Evidence of Echolocation in the Common Shrew from Molecular Convergence with Other Echolocating Mammals

© 2020 Academia Sinica, Taiwan Open Access Evidence of Echolocation in the Common Shrew from Molecular Convergence with Other Echolocating Mammals Simin Chai1, Ran Tian1, Xinghua Rong1, Guiting Li1, Bingyao Chen1, Wenhua Ren1, Shixia Xu1,*, and Guang Yang1,* 1Jiangsu Key Laboratory for Biodiversity and Biotechnology, College of Life Sciences, Nanjing Normal University, Nanjing 210023, China. *Correspondence: E-mail: [email protected] (Xu); [email protected] (Yang) E-mail: [email protected] (Chai); [email protected] (Tian); [email protected] (Rong); [email protected] (Li); [email protected] (Chen), [email protected] (Ren) Received 5 November 2019 / Accepted 22 January 2020 / Published 25 February 2020 Communicated by Benny K.K. Chan Along with sophisticated echolocation found in bats and toothed whales, the common shrew (Sorex araneus) was confirmed to possess echolocation ability based on behavioral and experimental evidence such as high-frequency twittering and close-range spatial orientation. However, whether echolocation in the common shrew is convergent with bats and dolphins at the molecular level remains poorly understood. In this study, we gathered the coding region sequences of 11 hearing-related genes from genome data and previous studies. Convergent evolutionary analyses identified 13 amino acid residues (seven in CDH23, five in OTOF, and one in PRESTIN) under strong convergent evolution shared among the common shrew and other echolocating mammals (bats and dolphins). Furthermore, a phylogenetic tree was constructed based on the combined amino acid dataset of convergent/parallel substitutions, sites with parallel radical property changes, and sites supporting echolocator-convergence; it supported the converged topology of the simple echolocator Sorex araneus and sophisticated echolocating bats with high posterior probability. This study gives evidence at the molecular level that the common shrew echolocate and provides novel insights into the convergent evolution between the common shrew and bats and dolphins. Key words: Convergent evolution, Echolocation, Sorex araneus, Radical property change, Amino acid substitution. Citation: Chai S, Tian R, Rong X, Li G, Chen B, Ren W, Xu S, Yang G. 2020. Evidence of echolocation in the common shrew from molecular convergence with other echolocating mammals. Zool Stud 59:4. doi:-10.6620/ZS.2020.59-04. BACKGROUND Adaptive convergent evolution, an important concept in Darwin’s theory of evolution, occurs when a similar trait evolves independently in more than one lineage from distantly diverged ancestors. In nature, convergent evolution commonly arises in the evolutionary trajectories of different species, includes the organic osmotic solute systems in bacteria, plants, and animals facing the challenges of water stress (Yancey et al. 1982); the ability to fly in birds and bats (Norberg and Rayner 1987); the reduced pigmentation of skin melanin among Europeans and Asians in adaptation to high-latitude environments (Wang et al. 2014); and the shared phenotypic traits of several unrelated marine mammals adapted to aquatic environments (Foote et al. 2015; Zhou et al. 2015). It has been suggested that convergent morphological evolution is attributed to convergent substitution at the amino acid layer (Zakon 2002; Wierer et al. 2012). Convergent amino acid substitutions refer to changes from different ancestral states to the same descendent amino acid within a pair of branches, whereas parallel substitutions are defined as independent changes from the same ancestral amino Zoological Studies 59: 4 (2020) doi:10.6620/ZS.2020.59-04 1 © 2020 Academia Sinica, Taiwan acid. Furthermore, multiple approaches to detect natural selection acting on genes and proteins have been put forward. Amino acid replacement diagnosis via testing radical physicochemical changes has been suggested as a sensitive method for detecting selection in proteincoding genes (McClellan and McCracken 2001; McClellan and Ellison 2010). Echolocation is a biological ability in animals to emit sonar pulses via clicking sounds and then utilize the returning echoes for spatial orientation, food