Connexin26 (Connexin26, Cx26), encoded by the genes GJB2, is one of the most important members in the connexin family. The mutations of GJB2 are related to nonsyndromic hearing loss (NSHL) and GJB2 is the most common deafness genes (Mahdieh et al., 2010; Wu et al., 2011). In the GJB2 genes, about 200 kinds of mutations are found to be associated with deafness (http://davinci.crg.es/deafness/) and they have ethnic and geographical differences. For example, 35delG, 167dlT, 235delC are more common in Caucasus, Jews and Asians, respectively, while the frequency of the mutation of 235delC is higher in China than in other Asian countries (Dai et al., 2007). Due to the diversity of the mutations of GJB2 gene and the continuous discoveries of new mutations, it has become a subject that is worth scientifically predicting neonatal disease-causing mutations and the correlations to disease phenotypes. Today, the bioinformatics methods that can conduct pathogenic prediction and possible functional analysis of newly discovered mutations have provided a useful platform for genomic research of diseases. Therefore, based on the fundamental principles, this article explored the relationship between the mutations of GJB2 gene and the genetic effects of hearing loss by analyzing sequence evolutions and molecular characteristics of GJB2 genes and referring to mutation database and related research reports through bioinformatical approaches.

1 Results and Analysis
1.1 17 species and their GJB2 protein sequence information
This study selected 17 species of GJB2 protein sequence information (Table 1), including 2 Amphibia species and 15 Mammalia which contains Primates, Carnivora, Perissodactyla, Mammalia, Artiodactyla and Rodents. The length of the 17 GJB2 protein sequences selected is 226, except the length of western clawed frog and African clawed frog whose GJB2 protein sequence length are 248.

Table 1 GJB2 protein sequence information of 17 species

1.2 Analysis of GJB2 phylogenetic tree
In the phylogenetic tree of GJB2 (Figure 1), GJB2 of mammals is genetically far away from GJB2 of amphibians. In addition, among the 15 species of mammals, 2 species of amphibians and ancestral sequences of amphibians and mammals, the similarity (the frequency of fixed amino acid residues) were 83.6% (189/226), 94.0% (233/248) and 70.7% (159/225), respectively. Thus, it comes to the conclusion that the ancestral sequences of mammals, compared with that of amphibian ancestors, are less similar (Figure 2), which matches the GJB2 phylogenetic tree.

Figure 1 The phylogenetic tree of GJB2 proteins

Figure 2 The results of multiple sequence alignment in human, mammalian ancestor and amphian ancestor

1.3 Analysis of GJB2 conserved regions
Comparing with the GJB2 sequence of 17 species (Figure 2), 65.93% (149/226) of the amino acid residues in human GJB2 are fixed, while 23.89% (54/226) of amino acid residues are conserved. Comparing the human GJB2 sequence with phylogenetic tree node at the ancestral sequence, the results showed that 70.35% (159/226) of human GJB2 amino acid residues are fixed, while 20.80% (47/226) of the amino acid residues are conserved. In addition, in the comparison of sequences of 17 species, the frequency of fixed amino acid residues in the ancestral sequences are the same. And there are 73.45% (166/226) of the sites are fixed when comparing with human GJB2 sequences, ancestral sequences of mammals and amphibians into multiple sequence alignment. Because of the genetic evolutionary role of GJB2 polymorphisms (SNPs), some of whose polymorphic loci compared with human and ancestral sequences come out to be fixed. Therefore, taking these loci of GJB2 sequences in human as fixed sites can improve the accuracy of the analysis and that helps get the final 166 fixed amino acid residues.

