The mitochondrial tumor suppressor 1 (MTUS1) gene acts as a crucial tumor suppressor by inhibiting growth and proliferation of eukaryotic cells including tumor cell lines. Down regulation of MTUS1 gene has been implicated in a wide range of cancers as well as various human diseases. Alteration through nsSNPs can potentially damage the structure and/or function of the protein. As characterization of functional SNPs in such disease linked genes is a major challenge, it is feasible to analyze putative functional SNPs prior to performing larger population studies. Hence, in this in silico study we differentiated the potentially harmful nsSNPs of the MTUS1 gene from the neutral ones by using various sequence and structure based bioinformatic tools. In a total of 215 nsSNPs, 9 were found to be most likely to exert deleterious effect using 7 prediction tools. From which, 5nsSNPs (S1259L, E960K, P503T, L1084V and L1143Q) were selected as potentially damaging due to their presence in the highly conserved region and ability to decrease protein stability. In fact, 2 nsSNPs (S1259L and E960K) among these 5 were found to be individually associated with two distinctive cancers named Stomach adenocarcinoma and Uterine corpus endometrial carcinoma. As this is the first comprehensive study analyzing the functional nsSNPs of MTUS1, the results of the current study would certainly be helpful in future prospects concerning large population-based studies as well as drug discovery, especially developing individualized medicine.
Citation: Rozario LT, Sharker T, Nila TA (2021) In silico analysis of deleterious SNPs of human MTUS1 gene and their impacts on subsequent protein structure and function. PLoS ONE 16(6): e0252932. https://doi.org/10.1371/journal.pone.0252932
Editor: Junwen Wang, Mayo Clinic Arizona, UNITED STATES
Received: December 17, 2020; Accepted: May 25, 2021; Published: June 14, 2021
Copyright: © 2021 Rozario et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: The author(s) received no specific funding for this work.
Competing interests: The authors have declared that no competing interests exist.
Cancer is known as one of the most dreadful and enigmatic diseases in today’s world. Continuous researches are going on to find out new approaches to combat cancers. Among these, identification and characterization of tumor suppressor genes play a crucial role as the loss of tumor suppressor activity is an important benchmark of cancer .
Our study focuses on MTUS1 gene that has been reported as a tumor suppressor gene at a chromosomal position 8p22. This gene encodes a family of microtubule-associated proteins named Microtubule-Associated Scaffold Protein 1(MTUS1, previously known as mitochondrial tumor suppressor 1 and microtubule associated tumor suppressor 1) which interacts with angiotensin II type 2 (AT2) receptor. Dispersing over 112 kb on the chromosome, the gene incorporates 17 exons, alternative splicing of which gives onto three major transcripts entitled as ATIP1, ATIP3 and ATIP4. Among them the transcripts of ATIP3 entail three variants- ATIP2, ATIP3a and ATIP3b [2,3]. These three transcripts of ATIP (Angiotensin II receptor-interacting proteins) show diverse tissue distribution along with subcellular compartmentalization, signifying their involvement in various cellular mechanisms . The interaction of ATIP1 with AT2 receptors induces apoptosis and inhibits cell proliferation as well as involves in cell division and migrations . ATIP3 controls critical steps of mitotic cell division reflecting its roles in cell cycle regulation .
Furthermore, MTUS1 gene has been reported to be underexpressed in various types of malignancies including ovarian, breast, head-and-neck, pancreas, colon, bladder and lung cancers [3,5]. The rationality for the lower expression of this gene in various cancers has been implied to mutations. Previous findings have revealed the correlation of mutation and reduced expression of MTUS1 gene with hepatocellular and squamous cell carcinoma of the tongue accordingly [6,7]. Ovarian carcinoma survival rate also increases after the restoration of ATIP3 which prevents tumor growth via down-regulation of the ERK/EMT pathway .
Among different types of mutations single nucleotide polymorphisms (SNPs) account 90% genetic variations in human. SNPs change single base pair in alleles which are the most common form of disparities in DNA sequence . SNPs that alter the encoded amino acids are referred to as nonsynonymous SNPs (nsSNPs), which can influence subsequent structure and/or function of the protein with either neutral or deleterious effects. These nsSNPs are accountable for about half of all genetic variations related to human diseases [10,11].
As human genome contain vast number of genetic polymorphisms, massive investigations are indispensable to explore the significance of each of them along with their association to disease susceptibilities as well as individualized drug designing . However, to lessen this enormous effort numerous computational based methods have been developed to identify prospective and potentially significant variants before testing in vitro or in vivo conditions. In this situation in silico approach is a convenient way to distinguish the deleterious SNPs from neutral ones using particular algorithms. The overall effect of polymorphism including functional and structural alterations can also be analyzed through several databases.
In present study, we use public datasets and freely accessible bioinformatics tools to identify the most deleterious nsSNPs of the MTUS1 gene and to analyze consequent protein structure and functions. Here, our focus is not only establishing the structural configuration but also finding the association of different cancers with particular nsSNPs. As no effort has been noticed till date, it is the first comprehensive study to analyze the SNPs of the MTUS1 gene systematically which will be helpful in future extensive studies in this regard.
Retrieval of SNPs
Data related to human MTUS1 gene and its protein sequence (FASTA format) was collected from NCBI (https://www.ncbi.nlm.nih.gov/) and UniProtKB (http://www.uniprot.org/uniprot/) respectively. SNPs located in MTUS1 gene were retrieved from dbSNP database (http://www.ncbi.nlm.nih.gov/SNP/) .
Identification and prediction of the effect of deleterious SNPs
To analyze the functional and structural consequences of deleterious SNPs of MTUS1 gene; SIFT, SNAP2, Align GVGD, PolyPhen-2, PROVEAN, PhD-SNP and PANTHER were used sequentially.
SIFT (Sorting Intolerant from Tolerant) (https://sift.bii.a-star.edu.sg/) determines the deleterious (probability score <0.05) and tolerated SNPs (probability score ≥ 0.05) based on sequence homology. This prediction helps to analyze the effect of amino acid variation on the phenotypic and functional changes upon protein molecule. The rsIDs retrieved from dbSNP were used as a input query for this server .
SNAP2 (Screening for Nonacceptable Polymorphisms) (https://www.rostlab.org/services/SNAP/) differentiates between effect and neutral variants by scrutinizing a variety of sequence and variant features. For SNAP2, FASTA format of protein sequences was used as input query. The result was obtained as a score, ranges from −100 (strong neutral prediction) to +100 (strong effect prediction) that reveals the likelihood of specific mutation to alter the native protein function with expected accuracy .
