In vertebrates, six connexin molecules assemble to form a hemi-channel (connexon), and subsequent docking of two connexons in adjacent cell membranes result in the formation of a complete intercellular channel or gap junction. Gap junctions facilitate the exchange of nutritive material, ions, secondary messengers and small molecules of up to 1 kDa in size directly between adjacent cells (Goodenough et al., Reference Goodenough, Goliger and Paul1996; Alexander & Goldberg, Reference Alexander and Goldberg2003). Gap junctions were thought to have a crucial role in the synchronized contraction of the heart and in embryonic development (Britz-Cunningham et al., Reference Britz-Cunningham, Shah, Zuppan and Fletcher1995). Also in the inner ear, gap junctions are divided into the epithelial cell gap junction system and the connective tissue gap junction system among the supporting cells of the organ of Corti. These gap junctions were considered to have a very important function, facilitating the recycling of potassium ions from the hair cells back into the cochlear endolymph during auditory transduction processes (Kikuchi et al., Reference Kikuchi, Kimura, Paul and Adams1995).
As a member of the connexin family, gap junction protein connexin 43 (Cx43) was encoded by the GJA1 gene that is located on 6q21-q23.2 (Fishman et al., Reference Fishman, Spray and Leinwand1990; Corcos et al., Reference Corcos, Meese and Loch-Caruso1993). The gene is broadly expressed in many different tissues and organs (Beyer et al., Reference Beyer, Kistler, Paul and Goodenough1989; Musil et al., Reference Musil, Beyer and Goodenough1990; Fishman et al., Reference Fishman, Eddy, Shows, Rosenthal and Leinwand1991; van der Heyden et al., Reference van der Heyden, Roeleveld, Peterson and Destree2001; Willingham-Rocky et al., Reference Willingham-Rocky, Golding, Wright, Kraemer, Westhusin and Burghardt2007). Interestingly, for the working myocardium, GJA1 gene expression was only found in mammalian species (Becker et al., Reference Becker, Cook, Davies, Evans and Gourdie1998). Moreover, connexin43 is the most abundant connexin in the cardiovascular system.
Many mutations occur in the different domains of connexin, most of them occurring in highly conserved sites across various species as well as in the human connexin isoforms, and have been linked to specific diseases. Homozygotes for a Leu→Phe substitution in the highly conserved codon 11 and one homozygote for a Val→Ala transversion at the highly conserved codon 24 of GJA1 gene caused recessive, pre-lingual, profound deafness in four patients, suggesting these mutations could be a common cause of deafness in African–Americans, presumably by disrupting the recycling of potassium to the cochlear endolymph (Liu et al., Reference Liu, Xia, Adams, Chen, Welch, Tekin, Ouyang, Kristiansen, Pandya, Balkany, Arnos and Nance2001). Several novel polymorphisms, although no disease-associated mutations, were identified in the GJA1 gene in Turkish families with autosomal-recessive non-syndromic hearing loss (Uyguner et al., Reference Uyguner, Emiroglu, Uzumcu, Hafiz, Ghanbari, Baserer, Yuksel-Apak and Wollnik2003). Substitution of proline for serine at position 364 of GJA1 gene in heart disease patients led to abnormally regulated cell–cell communication, associated with visceroatrial heterotaxia (Britz-Cunningham et al., Reference Britz-Cunningham, Shah, Zuppan and Fletcher1995). Mutations in the arginine–serine–serine fragment concentrating on the carboxy terminus altered potential phosphorylation sites of connexin43, subsequently changing the permeability of gap junctions and resulting in heart malformation (Duncan & Fletcher, Reference Duncan and Fletcher2002). Transgenic animal models also show cardiac malformations (Reaume et al., Reference Reaume, de Sousa, Kulkarni, Langille, Zhu, Davies, Juneja, Kidder and Rossant1995) and significant slowing of conduction velocity because of the absence of connexin43 (Beauchamp et al., Reference Beauchamp, Choby, Desplantez, de Peyer, Green, Yamada, Weingart, Saffitz and Kleber2004). There are also research attempts to reveal the relation between the GJA1 gene and hibernation (Beauchamp et al., Reference Beauchamp, Choby, Desplantez, de Peyer, Green, Yamada, Weingart, Saffitz and Kleber2004; van der Heyden et al., Reference van der Heyden, van Eijk, Wilders, de Bakker and Opthof2004; Yan et al., Reference Yan, Burman, Nichols, Alila, Showe, Showe, Boyer, Barnes and Marr2006).
