In silico analysis of deleterious single nucleotide polymorphisms in human BUB1 mitotic checkpoint serine/threonine kinase B gene.

One of the major challenges in the analysis of human genetic variation is to distinguish mutations that are functionally neutral from those that contribute to disease. BubR1 is a key protein mediating spindle-checkpoint activation that plays a role in the inhibition of the anaphase-promoting complex/cyclosome (APC/C), delaying the onset of anaphase and ensuring proper chromosome segregation. Owing to the importance of BUB1B gene in mitotic checkpoint a functional analysis using different in silico approaches was undertaken to explore the possible associations between genetic mutations and phenotypic variation. In this work we found that 3 nsSNPs I82N, P334L and R814H have a functional effect on protein function and stability. A literature search revealed that R814H was already implicated in human diseases. Additionally, 2 SNPs in the 5' UTR region was predicted to exhibit a pattern change in the internal ribosome entry site (IRES), and eight MicroRNA binding sites were found to be highly affected due to 3' UTR SNPs. These in silico predictions will provide useful information in selecting the target SNPs that are likely to have functional impact on the BUB1B gene.

Introduction

The spindle assembly checkpoint (SAC) is a cell-cycle surveillance mechanism that prevents premature anaphase entry until all chromosomes have completely aligned at the metaphase plate. SAC is composed of the checkpoint proteins BubR1, Bub3, and Mad2, associated with the APC/C coactivator Cdc20. The checkpoint system acts to inhibit the activity of the large multi-protein E3 ubiquitin ligase known as the anaphase promoting complex/cyclosome (APC/C), by binding to the co-activating subunit Cdc20. BubR1 is a key protein mediating spindle-checkpoint activation That directly binds to Cdc20 and inhibits APC/C activity (Kaisari et al., 2016). The corresponding BUB1B gene is located on chromosome 15q15 and is composed of 23 exons that encodes 1050 amino acids (Davenport et al., 1999, Hanks et al., 2012). BUB1B mutated in several cancers including colorectal, lung, breast, hematopoietic malignancies and in a rare human hereditary condition called premature chromatid separation syndrome (mosaic variegated aneuploidy) (Kapanidou et al., 2015, Kops et al., 2005, Hanks et al., 2004, Hanks et al., 2006, Matsuura et al., 2006, Suijkerbuijk et al., 2010, Ohshima et al., 2000).

Single nucleotide polymorphisms (SNPs) found in any position throughout the genome in exons, introns, intergenic regions, promoters and enhancers (Drazen et al., 1999). Many SNPs are phenotypically neutral. However, others could predispose human to disease or influence their response to a drug. Nonsynonymous SNPs (nsSNPs) that lead to an amino acid substitution in the corresponding protein product are of particular interest as they are responsible for nearly half of the known gene lesions responsible for human inherited disease (Krawczak et al., 2000).

Computational analyses of BUB1B gene for harmful nsSNPs have not been carried out until now; therefore, we applied different publicly available computational tools according to Fig. 1. The value and novelty of this study is to prioritize SNPs with functional significance from an enormous number of neutral non-risk alleles of BUB1B and provides new insights for further genetic association studies.

Fig. 1

Schematic representation of computational tools for in silico analysis of BUB1B gene.

Material and method

Dataset

The NCBI database of SNPs (Sherry et al., 2001), dbSNP available at http://www.ncbi.nlm.nih.gov/SNP and SWISSProt databases (Bairoch and Apweiler, 1996) were used to obtain the SNP information [SNP ID, amino acid position, mRNA accession number NM_001211, and Protein accession number NP_001202.4] of the human BUB1B gene for our computational analyses.

Predicting functional context of missense mutation

The functional context of nsSNPs was predicted using SIFT (Sorting Intolerant from Tolerant), PolyPhen 2.0, I-Mutant 3.0 and PROVEAN (Protein Variation Effect Analyzer) (Table 1).

Table 1

In silico approaches available as online tools.

