In-silico analysis of phylogenetic relationship and potentially damaging nsSNPs in human SLC2A2 gene

2.1 Dataset compilation

National centre for biological information; NCBI (http://www.ncbi. nlm.nih.gov) was accessed. Protein and nucleotide sequences were acquired in FASTA format. Public SNP database named as dbSNP (http://www. ncbi.nlm.nih.gov/SNP), Gene Cards (http://www.genecards.org/) and UniProt (http://www.uniprot.org) were searched to assemble information related to longest isoform of GLUT2 regarding protein sequence. (GLUT2 gene ID: 6514 SNPs NCBI Reference Sequence: NP_000331). The coding region nsSNPs were screened and were subjected to further computational analysis.

2.2 Phylogenetic analysis

To carry out phylogenetic analysis, protein sequence of human GLUT2 (GenBank Accession Number: NP 000331.1) along other 9 species of Hominidae family; Gorilla gorillagorilla (XP_004038053.1), Pongo abelli (XP_002814322.1), Rhinopithecus roxellana (XP_010359570.1), Microcebus murinus (XP_012591846.1), Bos taurus (DAA33241.1), Myotis brandtii (XP_005886638.1), Neomonachus schauinsland (XP_021558267.1), Galeopterus variegates (XP_008586350.1) and Ailuropoda melanoleuca (XP_002920886.1) were retrieved from the NCBI. Protein sequence in FASTA format of all 10 species was obtained and saved. Protein sequence alignment was performed using Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/). In Clustal Omega, output format “ClustalW with charater counts” was chosen and option of phylogenetic tree was clicked.

2.3 Membrane topology

The number of machine learning algorithms wre used to align the set of protein sequences and clustering of topological data. These approaches or algorithms are transmembrane helices prediction (PHDhtm) (Rost et al. 1996), statistical algorithm TMAP (Persson and Argos 1994) and Hidden Markov models (HMMs) (HMMTOP (Tusnady and Simon, 1998, 2001); TMHMM (Sonnhammer et al., 1998; Krogh et al., 2001). The membrane topology of GLUT2 protein was predicted with the TMHMM program, server 2.0 (http://www.cbs.dtu.dk/services/TMHMM-2.0/).

2.4 Analysis of nsSNPs by SIFT

Sorting Intolerant from Tolerant (SIFT) is an algorithm that forecasts whether substitution of an amino acid influences protein function or not. It is based on the principle that essential amino acids in a protein family remain conserved, and any replacement at these positions is deleterious, which influences function/activity of protein. Protein sequence in FASTA format along substitution of amino acids was used as input at (http://www.blocks.fhcrc.org/sift/SIFT.html). It then considered the normalized probability for the provided substitution at the related position in the alignment. Substitutions with the normalized probability i.e. Tolerance Index (TI) less than 0.05 were forecasted to be intolerant or harmful whereas substitutions with TI higher than 0.05 were predicted to be tolerant (Adzhubei et al., 2010).

2.5 Analysis of nsSNPs by using PolyPhen

PolyPhen-2 is a tool for forecasting of the feasible impact of an amino acid substitution on the protein structure and function (Adzhubei et al., 2010). Automated divinations of this type are vital for interpreting rare genetic variants, which may have potential approach in recent research of human genetics. Its importance in modern research involves identification of rare alleles that generate Mendelian disease (Bamshad et al., 2011), scanning for medically important alleles in an individual’s genome (Ashley et al., 2010), and profiling the spectrum of rare variation uncovered by deep sequencing of large populations (Tennessen et al., 2012).

PolyPhen (http://genetics.bwh.harvard.edu/pph2) utilizes UniProtKB database which is used as source of reference. Protein sequence of GLUT2 in FASTA format and position of protein sequence was entered. AA1 wild type (query sequence) as well as substitution reside AA2 were selected. Two pair’s datasets were used. First pair of HumDiv and the second pair; HumVar were compiled in UniProt database.

2.6 Analysis of nsSNPs by using PROVEAN

PROVEAN is not only effective to predict single amino acid change but it can also perform task for other changes in the protein sequence including indels (deletion and insertion) (Choi et al., 2012). The PROVEAN web server (http://provean.jcvi.org) gives online approach for the functioning of distribution for the software package of stand-alone. Its major role is to give divination from any of organism’s protein sequence. It takes the sequence of protein and variation of amino acid as input. BLAST search was done to recognize the homologous sequence and produce the PROVEAN result. Generally the BLAST search took 10–20 min to produce the divination for a provided protein query. Sequence identifiers list for supporting sequences and clustering data was stored in database. Focused on sequence of query protein consecutive prediction data for supporting sequences were indexed.