Align GVGD (http://agvgd.hci.utah.edu/) web server predicts whether missense substitution is deleterious or neutral. This prediction emphasizes on protein multiple sequence alignments and biophysical characteristics of amino acids. It forms a spectrum of classified variants (C0, C15, C25, C35, C45, C55, C65) where C65 is most likely to interfere with function and C15 is less likely to affect . Here, FASTA sequence of protein and amino acid substitution were used as input file.
For exploring the possible effect of an amino acid substitution on the structure and function of protein, PolyPhen-2 (Polymorphism Phenotyping v2) (http://genetics.bwh.harvard.edu/pph2/) was used. As input query for PolyPhen-2; protein sequence, database ID/ accession number and details of amino acids substitution were given to the server. The score of PolyPhen-2 ranges from 0.0 to 1.0 that indicates the particular amino acid substitution as tolerated or deleterious respectively .
By assessing the single amino acid substitutions PROVEAN (Protein variation effect analyzer) (http://provean.jcvi.org/index.php) predicts the functional impact of protein sequence variations as ‘deleterious’ or ‘neutral’. FASTA format sequence with substitutions predicted by the SIFT server was used as an input. If the PROVEAN score is ≤ 2.5 the protein variant is predicted to have a deleterious effect, otherwise the variant is predicted to have a neutral effect .
PANTHER cSNP (Protein analysis through evolutionary relationship-codingSNP) (http://pantherdb.org/tools/csnpScoreForm.jsp) classification system is established on evolutionary relationship, molecular functions and their interactions with other proteins. This tool gives position specific evolutionary conservation (PSEC) scores by estimating alignment of evolutionarily related various proteins. Plain protein sequence, amino acid variants and human organism were used as input query for this prediction .
PhD-SNP (Predictor of human deleterious single nucleotide polymorphism) (https://snps.biofold.org/phd-snp/phd-snp.html) classifies single point protein mutation as disease-related or as neutral polymorphism based on Support Vector Machine (SVM) method. The required input query was protein sequence, position of mutation and mutated residue .
Prediction of protein stability change by I-Mutant Suite
I-Mutant Suite (http://gpcr2.biocomp.unibo.it/cgi/predictors/I-Mutant3.0/I-Mutant3.0.cgi) predicts the changes in protein stability due to single point mutation. Both protein sequence and protein structure can be used as input query. Upon mutation this tool calculates the sign of the protein stability changes and related Delta Delta G values .
Estimation of conservation profile by ConSurf
Based on the phylogenetic relations between homologous sequences, the ConSurf (https://consurf.tau.ac.il/) server calculates the evolutionary conservation of amino acid positions in a protein. The evolutionary rate of an amino (or nucleic) acid position is strongly dependent on its structural and functional importance and is computed by using either an empirical Bayesian method or a maximum likelihood (ML) method. It uses a color scheme to present conservation score from 1 to 9 which is classified into variable, average and highly conserved. FASTA format of protein sequence was used as input query .
Prediction of solvent accessibility by NetsurfP-2.0
NetSurfP-2.0 (http://www.cbs.dtu.dk/services/NetSurfP/) server is used to predict the surface accessibility, secondary structure, disorder, and phi/psi dihedral angles of amino acids in an amino acid sequence. The FASTA sequence of MTUS1 protein was submitted to NetSurfP and the output of this server revealed buried and exposed region in protein structure .
Analysis of structural effect of nsSNPs
HOPE (Have (y) Our Protein Explained) (https://www3.cmbi.umcn.nl/hope/) assumes the effects of the mutation on the protein structure and the corresponding function. It produces a report, complete with results, figures and animations by collecting and combining available information from a series of web services and databases. Protein sequence and mutation were used as input query for HOPE .
Prediction of protein-protein interactions
STRING (Search Tool for the Retrieval of Interacting Genes/Proteins) (https://string-db.org/cgi/input?sessionId=croExd7fYyEe&input_page_active_form=single_sequence) database gives an important assessment and integration of protein-protein interaction as well as association from databases of physical interaction and databases of curated biological pathway knowledge. Protein sequence was used as input query for this tool .
Prediction and Evaluation of 3D structure of MTUS1 protein
I-TASSER (Iterative Threading ASSEmbly Refinement) (https://zhanglab.dcmb.med.umich.edu/I-TASSER/) is a hierarchical approach for protein structure prediction as well as a structure-based function annotation. It identifies structural templates from the PDB to provide the most appropriate protein structure and usually provides the top 5 models of the targeted protein. The FASTA sequence of MTUS1 was the input file for this server .
SWISS-MODEL (https://swissmodel.expasy.org/) is a fully automated protein structure homology-modeling server that uses the updated UniProtKB proteome for target-template alignment. As an input query FASTA sequence was used here. Not only the prediction of structures but also the validation of predicted structures can be estimated through the favored region of Ramachandran plot, QMEAN, and Molprobity score provided by this server .
PROCHECK (https://servicesn.mbi.ucla.edu/PROCHECK/) evaluates the stereochemical quality of a protein structure by analyzing residue-by-residue geometry and overall structural geometry . ERRAT (https://servicesn.mbi.ucla.edu/ERRAT/) validates overall model quality by the statistical relationship of non-bonded interactions between different types of atoms based on characteristic atomic interaction. In both cases, predicted models in PDB format was used as an input query .
ProSA-web (https://prosa.services.came.sbg.ac.at/prosa.php) is also widely used for the refinement and validation of the experimental protein structure. Predicted structures in PDB format was used for the input query for the estimation of the model quality of the protein .
Three-Dimensional modeling of the mutated protein
Point mutation at the particular position in the native protein sequence was made and SWISS-MODEL (https://swissmodel.expasy.org/) was used for the structural analysis of the mutated protein. The mutated model was then analyzed in TM-align tool (https://zhanglab.dcmb.med.umich.edu/TM-align/) that gives template modeling-score (TM score) and root mean square deviation (RMSD) value for the comparison of protein structures based on the superimposition of the structures to find the structural similarity. In this server, TM-score ranges from 0 to 1, where 1 denotes a perfect match between two structures. In detail, 0.0 < TM-score < 0.30 means random structural similarity, whereas 0.5 < TM-score < 1.00 means both structures are in the same fold .