SMART analysis of the human GJB2 sequence (Kelley et al., 1998) shows that, GJB2 consists of five non- transmembrane regions (two cells in the outer ring, an N-terminal and a C-terminal and an intracellular cyclic) and 4 transmembrane domains (Figure 3), and 53↔180, 60↔174 and 64↔169 are binding sites for the disulfide bonds. Studies have shown that the homology among the connexin family members mainly stays in the extracellular domain, followed by the transmembrane domain. There are lots of differences in the cytosolic domain. The conserved degrees of comparative analysis of the regional showed that the average conserved degrees of the outer ring, the transmembrane domain and cytoplasmic domain were 94.74%, 93.38% and 72.78% (Table 2), respectively. The high conservatism of GJB2 outer rings suggested that there may be an interaction between GJB2 and other connexin family members outside the cells. Meanwhile, the cytoplasmic CT domain is defined in this article as “the 1st non- conserved region” due to the low conservation.

Figure 3 The three-dimensional structure of human GJB2 protein

Table 2 The degrees of conservatism of human GJB2 protein domains and the distribution frequencies of GJB2 mutation sites

1.4 The comparison of the 3-D structure of human GJB2 protein
The 3-D structure of human GJB2 includes 6 α-helices, 5 β-strands, 10 corners and 7 curls (Figure 3). Comparing 3-D structure of human GJB2 with any one of the remaining 16 species, the scores are more than 98 almost completely overlapped with that of humans. They are numbered 2, 3, 4, 5, 6, 15, 16 and 17 (Figure 4). When spatially comparing with humans and the 3-D structure of the western clawed frog whose scores (score = 73) are lower, spatial differences (Figure 5) are found in the structure of 109-125 amino acid residues. Meanwhile, when comparing the 109-125 segments of human beings with the rest of the species, it turns out that 3-d structures of the species numbered 2, 3, 4, 5, 6, 9, 10, 11, 15, 16 and 17 almost completely overlap with human beings’, while those numbered 7, 8, 12, 13 and 14 differ from humans in the 109-125 segments (Figure 6). Analysis showed that, No.4, 5, 6, 9, 10, 11 differ either in the 113^th or in the 123^rd site, No.7, 8, 13, 14 differ both in the 113^th and the 123^rd sites, while No.12, which related little to human beings, is in the 123^rd ones (Figure 7). Therefore, it is speculated that, when mutations happened to the 113th and 123^rd of the sequences of human GJB2, the corresponding spatial structures of 109-125 segments are likely to change and as a result, the 113^th and 123^rd sites of the human GJB2 sequences may be the conserved sites of the spatial structures. And because 109-125 segments located in the 99-131 non- transmembrane region and the conserved degree (78.79%) is much lower than other regions (except the 1st non- conserved region), the region is defined in this article as “2^nd non- conservative zone”.

Figure 4 The alignment scores between human and any other species

Figure 5 The molecular overlap map of the three-dimensional structure alignment of human and western clawed frog GJB2 protein

Figure 6 The molecular overlap map of the three-dimensional structure alignment of human, sheep, cattle, domestic guinea pig, western clawed frog and African clawed frog GJB2 protein

Figure 7 Multi-alignment of GJB2 protein sequences from 17 species

1.5 Analysis of missense mutations of human GJB2
In the CRG (Center for Genomic Regulation) database, 77 kinds of GJB2 genes were related to NSHL pathogenic missense mutations, which included 60 kinds of autosomal recessive missense mutation, 8 kinds of autosomal dominant missense mutation and 9 kinds of pathogenic missense mutation whose functions remained unknown (Table 3). These mutations involved a total of 60 amino acids, of which 90% (54/60) were conserved. It can be concluded from the distribution of regional frequency (Table 2) that mutational sites mostly concentrated in the District 4 (47.8%) and the distribution of the mutational frequency in the 2^nd non- conserved region was lower (29.4%) while extremely low (0%) in the 1st non- conserved region. Therefore, it can be speculated that the probability of the appearance of the pathogenic missense mutation in the M2 domain is high while it’s low in the CT domain. Additionally, missense mutations (S113R and T123N) both appeared in the conserved sites of the two spatial structures. Some studies have shown that the S113R mutations fail to induce GJB2 to form the same type of connectional channels (Martínez et al., 2009), which influenced its transportational function of membrane channel. Bioinformatics predictive results have also confirmed that mutation of the sites will change the three-dimensional structures of the GJB2 protein.

