In silico Analysis of Deleterious SNPs of the FGFR2 Gene

MATERIALS AND METHODS

Datasets: Database of Single Nucleotide Polymorphism (dbSNP) (http://www.ncbi.nlm.nih.gov/SNP) were used to retrieve SNPs and their related protein sequences in FGFR2 gene for systematic approach.

Analysis of functional consequences of coding nsSNPs by SIFT method: SIFT (Ng and Henikoff, 2003) is a sequence homology based tool that sorts intolerant from tolerant amino acid substitutions and predicts whether an amino acid substitution at a particular position in a protein will have a phenotypic effect. It also predicts the conservation level of a particular position in a protein. SIFT takes a query sequence and uses multiple alignment information to predict tolerated and deleterious substitutions for every position of the query sequence. SIFT is a multistep procedure that, given a protein sequence (1) searches for similar sequences (2) chooses closely related sequences that may share similar function (3) obtains the multiple alignment of these chosen sequences and (4) calculated normalized probabilities for all possible substitutions at each position from the alignment. Substitutions at each position with normalized probabilities less than a chosen cutoff are predicted to be deleterious; those greater than or equal to the cutoff are predicted to be tolerated. The cutoff values ranges from 0.00 to 0.05 was predicted to be intolerance/deleterious. The higher the tolerance index (above 0.05); will have less functional impact of the particular amino acid substitution.

Simulation for functional change in coding nsSNPs by PolyPhen: PolyPhen (Polymorphism phenotyping) (Ramensky et al., 2002) were used to predict possible impact of amino acid substitution on the structure and function of human protein(s). Analyzing the damaged coding non-synonymous SNPs at the structural level is considered to be important factor for understanding the functional activity of the protein. PolyPhen server gets input either as amino acid sequence of a protein or the SWALL database ID or accession number together with sequence position and two amino acid variants characterizing the polymorphism. Sequence based homologous sequences and mapping of substitution site to a know protein’s 3-Dimensional structure are the parameters taken into account by the PolyPhen server. It calculates Position Specific Independent Counts (PSIC) scores for each of the two variants and then computes the difference between the PSIC scores. The higher the PSIC score difference will have higher the functional impact for the particular amino acid substitution.

Functional significance of noncoding SNPs in regulatory untranslated regions: FastSNP (function analysis and selection tool for single nucleotide polymorphism) were used to identify the SNPs that are most likely to have functional effects (Yuan et al., 2006). 5’ UTR regions were predicted by FastSNP server. Recent studies show that SNPs may have functional effects on protein structures, by changing single amino acids, transcriptional regulation and on alternative splicing regulations (Prokunina and Alarcon-Riquelme, 2004; Prokunina et al., 2002). The FastSNP server follows the decision tree principle with external web service access to TF search, to obtain the predicted transcription factor-binding sites of the gene. Thus the decision tree will assign risk rankings for SNP prioritization with a ranking of 0, 1, 2, 3, 4 and 5. This signifies the levels of no, very low, medium, high and very high effects, respectively.

Scanning of UTR SNPs in UTR site: The 5’ and 3’ UTR regions are involved in various biological processes such as post-transcriptional regulatory pathways stability and translational efficiency (Sonenberg, 1994; Nowak, 1994). We used the program UTRscan (Pesole and Liuni, 1999) server to predict 3’ UTR region alone. This allows the search of user-submitted sequences for any patterns collected in the UTR site. UTR site is a collection of functional sequence patterns located in 5 and 3 UTR sequences. UTRscan which looks for UTR functional elements by searching through user-submitted sequence data for the patterns defined in the UTR site and UTR database. If the different UTR SNP is found to have different functional pattern(s) the UTR SNP is predicted to have functional significance. The internet resources for UTR analysis are UTRdb and UTR site. UTRdb contains experimentally proven biological activity of functional sequence patterns of UTR sequences from eukaryotic mRNAs (Pesole et al., 2002).

Modeling of mutant structure and calculation of their RMSD difference: Structural analysis was performed in order to evaluate and compare the stability of native and mutant structures. The native structure of FGFR2 protein was available in the Protein Data Bank (PDB ID: 1djs) with highest resolution (2.40 Å) (Stauber et al., 2000). We confirmed the mutation positions and the mutation residues using the dbSNP database (Sherry et al., 2001). The mutation and energy minimization for the 3D structure was performed by using the SWISSPDB viewer. Divergence of mutant structure from native structure is due to mutation, deletions and insertions (Rajasekaran et al., 2008) and the deviation between the two structures is evaluated by their RMSD values which could affect stability and functional activity (Rasricha et al., 2011).

Computational analysis of stabilizing residues: We used the server SRide (Magyar et al., 2005) for identifying the stabilizing residues in native protein and in the mutant model to check their stability. The input option for this server is the atomic coordinate file of the protein to be analyzed. Alternatively we can upload the PDB format directly; this option is mainly to analyze structures obtained by homology modeling or other computational approaches. Stabilizing residues were computed using parameters such as surrounding hydrophobicity, long-range order, stabilization center and conservation score.