Server	Feature	URL	Reference
SIFT	Focuses more on the sequence preservation over the evolutionary time in predicting the effect of residue substitutions on function.	http://sift.bii.a-star.edu.sg/index.html	Kumar et al. (2009), Magesh and Doss (2014)
PolyPhen 2.0	Sequence and structure based method that predicts the possible impact of an amino acid substitution on the structure and function of a protein.	http://genetics.bwh.harvard.edu/pph2	Adzhubei et al. (2010)
I-Mutant 3.0	Support vector machine (SVM) based predictors of protein stability changes upon single amino acid substitution.	http://gpcr2.biocomp.unibo.it/cgi/predictors/I-Mutant3.0/I-Mutant3.0.cgi	Capriotti et al. (2008)
PROVEAN	Sequence based predictor that estimates whether a protein sequence variation affects protein function.	http://provean.jcvi.org	Choi et al. (2012)
SNP&GO	Support vector machine (SVM) based web server that combine protein structural/functional parameters and sequence analysis derived information.	http://snps.biofold.org/snps-and-go/snps-and-go.html	Magesh and Doss (2014)
PhD-SNP	SVM based on evolutionary information.	http://gpcr.biocomp.unibo.it/cgi/predictors/PhD-SNP/PhD-SNP.cgi	Magesh and Doss (2014)
PANTHER	Protein family and subfamily database that predicts the frequency of occurrence of amino acid at a particular position in evolutionary related protein sequences.	http://pantherdb.org/tools/csnpScoreForm.jsp	Mi et al. (2005)
UTRscan	Analyzing functional impacts of UTR SNPs.	http://www/.ba.itb.cnr.it/BIG/UTRScan	Pesole et al. (1999)
NetSurfP	Analysis of SNP effects on surface and solvent accessibility of protein.	http://www.cbs.dtu.dk/services/NetSurfP	Petersen et al. (2009)
I-TASSER	Protein structure prediction server.	http://zhanglab.ccmb.med.umich.edu/I-TASSER	Zhang (2008)
HOPE Project	An automatic mutant analysis server for studying the structural features of native protein and the variant models.	http://www.cmbi.ru.nl/hope/home	Venselaar et al. (2010)
STRING	Database of known and predicted protein-protein interactions.	http://string-db.org	Von Mering (et al. 2005)
PolymiRTS	The polymorphic microRNA target sites are classifies into four classes: ‘D’ (the derived allele disrupts a conserved microRNA site), ‘N’ (the derived allele disrupts a nonconserved microRNA site), ‘C’ (the derived allele creates a new microRNA site) and ‘O’ (other cases when the ancestral allele cannot be determined unambiguously).	http://compbio.uthsc.edu/miRSNP/	Bhattacharya et al. (2013)
1000 Genomes Project	A resource about human genetic variation that will be used in many studies of particular phenotypes, such as complex diseases or drug response.	http://www.1000genomes.org	Via et al. (2010)

SIFT predict whether an amino acid substitution in a protein would be tolerated or damaging. The amino acid substitution is predicted damaging when the score is below or equal to 0.05, and tolerated if the score is greater than 0.05 (Ng and Henikoff, 2003). PolyPhen input is the amino acid sequence of protein or SNP identifier with the nsSNP. The output levels of probably damaging and possibly damaging were classified as functionally significant (≤ 0.5) and the benign level being classified as tolerated (≥ 0.5) (Ramensky et al., 2002). I-Mutant 3.0 performed analyses based on the protein sequence combined with mutational position and correlated new residue and the output result of the predicted free energy change (DDG) classifies the prediction into one of three classes: largely unstable (DDG < − 0.5 kcal/mol), largely stable (DDG > − 0.5 kcal/mol), or neutral (− 0.5 ≤ DDG ≥ 0.5 kcal/mol) (Capriotti et al., 2008). PROVEAN is able to provide predictions for any type of protein sequence variations including amino acid substitutions, and in-frame insertions and deletions (Choi et al., 2012). The PROVEAN predict a protein variant to be neutral if the score is above the threshold. The cutoff score − 2.5 indicates a deleterious substitution (Manickam et al., 2014).

Furthermore, we used SNP&GO, PHD-SNP Predictor of human deleterious single nucleotide polymorphisms and PANTHER (Protein Analysis Through Evolutionary Relationships) tools to filter the disease-associated nsSNPs (Table 1).

SNP&GO predict SNPs are or are not disease-associated with including the protein FASTA sequence and Gene Ontology terms. The probability score higher than 0.5 indicates the disease related effect of mutation on the parent protein function (Calabrese et al., 2009). PhD-SNP predicts whether the given amino acid substitution leads to disease associated or neutral along with the reliability index score (Capriotti et al., 2006). PANTHER comprehensive software system predicts the likelihood of a particular nsSNP to cause a functional impact on a protein. The cutoff subPSEC score − 3 indicates a deleterious substitution (Thomas et al., 2003).

Biophysical validation of nsSNPs

NetSurfP predicts the surface and, solvent accessibility of amino acids, using the amino acid FASTA sequence. The solvent accessibility has been predicted in two classes as either buried or exposed, based on the accessibility of the amino acid residues to the solvent, respectively. The reliability of this prediction method is in the form of Z-score. The Z-score highlights the surface prediction reliability, but is not associated with the secondary structure (Petersen et al., 2009).

Finding 3D structure of proteins is helpful in predicting the impact of SNPs on the structural level and in showing the degrees of alteration. I-TASSER generates a full length model of proteins by excising continuous fragments from threading alignments and then reassembling them using replica-exchanged Monte Carlo simulations. Low temperature replicas (decoys) generated during the simulation are clustered by SPICKER and the top five cluster centroids are selected for generating full atomic models. The quality of prediction models was reflected in the form of c-scores (− 5 to 2) (Roy et al., 2010, Roy et al., 2012). The native structure was mutated with the most deleterious amino acid substitution predicted in this study, using Swiss PDB viewer and Chimera (Kaplan and Littlejohn, 2001, Pettersen et al., 2004).

