are indicated. RSV, Rous sarcoma virus promoter. B, examination of α-exon inclusion in RNA isolated from transfected cells. The specific deletions (−) and mutations (uppercase letters) in the constructs containing modifications of the ISS-1 element are depicted on the right. A representative autoradiograph is shown on the left; arrows indicate the RT-PCR products that include (I) or exclude (E) the α exon. The percentages of α-exon inclusion ± the SD were derived from phosphor counts from three independent transfections.

To identify factors associated with ISS-1, we performed UV cross-linking experiments. The RNA probes used all contain the core 40-nt region and 28 additional nt of flanking sequence (see“ Materials and Methods”). Both SNB-19 and JEG-3 extracts were capable of forming splicing complexes on an adenovirus substrate (Ref. 17 , and data not shown). Several proteins were found to UV cross-link to the wild-type sequence in both SNB-19 and JEG-3 extracts (Fig. 2A) ⇓ . Prominent bands of ∼40, 44, and 60 kDa were observed in both extracts, and additional ∼50- and 130-kDa bands were specific to JEG-3 cells. Introduction of the M2 or M4 mutations into the RNA probe reduced the cross-linking of the 60-kDa protein to both SNB-19 and JEG-3 nuclear extracts, but the reduction was consistently greater in SNB-19 extract (Fig. 2A ⇓ , and data not shown). No reduction in the cross-linking of other major bands was observed, but additional proteins were seen to bind the mutant RNA. To examine the specificity of the 60-kDa binding protein, we performed competition experiments with SNB-19 extract (Fig. 2B) ⇓ . The wild-type but not the M2 sequence was able to specifically compete for binding with the 60-kDa protein, whereas the wild-type and mutant sequences competed equally for binding with the other proteins (Fig. 2B) ⇓ .

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

PTB specifically bound ISS-1. A, UV cross-linking of SNB-19 and JEG-3 cell nuclear extract to labeled wild-type (WT), M2 mutant, or M4 mutant probes (see “Materials and Methods” and Fig. 1 ⇓ B). B, UV cross-linking to wild-type probe was performed with SNB-19 cell nuclear extract that was preincubated with increasing amounts (20-, 50-, 100-, and 175-fold molar excesses) of unlabeled self competitor or with M2 mutant competitor. Cross-linking in the absence of competitor (NC) served as a control. C, immunoprecipitation (IP) with PTB antibody or antimouse immunoglobulin antibody (Ig) after UV cross-linking of SNB-19 cell nuclear extract to wild-type (WT) and M2 mutant RNA. Total UV cross-linking of SNB-19 cell nuclear extract is provided as a control (T). In A and B, masses of molecular markers are on the left.

The UV cross-linking results provided evidence that a 60-kDa protein interacts specifically with the ISS-1 regulatory region. The protein’s molecular mass and proposed role as a splicing inhibitor suggested that the protein might be PTB (18 , 19) . PTB has been shown to mediate exon skipping for a growing number of mammalian genes, including GABA_A receptorγ 2, c-src, fibronectin, and α- and β-tropomyosin (15 , 16 , 18 , 19) . To test whether the 60-kDa protein is PTB, we immunoprecipitated UV cross-linked complexes with PTB antibody (a generous gift from David Helfman). The 60-kDa UV cross-linked protein was indeed precipitated by PTB antibody (Fig. 2C) ⇓ . As observed previously, significantly more protein was associated with the wild-type sequence than with the M2 mutant RNA. No immunoprecipitated protein was seen when a control antibody was used (Fig. 2C) ⇓ .

Our results clearly demonstrated that PTB can specifically bind to the 40-nt ISS-1 regulatory element, although in vitro binding was not cell specific. This raises questions about the in vivo function of PTB. To better address whether PTB plays a role in the dysregulation of FGFR-1 splicing in glioblastomas, we examined the expression of PTB in a series of graded brain tumors. Total cellular protein was prepared from tumor and adjacent normal tissue obtained from patients diagnosed with low-grade astrocytoma, anaplastic astrocytoma, or glioblastoma multiforme. These data clearly indicated that the level of PTB was significantly elevated in tumor samples relative to adjacent normal tissue in all seven glioblastoma multiforme patients (Fig. 3) ⇓ . Elevated PTB expression was also observed in the single patient with anaplastic astrocytoma, but not in a patient with low-grade astrocytoma. The level of PTB expression in normal brain varied between patient samples and may reflect infiltration of tumor cells. One patient with a prior history of oligodendroglioma was found to have radionecrosis with a lesion mimicking a tumor. Interestingly, no difference was seen in the level of PTB expression in the tissues obtained from this patient.

To correlate PTB expression with α-exon skipping, we examined FGFR-1 splicing in tissues from which sufficient sample was available to isolate RNA (Fig. 3) ⇓ . As reported previously (3) , we found a high level of α-exon exclusion (∼70–90%) in the RNA samples from glioblastomas. In normal tissue, the level of FGFR-1 expression was consistently lower than in the tumor samples, but the methodology used did not allow us to directly quantitate these differences. However, the level of α-exon inclusion in normal tissue was clearly higher (∼30–55%) than that observed in tumor samples. Therefore, elevated PTB expression correlated with α-exon exclusion in tumors. In normal tissues, however, the relationship was less clear. Decreased levels of PTB correlated with increased α-exon inclusion. However, normal samples expressing high PTB levels did not necessarily have high levels ofα -exon exclusion. This may reflect differences in the cell composition of the normal tissue or may indicate that additional factors are involved in the regulation of FGFR-1 α-exon splicing. Tumor cells and gray matter express significantly higher levels of FGFR-1 than does normal white matter (3) , and contamination with these cells types could contribute to an artificially high level of exon skipping. Alternatively, the observed increase in PTB expression in glioblastomas may be coupled with a loss of factors required for α-exon inclusion. We previously have reported the identification of exonic sequences that are required for α-exon inclusion in JEG-3 cells (7) . The UV cross-linking results suggest that in JEG-3 cells, an inclusion pathway may be the dominant splicing pathway or that PTB alone is not sufficient to cause α-exon skipping. However, it is clear from this study that PTB interacts with the ISS-1, an element required for α-exon skipping, and that PTB protein levels are markedly elevated in glioblastomas. Given the ability of PTB to affect RNA processing decisions, our results suggest that in addition to FGFR-1, a number of other pre-mRNAs may undergo changes in alternative splicing during the early and intermediate stages of brain cancer.