acquisition, and predator avoidance. It is widely accepted that bats (Chiroptera, except for Pteropodidae) and toothed whales (Odontoceti) independently acquired the ability to echolocate over their evolutionary histories (Schnitzler et al. 2003; Eick et al. 2005; Steeman et al. 2009). In earlier studies, the peak frequencies of bats’ echolocation signals were found to range from 8 to > 200 kHz, with an effective working range of ≥ 20 meters (Fenton and Bell 1981; Fenton et al. 2014). As for bats’ echolocating counterparts underwater, toothed whales have the highest frequency of narrowband clicks on record (130–135 kHz) (Fenton et al. 2014; Kyhn et al. 2010). In recent studies, some hearing-related proteins have been identified as potentially essential to echolocation, such as the motor protein PRESTIN, which is expressed in the cochlear outer hair cells (OHCs) and charges voltage motility; homodimers of protein CDH23 and PCDH15, involved in hair bundle motility; transmembrane protein TMC1, expressed in the inner hair cells (IHCs) and OHCs, which regulates hair cell maturation. These proteins have been reported to be convergent in bats and toothed whales (Li et al. 2008; Liu et al. 2010 2011; Davies et al. 2012; Ru et al. 2012; Shen et al. 2012; Kirwan et al. 2013). Similar to other convergent evolution phenomena, convergence of echolocation has been suggested to be the result of convergence and/or parallelism in protein-coding sequences that alter hearing-related genes to some extent (Zakon 2002; Parker et al. 2013; Stern 2013). Evidence that echolocation is convergent can be found in the topologies in which echolocating bats and toothed whales were grouped together based on the amino acid sequences of PRESTIN. However, these topologies of convergence were not supported after excluding convergent substitution amino acid sites (Li et al. 2008; Liu et al. 2010). It is noteworthy that, in addition to the aforementioned well-known echolocating animals, a small insectivorous mammal, the common shrew (Sorex araneus), very likely possesses the ability to echolocate (Buchler 1976; Forsman and Malmquist 1988; Siemers et al. 2009). The frequency of ultrasonic calls from the common shrew was recorded at 50 kHz (Thomas and Jalili 2004). The low-amplitude, broadband, and frequency-modulated high-pitched laryngeal twittering ultrasonic calls from the common shrew are proposed to be used for simple and close-range spatial orientation, such as to detect habitat type and locate protective cover to minimize the risk of predation (Forsman and Malmquist 1988; Siemers et al. 2009). With the exception of behavioral experiments, ours is the first molecular study address the ultrasonic call emitting capability or the underlying mechanism of echolocation in the common shrew. We collected 11 hearing-related genes from different mammalian lineages under an enlarged toothed whale dataset. We present this evolutionary study with the goal of investigating whether the common shrew, as an ultrasonic call emitter and potential echolocator, has experienced similar adaptive evolution at the molecular level. To our knowledge, this is the first study to provide a novel insight into the molecular advancement of unexplored echolocation behavior in the common shrew. Further, it could expand our understanding of the evolution of complex diverse mechanisms involved in mammal echolocation. MATERIALS AND METHODS Data collection and sequence alignment The coding sequences of mammalian hearingrelated genes from previous studies (Liu et al. 2010 2011; Davies et al. 2012; Ru et al. 2012; Shen et al. 2012; Kirwan et al. 2013) were obtained according to accession numbers. Sequences of the common shrew (Sorex araneus) and hedgehog (Erinaceus europaeus) were downloaded from the Ensembl Genome Database (http://www.ensembl.org) and GenBank (http://www. ncbi.nlm.nih.gov). Sequences of the cetaceans used in this study—killer whale (Orcinus orca), Yangtze finless porpoise (Neophocaena asiaeorientalis), Yangtze river dolphin (Lipotes vexillifer), sperm whale (Physeter microcephalus), bowhead whale (Balaena mysticetus), minke whale (Balaenoptera acutorostrata)—were retrieved by BLAST using the cow (Bos taurus) sequence as the query, from