Owing to the extreme demands of echolocation and flight, bats are among the most unusual and specialized of all mammals. Flight requires three times as much energy output per unit time as does walking or running. To solve this problem, bats have the largest and most muscular heart of any mammal so that the heart can provide enough oxygen for flying (Neuweiler, Reference Neuweiler2000). The specialized hearing system of bats has an important role in receiving and processing echo information during orientation, and for many species during foraging (Siemers & Schnitzler, Reference Siemers and Schnitzler2004; Jones & Teeling, Reference Jones and Teeling2006). The origination and evolution of echolocation in bats have been studied for long time. The traditional morphological tree and the advanced molecular phylogenetic tree, which is based on the nuclear and mitochondrial sequences, support the different hypotheses of gain or loss of echolocation of bats. The former indicated the independent origination of echolocation in microbats, but the latter suggested there is one gain of echolocation in the ancestor of all bats and one loss in the megabats (fruit bats) (Jones & Teeling, Reference Jones and Teeling2006). More evolution analyses about the functional gene sequences can benefit the better understanding of the natural history about the origination and evolution of the fascinating bats' ecological characters. We cloned the full-length coding region of GJA1 gene from 12 bat species and obtained other full-length GJA1 gene sequences of mammals from GenBank and try to test the evolution of this protein in bats' lineage. Phylogeny reconstruction and evolution rate estimation are employed to reconstruct the course of evolution for the GJA1 gene in bats and other mammals. We also used the maximum likelihood method to identify signatures of natural selection.
2. Materials and methods
The cDNA of bats came from the former work about the Foxp2 gene studies of bats (Li et al., Reference Li, Wang, Rossiter, Jones and Zhang2007). Then, the coding region sequences of GJA1 gene were amplified with the PCR conditions as 95°C for 5 min, then 35–39 cycles of 95°C for 35 s, 55°C for 45 s and 72°C for 2·5 min, with a final extension for 10 min at 72°C. A pair of primers that were designed from conserved flanking untranslated region (UTR) conserved regions of connexin43 gene of other mammals is forward primer 5′-CGAGGTATCAGCACTT-TTCTTTCATTAGG-3′ and reverse primer 5′-GGCTGTTGAGTACCACCTCCAC-3′.
Cloning and sequencing
PCR products were analysed and isolated from 1% agarose gels and purified using the TaKaRa Agarose Gel DNA Purification Kit Ver.2.0 (TaKaRa, Japan), followed by ligation with the pGEM-T-easy vector (Promega, USA) and transformation into the DH5α competence cell (TaKaRa, Japan). The clones' identity and orientation were verified by the universal Sp6/T7 primer. The clones were cycle sequenced from both directions using BigDye sequencing kits (Applied Biosystems) on an ABI 3730A automated DNA sequencer. To avoid artefacts, multiple clones were sequenced for every specimen.
