Mutational analysis of the GAP-related domain of the neurofibromatosis type 1 gene in Brazilian NF1 patients

Neurofibromatosis type 1 (NF1) is a common autosomal dominant disorder caused by mutations in the NF1 gene. In the present study, a total of 55 unrelated NF1 patients were screened for mutations in the GAP-related domain/GRD (exons 20-27a) by single-strand conformation polymorphism (SSCP). Four different mutations were identified and, taken together, they comprise one nonsense substitution (Q1189X), one deletion (3525-3526delAA), one missense substitution (E1356G) and one mutation in the splice acceptor site (c.4111-1G>A). One novel polymorphism (c.4514+11C>G) and other three putative polymorphisms were also found (c.3315-27G>A, V1146I and V1317A). Genotype-phenotype correlations were investigated, but no particular association was detected.


Introduction
Neurofibromatosis type 1 is one of the most common autosomal dominant disorders, affecting approximately 1:3,000 individuals.The main characteristics of the disease comprise multiple neurofibromas, cafe-au-lait skin spots, Lisch nodules and freckling, but, in a minority of patients, other features, such as scoliosis, macrocephaly, short stature, malignancies, and learning disabilities, are also found (Huson and Hughes, 1994).
The NF1 gene, mapped to 17q11.2 (Barker et al., 1987), contains 60 exons and has one of the highest mutation rates described for human genes (Huson and Hughes, 1994).It is transcribed to an mRNA that encodes a protein, neurofibromin.A central region of this protein (exons 20-27a) shows functional and structural homology to the mammalian GTPase-activating protein (GAP).The GAPrelated domain (GRD) has been shown to down-regulate p21 ras by accelerating the rate of GTP hydrolysis and inhibiting Ras-mediated signal transduction (Martin et al., 1990).
Ten years after the cloning of the NF1 gene, mutation analysis and diagnostic testing remain a challenge, mainly due to the large size of the gene, the presence of several pseudogenes and the lack of mutational hotspots.Despite these difficulties, more than 400 different NF1 mutations have been reported (http://www.hgmd.org;http://www.clam.com/nf/nf1gene).Different methodologies have been used to screen mutations in the NF1 gene.Denaturing high-performance liquid chromatography (DHPLC) was used by Han et al. (2001) andDe Luca et al. (2003), and the mutation detection rates were 97% and 72.5%, respectively.Heteroduplex, FISH, Southern blot, PTT, TGGE and genomic sequencing were also used by Fahsold et al. (2000) and Messiaen et al. (2000), and up to 95% of mutations were identified.Mutations in the GAP-related domain in lung cancer samples have been screened by Furukawa et al. (2003) using SSCP and sequencing, with a mutation and polymorphism detection rate of 8%.
No genotype-phenotype correlation has been detected, except for microdeletions and deletions encompassing the entire gene, or perhaps contiguous genes also, in patients with facial anomalies, great number of neurofibromas and severe developmental delay (Wu et al., 1995;Leppig et al., 1996;Lopez-Correa et al., 2001).
Learning disabilities have also been observed in patients with different inactivating mutations and even in NF1 mutant flies and mice (Guo et al., 2000;Costa et al., 2001;Costa and Silva, 2003).Relatively few learning disability data based on IQ tests are reported in molecular studies on neurofibromatosis.
In the present study, we analyzed 55 patients with NF1 for mutations in the GAP-related domain (GRD) of the NF1 gene.

Materials and Methods
Fifty-five unrelated Brazilian NF1 patients were examined within our multidisciplinary NF1 Program.Thirtythree were familial cases and twenty-two were sporadic.Clinical details were fully documented.Cognitive functions were evaluated using either Wechsler's Intelligence Scale for Children (WISC) or Wechsler's Adult Intelligence Scale (WAIS) tests.All patients gave their informed consent prior to inclusion in the study.
Amplified fragments were then subjected to SSCP analysis.All aberrant SSCP mobility patterns were verified on a second PCR-SSCP analysis, and the DNA samples were automatically sequenced in an ABI 377 PRISM DNA Sequencer (Applied Biosystems).The mutations were also investigated on 100 unrelated alleles, and polymorphisms in 200 control samples.