2.7 Analysis of nsSNPs by using SNPeffect

SNPeffect (De Baets et al., 2012) was used to predict the molecular phenotypic influence of nsSNPs lying in coding region of GLUT2 protein. It works beyond the scores got on conservational basis and mainly emphasizes to map the effect of SNPs on the capability of cells to uphold suitable concentration of the properly folded proteins in appropriate cellular region i.e. protein homeostasis landscape (Powers et al., 2009). For this assessment, wild type protein sequence in FASTA format along with each of its variants was given to the SNPeffect server (http://snpeffect.switchlab.org) for analysis. Homology threshold was fixed to 90%. It uses TANGO (Fernandez-Escamilla et al., 2004) which forecasts the regions in given protein sequence that are prone to aggregation and calculated TANGO score with wild and variant amino acid. On the basis of TANGO score difference (dTANGO), the server assesses effect of these variants on protein aggregation. Aggregate morphology is more distinctively evaluated by WALTZ server (Maurer-Stroh et al., 2010) that forecasts amyloid-forming regions in given protein sequence with accuracy and specificity on basis of dWALTZ score. Chaperone binding propensity is forecasted by LIMBO (Van Durme et al., 2009) for the Hsp70 chaperones and effect of variant is determined by dLIMBO score. SNPeffect also uses high resolution crystal structure of proteins from Protein Data Bank (PDB) (Deshpande et al., 2005) and models the variants using the empirical force field FoldX (version 2.5) (Schymkowitz et al., 2005) to determine possible effects on stability and binding properties of given protein.

3.1 Phylogenetic analysis

Evolutionary relationship of GLUT2 protein among 10 different species of primates was carried out using human GLUT2 protein as a reference sequence (Fig. 2). The phylogenetic tree generated from Clustal Omega helps to understand the evolutionary relationship of GLUT2 among different species. Evolutionary tree demonstrates that GLUT2 of human and Pongo abelli lie close to each other whereas that of Bos taurusis most distant from human.

Close

Fig. 2

Phylogenetic analysis of human GLUT2 protein. A phylogenetic tree of the amino acid sequences of GLUT2 protein was constructed using the Clustal Omega.

Export to PPT

3.2 Membrane topology

A significant server TMHMM was used to predict the significant features of helices in transmembrane protein and GLUT2 was found to comprise of 12 predicted transmembrane helices with intracellular amino (N-terminal) and carboxyl (C-termini) in cytosol. It also contains extracellular and intracellular loops which are located between the segments of 1st and 2nd transmem brane domains and between 6th and 7th transmembrane segments (Fig. 3). While, the remaining transmembrane segments from seven to twelve have ability to perform transportation of fructose and glucose to the membrane.

Close

Fig. 3

Toplogy prediction of human GLUT2 protein. The human GLUT2 protein comprises of 12 predicted transmembrane helices with intracellular amino (N-terminal) and carboxyl (C-termini) in cytosol. It also contains extracellular and intracellular loops which are located between the segments of 1st and 2nd transmembrane domains and between 6th and 7th transmembrane segments.