RESULTS AND DISCUSSION

SNP dataset: SNPs have become marker of choice for many applications in genome analysis because SNPs are abundance, stable, ubiquity and interspersed in nuclear genome (Tchin et al., 2011). Single nucleotide polymorphism data of FGFR2 gene was retrieved from the dbSNP database. It contained a total of 2255 SNPs, of which 58 were non-synonymous, 39 were in non-coding region which comprises 11 SNPs in 5’ UTR, 28 SNPs in 3’ UTR and the rest SNPs were found in the intronic region. We took non-synonymous, 5’UTR and 3’UTR SNPs for our computational analysis. The distribution of these SNPs in coding and noncoding regions was illustrated in Fig. 1.

Deleterious nsSNPs by SIFT program: The conservation level of the particular position of the protein was determined using the sequence homology based method. The protein accession number with mutational position with two amino acid variants associated to the 58 nsSNPs were submitted as input to the SIFT server to check its tolerance index.

Fig. 1:	Distribution of nonsynonymous, 5’ UTR and 3’ UTR SNPs

Table 1:	List of nsSNPs that were predicted to have functional significance by SIFT

Lower the tolerance index, higher will be the functional impact of the amino acid substitution (Ng and Henikoff, 2001). Among the 58 nsSNPs, 22 were found to be deleterious, having a tolerance index score of p = 0.05. The results are summarized in Table 1. We observed that of 58 deleterious nsSNPs, 10 showed a high deleterious tolerance index score of 0.00. The remaining 5 nsSNPs showed a tolerance index score of 0.01, followed by 1, 1, 1 and 4 nsSNPs with a tolerance index of 0.02, 0.03, 0.04 and 0.05, respectively. On analyzing the nucleotide change, seven nsSNPs showed a nucleotide change of C/G, four a change from G/A, three a change from C/T and T/C, two a change from A/G and T/A and one a change from A/C. The nucleotide changes from C/G occurred a maximum number of times which accounts for the highest number of deleterious nsSNPs with a SIFT tolerance index of 0.00.

Damaged nsSNP found by the PolyPhen server: The structural level of the protein was analyzed using the PolyPhen server. All the 58 protein sequences of non-synonmous SNPs submitted to SIFT were also submitted to the PolyPhen server. A Position-Specific Independent Count (PSIC) score difference of 1.5 and above is considered to be damaging (Ramensky et al., 2002). Out of 58 nsSNPs, 23 nsSNPs were considered to be damaging and exhibited a PSIC score difference between 1.566 and 4.443 was showed in Table 2. Ten nsSNPs that was observed to be deleterious by the SIFT program was also found to be damaging according to PolyPhen. Hence, we could infer that the results obtained from SIFT shows a significant correlation with the results obtained from the PolyPhen which specifies the structural details. From Table 1 and 2, it was clear that the non-synonmous SNPs (rs121918501 and rs121918499) may disrupt both the structure and function of the protein coded by FGFR2 gene and it could be précised important for the recognition of various disorders such as Crouzon and Jackson-Weiss syndromes, breast, ovarian and lung cancers. The most damaging non-synonmous SNPs (rs121918501 and rs121918499) showed a PSIC score of 4.443 and 4.189 with a SIFT tolerance index of 0.00, due to the mutation from Tryptophan to Arginine and Tryptophan to Cysteine. Meanwhile, other studies had also reported on the association of SNPs with various applications such as the search for SNPs in PS1 in AD patients (Al-Khedhairy et al., 2005), identification of polymorphic sites (1236 and 3435) in multi drug resistance gene 1 Influencing drug response in breast cancer patients (Khan et al., 2007), genetic polymorphisms in the ESR genes and the risk of breast cancer among Iranian women (Abbasi et al., 2009a), genome wide SNP analysis in Mycobacterium sp. (Srivastava et al., 2009).

Table 2:	List of nsSNPs that were predicted to have functional significance by PolyPhen

Functional SNPs in UTR found by the FastSNP: We used FastSNP server to analyze the SNPs in 3’ and 5’ untranslated region that are predicted to be functionally significant (Yuan et al., 2006). According to this server, of 11 5’ UTR SNPs present in the FGFR2 gene, two 5’ UTR namely rs3135721 and rs41258305 were predicted to be damaging with a risk ranking of 1-3 (Table 3). The nucleotide changes were G/T for SNP ID rs3135721 and C/T for SNP ID rs41258305. 3’UTR SNPs were not predicted by this server.