Table 3 The pathogenic missense mutations of GJB2 in nonsydromic hearing loss

1.6 SOSUI analysis of human GJB2 protein
In the SOSUI analysis of human GJB2 protein, the results of the transmembrane region are similar to that of the SMART analysis. Nearly half of the 60 mutations concentrated in the transmembrane region (Figure 8), of which 23 were located in hydrophobic residues and 37 were in the hydrophilic residues. In the residues consisting the transmembrane region, 28.3% (26/92) of which are hydrophilic residues while in the mutations located of the transmembrane region, the frequency becomes 46.2% (12/26). Thus, it is speculated that there are higher probability of the transmembrane region mutations occurring in the hydrophilic residues. Yang et al (2010) did a research on the mutations of GJB2 protein transmembrane district R32H, V95L, R143W, S199F, respectively, and found that the mutations in R32H, R143W, S199F led to a lack in expression of GJB2 protein. Although V95M mutant protein in the membrane formed a gap junction protein channels, there is no biochemical coupling permeable functions (Yang et al., 2010). Choi et al (2009) also proved that mutations on T86R in the transmembrane domain would cause proteins fail to form a gap junction; Ambrosi et al (2010) who studied four kinds of missense mutation of the transmembrane region, found that the T135A mutation would cause the membrane channel to be closed and the M34A mutation would result in reducing the membrane channel’s activity. In addition, Oshima et al (2011) who studied the electronic crystallographic structure of the M34A mutant protein has found that M34A would have an influence on the formation of the N-terminal of the channel that the channel was almost off. Therefore, it is reliable that missense mutations will affect the expression of the protein on the cell membrane or the membrane channel permeability when it happens in the transmembrane region and changes the characteristics of amino acids.

Figure 8 The result of SOSUI analysis in human GJB2protein

2 Discussions
The gap junction is the only channel of direct exchange of substances between cells, no matter in invertebrates or in the vertebrates. In vertebrates, gap junction channels are constructed by connexin (Cx) family. There are about 20 kinds of gap junction protein genes in mice and 21 in humans. All Cx peptides go 4 times across the cell membrane, forming three structural segments of the outer ring, the inner ring and the transmembrane region. Though highly conserved in transmembrane and extracellular domains, the diversity of the function of Cx is caused by the different amino acid sequences of these regions. The abnormalities of gap junction-mediated intercellular signals can cause multi-systemic diseases, which differs due to the different Cx distribution in various organs, tissues. Cx26 expressed in most tissues and organs, especially in the skin and inner ear. The Cx26 protein shows a high level of expression in the human cochlear striavascularis, basilar membrane, and spiral in the inner ear. GJB2 gene mutations can lead to the Cx26 protein structural or functional disorders which become the reason for hereditary NSHL. Therefore, different GJB2 gene mutation spectrum (race specific) in different deaf populations and the diversity of clinical phenotypes caused by GJB2 gene mutations has been one of the hotspots of the study of deafness-inducing gene mutations.

We used bioinformatics methods to analyze the correlation analysis of GJB2 molecular evolution characteristics and the genetic effects of NSHL, the article shows that the human GJB2 amino acid sequence is highly conserved and there are about 166 fixed amino acid sites. The conserved degree of the ecyto-domain of GJB2 sequence is much higher in the transmembrane region and is higher than that in the cytoplasmic domain, but is extremely low in CT domain. In sequence 109-125, because of the low conservative in spatial structure, the probability of pathogenic missense mutations in the two non-conserved region is low. The 113^th and 123^rd sites are predicted to affect the function of the protein. If the missense mutations occurred in the transmembrane domain, it will change the characteristics of the amino acid; they will affect the expression of proteins on the membrane or change the permeability of the membrane channel. Referring to Dai et al (2009) research on the pathogenic missense mutations of GJB2 genes in 2063 NSHL cases and 10 kinds of missense mutations in patients was obtained from 58 cases of pathogenic missense mutations. There is a simultaneous mutation of W3X and R143W in one patient and a simultaneous mutation of Y152X and R127H in another. These 10 missense mutations are all located in the conserved region, and those in the transmembrane region (R32C, S85P, T86R, R143W) have changed the characteristics of amino acids. Meanwhile, pathogenic missense mutations in the 1^st and the 2^nd non-conserved region cannot be found. Such results also verified that mutations on conserved sites were more pathogenic than those on non-conserved sites and the possibility of non-conservative amino acid becoming pathogenic is rather low.