Method for the analyses
We amplified and sequenced the complete coding sequences of GJA1 gene from 12 species of bats and obtained 13 sequences of other mammals from GenBank (the species name and GenBank numbers are human, NM_000165; cattle, NM_174068; dog, NM_001002951; mouse, NM_010288; rat, NM_012567; African green monkey, AY382588; Chinese dwarf hamster, AY206456; European hedgehog, AY382589; Syrian hamster, AY206455; rabbit, AY382590; Russian dwarf hamster, AY382591; European ground squirrel, AY382592; pig, AY382593) for analysis. We used the software CLUSTALX 1.81 (Thompson et al., Reference Thompson, Gibson, Plewniak, Jeanmougin and Higgins1997) to align the nucleotide and amino acid sequences. MEGA3 (Kumar et al., Reference Kumar, Tamura and Nei2004) was used to analyse the divergence of nucleotides and amino acids among the different mammalian lineages with the Kimura 2-parameter model to calculate pairwise comparisons of genetic distances and the Poisson correction model for amino acid data. The Nei–Gojobori method (Nei & Gojobori, Reference Nei and Gojobori1986) was employed to calculate the synonymous and non-synonymous substitutions per synonymous and non-synonymous sites (dS and dN) among different lineages using MEGA3 and 500 replications for bootstrap tests. We also used MEGA3 to estimate the ratio of transition and transversion, as well as the saturation of the transition and transversion on three codon sites and the third codon sites, respectively. Before phylogenetic reconstruction, the program MODELTEST 3.6 (Posada & Crandall, Reference Posada and Crandall1998) was used to estimate the most appropriate nucleotide substitution model and parameters for the maximum likelihood method. We used MrBayes 3.1 (Huelsenbeck & Ronquist, Reference Huelsenbeck and Ronquist2001) to reconstruct the phylogenetic tree of GJA1 gene from 25 species of mammals. In the control block of the input file for MrBayes 3.1, we set the generations equal to 1 000 000 and six Markov chains for simulation. A total of 500 000 generations were discarded before the simulation was predicted to reach to a steady condition.
After obtaining the topologies of the GJA1 gene, which were supported in a robust statistical manner, we used the branch-specific models to test for different selection pressures among the different lineages of bats and other mammals using the program CODEML of the PAML package (Yang, Reference Yang1997). Secondly, we calculated and compared the site-specific models to test the selection pressure on the different codon sites of the GJA1 gene during the divergence and evolution of the different lineages of mammals. The site-specific models involved model M0 (One Ratio), which assumed the equal dS/dN ratio among all the amino acid sites. This model can be compared with M3 (discrete), which assumed the variable dS/dN ratio among different parts of amino acid sites, to test the possibility of positive natural selection affecting the protein evolution. Moreover, the comparison between the model M7 (beta) and model M8 (beta and ω) was used to identify the sites, which were selected positively. The likelihood ratio tests (LRTs) were used to appraise the null hypothesis and alternative hypothesis. The tests were done using the χ2 distribution and the degrees of freedom were the parameter differences between two models tested.
To compare the GJA1 gene sequences among lineages of mammals, we obtained the complete sequences of 12 different species of bats, involving three species of fruit bats (Family Pteropodidae: Eonycteris spelaea, Cynopterus sphinx and Rousettus leschenaulti), four species of horseshoe bats (Rhinolophidae: Rhinolophus pearsoni, Rhinolophus rex, Rhinolophus macrotis and Rhinolophus ferrumequinum) and each one species of Old World leaf nosed bats (Hipposideridae: Hipposideros armiger), a free-tailed bat (Molossidae: Chaerephon plicata), a false vampire (Megadermatidae: Megaderma spasma), a sac-winged bat (Emballonuridae: Taphozous melanopogon) and a vesper bat (Vespertilionidae: Myotis ricketti), respectively. The sequences are uploaded to GenBank with the accession numbers from EU195811 to EU195822.
After alignment of all the amino acid sequences of the GJA1 gene from different mammals involved in this paper, the results show strong conservation of the GJA1 gene. Except for an insertion amino acid in the cow and an amino acid absence in T. melanopogon, there are no variations in the length of the amino acid sequences among six orders of mammals analysed in this paper. Moreover, the summation of the amino acid sequence variations from eight orders of mammals showed that 348 out of 383 amino acid sites were conserved sites, which took more than 90·9% of whole gene. Among the sequences from bats, 366 amino acid sites were identified as conserved sites and less than 5% of amino acids were variable sites. The calculation of the pairwise distance of the amino acid distance using the Poisson correction model in software MEGA3 showed that the most amino acid difference value among bats were quite low (0·027). At the nucleotide level, 802 out of 1149 nucleotide sites were identical and 347 nucleotide sites were variable comprising 30·2% of complete sequences from all mammals discussed in this paper. A total of 204 nucleotide sites were identified as variable sites among bats. The overall average calculation of the transition/transversion based on the GJA1 gene sequences from different lineages of mammals give the value of 2·258. The evaluation of the degree of saturation showed the linear increase of transition and transversion changes with increasing sequence divergence for complete sequences and the third codon sites data indicating the unsaturated of transition and transversion substitutions.