In addition, we used HOPE Project that provides the 3D structural visualization of mutated proteins, and gives the results by using UniProtKB and predictions from DAS-servers. FASTA sequence of whole protein and selection of mutant variants is considered to be an input option, the output is based on the structural variation between the mutant and the wild-type residues (Venselaar et al., 2010).

Predictions of protein-protein interactions

STRING (Search Tool for the Retrieval of Interacting proteins) is a database and web resource dedicated to protein–protein interactions, including direct (physical) and indirect (functional) interactions (Jensen et al., 2009); the database contains information from: genomic context, experimental repositories, co-expression and public text collections (Szklarczyk et al., 2011).

Functional SNPs in UTR found by the UTRscan

The UTRscan program allows one to search the user-submitted sequences for any of the patterns collected in the UTR site (Grillo et al., 2010). If different sequences for each UTR SNP are found to have different functional patterns, that the particular UTR SNP is predicted to have functional significance.

PolymiRTS database (version 3.0) for polymorphism in microRNA target site

PolymiRTS database was designed specifically for the analysis of non-coding SNPs namely 3′ UTR. The polymorphic microRNA target sites are classified into four classes according to Table 1 (Bhattacharya et al., 2013). PolymiRTS of ‘D’ may cause loss of normal repression; PolymiRTS of class ‘C’ may cause abnormal gene repression control. Therefore, these two classes of PolymiRTS are most likely to have functional impacts.

1000 Genomes Project

The 1000 Genomes Project (Consortium, 2010) is sequencing the entire genome of approximately 2500 individuals from different worldwide populations. The aim of the1000 Genomes Project is to determine most of the genetic variation that occurs at a population frequency greater than 1%.

Results

SNP dataset from dbSNP

The BUB1B gene investigated in this work was retrieved from dbSNP database (Table 4). It contained a total of 827 SNPs: 90 were non-synonymous SNPs (nsSNPs), 57 were in non-coding regions, which comprises of 21 SNPs in 5′ UTR region and 36 SNPs in 3′ UTR region. The rest were in the intron region. We selected non-synonymous coding SNPs, 5′ and 3′ UTR region SNPs for our investigation.

Table 4

Surface accessibility of wild-type and mutant variants in BUB1B.

Amino acid	Class assignment	Position	Relative surface accessibility	Absolute surface accessibility	Z-fit score for RSA prediction
I	Buried	82	0.02	3.829	1.418
N	Buried		0.022	3.177	1.383
P	Exposed	334	0.546	77.520	− 1.915
L			0.349	63.975	− 0.779
R	Exposed	814	0.436	99.798	− 0.340
H	Exposed		0.423	76.998	− 0.280

Prediction of functional mutations

Of the 90 nsSNPs used in our analysis, 18 nsSNPs were identified to be deleterious with SIFT and the results were listed in Table 2.

Table 2

List of nsSNP analysis by SIFT, PolyPhen-2, I-Mutant 3.0 PROVEAN respectively.