Through the further understanding of phylogenetic analysis, three-dimensional structural analysis and SOSUI analysis of GJB2, it is speculated for the first time and take out the concept of “conserved spatial structural sites” for the 113^th and 123^rd sites of the human GJB2 sequence which will continue to be verified. In multi-species comparison of sequences, we found that the score between sheep and human sequences is higher (score = 96), but there are some varieties between both spatial structures, which is parallel to the evolutionary tree that sheep and human kinships are far from each other. Similarly, though the comparison of Rattusnorvegicus and human sequences gets a lower score (score = 93), their three-dimensional structure almost completely overlap, which is also in line with the phylogenetic relationship in the evolutionary tree. These results further enrich the bioinformatic information of GJB2.

At present, most of approaches for studying GJB2 pathogenic mutations relies on genetic sequence analysis, electronic and crystal structural analysis, and cell and molecular biological methods (Choi et al., 2009; Ambrosi et al., 2010; Oshima et al., 2011; Yang, 2010) while approaches combining systematic bioinformatics haven’t been reported. In this paper, by applying to bioinformatics methods, the author made sequence homological search of the GJB2 gene, built a species sequence library of the gene, compared the various species to get an analysis of multi-sequence comparison, the phylogenetic and 3-D protein structural spacial comparison and identified the conserved point and region of the sequence. Further more, the author carried out a correlation analysis of conserved sites (regions), functional sites (regions) and the law of pathogenic mutation distribution, which provided a useful forward-looking information for the forecast of new pathogenic mutations of GJB2 and the research of mutant pathogenesis and also a new approach for other diseases.

3 Materials and Methods
3.1 Data
We obtained the complete amino acid sequences of 17 species of GJB2 proteins by searching in the National Center for Biotechnology Information (NCBI), http://blast.ncbi.nlm.nih.gov/Blast.cgi.

3.2 Bioinformatic analysis of GJB2 protein sequences
3.2.1 Alignment of multiple sequences in ClustalW
Use the software ClustalW2 (www.ebi.al.uk/Tools/mas/clustalw2) to get a multiple sequence analysis of 17 species of GJB2 protein sequences (Thompson et al., 1994). ClustalW is currently the most widely used multiple sequence alignment software, which is based on a progressive method. There are three steps of multiple sequence comparison: pair-wised comparison, building wizard tree and then added to the sequence. Firstly, make a pair-wised comparison of multiple sequences and then, build the systematic evolutionary tree according to the distance matrix calculation, and followed by weighing sequences according to the degree of relationship, and then start from the closest sequences, gradually add neighboring sequence and continue to rebuild comparison until all sequences have been added.

3.2.2 Analysis of the evolution of GJB2
Using MEGA5.0 software to map out the GJB2 evolutionary tree of the 17 species, we got the ancestral sequences at the phylogenetic tree node for further study (Jones et al., 1992; Tamura et al., 2011). Mega is currently one of the most widely used evolutionary analysis software, the Maximum Likelihood method was used in this study.

3.2.3 Analysis of protein transmembrane domain and structural domain
We used SMART (http://smart.embl-heidelberg.de) to analyze GJB2 protein domain. SMART is able to identify and annotate protein domains online, which can study on the structure of the protein domain and compared with each other (Schultz Jörg et al., 1998; Letunic Ivica, 2009).