Identification of cancer association with nsSNPs
cBioPortal (https://www.cbioportal.org/) is an open-access database which allows exploration, visualization and analyzation of multidimensional cancer genomics data. Distribution of MTUS1 gene mutations in the lollipop plot for mutation frequency in the server was searched to find association of the specific nsSNPs with cancer .
canSAR Black (https://cansarblack.icr.ac.uk/) is an integrative translational research and drug discovery web resource for oncology. The mutation profile for MTUS1 provided by this database was searched for finding specific cancers linked to deleterious nsSNPs .
Analysis of gene expression and overall survival rate
GEPIA (Gene Expression Profiling Interactive Analysis) (http://gepia.cancer-pku.cn/) is an interactive database that is used to analyze the RNA sequencing expression data from the Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) projects. It provides expression profiles of a given gene in dot plots or box plots and survival analysis by using log-rank test. For both analysis the name of the candidate gene (MTUS1) was used as input and specific cancer name (i.e., STAD, UCEC) was selected .
Analysis of correlation of gene expression and mutation
muTarget (https://www.mutarget.com/) is a tool based on the TCGA that correlates somatic mutations and gene expression in cancer. Correlations can be analyzed in two ways- The ‘Genotype’ run is for finding changes in gene expression that are related to a specific mutation and the ‘Target’ run is for finding mutations that alter the expression of target genes. The name of the candidate gene MTUS1 was used as input and the specific cancers associated with the significant nsSNPs were selected .
SNPs of the human MTUS1 gene were retrieved from dbSNP database (dbSNP NCBI: https://www.ncbi.nlm.nih.gov/snp/?term=chek2). It comprised a total of 55120 SNPs, out of which 1453 were missense (nsSNP), 3400 non-coding transcript, 558 synonymous, 52194 intronic, 9 initiator codon variant, 1 inframe insertion, 4 inframe indel and 20 inframe deletions (Fig 1). For this study only nsSNPs of MTUS1 were selected, which contributed to only 2.63% of all SNPs known in human MTUS1 gene.
Screening of functional nsSNPs in MTUS1
Single nucleotide variants of the MTUS1 attained from dbSNP analysis were subjected to in silico analysis through variety of tools such as SIFT, SNAP2, Align GVGD, Polyphen-2, PROVEAN, PANTHER and PhD-SNP.
For initial screening we used SIFT that predicted a total 215 nsSNPs as tolerated or deleterious out of 1453 whereas the rest of nsSNPs were not found. Among these 215 nsSNPs, SIFT classified 108 nsSNPs as deleterious, 107 as tolerated (S1 Table). To filter the SIFT result, we performed SNAP2, Align GVGD, PolyPhen-2, PROVEAN, PANTHER and PhD-SNP. According to SNAP2, 104 variants (48%) were significant while the rest (52%) showed no effect (S1 Table). Whereas Align GVGD predicted 194 SNPs as most likely affected and 21 nsSNPs as less likely affected out of 215 nsSNPs (S1 Table).
PolyPhen-2 speculated 67(31%) as probably damaging, 37(17%) as possibly damaging and the remaining (52%) as benign (S1 Table). PROVEAN suggested that, 178 amino acid substitutions (83%) were neutral (score is above-2.5) and the remaining 37 (17%) were diseases associated (score below or equal -2.5) (S1 Table). Interestingly all of the nsSNPs (215) were found to be probably damaging by PANTHER (S1 Table). In addition, out of the 215 nsSNPs of MTUS1 gene, PhD-SNP revealed only 20 (9%) nsSNPs as diseased and rest of 195 (81%) nsSNPs as neutral (S1 Table) (Fig 2).
By using 7 Prediction tools (SIFT, SNAP2, Align GVGD, Polyphen-2, PROVEAN, PANTHER and PhD-SNP), eventually 9 significant nsSNPs (E934V, C264Y, S1259L, E960K, L1084V, P503T, H1077L, S1245Y and L1143Q) were selected as most deleterious (Table 1). To investigate the impact of these nsSNPs on the structure and function of the MTUS1 protein additional analysis were performed.
Analysis of protein structural stability
The selected 9 nsSNPs were analyzed by I-mutant Suite to reveal the effect of point mutation on protein stability based on free energy change value. Out of 9, 6 variants (C264Y, S1259L, E960K, L1084V, P503T and L1143Q) were predicted to decrease stability whereas others (E934V, H1077L and S1245Y) were found to increase protein stability (Table 2).
Conservation profile of deleterious nsSNPs in MTUS1
Through the ConSurf web server, all the 9nsSNPs were analyzed to evaluate evolutionary conservation and find putative structural and functional residues. According to ConSurf output, out of 9nsSNPs 7 variants (E934V, S1259L, E960K, L1084V, P503T, S1245Yand L1143Q) were highly conserved residues (conservation score of 9) and remaining H1077L and C264Y were predicted moderately conserved (conservation score of 7) and variable (conservation score of 1) residue respectively. Among these 7 highly conserved residues, 5 variants (E934V, S1259L, E960K, P503T and S1245Y) were predicted as functional and exposed whereas remaining 2 (L1084V and L1143Q) were structural and buried. The result of ConSurf is shown in Table 2. The summary of deleterious prediction for each SNP by ConSurf is shown in S1–S3 Figs.
For the identification of the most deleterious nsSNPs, I-mutant and ConSurf output were compared and scrutinized. 5nsSNPs (S1259L, E960K and P503T, L1084V and L1143Q) were selected as potentially damaging based on this comparison which were subjected to further analysis.
Prediction of solvent accessibility
Solvent accessibility and stability were assessed for the 5 variants (S1259L, E960K, P503T, L1084V and L1143Q) by NetsurfP. Among these deleterious nsSNPs, 3 variants (S1259L, E960K P503T) and their respective wild variants were exposed to the surface whereas 2 variants (L1084V and L1143Q) and their respective wild variants were buried (Table 3).
The structural impacts of functional MTUS1 mutations
HOPE speculated the effect of amino acid substitution on the physical and chemical properties, hydrophobicity, spatial structure, and function of protein (Table 4). Based on HOPE, the mutant residue S1259L, E960K and L1143Q were bigger than the wild type whereas the mutant residue L1084V was smaller than wild type. In addition, the mutant S1259L increased hydrophobicity whereas the mutant E960K and L1143Q decreased hydrophobicity. And the differences of the hydrophobicity and size between wild-type and mutant residue can disrupt the H-bonding interactions with the neighboring residues, and hence the protein framework. Moreover, for the mutant residue P503T, the substitution of proline might harshly disrupt a special conformation of protein as well as decrease hydrophobicity (Fig 3).
Schematic structures of the original (left) and the mutant (right) amino acid.