rsID	Amino acid change	SIFT	Score	PolyPhen-2	Score	I-Mutant 3.0	Score	PROVEAN	Score
rs38678332	G37V	Tolerated	0.17	Benign	0.181	Large decrease	− 0.46	Neutral	− 1.187
rs52798733	E641V	Damaging	0.03	Probably damaging	0.981	Large decrease	− 0.04	Neutral	− 2.256
rs53178613	R256K	Tolerated	0.35	Probably damaging	0.992	Large decrease	− 0.52	Neutral	− 0.978
rs53231959	K170E	Damaging	0.05	Possibly damaging	0.897	Neutral	− 0.53	Deleterious	− 3.155
rs53396744	I272N	Damaging	0.02	Probably damaging	0.980	Large decrease	− 1.75	Deleterious	− 3.292
rs53429711	R36Q	Damaging	0	Probably damaging	1.000	Neutral	− 0.65	Deleterious	− 3.718
rs54188126	I625M	Tolerated	0.11	Probably damaging	0.980	Large decrease	− 1.71	Neutral	− 0.454
rs54578440	A348V	Tolerated	0.25	Probably damaging	1.000	Large decrease	0.01	Neutral	− 1.909
rs54653854	I156V	Tolerated	0.93	Probably damaging	0.966	Large decrease	− 1.01	Neutral	− 0.489
rs54660763	H836Y	Tolerated	1	Benign	0.004	Neutral	0.44	Neutral	− 2.169
rs54865001	M353T	Tolerated	0.48	Probably damaging	0.992	Large decrease	− 0.62	Neutral	− 2.259
rs55238070	A173V	Tolerated	0.27	Benign	0.034	Large decrease	− 0.45	Neutral	− 0.601
rs55342059	K539Q	Tolerated	0.24	Possibly damaging	0.682	Large decrease	− 0.79	Neutral	− 0.111
rs55355571	I82N	Damaging	0	Probably damaging	1.000	Large decrease	− 2.12	Deleterious	− 6.284
rs55478232	G8A	Tolerated	0.09	Benign	0.018	Large increase	− 0.11	Neutral	− 0.595
rs55619315	R616H	Tolerated	0.52	Benign	0.003	Large decrease	− 1.29	Neutral	1.972
rs55752197	R677H	Tolerated	0.12	Benign	0.002	Large decrease	− 1.39	Neutral	− 0.882
rs55935830	D576E	Tolerated	1	Benign	0.003	Neutral	− 0.14	Neutral	0.056
rs56791614	R194Q	Tolerated	0.2	Probably damaging	1.000	Neutral	− 0. 45	Neutral	− 0.742
rs57105655	T100M	Tolerated	0.06	Probably damaging	0.998	Large increase	− 0.16	Neutral	− 1.801
rs57153880	G316D	Tolerated	0.37	Benign	0.024	Large decrease	− 0.76	Deleterious	− 3.017
rs57759191	R727C	Tolerated	0.09	Probably damaging	1.000	Large decrease	− 0.69	Neutral	− 2.061
rs28989181	L844F	Damaging	0	Probably damaging	0.998	Large decrease	− 0.82	Neutral	2.454
rs28989182	R814H	Damaging	0	Probably damaging	1.000	Large decrease	− 1.43	Deleterious	− 2.880
rs28989187	R550Q	Tolerated	0.86	Benign	0.001	Large decrease	− 0.78	Neutral	0.332
rs56079734	T40M	Damaging	0.05	Probably damaging	1.000	Neutral	0.01	Neutral	− 1.811
rs1017842	E390D	Tolerated	0.46	Benign	0.102	Large decrease	− 0.31	Neutral	− 0.530
rs1801528	V618A	Tolerated	1	Benign	0.000	Large decrease	− 1.43	Neutral	1.441
rs17851677	P378S	Damaging	0.04	Possibly damaging	0.804	Large decrease	− 1.10	Deleterious	− 3.228
rs28989188	E409D	Tolerated	0.24	Probably damaging	1.000	Large decrease	− 0.49	Neutral	− 1.355
rs35923791	N133S	Tolerated	1	Benign	0.248	Large decrease	− 0.35	Neutral	− 0.664
rs56158360	R244H	Damaging	0.02	Probably damaging	1.000	Large decrease	− 1.08	Deleterious	− 4.461
rs75763304	Q460K	Tolerated	0.72	Benign	0.072	Neutral	0.13	Neutral	− 1.236
rs76546181	F531S	Tolerated	0.56	Probably damaging	1.000	Large decrease	− 1.68	Neutral	− 2.074
rs77520855	Y162H	Damaging	0.04	Probably damaging	0.960	Large decrease	− 1.09	Neutral	− 2.414
rs117485407	T471M	Tolerated	0.1	Possibly damaging	0.579	Neutral	0.32	Neutral	− 0.615
rs138332995	P544S	Tolerated	0.74	Benign	0.181	Large decrease	− 1.59	Neutral	− 1.109
rs139226455	P800S	Tolerated	0.07	Possibly damaging	0.839	Large decrease	− 1.85	Deleterious	− 3.780
rs140368608	K779R	Tolerated	0.64	Benign	0.073	Neutral	− 0.30	Neutral	− 1.176
rs141953425	P334L	Damaging	0.01	Possibly damaging	0.453	Large decrease	− 0.80	Deleterious	− 5.244
rs142705245	A784V	Tolerated	0.38	Benign	0.019	Large decrease	− 0.20	Neutral	0.604
rs143346774	H850R	Tolerated	0.61	Probably damaging	0.998	Neutral	0.16	Neutral	− 1.165
rs143559902	D675E	Tolerated	1	Benign	0.003	Neutral	− 0.37	Neutral	− 0.008
rs145026343	C825F	Tolerated	0.35	Probably damaging	0.966	Large decrease	− 0.13	Deleterious	− 4.601