Results of the phylogenetic reconstruction
MODELTEST selected the six parameters nucleotide substitution model (GTR) with a gamma distribution (G) and invariable sites involved (I), and the estimated values of other relative parameters were base frequencies: A=0·2302, C=0·2895, G=0·2609, T=0·2194; rate matrix of substitution model: A–C=1·9176, A–G=4·9434, A–T=0·4246, C–G=0·7914, C–T=7·2726; proportion of invariable sites (I)=0·5707; gamma distribution shape parameter=1·0709. According to the topologies of the tree (Fig. 1), the evolution of the GJA1 gene in mammals corresponds with the advanced molecular phylogenetic reconstruction of the different lineages of mammals (Murphy et al., Reference Murphy, Eizirik, O'Brien, Madsen, Scally, Douady, Teeling, Ryder, Stanhope, de Jong and Springer2001). Moreover, the monophyly of the GJA1 gene in bats was highly supported. Within the lineage of Chiroptera, two main clades diverged from each other. One is composed of bats, which were assigned to the Pteropodidae, Rhinolophoidea, Hipposideridae and Megadermatidae, and the other clade included all the other bats sequenced. This topology corresponds with the results of the most advanced molecular phylogenetic reconstruction of bats with multiple-molecular markers (Teeling et al., Reference Teeling, Scally, Kao, Romagnoli, Springer and Stanhope2000), which showed divergence between two major lineages of bats, the Yinpterchiroptera (Pteropodidae, Rhinolophoidea, Hipposideridae and Megadermatidae covered in this study) and the Yangochiroptera (Emaballonuridae, Molossidae and Vespertilionidae sampled here). The concordance between the gene tree of GJA1 gene from bats and the species tree of Teeling et al. (Reference Teeling, Scally, Kao, Romagnoli, Springer and Stanhope2000) suggests that the divergence of the GJA1 gene among the bats reflects the evolutionary history of this clade.
Results of the selection signature test of GJA1 gene
The Nei–Gojobori method (Nei & Gojobori, Reference Nei and Gojobori1986) was used to calculate the average rate of evolution in bats, rodents, artiodactyls, and in all non-bat species involved in this paper and in all species samples (Table 1). The results showed significantly higher dS than dN rates in each lineage and the ratio of dN/dS was significantly less than 1 for all analyses. Except for the lower ratio of rodents, the ratio of bats was quite similar to that of other mammals at about 0·02. Furthermore, we used the maximum likelihood method to estimate the ω ratio of the GJA1 gene among the different lineages of mammals, which was regarded as the branch-specific model. Table 2 presents the log likelihood values and the relative parameters estimated by maximum likelihood under different models. The One Ratio Model assumed the equal dN/dS ratio of each branch to calculate the average dN/dS ratio for each branch and site. The One Ratio Model estimated the dN/dS ratio at 0·024 showing that the average synonymous substitution rate is much higher than the rate of non-synonymous substitution and that the GJA1 gene has undergone elementary purifying selection in general during mammal evolution. This model gave the log maximum likelihood value as −6049·30 (Table 2). The Free Ratio Model assumed an independent ω ratio for each branch in the tree. In the tree of Fig. 1, 54 branches were assumed one ω ratio under the free ratio estimation, which leads to a log maximum likelihood value of −5996·85. In comparison with the One Ratio Model, the Free Ratio Model involved 53 extra parameters. Therefore, we used LRT to compare these two models with a χ2 distribution with 53 degrees of freedom. The result showed that the Free Ratio Model fits significantly better than the One Ratio Model (2Δℓ=104·9, df=53, P<0·001), which indicated the indeed inconsistent dN/dS ratios among different lineages of mammals. But the branch-specific model does not support the hypothesis that there is positive selection to accelerate the evolutionary rate of the GJA1 gene in specific mammal lineages.