Results and Discussion
In the 55 unrelated patients screened, four different mutations were identified, of which three are novel: one missense and one nonsense substitutions, and one mutation in the splice acceptor site.Four polymorphisms were also found (Table 1).
The missense mutation (E1356G) of patient 26 (exon 23.2) changed a negatively charged polar for a non-polar amino acid in the peptide.Such a substitution may alter the protein structure and is, thus, likely to be disease-causative.From the information available on neurofibromin sequences of mouse and rat (Genbank L10370 and D45201), this amino acid substitution occurred in a highly conserved position of GRD (identical in human, mouse and rat) and may be essential for GAP function.However, an analysis of the molecular model of the complex NF1(E1356G)-Ras strongly indicated that this mutation has no influence on the interaction between NF1 and Ras, since it is far from the protein-protein interface of the complex (Figure 1).
While the cause of disease in patients with missense mutations is unclear, the significance of stop codons is obvious.In the present study, we identified one nonsense mutation (Q1189X).This mutation (case 28) was identified in exon 21 and is predicted to lead to a severely truncated protein, with only 1,188 amino acids instead of the normal 2,818.The deletion of two adenines, at nucleotide 3,525 in exon 21 (patient 32), caused a shift in the reading frame, leading to the creation of a premature stop codon at nucleotide 3,580.This mutation may result in the generation of a shortened, non-functional protein of 1,192 amino acids.Recently, the same mutation was reported by Fahsold et al. (2000) and Serra et al. (2001), but, as in the case of the nonsense mutation, the clinical features of these patients are not available for comparison.
One mutation (case 18) was found in a splice site (c.4111-1G>A) and probably destroyed the acceptor consensus sequence of intron 23.2.Family members of this patient were available for analysis, and the presence of the mutation segregated with the disease.Upadhyaya et al. (1997) and Fahsold et al. (2000) found different splicing mutations at the donor site of the same intron.
A novel polymorphism (c.4514+11C>G) was detected in intron 26 (patients 16 and 36).This polymorphism was also observed in 3 out of 200 control samples.Other three putative polymorphisms were found in exons 20 (V1146I) and 23.1 (V1317A) and in intron 19b   (c.3315-27G>A) in patients 24, 8 and 34, respectively, but in none of the 200 control samples.Both V1146I and V1317A replace a non-polar R group by another non-polar R group and are substitutions often found as polymorphisms in other genes (Miller and Kumar, 2001).The latter polymorphism segregated with the disease in the patient's family and is therefore likely to be located on the same chromosome as the NF1 mutation.Molecular modeling might clarify if V1146I and V1317A affect the structure of the neurofibromin GAP-related domain.However, it is not possible to generate a molecular model for both, because the template NF1GRD (NF1-333; residues 1198-1304 and 1331-1551; PDB access code: 1NF1) has no atomic coordinates in these regions (Scheffzek et al., 1998).This is the first report on mutational screening in a large number of Brazilian NF1 patients.As in previous studies, we were unable to correlate the presence or severity of clinical features and/or mental retardation with the site of the mutation (Table 2).This is not unexpected, given the wide range of clinical variability that has previously been reported, even in different members of the same family.As Messiaen et al. (2000) questioned: Do genotype-phenotype correlations in NF1 exist?It is possible that the clinical picture is dependent on many endogenous and exogenous, transitory or lasting, environmental factors.Therefore, in addition to more sensitive methods of mutation detection, studies on gene function are necessary to understand the pathogenesis of this disease.The greatest challenge for the next years will undoubtedly be to link the phenotypic features to the role of neurofibromin and related proteins in growth control and cell differentiation.

328
Figure 1 -Cα-backbone for the complex NF1(E1356G)-Ras, showing the G1356 position in the structure.NF1 is drawn with thick lines and Ras with thin lines.

Table 1 -
NF1 mutations and polymorphisms in nine unrelated patients with neurofibromatosis type 1.