The backbone, which is the same for each amino acid, is colored red. The side chain, unique for each amino acid, is colored black; (a) S1259L (b) E960K (c) L1084V (d) P503T (e) L1143Q.
Analysis of protein-protein interaction
STRING revealed that products of MTUS1 interacts with Angiotensin II Receptor Type 2(AGTR2), CD274, also commonly referred to as PDL1, Bradykinin Receptor B2 (BDKRB2), Kinesin Family Member 2C(KIF2C), Aurora kinases (AURKB), Leucine Rich Repeat LGI Family Member 3 (LGI3), ELMO Domain Containing 2 (ELMOD2), Polyamine Modulated Factor 1 Binding Protein 1 (PMFBP1), Spermatogenesis Associated 4 (SPATA4), and RIO Kinase 3 (RIOK3) (Fig 4).
3D structure of MTUS1 protein
The three-dimensional structure of the human MTUS1 protein was modeled using I-TASSER. The top 10 structural analogs in PDB were used as templates for such modeling, from which the topmost template (PDB ID: 6w1sr) covered 93% of the human MTUS1 query sequence. The server provided the top 5 models for the targeted protein though none of them were validated according to the scores of structure validation software due to their poor stereochemical properties (S2 Table).
Hence, we used SWISS-MODEL for the three-dimensional structural analysis of MTUS1 protein. Unlike I-TASSER, SWISS-MODEL provided 9 structures and all of them were a partial structure of our targeted protein based on the best-aligned template from UniProtKB proteome. Further analyses were done using PROCHECK, QMEAN, Molprobity, ERRAT, and ProSA programs to estimate the quality of the models (Table 5), and model number 8 was selected as the best one (Fig 5).
Homology models from SWISS-MODEL server; (a) Structure of Model-8 (b) Mutated Model-8 structure.
For model 8- Ramachandran favored region was 98.04% by SWISS-MODEL and 95.1% core region by PROCHECK analysis. Molprobity and QMEAN scores were 1.08 and 0.82 respectively. ERRAT showed a higher score of 96.875% and ProSA-Z score was -0.83 for the best-predicted model.
Structure and comparison of the mutated protein
In our study, the most suitable model (model number 8) covered the amino acid residues from 1133–1236 and it contained only one mutation (L1143Q) out of the investigated 5 most deleterious nsSNPs. Instead of Leucine, Glutamine was introduced at the position of 1143 in the native protein sequence and SWISS-MODEL was used to determine the mutated structure. While verifying the mutated model quality by SWISS-MODEL structure assessment, PROCHECK, QMEAN, Molprobity, and ERRAT similar scores were found as its native structure (Table 5).
TM-align, an algorithm for protein structure alignment and comparison was used for comparing the structural analog. For mutated structure, we got 0.99982 and 0.05 as TM-score and RMSD value respectively. These values indicate that both the structures are in the same fold.
Association of the damaging nsSNPs with cancer
The mutation profile of MTUS1 in both cBioPortal and canSAR Black webserver revealed the association of S1259L with Stomach adenocarcinoma (STAD) and the substitution of glutamate at 960 position with Uterine corpus endometrial carcinoma (UCEC). The canSAR black webserver also revealed that S1259L may result in moderate STAD whereas mutation at 960 may result in severe UCEC. However, the remaining nsSNPs were not found in the mutation profile presented by the web servers.
Expression analysis of MTUS1 gene
The box plot analysis revealed that, Stomach adenocarcinoma (STAD) results due to overexpression of MTUS1 whereas Uterine corpus endometrial carcinoma (UCEC) ensues due to underexpression of MTUS1 (Fig 6).
Survival analysis in STAD and UCEC patients
We used the GEPIA database for survival analysis of patients with Stomach adenocarcinoma (STAD) and Uterine corpus endometrial carcinoma (UCEC). The patients were categorized into high-expression and low-expression groups based on the median expression level of MTUS1. Abnormal expression of MTUS1 displayed association with poor prognosis of patients with STAD and UCEC. In STAD, overexpression of MTUS1 showed shorter survival time compared to patients with lower expression levels. In contrast, UCEC patients with high MTUS1 expression had longer survival time than patients with lower MTUS1 expression (Fig 7).
Relationship between mutation and expression of MTUS1 gene with other significant genes
The muTarget links mutation with gene expression alteration and provides expression plots. We used it for two independent analysis. Firstly, the ‘Genotype’ run with MTUS1 as mutated gene showed that for Gastric cancer, mutation in MTUS1 can affect the expression of other genes such as CCNB2, OIP5 etc. This same analysis for Uterine cancer showed that expression of FAXDC2, MYCL etc. altered due to MTUS1 mutation (Fig 8). Secondly, the ‘Target’ run with MTUS1 as the target gene represented that altered genes of Gastric cancer (i. e., TEX15, RECQL4) can change expression of MTUS1. Similarly, expression of MTUS1 can be changed due to the mutation of genes responsible for Uterine cancer (i.e., TP53, ZNF18). Though the server provides up to top 30 models, we represented expression of two genes for each analysis (Fig 9).
Genotype run for the changes in MTUS1 gene expression; (a) Gastric Cancer (b) Uterine cancer.
Mitochondrial tumor suppressor 1 (MTUS1) is a tumor suppressor gene which has been associated with various cancers i.e. breast, colon, lung, pancreas, bladder and prostate cancer [3,4]. The gene encodes a family of angiotensin II type 2 (AT2) receptor interacting proteins (ATIPs) which exert anti-tumor effects by inhibiting cell proliferation as well as differentiation .
In recent years presence of damaging SNPs in several oncogenes [35,36] has made in silico investigation of deleterious SNPs from large datasets a major concern. Despite the impact of MTUS1 has been annotated in the cell proliferation of several carcinomas, the in silico analysis of deleterious nsSNPs in our candidate MTUS1 gene has remained uncharacterized yet.
The effectiveness of the most deleterious SNPs prediction can be made more valid by combining several computational based approaches together. In this study, we performed the analysis with 7 prediction tools (SIFT, SNAP2, Align-GVGD, Polyphen-2, PROVEAN, PANTHER and PhD-SNP) to get an integrated picture of pathogenic SNPs of MTUS1 gene. Owing to the reliance of each algorithm on different parameters, in this study we selected 9 nsSNPs (Table 1) as high-risk that were predicted to be deleterious by all SNP prediction algorithms for further analysis.