rs145028054	E184Q	Damaging	0.01	Benign	0.362	Neutral	0.07	Neutral	− 2.086
rs145184714	A335T	Tolerated	0.84	Benign	0.005	Large decrease	− 0.78	Neutral	0.308
rs145578529	I567V	Tolerated	0.3	Possibly damaging	0.512	Large decrease	− 1.10	Neutral	− 0.081
rs146387899	L258F	Tolerated	0.81	Benign	0.074	Large decrease	− 0.64	Neutral	− 1.786
rs146795655	T493I	Tolerated	0.16	Benign	0.001	Neutral	− 0.26	Neutral	− 0.913
rs146821149	R886S	Damaging	0.03	Benign	0.002	Large decrease	− 1.25	Neutral	− 0.328
rs147150527	G376V	Tolerated	0.17	Benign	0.181	Large decrease	− 0.46	Neutral	− 1.187
rs147549987	V4M	Tolerated	0.23	Benign	0.000	Neutral	− 0.68	Neutral	− 0.282
rs147832586	S83G	Damaging	0.01	Benign	0.061	Large decrease	− 0.86	Neutral	− 1.807
rs148159407	N26D	Tolerated	0.52	Possibly damaging	0.913	Neutral	− 0.43	Deleterious	− 2.827
rs148348158	T648I	Tolerated	0.19	Benign	0.000	Large decrease	0.25	Neutral	− 1.427
rs149628229	D579G	Tolerated	0.1	Benign	0.328	Large decrease	− 1.20	Neutral	− 1.894
rs149955447	E813A	Tolerated	0.07	Probably damaging	0.997	Large decrease	− 0.69	Neutral	− 2.258
rs150707631	S797A	Tolerated	0.13	Possibly damaging	0.495	Large decrease	− 0.79	Neutral	− 1.043
rs150983783	R421Q	Tolerated	0.15	Benign	0.178	Neutral	− 0.79	Neutral	− 1.197
rs181352808	H836Q	Tolerated	0.29	Probably damaging	0.959	Neutral	− 0.22	Deleterious	− 2.528
rs184449375	M626V	Tolerated	0.61	Benign	0.000	Large decrease	− 0.84	Neutral	0.072
rs190909040	Y343F	Tolerated	0.14	Possibly damaging	0.925	Large decrease	− 0.13	Neutral	− 1.814
rs199509124	P222L	Damaging	0.01	Possibly damaging	0.774	Large increase	− 0.20	Deleterious	− 4.125
rs199743655	V274A	Damaging	0	Possibly damaging	0.866	Large decrease	− 1.00	Deleterious	− 3.319
rs200060772	S691L	Tolerated	0.08	Probably damaging	0.999	Neutral	− 0.32	Neutral	− 2.173
rs200788206	Q350K	Tolerated	0.79	Probably damaging	0.984	Large decrease	0.10	Neutral	− 1.593
rs200997833	K542R	Tolerated	0.35	Benign	0.055	Neutral	− 0.28	Neutral	− 0.375
rs201251790	R421W	Tolerated	0.09	Benign	0.021	Neutral	− 0.38	Neutral	− 1.629
rs201360106	E21K	Tolerated	0.09	Probably damaging	1.000	Large decrease	− 0.63	Deleterious	− 2.504
rs202114756	S384G	Tolerated	0.31	Possibly damaging	0.860	Large decrease	− 0.57	Neutral	− 1.503
rs202132335	A739G	Damaging	0.0	Benign	0.072	Large decrease	− 1.08	Neutral	− 1.113
rs367543489	Q829E	Tolerated	0.11	Probably damaging	0.999	Neutral	− 0.10	Neutral	− 1.197
rs368023159	K488N	Tolerated	0.15	Benign	0.004	Neutral	− 0.60	Neutral	− 0.119
rs368079817	Q42R	Tolerated	0.25	Probably damaging	0.999	Neutral	− 0.06	Neutral	− 1.985
rs368996088	F781L	Tolerated	0.16	Benign	0.075	Large decrease	− 1.32	Neutral	− 1.964
rs370388424	P640L	Tolerated	0.29	Benign	0.020	Large decrease	− 0.60	Neutral	− 0.985
rs370655726	C356S	Tolerated	0.74	Probably damaging	1.000	Large decrease	− 0.69	Deleterious	− 3.683
rs371124423	C700R	Tolerated	0.38	Benign	0.068	Large decrease	− 0.26	Neutral	− 1.228
rs371305662	T291K	Tolerated	0.84	Benign	0.100	Large decrease	− 0.46	Neutral	− 1.934
rs372003254	D846E	Tolerated	0.1	Possibly damaging	0.626	Neutral	0.24	Neutral	− 1.695
rs372569297	I755T	Tolerated	0.75	Benign	0.000	Large decrease	− 1.59	Neutral	− 0.814
rs373256667	K454R	Tolerated	0.21	Probably damaging	1.000	Neutral	0.12	Neutral	− 0.935
rs373789523	T658I	Tolerated	0.16	Benign	0.000	Large decrease	0.25	Neutral	− 1.951
rs373830262	A108T	Tolerated	0.49	Probably damaging	0.982	Large decrease	− 0.71	Neutral	− 0.949
rs374682772	V333I	Tolerated	0.35	Benign	0.002	Large decrease	− 0.84	Neutral	− 0.423
rs375105548	I854V	Tolerated	0.44	Benign	0.007	Large decrease	− 0.65	Neutral	− 0.137
rs375388175	I703T	Tolerated	0.76	Probably damaging	0.985	Large decrease	− 1.94	Neutral	− 1.952
rs375798678	Q181R	Tolerated	1	Benign	0.000	Neutral	0.06	Neutral	0.416
rs375885859	C51R	Tolerated	0.11	Benign	0.000	Neutral	0.07	Neutral	− 1.379
rs376072541	P632L	Tolerated	0.1	Benign	0.001	Large decrease	− 0.40	Neutral	− 2.406