We also used the site-specific models tests to show that the LRT between model M3 and model M0 is significant (2Δℓ=127·84, df=4, P<0·001), suggesting that discrete selection pressures act on the different sites of the GJA1 gene. The estimation of model M3 showed that 90·4% (Table 2) of sites were under purifying selection having an extremely low evolution ratio of 0·004. Moreover, 8·4 and 1·2% of sites also had more synonymous substitutions than non-synonymous substitutions with the dN/dS ratios as 0·18 and 0·57, respectively. The LRT between model M8 and M7 did not support the hypothesis that model M8 was better than model M7 (2Δℓ=2·28, df=3, P>0·05) and none of the sites were identified being under positive selection.
The members of gap junction gene families have been widely found in vertebrate species, as well as being well studied in human and mouse (Willecke et al., Reference Willecke, Eiberger, Degen, Eckardt, Romualdi, Guldenagel, Deutsch and Sohl2002; Desplantez et al., Reference Desplantez, Marics, Jarry-Guichard, Veteikis, Briand, Weingart and Gros2003). The phylogenetic reconstruction of the gap junction gene super families shows a monophyletic clade for the GJA1 (connexin43) gene, which included human, mouse and zebra fish (Eastman et al., Reference Eastman, Chen, Falk, Mendelson and Iovine2006). Other researchers identified conserved characters in amphibians (van der Heyden et al., Reference van der Heyden, Roeleveld, Peterson and Destree2001) and birds (Musil et al., Reference Musil, Beyer and Goodenough1990), suggesting that GJA1 diverged from other isogenous genes an extraordinarily long time ago, at least before the radiation of vertebrate species. Previous work compared differences among vertebrate lineages at nucleotide and amino acid levels and showed the highest divergences of 31·3 and 28·4% at nucleotides and amino acids, respectively (van der Heyden et al., Reference van der Heyden, van Eijk, Wilders, de Bakker and Opthof2004). The saturation tests give the results that the transition and transversion of whole data and third codon sites data do not reach saturation, which could bring the mis-estimation in phylogenetic reconstruction. In this paper, we sequenced the GJA1 gene from 12 bat species from seven families and obtained orthologous sequences of mammals from GenBank to investigate GJA1 gene evolution. In a manner similar to other mammals, the bat sequences showed high conservation. The phylogenetic reconstruction results showed high concordance between the GJA1 gene tree and the species tree based on the combined multiple-markers data on the higher-level classification of mammals. In the Bayesian tree of the GJA1 gene, high statistical support (Bayesian posterior probabilities values higher than 0·95) was obtained for the monophyly of the major clades of the mammals as suggested by the advanced molecular phylogenetic work on Placentalia (Madsen et al., Reference Madsen, Scally, Douady, Kao, DeBry, Adkins, Amrine, Stanhope, de Jong and Springer2001; Murphy et al., Reference Murphy, Eizirik, O'Brien, Madsen, Scally, Douady, Teeling, Ryder, Stanhope, de Jong and Springer2001). Chiroptera, Carnivora, Artiodactyla and Insectivora were suggested as orders belonging to Laurasiatheria (Murphy et al., Reference Murphy, Eizirik, O'Brien, Madsen, Scally, Douady, Teeling, Ryder, Stanhope, de Jong and Springer2001), which were highly supported as a monophyletic group in our data. Moreover, the monophyly of Chiroptera was highly supported and matched the results of molecular phylogenetic and morphological research (Simmons et al., Reference Simmons, Novacek and Baker1991; Simmons, Reference Simmons1994; Teeling et al., Reference Teeling, Scally, Kao, Romagnoli, Springer and Stanhope2000). Our results support recognition of the Yinpterchiroptera and Yangochiroptera clades based on combined nuclear and mitochondrial markers (Teeling et al., Reference Teeling, Scally, Kao, Romagnoli, Springer and Stanhope2000). Gene duplication and extinction can create discrepancies between gene trees and species trees (Page, Reference Page1998). The high comparability of the GJA1 gene tree and the published species tree gives information about the evolutionary stability of this gap junction protein in mammals and the divergence patterns of the GJA1 gene correspond closely with the species' origins and divergence patterns. According to all above, we suggested that GJA1 gene should be a good molecular marker for phylogenetic reconstruction, especially for investigation of the high-level relationships in mammals.