Export to PPT

3.3 Distribution of SNPs in GLUT2 protein

In the human SLC2A2 gene the non coding region of gene contains 6,628 SNPs in which 83 SNPs were near 3′gene, 340 in 3′UTR, 427 SNPs in 5′ near gene, 62 in 5′UTR and in the intron 5716 SNPs were found (Fig. 4). While the coding region contained 484 SNPs in which 137 SNPs were synonymous, 293 were nsSNPs, 21 were frameshifts; (rs766082034), (rs1447936042), (rs769888108), (rs1255595607), (rs776307487), (rs1162215911), (rs771361095), (rs1326032349), (rs1350704340), (rs1290975016), (rs765132996), (rs1181030797), (rs771799491), (rs746178753), (rs1316522125), (rs748296868), (rs34066960), (rs1174349159), (rs1384674663),(rs1386374799) and (rs1290412048). 11 were nonsense; (rs774841662), (rs771477447), (rs1475086161), (rs1114167428), (rs753629940), (rs773581866), (rs121909746), (rs121909743), (rs121909742), (rs121909745) and (rs1379944645). Out of 7 SNPs, 4 were inframe deletion; (rs772999215), (rs1169887677), (rs763620441), (rs774721090) and 3 were inframe insertion; (rs1161394690), (rs1463507753) and (rs749789723). 10 (rs1294679246), (rs756163471), (rs371977235), (rs1240053337), (rs756874949), (rs1303795800), (rs1281471314), (rs985090030), (rs757587931), (rs867530965) were donor and 4 (rs754220999), (rs1318756243), (rs776248984), (rs1230247311) were acceptor shown (Fig. 5). (rs749789723) SNPs also contain stop gained. 1 SNPs was stop lost (rs750463981). 10 additional SNPs were detected by dbSNP in which 4 SNPs; (rs766082034), (rs1255595607), (rs1162215911), (rs746178753) were frameshifts and 6; (rs766762468), (rs1800572), (rs777020657), (rs867996396), (rs121909747), (rs1309197020) were nsSNPs (Table 1)

Close

Fig. 4

Distribution of SNPs in human GLUT2 Protein.

Export to PPT

Close

Fig. 5

Distribution of nsSNPs in coding region of GLUT2 protein.

Export to PPT

3.4 Analysis of nsSNPS using SIFT

SIFT forecasts whether amino acid substitution in protein has phenotypic effect or not. Alignment also gave forecasted changes of all amino acid locations. The SIFT intolerance index starting point was 0.05. The color code of non polar, basic, uncharged polar and acidic amino acids were shown by black, red, green and blue color respectively. On the other hand, the amino acid alignment is shown by capital letter while the results of prediction are shown by small letter. 'Seq Rep' was the proportion of sequences that has one of the important amino acids. The position either severely gapped or unalignable was determined by low fraction and it has little information which has poor prediction. Its prediction score ranges from 0 to 1. On phenotypic substitution by aligning orthologous and paralogous protein sequence, this study mainly considers the physical properties of amino acids, effect of natural nsSNPs and alignment of homologous sequences (Noreen et al., 2015a, 2015b). For positions 1–524 of amino acids, the prediction of possible substitutions was made (Figure 6). It was used to predict damaging and tolerated effect of 293 nsSNPs that occurs in coding region of GLUT2 protein. As a consequence, 167 were predicted to be damaging (Table 2).

3.5 Analysis of nsSNPS using PolyPhen

To forecast the effects of function and structure 293 nsSNPs lying in the coding region of GLUT2 protein, PolyPhen-2 server was used. It permits three types of prediction score which is based on a “Sensitivity” as well as “Specificity”. Three prediction types are given by PolyPhen benign, possibly damaging or probably damaging. Out of 293 nsSNPs, 165 were considered to be benign. 65 were predicted to be possibly damaging and 77 were considered to be probably damaging (Table 2).

3.6 Analysis of SNPs using PROVEAN

PROVEAN gave score to predict neutral and deleterious effect of 293 nsSNPs present in coding region of GLUT2 protein. Score equal to or below −2.5 was considered as deleterious and above −2.5, as neutral. Out of 293 nsSNPs, 162 were predicted to be neutral and 138 were considered to be deleterious. It also gave the results of inframe deletion and insertion. Out of 4 inframe deletion 2 (rs772999215 and rs774721090) were shown to be deleterious 3 SNPs were inframe insertion (rs1161394690), (rs1463507753), (rs749789723) and no results were found for them (Table 2).

3.7 Analysis of nsSNPS using SNPeffect

The SNPeffect server was used to predict the effect of molecular phenotype of nsSNPs found in coding region of GLUT2 protein. Effect of 293 nsSNPs on aggregation tendency, amyloid propensity and chaperon binding tendency was specified by dTANGO, dWALTZ and dLIMBO score. Two significant effects, intrinsic aggregation propensity and protein stability were shown. In a protein sequence, aggregation propensity was identified by dTANGO. Out of 293 nsSNPs, 61 were detected to decrease the aggregation propensity, 27 to increase the aggregation propensity while 211 had no effect on aggregation propensity. dWALTZ indicated the score of amyloid propensity. 23 nsSNPs were shown to increase the amyloid propensity, 16 decreased the amyloid propensity and 257 had no effect on the amyloid propensity. According to dLIMBO, the chaperon binding tendency was given. None of them had any effect on chaperon binding tendency (Table 3).