A total of 46 nsSNPs was predicted to be damaging and the remaining 44 nsSNPs were categorized as benign with Polyphen 2.0 and the results were listed inTable 2. Out of 90 nsSNPs, 27 nsSNP were predicted to be neutral mutation (− 0.5 ≤ DDG ≤ 0.5 kcal/mol), 60 nsSNP were predicted to be “large decrease” (≤−0.5 kcal/mol) and 3 nsSNP were predicted to be “large increase” (> 0.5 kcal/mol). I-Mutant 3.0 predicted 63 of SNPs to affect the stability of protein structure (Table 2). All the nsSNPs submitted to SIFT and PolyPhen 2.0 and I-Mutant 3.0 were submitted as input to PROVEAN. Out of 90 nsSNPs, 17 nsSNP were predicted to be deleterious and 73 were found to be neutral (Table 2).

The accuracy of the in silico techniques for prioritizing deleterious SNPs can be increased by combining different computational methods. Out of 90 nsSNPs, SIFT, PolyPhen, I-Mutant 3.0 and PROVEAN predicted 8 nsSNPs as deleterious (Fig. 2).

Fig. 2

List of functionally significant mutations.

SNPs&GO, PhD-SNP and PANTHER were performed to validate the results obtained from four tools. Out of 8 nsSNPs that predicted to be deleterious with SIFT, Polyphen, I-Mutant and PROVEAN; SNP&GO predicted 3 nsSNP, PhD-SNP predicted 4 nsSNP and PANTHER predicted 5 nsSNP to be associated with disease (Table 3).

Table 3

List of nsSNP predicted as disease associated byPHD-SNP,SNP&GO and PANTHER server.

rsID	Amino acid change	PHD-SNP	SNP&GO	PANTHER	subPSEC score
rs527987333	I82N	Disease	Disease	Deleterious	− 5.89613
rs53396744	I272N	Disease	Neutral	Tolerated	− 2.75316
rs199743655	V274A	Neutral	Neutral	Tolerated	− 2.50206
rs199509124	P222L	Neutral	Neutral	Deleterious	− 3.07979
rs141953425	P334L	Disease	Disease	Deleterious	− 3.73233
rs56158360	R244H	Neutral	Neutral	Deleterious	− 4.34981
rs17851677	P378S	Neutral	Neutral	Deleterious	− 3.97828
rs28989182	R814H	Disease	Disease	Deleterious	− 7.86508

Finally out of 90 nsSNP, we found 3 nsSNPs namely rs55355571 (I82N), rs141953425 (P334L) and rs28989182 (R814H) that are common in all (SIFT, Polyphen, I-Mutant, PROVEAN, PHD-SNP, SNP&GO, PANTHER) prediction.

In silico biophysical validation of nsSNPs

Based on the in silico analyses performed, 3 nsSNPs was selected for further analyses. The location and the type of a mutated residue affect the stability of the protein. In particular, as the solvent accessibility of a residue decreases, the stability of the protein due to mutation decreases. NetSurfP Z-score allows the identification of the most reliable/unreliable predictions for both buried and exposed amino acids. A huge drift in the Z-score was not observed for 3 nsSNPs as given in Table 5. For any of 3 nsSNPs, the class assignment does not change.

Table 5

Top 10 templates used by I-TASSER to create the high quality models for human BUB1B secondary structure.

Rank	PDB hit	Iden1	Iden2	Cov.	Norm. Z-score
1	3e7eA	0.22	0.08	0.31	1.71
2	1vw1A	0.08	0.20	0.87	2.36
3	3e7eA	0.22	0.08	0.31	1.40
4	4jspB	0.09	0.19	0.92	1.44
5	3e7eA	0.23	0.08	0.30	2.68
6	3cm9S	0.07	0.11	0.55	1.19
7	3e7eA	0.22	0.08	0.31	2.84
8	4jspB	0.08	0.19	0.89	2.06
9	3e7eA	0.22	0.08	0.31	1.51
10	4kf7A	0.10	0.19	0.85	2.31

The I-TASSER tool created the 5 full-length models for BubR1 protein (with C-scores: − 0.24, − 1.01, − 2.83, − 2.95 and − 3.00) by excising top 10 structures with C-scores after targeting the PDB library hits (Table 5).Top 10 proteins in the PDB which are structurally closest to the predicted models. Among the 5 predicted models, model 1 (Fig. 3) carried the high-quality confidence in the form of C-score (− 0.24), TM-score (0.68 ± 0.12), and the RMSD (9.6 ± 4.6 Å) (Table 6). We did not perform any molecular dynamics structure optimization; therefore, our 3D homology model is a preliminary model implicating the disruptive role of the SNPs.

Fig. 3

3D structure of BubR1 predicted with I-TASSER.

Table 6

I-TASSER results carrying C-score, TM-score and RMSD regarding selected secondary structure (native protein model 1).

Model	C-score	Exp. TM-score	Exp. RMSD	No. of decoys	Cluster density
Model 1	− 0.24	0.68 ± 0.12	9.6 ± 4.6 Å	312	0.3039
Model2	− 1.01			258	0.1404
Model3	− 2.83			60	0.0228
Model4	− 2.95			57	0.0202
Model5	− 3.00			57	0.0191

Project Hope revealed the 3D structure of the proteins with its new residue. Furthermore, it described the reaction and physiochemical properties of these candidates. Here we present the results upon each candidate and discuss the conformational variations and interactions with the neighboring amino acids:

A/G Mutation (rs28989182) caused substitution of the amino acid from Arginine into a Histidine at position 814 (R814H). For this variant the mutated residue is smaller (Fig. 1); this might lead to loss of interactions. The wild-type residue was positively charged, the mutated residue is neutral. Only this residue type was found at this position. Mutation of a 100% conserved residue is usually damaging for the protein. Additionally, the structural analysis of H814 showed some clashes for Phe822 which may contribute to the extra energy in the protein structure, and hence the decrease in stability (Fig. 4).