The estimated evolution ratios of all the branches in the phylogenetic tree show that strong purifying selection is the predominant factor in the evolution of the GJA1 gene in mammals. Although the ancestral sequence reconstruction gave different evolutionary rates among different lineages of mammals and significant results of different selection powers among the different locations of this protein, the average synonymous mutation rate is much higher than the non-synonymous mutation rate, meaning that natural selection tends to stabilize this protein and eliminate nucleotide mutations caused by the deleterious changes to the protein's functions.
The GJA1 gene codes for the protein connexin43, which is a critical factor necessary to maintain cardiac electrical conduction. In the cardiac working myocardium, three connexin proteins have been identified (connexin40, connexin43 and connexin45). The expression of each gene is in different locations in the heart. With a few exceptions, such as nodal tissues and the sinuatrial node, connexin43 is the most abundant connexin in the heart. Connexin40 and connexin45 were expressed in atrial tissue, nodal tissues and in the conduction system (Gros & Jongsma, Reference Gros and Jongsma1996). Connexin43 is the exclusive connexin protein in the intercalated disk (ID) region in adult ventricles (van Kempen et al., Reference van Kempen, ten Velde, Wessels, Oosthoek, Gros, Jongsma, Moorman and Lamers1995). According to the disease-related research, changes in connexin43 can disrupt the normal distribution of the gap junction in the ID of patients, and the amount of connexin43 protein was reduced by almost 40% in the heart of the people when the left ventricle is hypertrophied (Peters et al., Reference Peters, Green, Poole-Wilson and Severs1993). All these results suggest that connexin43 has a vital function in the heart involving conduction and ventricular action potentiality. The importance of the GJA1 gene in maintaining heart function was also supported by knockout work on the mouse. Knockout of the GJA1 gene causes death because of malignant ventricular arrhythmias (Danik et al., Reference Danik, Liu, Zhang, Suk, Morley, Fishman and Gutstein2004). Among the mammals, bats are faced with extreme demands imposed by flight and echolocation. Flying requires more than three times the energy cost of running (Neuweiler, Reference Neuweiler2000), bringing major challenges to the circulatory and respiratory systems. The heart is the engine of the circulatory system, and shows extreme adaptation in bats compared with other mammals. Given similarities in size and body weight, the heart proportion of bats is two to three times bigger than that of a mouse (Wachtlova et al., Reference Wachtlova, Poupa and Rakusan1970). Moreover, bats possess an extreme range of heart rates. The average heart beat frequencies of bats can be 500 beats per minute, rising to more than 1000 beats per minute in flight, as well as the lowest 4 beats per minute in hibernation (Neuweiler, Reference Neuweiler2000). This variation is probably the largest among the mammals and brings large stress on cardiac electrical conduction.