Generally, the function, activity and regulation of a protein significantly depend on the structural stability of the molecule. Decrease in protein stability causes degradation, misfolding and aggregation of proteins leading to subsequent dysfunction [37,38]. To determine the effect of the above mentioned 9 deleterious nsSNPs on the stability of the MTUS1 protein, I-mutant suite was used. Out of these 9 nsSNPs, 6 decreased the stability of the protein hence may have impact on protein dysfunction.
The evolutionary conservation profile of a protein helps to determine the severity of a detrimental mutation. The nsSNPs that are located at highly conserved region are more likely to be detrimental than nsSNPs that are located at variable regions . We used ConSurf web server to inspect potential effects of these 9 most deleterious nsSNPs (Table 2). For finding putative structural and functional sites, ConSurf offers evolutionary conservation data with solvent accessibility predictions . Furthermore, based on their position relative to the protein surface or protein core, highly conserved residues are subjected to be either structural or functional respectively . According to ConSurf, 7 deleterious nsSNPs among 9 nsSNPs had high conservation score and out of these 7 highly conserved nsSNPs, 5 were predicted as functional (exposed) whereas the rest of them were predicted to be structural (buried).
By analyzing the result of I mutant and ConSurf, we can conclude that out of these 9 nsSNPs, 5nsSNPs (S1259L, E960K, P503T, L1084V and L1143Q) are potentially vulnerable because of their higher conservancy as well as ability to decrease protein stability.
Moreover, to analyze the effect of these above-mentioned 5 high-risk nsSNPs on the structure of MTUS1 protein we performed NetsurfP and HOPE. NetsurfP speculates solvent accessibility and secondary structure of the protein . It presented 3 variants as exposed (S1259L, E960K and P503T) and other 2 variants as buried (L1084V and L1143Q), which can be verified by the similar findings from ConSurf web server.
Determination of protein stability is quite difficult as it cannot be predicted by considering a single factor only. Physiochemical properties of amino acid residues such as-polarity, charge, hydrophobicity etc. play crucial role for the overall stability of the protein molecule. It has been shown that protein stability is linked with the presence of hydrophobic residues at its surface area because these residues hinder accessibility of water molecule to the protein . Among the exposed variants, in case of S1259L, the exposure of the non-polar residue (Leucine) in place of the polar one (Serine) on the surface area may decrease protein stability. Again, in E960K the substitution of a negatively charged molecule (Glutamic acid) with a positively charged one (Lysine) can cause repulsion which may severely hamper the interaction of the protein with other molecules. However, in P503T, a nonpolar group (Proline) is replaced by a polar group (Threonine) which may decrease the hydrophobicity of surface area. Between the 2 buried variants, in case of L1084V there is no polarity change, however in L1143Q, a nonpolar group (Leucine) is substituted by a polar group (Glutamine) which may decrease the hydrophobicity of the protein core. These findings can be further validated by HOPE result.
Hope server predicted that these 5 highly risky nsSNPs might have damaging effect on the structure of the protein (Table 4) among which, 2 nsSNPs were structural and 3 nsSNPs were functional residues according to ConSurf.
For the understanding of cellular processes protein–protein interaction network is a vital factor. STRING plays a critical role to filter and assess functional genomics data as well as to provide an intuitive platform for interpreting structural, functional and evolutionary properties of proteins . In the present study, this database revealed the interaction of MTUS1 protein with other proteins which may involve in different pathways, and disruption of these pathways may result in diseases.
To determine the three-dimensional structure of human MTUS1 protein, structural prediction tools were used due to the unavailability of the PDB ID (Protein Data Bank ID) of this protein. Automated protein structure prediction tool, I-TASSER was used where FASTA sequence of the whole protein was given as an input file and the server came up with the top 5 final structural models. The quality of the predicted models is preliminarily estimated by the confidence score (C-score) given by the server that ranges from -5 to 2 and the highest score depicts the most compatible ones. According to this, the first model (highest C-score = -1.67) was selected as the most suitable structure . As there is a possibility of having a better quality of low-ranking models than that of high-ranking in rare cases, hence, all the models were further analyzed for structural validation.
Validation of experimental models is indispensable to obtain a better quality of targeted protein structure. To ensure it several computational programs- SWISS-MODEL, PROCHECK, QMEAN, Molprobity, ERRAT, and ProSA were used. Among all the verification matrices Ramachandran plot is the most prioritized as it represents the φ-ψ torsion angles of the protein backbone of predicted models. SWISS-MODEL gives the idea about the favored region of the Ramachandran plot and PROCHECK reveals the stereochemical quality of a given protein structure by assessing the Ramachandran plot into different regions—core, allowed, generously allowed, and disallowed region. More than 90% of residues in the core region or in the most favored region can be preferred as a favorable structure .
Other computational tools provide scores for the estimation of protein model quality. QMEAN-Z score of -4.0 or below -4.0 indicates the model with low quality and higher score points to the favorable states of the structure . ProSA-Z score of -4.0 or below depicts the poor quality of models like QMEAN . Around 95% or higher value of ERRAT score generally provides high standard resolution of structure . On the other hand, Molprobity score closer to zero represents the higher quality of a structure .
In our study, all the 5 models from I-TASSER were invalidated according to the standard score of validation tools mentioned above (S2 Table). However, we provided our query sequence to SWISS-MODEL, a homology-modeling server. Based on the best-aligned template, the server came up with 9 models, although none of them covered the whole sequence of our targeted protein. Interestingly, all of the 9 models fell into the range of amino acids from 906 to 1239 which might indicate that the protein sequence around this region could be a conserved region. Our findings from ConSurf analysis also showed a similar result. All of our predicted 5 significantly damaging nsSNPs (S1259L, E960K, L1084V, P503T, and L1143Q) were highly conserved and were in between this region.
All these 9 models were then evaluated for structural validation by PROCHECK, QMEAN, Molprobity, ERRAT, and ProSA programs. According to the standard score of all validation software, model number 8 was selected as the best presumptive structure of MTUS1 protein (Table 5). This model comprises 104 amino acid residues that range from 1133–1236 based on template alignment and only one mutation (L1143Q) out of the 5 most deleterious nsSNPs fell into this range. To model the mutant structure, a point mutation at this specific position was made in the native protein sequence and provided to SWISS-MODEL. Only one mutated model was provided by the server and it showed similar scores as model 8 while verified by PROCHECK, QMEAN, Molprobity, and ERRAT programs (Fig 5 and Table 5).
Structural comparison between wild type and mutant structure was analyzed by TM-align tool. Low TM score and high RMSD value indicate structural dissimilarity whereas we found a high TM-score (0.99982) and low RMSD value (0.05); indicate both the structures are on the same fold . The wild-type structure is of a snippet of the whole structure and only one mutation that falls into that region might be the reason for showing the similarity between native and mutant structures.