Fig. 4

Deep view of superimposed structure of wild and mutant residue at 814 position. The main protein core is shown in gray color while the wild type and mutated residues are shown in green and red colors respectively. SNP ID: rs28989182, protein position 814 changed from Arginine to Histidine.

A/T mutation (rs553555716) resulted in a change of the Isoleucine to Aspargine at position 82 (I82N). The wild type residue is smaller and more hydrophobic than the mutated residue (Fig. 6). The residue is buried in the core of a domain, annotated in UniProt as: “BUB1 N terminal”. The mutation will cause loss of hydrophobic interactions in the core of the protein. This residue is part of an interprotein domain named “Mitotic checkpoint serine/threonine protein kinase Bub1/Mitotic spindle checkpoint component Mad3” (IPR015661).

Fig. 6

Superimposed structure of (wild type) Ile and (mutant) Asn residues at 82 position.

C/T (rs141953425) leads to conversion of Proline into a Leucine at position 334 (P334L). The wild-type and mutant amino acid differs in sizes; the mutated residue is bigger and this might lead to displace of the mutant residue. Prolines are known to have a very rigid structure, sometimes forcing the backbone in a specific conformation. The mutation can disturb this special conformation. This residue is part of an inter protein domain named “Mitotic checkpoint serine/threonine protein kinase Bub1/Mitotic spindle checkpoint component Mad3”.

Chimera (Fig. 5, Fig. 7, Fig. 8) and Swiss PDB viewer were used to visualize the structural features of amino acids in native and mutant protein chains. During structural visualization for all 3mutations, only mutant residue (Histidine) at 814 position showed a network of clashes with Phe822 (Fig. 9).

Fig. 5

SNP ID: rs28989182, protein position 814 changed from Arginine (green) to Histidine (red).

Fig. 7

SNP ID: rs553555716, protein position 82 changed from Isoleucine (left image) to Aspargine (right image).

Fig. 8

SNP ID: rs141953425, protein position 334 changed from Proline (left image) to Leucine (right image).

Fig. 9

H-bonding (green discontinuous line) interactions and clashes (pink discontinuous line) of wild type and mutant analogues with the vicinal amino acid residues. (a) At 814 position 2 H-bond is observed with Leu811and Glu817 in both native (Arg) and mutant (His814) structures, but a network of clashes appeared between His814and Phe822. (b) At 82 position, 4 H-bond is observed with Trp78, Asp79, Thr85 and Glu86 in both native (Ile) and mutant (Asn) structures. (c) At 334 position, 2 H-bond is observed with Leu330 and Pro338 in both native (Pro) and mutant (Leu) structures.

Protein-protein interactions analysis

The interaction analysis revealed that BUB1B is related to Cell Division Cycle 20 homolog (CDC20), Budding Uninhibited By Benzimidazoles 3 (BUB3), Cancer Susceptibility Candidate 5 (CASC5), MAD2 Mitotic Arrest Deficient-Like 1 (MAD2L1), Cell Division Cycle 27 homolog (CDC27), Centromere Protein E (CENPE), BUB1 Mitotic Checkpoint Serine/Threonine Kinase (BUB1), ZW10 Interacting Kinetochore Protein (ZWINT), Anaphase Promoting Complex Subunit 2 (ANAPC2), Cell Division Cycle 16 homolog (CDC16) (Fig. 10). Furthermore, our literature search demonstrated that BubR1 interacts with Bub3, Cdc20, and Mad2 (Kapanidou et al., 2015).

Fig. 10

Protein–protein interaction network of BUB1B using STRING 9.0 server.

Functional SNPs in UTR found by UTRscan server

All of the 57 UTR SNPs was analyzed using UTRscan. After comparing the functional elements for each UTR SNP, we predicted that two SNPs, namely rs375434078 and rs538302864 in 5′ UTR are related to the functional pattern change of IRES (Table 7). Internal ribosome entry site (IRES) is bound by an internal mRNA ribosome that is an alternative mechanism of translation initiation compared to the common 5′-cap dependent ribosome scanning mechanism (Pickering and Willis, 2005).

Table 7

List of mRNA UTR SNPs that were predicted to be of functional significance by UTRscan server.

SNP ID	Nucleotide change	UTR position	Functional element change
rs375434078	C/T	5′ UTR	IRES → no pattern
rs538302864	A/G	5′ UTR	IRES → no pattern

Functional SNPs in 3′ untranslated regions (UTR) predicted by PolymiRTS database 3.0

Among 36 SNPs in 3′ UTR region of BUB1B gene, 3 functional SNPs was predicted that among them, only one SNP disrupts 8 miRNAs conserved site (ancestral allele with support ≥ 2), while all of them create 8 new miRNA site. The results are listed in Table 8.