On the other hand, research shows that the GJA1 gene impacts upon the growth and maturation of follicles in the ovary (Teilmann, Reference Teilmann2005). Moreover, communication changes induced by expression changes of GJA1 mRNA and connexin43 protein levels influence the meiotic maturation and the change of atretic follicles (Nuttinck et al., Reference Nuttinck, Peynot, Humblot, Massip, Dessy and Flechon2000). Bats are among the few mammals that show delayed ovulation and fertilization (Oxberry, Reference Oxberry1979, and the unpublished data of Zhe Wang et al.). Although we found the amino acid Arg109 as the bats-specific amino acid substitution, which is caused by a nucleotide mutation (A to G) at the second codon position in the ancestral bats, more experimental work is needed to test the relationship between the substitution of this amino acid and bats-specific ecological characters. Like other connexins, connexin43 consists of the following topological domains: an amino terminus, two extracellular loops, four membrane-spanning domains, one cytoplasmic loop and a carboxy terminus. The parts of cytoplasmic loop and carboxy terminus are highly variable among the different connexins and were thought to be important for regulation (Kumar & Gilula, Reference Kumar and Gilula1996). Except for this, every domain has special amino acid sites and specific functions (Krutovskikh & Yamasaki, Reference Krutovskikh and Yamasaki2000). The 109 site is located on the inner loop between the second and third transform membrane domains, an area suggested as having an important influence on pH and voltage gating sensitivity in cells (Wang et al., Reference Wang, Li, Peracchia and Peracchia1996). Previous work compared the sequences of the GJA1 gene from some hibernating and non-hibernating mammals, and tried to investigate possible contribution of this gene to the cardiovascular physiological adaptation to hibernation (van der Heyden et al., Reference van der Heyden, van Eijk, Wilders, de Bakker and Opthof2004). The research of van der Heyden et al. (Reference van der Heyden, van Eijk, Wilders, de Bakker and Opthof2004) identified six amino acid sites (A116, T118, S244, H248, L254 and A349), where mammals differed from other non-mammalian vertebrates. However, in our alignment of amino acid sequences, we find that M. spasma did not fit with this pattern. M. spasma has a substitution at M254 that is shared with the chicken. This site is in the carboxy terminus of connexin43. Although there are some phosphorylation sites (Swenson et al., Reference Swenson, Piwnica-Worms, McNamee and Paul1990; Kanemitsu & Lau, Reference Kanemitsu and Lau1993; Saez et al., Reference Saez, Nairn, Czernik, Fishman, Spray and Hertzberg1997; Lampe et al., Reference Lampe, TenBroek, Burt, Kurata, Johnson and Lau2000) in this domain, the potential effect of this amino acid substitution is still not clear.
In the alignment of the amino acid sequences presented here, we also found several bats having distinct amino acids. Most of these amino acids are located in carboxy terminus, which is the most variable region among the different connexins and was also thought to be important for regulation (Kumar & Gilula, Reference Kumar and Gilula1996). Although carboxy terminus of connexin43 is longer than other connexins, only two functional roles have been suggested for this domain so far, one of which is that it contains the phosphorylation sites for different kinases (Krutovskikh & Yamasaki, Reference Krutovskikh and Yamasaki2000). The other suggested function is that the last six residues of connexin43 directly interact with the ZO-1 protein of tight junctions (Giepmans & Moolenaar, Reference Giepmans and Moolenaar1998; Toyofuku et al., Reference Toyofuku, Yabuki, Otsu, Kuzuya, Hori and Tada1998). Moreover, the rather short size of the carboxy terminus of connexin26 suggests that it might not be essential for normal function of connexin (Krutovskikh & Yamasaki, Reference Krutovskikh and Yamasaki2000). Several mutations in the carboxy terminus sites of connexin43 can cause heart disease, such as malformation and defects of laterality in patients (Britz-Cunningham et al., Reference Britz-Cunningham, Shah, Zuppan and Fletcher1995, but see Casey & Ballabio, Reference Casey and Ballabio1995; Penman Splitt et al., Reference Penman Splitt, Tsai, Burn and Goodship1997).
In conclusion, we firstly studied the GJA1 gene of bats. Using the molecular experiment method, we cloned and sequenced the GJA1 gene from 12 bats. We found the novel amino acid substitutions on the specific branches of bats which have the potentially affection of function. Using phylogenetic methods, the evolutionary analyses showed the high power of purifying selection in mammals, as well as in bats. The gene tree of GJA1 gene from mammals gives the high accordant topologies with good statistic support to that of the species tree based on the multiple molecular markers. This suggested the high evolutionary stability of GJA1 gene, which should be a good molecular marker to study the phylogenetic relationships among the mammals.
This work was funded by grants awarded to S. Zhang under the Key Construction Program of the National “985” Project and “211” Project. We thank Z. Wang for her data on procreation of bats.