Functional SNPs in UTR found by the UTRscan server: Polymorphisms in the 3’ UTR affect gene expression by affecting the ribosomal translation of mRNA or by influencing the RNA half-life (Deventer, 2000). We analyzed 28 SNPs in the 3’ untranslated region using the UTRscan server. The UTRscan server finds patterns of regulatory region motifs from the UTR database and gives information about whether the matched pattern is damaged. About fifteen UTR SNPs were predicted to be damaging by this server and presented in Table 4. The 15-lipoxygenase (15-LOX) differentiation control element (15-LOX-DICE) controls 15-LOX synthesis which catalyzes the degradation of lipids and is an important factor responsible for the degradation of mitochondria during reticulocyte maturation (Rajasekaran et al., 2007). This 15-LOX-DICE exists in the fifteen 3’ UTR SNPs that were considered to be of functional significance and it may lead to FGFR2 gene damage. No reports have been analysed so far relating the deleterious nature of SNPs with untranslated regions of FGFR2 gene.

Modeling of mutant structure and calculation of their RMSD difference: Single amino acid mutations can significantly change the stability of a protein structure. So the knowledge of a protein’s three-dimensional (3D) structure is essential for understanding of its functionality (Chen and Shen, 2009). Information about mapping the deleterious nsSNPs in the protein structure was obtained from dbSNP. The available structure for the FGFR2 gene has the PDB ID 1djs (Fig. 2). Two nsSNPs (rs121918501 and rs121918499) were selected for structural analysis. The mutational position and amino acid variant associated with these nsSNPs is W/R at the residue position 290 and W/C at the residue position 290 was mapped in the 1djs native structure.

Table 3:	List of SNPs (UTR mRNA) predicted to be functionally significant by FastSNP

Table 4:	List of SNPs (UTR mRNA) predicted to be functionally significant by UTRscan

Table 5:	Stabilizing residues in native and mutant models of 1djs by SRide

Fig. 2:	Native structure of FGFR2 gene (1djs)

Mutation and energy minimization at the specified position was performed by SWISSPDB viewer to obtain a modeled structure for both the native structure (1djs) and mutant modeled structures (W290R) and (W290C). It can be seen that the total energy for the mutant modeled structures were found to be -15249.33 and -14932.46 Kcal moL^-1, respectively. The super impositions between the native and mutant structures were performed using Pymol suite and RMSD values was found to be 0.40Å and 0.45Å, respectively. Since these values are low, we can suggest that these mutations do not cause a significant change in the mutant structures with respect to native protein structure. Mutation and polymorphism of cancer-associated genes have been found to predict tumor formation and prognosis. It is also considered as an effective risk factor with positive effects and negative effects in different studies (Abbasi et al., 2009b). The superimposed structures of the native protein 1djs and the two mutant type proteins are shown in Fig. 3a and b, respectively.

Computing stabilizing residues between native structure and mutant modeled structures: We used the SRide server to identify the stabilizing residues of the native-type structure and mutant modeled structures. The results are summarized in Table 5. Six residues were identified in the native-type 1djs structure. Four stabilizing residues were identified in mutant model 1djs (W290R) and seven were identified in mutant model 1djs (W290C). Three stabilizing residues, namely Ala (181), Val (248) and Cys (342) were found to be common to both the native structure 1djs and mutant models. The remaining residues were not found common in mutant and native structures. Higher number of stabilizing residues in the mutant type 1djs (W290C) matched with the native protein structure compared to the mutant type (W290R). Therefore we predict that the mutation from Tryptophan to Cysteine in the native type protein will be more deleterious than Tryptophan to Arginine and hence both SNPs could be important candidates for various disorders in the FGFR2 gene.

Fig. 3(a-b):	(a) Superimposed structure of native protein 1djs (cyan color) with mutant protein W290R magenta color (b) Superimposed structure of native protein 1djs (cyan color) with mutant protein W290C orange color

CONCLUSION

In the present study, we investigated the functional and structural impact of SNPs in the FGFR2 gene by computational prediction tools. Out of a total 2255 SNPs in the FGFR2 gene, we found 58 non-synonmous SNPs, 28 3’ UTR and 11 were found in 5’ UTR region and the remaining 2158 were found in intronic region. Out of 58 nsSNPs, 10 nsSNPs were found to be deleterious by both SIFT and PolyPhen server. Two SNPs in the 5’ UTR and 15 SNPs in the 3’ UTR were found to be of functional significance. It was found that the major mutation in the native protein of the FGFR2 gene was from Tryptophan to Arginine and Tryptophan to Cysteine. Of two nsSNPs that had this mutation, we conclude that rs121918501 with a mutation of Tryptophan to Arginine and rs121918499 with a mutation of Tryptophan to Cysteine at position 290 in the native protein 1djs could be the main target for the diseases caused by the FGFR2 gene.

ACKNOWLEDGMENT

The authors gratefully acknowledge the computational and bioinformatics facility provided by the Department of Bioinformatics, Alagappa University, Karaikudi-630003, Tamilnadu, India.