Mutation of a protein leads to genomic instability which may result in various types of cancers. To investigate such correlations different cancer prognostic tools were used. The Cancer genomics database cBioPortal provided us with a summary of types of cancers associated with MTUS1. According to this study, more than 30 cancers can occur due to various anomalies (i. e., mutation, structural variant, amplification, deep deletion and multiple alterations) in MTUS1 gene. Interestingly, mutation is the sole reason for several cancers and high frequency of mutation has been observed in Uterine corpus endometrial carcinoma (UCEC) and Stomach adenocarcinoma (STAD) which account 6.99% and 3.86% respectively (S4 Fig).
We meticulously searched both cBioPortal and canSAR Black webserver for the 5 deleterious nsSNPs in the Lollipop plot that showing mutation frequency along the MTUS1 amino acid sequence. Among these 5 nsSNPs, the mutation S1259L was found to be associated with Stomach adenocarcinoma (STAD) and substitution of glutamate at 960 position was linked to severe Uterine corpus endometrial carcinoma (UCEC).
GEPIA provided an expression dot plot for all caner types related to MTUS1(S5 Fig). This revealed that most of the cancers occurred due to the downregulation of MTUS1 though some of them resulted from upregulation of MTUS1. To represent the opposite expression of MTUS1 gene more clearly, we performed box plot analysis of MTUS1 gene expression in STAD and UCEC where these cancers resulted from upregulation and downregulation of MTUS1 accordingly (Fig 6).
The survival curve is a plot of the survival probability (percentage) against time which provides crucial summary of data that is utilized to calculate measures i.e., median survival time [33,46]. In this study, survival analysis revealed that STAD patients with high levels of MTUS1 had shorter survival time whereas UCEC patients with reduced expression of MTUS1 had decreased survival time. This finding is consistent with the results we found in MTUS1 gene expression analysis. So, both upregulation  and downregulation of MTUS1 can lead to cancer progression though the molecular mechanisms underlying this process is still uncharted and can be a new agenda of research.
The muTarget is a tool that is designed to find out not only the genes whose expression are changed due to mutation in the query gene but also those genes whose mutations affect the expression of the target gene. From our analysis, it has been found that the expression of various genes was changed due to mutation in MTUS1. For instance, in case of gastric cancer, the expression of CCNB2, OIP5 etc. genes were found to be altered along with MTUS1 mutation. The CCNB2 gene is involved in cell cycle regulation  whereas OIP5 gene is associated with chromosomal segregation during mitosis . For Uterine cancer, FAXDC2 and MYCL gene’s expression changed due to MTUS1 mutation. The FAXDC2 gene induces megakaryocyte differentiation  whereas MYCL encodes a transcription factor concerned in lung cancer .
Again, when MTUS1 was used as the target gene, its expression was changed by numerous mutated genes. For instance, the server represented that mutation in TEX15, RECQL4 etc. genes altered the expression of MTUS1 in STAD patients. TEX15 is engaged in normal chromosome synapsis and metabolic recombination during spermatogenesis  whereas RECQL4 encodes a DNA- dependent ATPase which may regulate chromosome segregation . The expression of our target gene changed in the UCEC patients likewise owing to the mutations in genes like TP53 and ZNF18. Interestingly, TP53 is a tumor suppressor gene  like our candidate MTUS1 gene. This interaction indicates that aberration in one tumor suppressor may affect other gene of similar category which might worsen the state of the disease . However, ZNF18 is a zinc finger protein involved in transcriptional regulation . Such gene-gene interactions can be used to identify cancer biomarkers along with therapeutic drug targets and different treatment options in various cancers.
Two nsSNPs from the five deleterious nsSNPs have been found to be associated with cancers. It is highly likely that the other three nsSNPs might be involved in different cancers. However, robust in vivo investigation is needed to establish the association with specific cancer in future.
As a tumor suppressor, the MTUS1 gene products play critical roles in various cellular mechanisms and prevent uncontrolled cell growth and proliferation. Consequently, alteration of this gene has been associated with different types of diseases including various cancers.
This study is the first systematic and extensive in silico analysis of functional SNPs in the MTUS1 gene. We reported 5nsSNPs (S1259L, E960K, P503T, L1084V and L1143Q) as potentially damaging due to their presence in the highly conserved region and ability to affect protein stability. And two of them, S1259L and E960K were found to be associated with Stomach adenocarcinoma and Uterine corpus endometrial carcinoma accordingly. The findings of this study will hopefully provide a guideline in extricating the damaging SNPs which increase the risk of cancers and disease susceptibilities. However, extensive population-based studies along with clinical trials are essential to characterize the effects of these polymorphisms on structure and function of the protein as well as to develop effective and individualized treatment option.
S1 Fig. Conservation profile of MTUS1 (Amino Acid 1–600).
S2 Fig. Conservation profile of MTUS1 (Amino Acid 601–1200).
S3 Fig. Conservation profile of MTUS1 (Amino Acid 1201–1270).
S4 Fig. Cancer types summary that are related to MTUS1.
S5 Fig. Profile of MTUS1 from GEPIA for gene expression related to all cancer types.
S1 Table. Identification of the impact of nsSNPs of the MTUS1 gene.