3.2.4 Analysis of protein transmembrane domain and the domain
We employed SOSUI to analyze transmembrane region of the GJB2 protein. SOSUI (http://bp.nuap.nagoya-u.ac.jp/sosui) is used to identify membrane protein and to predict its transmembrane helix wheel, the identification and prediction accuracy of which is up to 99% and 97 %, respectively (Takatsugu Hirokawa et al., 1998).

3.2.5 Building the 3-D protein homological model and analysis of 3-D structure
We also utilized SWISS-MODEL (http://swissmodel.expasy.org/) to homologically build models for 17 GJB2 protein sequences (Arnold et al., 2006; Schwede et al., 2003; Guex et al., 1997). Then, VMD1.9 software (Board of Trustees, of the University of Illinois and others) was used to analyze sequences and their corresponding three-dimensional structures. Protein three-dimensional homological model building method is one of the most common three-dimensional protein structure prediction methods, the principle of which is to find homologous configurations of the unknown structure of proteins in the protein structure database, and then optimize the homologous protein structure to build a predicted 3-D protein.

Authors’ Contribution
YF and AZY completed the data analysis, essayed and modify the manuscript; YY assisted with data analysis and modified part of the data. XC provided the techniques for data analysis of hereditary effect with deafness. XHH designed the research, guided the process and modified this paper. All authors had read and agreed the final manuscript.

Acknowledgements
This research was supported by the Research Grant Award from the National Natural Science Foundation of China (No.31171217). Thank you for the help from Associate Professor Zheng Heng of China Pharmaceutical University.

References
Ambrosi C., Boassa D., Pranskevich J., Smock A., Oshima A., Xu J., Nicholson B.J., and Sosinsky G.E., 2010, Analysis of four connexin26 mutant gap junctions and hemichannels reveals variations in hexamerstability, Biophys J., 98(9): 1809-1819
http://dx.doi.org/10.1016/j.bpj.2010.01.019 PMid:20441744    PMCid:2862186

Arnold K., Bordoli L., Kopp J., and Schwede T., 2006, The swiss-model workspace: A web-based environment for protein structure homology modeling, Bioinformatics, 22(2): 195-201
http://dx.doi.org/10.1093/bioinformatics/bti770 PMid:16301204

Choi S.Y., Park H.J., Lee K.Y., Dinh E.H., Chang Q., Ahmad S., Lee S.H., Bok J., Lin X., and Kim U.K., 2009, Different functional consequences of two missense mutations in the GJB2 gene associated with non- syndromic hearing loss, Hum. Mutat., 30(7): 716-727
http://dx.doi.org/10.1002/humu.21036 PMid:19384972

Dai P., Yu F., Han B., Yuan Y., Li Q., Wang G., Liu X., He J., Huang D., Kang D., Zhang X., Yuan H., Schmitt E., Han D., and Wong L.J., 2007, The prevalence of the 235delC GJB2 mutation in a Chinese deaf population, Genet. Med., 9(5): 283-289
http://dx.doi.org/10.1097/GIM.0b013e31804d2371 PMid:17505205

Dai P., Yu F., Han B., Liu X., Wang G., Li Q., Yuan Y., Liu X., Huang D., Kang D., Zhang X., Yuan H., Yao K., Hao J., He J., He Y., Wang Y., Ye Q., Yu Y., Lin H., Liu L., Deng W., Zhu X., You Y., Cui J., Hou N., Xu X., Zhang J., Tang L., Song R., Lin Y., Sun S., Zhang R., Wu H., Ma Y., Zhu S., Wu B.L., Han D., and Wong L.J., 2009, GJB2 mutation spectrum in 2 063 Chinese patients with non syndromic hearing impairment, Journal of Translational Medicine, 7(26): 1-12
PMid:19123955    PMCid:2627828

Guex N., and Peitsch M.C., 1997, SWISS-MODEL and the Swiss-PdbViewer: An environment for comparative protein modeling, Electrophoresis, 18(15): 2714-2723
http://dx.doi.org/10.1002/elps.1150181505 PMid:9504803