Table 8

Prediction result of PolymiRTS database.

dbSNP ID	miR ID	Conservation	miRSite	Function class
rs149437374	hsa-miR 130a -3p	2	ATGCACTAccatt	D
hsa-miR-130b-3p
hsa-miR-301a-3p	2	ATGCACTAccatt	D
hsa-miR-301b
hsa-miR-3666	2	ATGCACTAccatt	D
hsa-miR-4295
hsa-miR-454-3p	2	ATGCACTAccatt	D
hsa-miR-4671-3p	2	ATGCACTAccatt	D
hsa-miR-323a-5p	2	ATGCACTAccatt	D
hsa-miR-876-3p	2	ATGCACTAccatt	D
	2	ATGCACTAccatt	D
	5	atgcACCACCAtt	C
	5	atgcACCACCAtt	C
rs143807849	hsa-miR-539-5p	4	CcATTTCTCtcta	C
hsa-miR-5680
hsa-miR-6758-5p	5	CCATTTCTctcta	C
hsa-miR-6856-5p	4	CcatttCTCTCTA	C
	4	CcatttCTCTCTA	C
rs1047193	hsa-miR-4450	4	atgATCCCCAtgt	C
hsa-miR-6857-5p	4	atgATCCCCAtgt	C

Discussion

The identification of SNPs responsible for specific phenotypes with molecular approaches seems to be expensive and time-consuming (Chen and Sullivan, 2003), hence computational approaches can help in narrowing down the number of missense mutations to be screened in genetic association studies and for a better understanding of the functional and structural aspects of the parent protein.

Previous studies on polymorphisms screening using in silico analysis helped in predicting the functional nsSNPs associated with genes such as G6PD (Rajith, 2011), ATM (Doss and Rajith, 2012), PTEN (Doss and Rajith, 2013), BRAF (Hussain et al., 2012). Our results also revealed that implementations of different algorithms often serve as powerful tools for prioritizing candidate functional nsSNPs. Recent work by Thusberg and Vihinen (2009) compared different in silico tools, out of which SIFT and PolyPhen were reported to have better performance in identifying deleterious nsSNPs. The accuracy of SIFT and PolyPhen 2.0 was further validated by Hicks et al. (2011), which makes these tools more applicable for the prediction. I-Mutant 3.0 was used which evaluate the stability change upon single amino acid mutation that ranked as one of the most reliable predictor based on the work performed by Khan and Vihinen (2010).

Based on these in silico studies, we select SIFT, PolyPhen, I-Mutant, PROVEAN, SNP&GO, PHD-SNP and PANTHER for the screening of functional mutation in BUB1B gene. By comparing the scores of all 7 methods, 3 nsSNPs with positions I82N, P334L and R814H were found to be highly significant.

The 5′ and 3′ UTR SNPs was analyzed using UTRscan. Due to the importance of the translational regulation of microRNAs, we further studied whether the 3′ UTR SNPs changes the profile of microRNA binding to the BUB1B gene using PolymiRTS. Two SNPs in the 5′ UTR was predicted to influence the translation pattern of the BUB1B gene through UTRscan analysis, and three 3′ UTR SNPs may affect microRNA binding sites, as determined through PolymiRTS. Protein-protein interaction analysis showed the interaction of BUB1B with ten different genes. Therefore, any changes in the protein function would have an impact on many pathways involved in disease.

In conclusion, we surveyed and compared available databases such as NCBI, dbSNP, 1000 genome project along with in silico prediction programs to assess the effects of deleterious functional variants on the protein functions. Analyzing deleterious nsSNPs by both sequence and structure level has the added advantage of being able to assess the reliability of the generated prediction results by cross-referencing the results from both approaches. One striking observation was the identification of rs28989182 (R814H), that associated with Mosaic Variegated Aneuploidy Syndrome (Bairoch and Apweiler, 1996), lie within a serine/threonine kinase domain of BubR1 protein. Only this residue type was found at this position. Mutation of a 100% conserved residue is usually damaging for the protein. Both I82N and P334L mutations occurred in the N terminal region of BubR1; Therefore, these mutations may compromise its binding to Bub1, Mad2 and cdc20 resulting plausible failure of the corresponding checkpoint. In addition rs149437374 and rs143807849 in 3′ UTR that disrupts a conserved of 8 miRNAs site are genotyped by 1000 genome project; Based on the data obtained through determining the allele frequency in 1000 genome populations, it is observed that the frequency of normal allele is more than the mutant allele. Therefore, it is concluded that rs149437374 and rs143807849 in 3′ UTR are deleterious, so that in different 1000 genome population, it has a low frequency; hence allele frequency reported in 1000 genome project confirmed our results. These results indicate that our approach successfully allowed us in selecting the deleterious SNPs that are likely to have functional impact on the BUB1B gene and contribute to an individual's susceptibility to the disease.