- 1. Guo XE, Ngo B, Modrek AS, Lee W-H. Targeting Tumor Suppressor Networks for Cancer Therapeutics. Curr Drug Targets. 2014 Jan;15(1):2–16. pmid:24387338
- 2. Seibold S, Rudroff C, Weber M, Galle J, Wanner C, Marx M. Identification of a new tumor suppressor gene located at chromosome 8p21.3–22. FASEB J. 2003;17(9):1180–2. pmid:12692079
- 3. Di Benedetto M, Bièche I, Deshayes F, Vacher S, Nouet S, Collura V, et al. Structural organization and expression of human MTUS1, a candidate 8p22 tumor suppressor gene encoding a family of angiotensin II AT2 receptor-interacting proteins, ATIP. Gene. 2006 Oct 1;380(2):127–36. pmid:16887298
- 4. Nouet S, Amzallag N, Li J-M, Louis S, Seitz I, Cui T-X, et al. Trans-inactivation of Receptor Tyrosine Kinases by Novel Angiotensin II AT2 Receptor-interacting Protein, ATIP. J Biol Chem. 2004 Jul 9;279(28):28989–97. pmid:15123706
- 5. Rodrigues-Ferreira S, Tommaso AD, Dimitrov A, Cazaubon S, Gruel N, Colasson H, et al. 8p22 MTUS1 Gene Product ATIP3 Is a Novel Anti-Mitotic Protein Underexpressed in Invasive Breast Carcinoma of Poor Prognosis. PLOS ONE. 2009 Jan 10;4(10):e7239. pmid:19794912
- 6. Di Benedetto M, Pineau P, Nouet S, Berhouet S, Seitz I, Louis S, et al. Mutation analysis of the 8p22 candidate tumor suppressor gene ATIP/MTUS1 in hepatocellular carcinoma. Mol Cell Endocrinol. 2006 Jun 27;252(1):207–15. pmid:16650523
- 7. Ding X, Zhang N, Cai Y, Li S, Zheng C, Jin Y, et al. Down-regulation of tumor suppressor MTUS1/ATIP is associated with enhanced proliferation, poor differentiation and poor prognosis in oral tongue squamous cell carcinoma. Mol Oncol. 2012 Feb 1;6(1):73–80. pmid:22153618
- 8. Huang P, Guo L, Xi J, Wang D. Angiotensin II type 2 receptor–interacting protein 3a inhibits ovarian carcinoma metastasis via the extracellular HMGA2-mediated ERK/EMT pathway. Tumor Biol. 2017 Jun 1;39(6):1010428317713389.
- 9. Dakal TC, Kala D, Dhiman G, Yadav V, Krokhotin A, Dokholyan NV. Predicting the functional consequences of non-synonymous single nucleotide polymorphisms in IL8 gene. Sci Rep. 2017 Jul 26;7(1):6525. pmid:28747718
- 10. Cargill M, Altshuler D, Ireland J, Sklar P, Ardlie K, Patil N, et al. Characterization of single-nucleotide polymorphisms in coding regions of human genes. Nat Genet. 1999 Jul;22(3):231–8. pmid:10391209
- 11. Yue P, Moult J. Identification and Analysis of Deleterious Human SNPs. J Mol Biol. 2006 Mar 10;356(5):1263–74. pmid:16412461
- 12. Sherry ST, Ward M-H, Kholodov M, Baker J, Phan L, Smigielski EM, et al. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res. 2001 Jan 1;29(1):308–11. pmid:11125122
- 13. Kumar P, Henikoff S, Ng PC. Predicting the effects of coding non-synonymous variants on protein function using the SIFT algorithm. Nat Protoc. 2009 Jul;4(7):1073–81. pmid:19561590
- 14. Bromberg Y, Yachdav G, Rost B. SNAP predicts effect of mutations on protein function. Bioinformatics. 2008 Oct 15;24(20):2397–8. pmid:18757876
- 15. Tavtigian SV, Deffenbaugh AM, Yin L, Judkins T, Scholl T, Samollow PB, et al. Comprehensive statistical study of 452 BRCA1 missense substitutions with classification of eight recurrent substitutions as neutral. J Med Genet. 2006 Apr 1;43(4):295–305. pmid:16014699
- 16. Adzhubei IA, Schmidt S, Peshkin L, Ramensky VE, Gerasimova A, Bork P, et al. A method and server for predicting damaging missense mutations. Nat Methods. 2010;7(4):248–9. pmid:20354512
- 17. Choi Y, Sims GE, Murphy S, Miller JR, Chan AP. Predicting the Functional Effect of Amino Acid Substitutions and Indels. PLOS ONE. 2012 Aug 10;7(10):e46688. pmid:23056405
- 18. Tang H, Thomas PD. PANTHER-PSEP: predicting disease-causing genetic variants using position-specific evolutionary preservation. Bioinformatics. 2016;32(14):2230–2. pmid:27193693
- 19. Capriotti E, Calabrese R, Casadio R. Predicting the insurgence of human genetic diseases associated to single point protein mutations with support vector machines and evolutionary information. Bioinformatics. 2006 Nov 15;22(22):2729–34. pmid:16895930
- 20. Capriotti E, Fariselli P, Casadio R. I-Mutant2. 0: predicting stability changes upon mutation from the protein sequence or structure. Nucleic Acids Res. 2005;33(suppl_2):W306–10.
- 21. Ashkenazy H, Erez E, Martz E, Pupko T, Ben-Tal N. ConSurf 2010: calculating evolutionary conservation in sequence and structure of proteins and nucleic acids. Nucleic Acids Res. 2010;38(suppl_2):W529–33. pmid:20478830
- 22. Petersen B, Petersen TN, Andersen P, Nielsen M, Lundegaard C. A generic method for assignment of reliability scores applied to solvent accessibility predictions. BMC Struct Biol. 2009 Jul 31;9(1):51. pmid:19646261
- 23. Venselaar H, te Beek TA, Kuipers RK, Hekkelman ML, Vriend G. Protein structure analysis of mutations causing inheritable diseases. An e-Science approach with life scientist friendly interfaces. BMC Bioinformatics. 2010 Nov 8;11(1):548.
- 24. Jensen LJ, Kuhn M, Stark M, Chaffron S, Creevey C, Muller J, et al. STRING 8—a global view on proteins and their functional interactions in 630 organisms. Nucleic Acids Res. 2009;37(suppl_1):D412–6. pmid:18940858
- 25. Yang J, Yan R, Roy A, Xu D, Poisson J, Zhang Y. The I-TASSER Suite: protein structure and function prediction. Nat Methods. 2015 Jan;12(1):7–8. pmid:25549265
- 26. Waterhouse A, Bertoni M, Bienert S, Studer G, Tauriello G, Gumienny R, et al. SWISS-MODEL: homology modelling of protein structures and complexes. Nucleic Acids Res. 2018 Jul 2;46(W1):W296–303. pmid:29788355
- 27. Laskowski RA, MacArthur MW, Moss DS, Thornton JM. PROCHECK: a program to check the stereochemical quality of protein structures. J Appl Crystallogr. 1993 Apr 1;26(2):283–91.