Jones D.T., Taylor W.R., and Thornton J.M., 1992, The rapid generation of mutation data matrices from protein sequences, Comput. Appl. Biosci., 8(3): 275-282
PMid:1633570

Kelley P.M., Harris D.J., Comer B.C., Askew J.W., Fowler T., Smith S.D., and Kimberling W.J., 1998, Novel mutations in the connexin 26 gene (GJB2) that cause autosomal recessive (DFNB1) hearing loss, Am. J. Hum. Genet., 62(4): 792-799
http://dx.doi.org/10.1086/301807 PMid:9529365

Letunic I., Doerks T., and Bork P., 2009, SMART 6: recent updates and new developments, Nucleic Acids Research, 37(1): D229-D232
http://dx.doi.org/10.1093/nar/gkn808 PMid:18978020    PMCid:2686533

Mahdieh N., Rabbani B., Wiley S., Akbari M.T., and Zeinali S., 2010, Genetic causes of nonsyndromic hearing loss in Iran in comparison with other populations, The American Journal of Human Genetics, 55(10): 639-648
http://dx.doi.org/10.1038/jhg.2010.96 PMid:20739942

Martínez A.D., Acuña R., Figueroa V., Maripillan J., and Nicholson B., 2009, Gap-junction channels dysfunction in deafness and hearing loss, Antioxid Redox Signal, 11(2): 309-322
http://dx.doi.org/10.1089/ars.2008.2138

Oshima A., Tani K., Toloue M.M., Hiroaki Y., Smock A., Inukai S., Cone A., Nicholson B.J., Sosinsky G.E., and Fujiyoshi Y., 2011, Asymmetric configurations and N-terminal rearrangements in connexin26 gap junction channels, Journal of Molecular Biology, 405(3): 724-735
http://dx.doi.org/10.1016/j.jmb.2010.10.032 PMid:21094651    PMCid:3026138

Schultz J., Milpetz F., Bork P., and Ponting C.P., 1998, SMART, a simple modular architecture research tool: Identification of signaling domains, Proc. Natl. Acad. Sci., USA, 95(11): 5857-5864
http://dx.doi.org/10.1073/pnas.95.11.5857

Schwede T., Kopp J., Guex N., and Peitsch M.C., 2003, SWISS-MODEL: an automated protein homology-modeling server, Nucleic Acids Research, 31(13): 3381-3385
http://dx.doi.org/10.1093/nar/gkg520 PMid:12824332    PMCid:168927

Hirokawa T, Boon-Chieng S, and Mitaku S., 1998, SOSUL: classification and secondary structure prediction system for membrane proteins, Bioinformatics, 14(4): 378-379
http://dx.doi.org/10.1093/bioinformatics/14.4.378 PMid:9632836

Tamura K., Peterson D., Peterson N., Stecher G., Nei M., and Kumar S., 2011, MEGA5: molecular evolutionary genetics analysis using maximum likelihood, evolutionary distance, and maximum parsimony methods, Molecular Biology and Evolution, 28(10): 2731-2739 Thompson J.D., Higgins D.G., and Gibson T.J., 1994, CLUSTAL W: Improving the ensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice, Nucleic Acids Research, 22( 22): 4673-4680

Wu C.C., Hung C.C., Lin S.Y., Hsieh W.S., Tsao P.N., Lee C.N., Su Y.N., and Hsu C.J., 2011, Newborn genetic screening for hearing impairment: a preliminary study at a tertiary center, PLoS One, 6(7): e22314
http://dx.doi.org/10.1371/journal.pone.0022314 PMid:21811586    PMCid:3139636

Yang Z.C., Xiao Z.A., Xie D.H., and Xia K., 2010, Aberrant expression pattern of a novel mutation in connexin 26 gene resulting in autosomal recessive deafness, Zhonghua Yixue Yichuanxue Zazhi (Medical genetics in China), 27(3): 241-245