- 28. Colovos C, Yeates TO. Verification of protein structures: patterns of nonbonded atomic interactions. Protein Sci. 1993;2(9):1511–9. pmid:8401235
- 29. Wiederstein M, Sippl MJ. ProSA-web: interactive web service for the recognition of errors in three-dimensional structures of proteins. Nucleic Acids Res. 2007 Jul 1;35(suppl_2):W407–10. pmid:17517781
- 30. Zhang Y, Skolnick J. TM-align: a protein structure alignment algorithm based on the TM-score. Nucleic Acids Res. 2005 Apr 1;33(7):2302–9. pmid:15849316
- 31. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio Cancer Genomics Portal: An Open Platform for Exploring Multidimensional Cancer Genomics Data. Cancer Discov. 2012 May 1;2(5):401–4. pmid:22588877
- 32. Mitsopoulos C, Di Micco P, Fernandez EV, Dolciami D, Holt E, Mica IL, et al. canSAR: update to the cancer translational research and drug discovery knowledgebase. Nucleic Acids Res. 2021 Jan 8;49(D1):D1074–82. pmid:33219674
- 33. Tang Z, Li C, Kang B, Gao G, Li C, Zhang Z. GEPIA: a web server for cancer and normal gene expression profiling and interactive analyses. Nucleic Acids Res. 2017 Jul 3;45(W1):W98–102. pmid:28407145
- 34. Nagy Á, Győrffy B. muTarget: A platform linking gene expression changes and mutation status in solid tumors. Int J Cancer. 2021;148(2):502–11. pmid:32875562
- 35. Doss CGP, Nagasundaram N, Chakraborty C, Chen L, Zhu H. Extrapolating the effect of deleterious nsSNPs in the binding adaptability of flavopiridol with CDK7 protein: a molecular dynamics approach. Hum Genomics. 2013;7(1):10.
- 36. Kamaraj B, Rajendran V, Sethumadhavan R, Purohit R. In-silico screening of cancer associated mutation on PLK1 protein and its structural consequences. J Mol Model. 2013;19(12):5587–99. pmid:24271645
- 37. Deller MC, Kong L, Rupp B. Protein stability: a crystallographer’s perspective. Acta Crystallogr Sect F Struct Biol Commun. 2016 Feb 1;72(2):72–95. pmid:26841758
- 38. Hossain MS, Roy AS, Islam MS. In Silico Analysis Predicting Effects of Deleterious SNPs of Human RASSF5 Gene on its Structure and Functions. bioRxiv. 2020. pmid:32884013
- 39. Miller MP, Kumar S. Understanding human disease mutations through the use of interspecific genetic variation. Hum Mol Genet. 2001;10(21):2319–28. pmid:11689479
- 40. Berezin C, Glaser F, Rosenberg J, Paz I, Pupko T, Fariselli P, et al. ConSeq: the identification of functionally and structurally important residues in protein sequences. Bioinformatics. 2004;20(8):1322–4. pmid:14871869
- 41. Strub C, Alies C, Lougarre A, Ladurantie C, Czaplicki J, Fournier D. Mutation of exposed hydrophobic amino acids to arginine to increase protein stability. BMC Biochem. 2004;5(1):1–6. pmid:15251041
- 42. Zobayer N, Hossain ABMA. In silico Characterization and Homology Modeling of Histamine Receptors. J Biol Sci. 2018 Apr 15;18:178–91.
- 43. Benkert P, Biasini M, Schwede T. Toward the estimation of the absolute quality of individual protein structure models. Bioinformatics. 2011 Feb 1;27(3):343–50. pmid:21134891
- 44. Adiba M, Das T, Paul A, Das A, Chakraborty S, Hosen MI, et al. In silico characterization of coding and non-coding SNPs of the androgen receptor gene. Inform Med Unlocked. 2021 Jan 1;24:100556.
- 45. Davis IW, Leaver-Fay A, Chen VB, Block JN, Kapral GJ, Wang X, et al. MolProbity: all-atom contacts and structure validation for proteins and nucleic acids. Nucleic Acids Res. 2007 Jul 1;35(suppl_2):W375–83. pmid:17452350
- 46. Clark TG, Bradburn MJ, Love SB, Altman DG. Survival Analysis Part I: Basic concepts and first analyses. Br J Cancer. 2003 Jul;89(2):232–8. pmid:12865907
- 47. Zindy PJ, L’Helgoualc’h A, Bonnier D, Le Béchec A, Bourd-Boitin K, Zhang CX, et al. Upregulation of the tumor suppressor gene menin in hepatocellular carcinomas and its significance in fibrogenesis. Hepatol Baltim Md. 2006 Nov;44(5):1296–307. pmid:17058241
- 48. Wu T, Zhang X, Huang X, Yang Y, Hua X. Regulation of Cyclin B2 Expression and Cell Cycle G2/M Transition by Menin*. J Biol Chem. 2010 Jun 11;285(24):18291–300. pmid:20404349
- 49. Fujita Y, Hayashi T, Kiyomitsu T, Toyoda Y, Kokubu A, Obuse C, et al. Priming of Centromere for CENP-A Recruitment by Human hMis18α, hMis18β, and M18BP1. Dev Cell. 2007 Jan 1;12(1):17–30. pmid:17199038
- 50. Jin Q, Ren Y, Wang M, Suraneni PK, Li D, Crispino JD, et al. Novel function of FAXDC2 in megakaryopoiesis. Blood Cancer J. 2016 Sep;6(9):e478–e478. pmid:27689744
- 51. Ikegaki N, Minna J, Kennett RH. The human L-myc gene is expressed as two forms of protein in small cell lung carcinoma cell lines: detection by monoclonal antibodies specific to two myc homology box sequences. EMBO J. 1989 Jun 1;8(6):1793–9. pmid:2548855
- 52. Yang F, Eckardt S, Leu NA, McLaughlin KJ, Wang PJ. Mouse TEX15 is essential for DNA double-strand break repair and chromosomal synapsis during male meiosis. J Cell Biol. 2008 Feb 18;180(4):673–9. pmid:18283110
- 53. Yin J, Kwon YT, Varshavsky A, Wang W. RECQL4, mutated in the Rothmund–Thomson and RAPADILINO syndromes, interacts with ubiquitin ligases UBR1 and UBR2 of the N-end rule pathway. Hum Mol Genet. 2004 Oct 15;13(20):2421–30. pmid:15317757
- 54. Surget S, Khoury MP, Bourdon J-C. Uncovering the role of p53 splice variants in human malignancy: a clinical perspective. OncoTargets Ther. 2013 Dec 19;7:57–68. pmid:24379683
- 55. Wang L-H, Wu C-F, Rajasekaran N, Shin YK. Loss of Tumor Suppressor Gene Function in Human Cancer: An Overview. Cell Physiol Biochem. 2018;51(6):2647–93. pmid:30562755
- 56. Guo L-L, Ci H-L, Shan H-S, Zou X, Zhai Y-G, Li Y-P. [Molecular cloning and expression analysis of a novel human gene ZNF18]. Yi Chuan Hered. 2005 Jul;27(4):523–30. pmid:16120570