the available genome data, Ensembl Genome Database and GenBank, or The Bowhead Whale Genome Resource (http://www. bowhead-whale.org/) (Keane et al. 2015). All taxonomy and sequence information are in table S1. Sequence alignments were done using the MUSCLE algorithm (Edgar 2004) in MEGA5 (Tamura et al. 2011). DAMBE (Xia 2013) was employed to examine the degree of nucleotide substitution saturation. page 2 of 11Zoological Studies 59: 4 (2020) © 2020 Academia Sinica, Taiwan Convergent and parallel amino acid substitutions Convergent and parallel amino acid substitutions between the common shrew and branches of sophisticated echolocators (echolocating bats and toothed whales) were conducted separately through pairwise comparison with their most recent common ancestors. Sequences of all ancestral nodes were inferred by the maximum likelihood method in PAML 4.7 (Yang 2007). To maximize the effectiveness of the control group and take phylogenetic relationships into consideration, we chose the hedgehog (Erinaceus europaeus), cow (Bos taurus), and Pteropodidae bats as representative non-echolocators, i.e., the control group. We employed the same analytical method to investigate the convergent and parallel amino acid substitutions between representative non-echolocating branches and two groups of advanced echolocating mammals. The true positive convergent and parallel substitutions denote substitutions that were only shared between the common shrew and advanced echolocating lineages, and were missing in the control group. Site-wise log-likelihood support Site-wise differences were measured for different topology hypotheses—i.e., hypotheses of the wellaccepted species tree and constrained echolocatorconverged trees—using RAxML version 8 (Stamatakis 2014) (Fig. 1). In this study, we computed respective site-wise log-likelihood support (SSLS) for a commonly accepted species tree (referred as H0) and three alternative constrained echolocator-converged topologies HA (H1 corresponds to all convergences in echolocators, H2 to the shrew convergences with the toothed whales, and H3 to the convergence between the common shrew and echolocating bats). Site-wise Fig. 1. Species trees and alternative constraint convergence topologies. H0 is the well-accepted species tree. H1, H2, and H3 are three alternate echolocator-converged topologies. H1-control, H2-control and H3-control refer to the constraint convergent topologies of representative non-echolocators (cow, hedgehog, and non-echolocating bats). Other mammals Other mammals Cow Other mammals Toothed whales Toothed whales Echolocating bats Echolocating bats Common shrew Hedgehog Other mammals Toothed whales H0 H1 Other mammals Cow Non-echolocating bats Hedgehog H1-control Other mammals Echolocating bats H3 Other mammals Non-echolocating bats Non-echolocating bats Hedgehog H3-control H2 Other mammals Cow Hedgehog H2-control Common shrew Common shrew Common shrew page 3 of 11 Zoological Studies 59: 4 (2020) © 2020 Academia Sinica, Taiwan differences between each pair of null and alternative hypotheses were calculated using: ΔSSLSi = SSLSi, HA—SSLSi, H0 where positive ∆SSLS for the ith amino acid in a locus indicates better fitness for the convergence in echolocators hypotheses. Likewise, there were three alternative hypotheses for non-echolocators in the control group, collectively referred to as HA-control. H1-control corresponds to hedgehog, cow, and nonecholocating bat convergence. H2-control is hedgehog and cow convergence, and H3-control is convergence of hedgehog and non-echolocating bat. Positive ΔSSLSi-control indicates more support for the convergence hypotheses in the control group, and negative ΔSSLSi-control indicates no support. ΔSSLSi-control = SSLSi, HA-control—SSLSi, H0 Amino acid sites with positive ΔSSLSi and negative ΔSSLSi-control values are regarded as true positive sites supported the echolocators clustering topologies. Analysis of radical property changes in amino acid residues Amino acid residues that might undergo adaptive changes were identified from radical changes in physiochemical properties (McClellan and McCracken 2001). The 31 distinguished structural and physicochemical properties of amino acid residues were analyzed in the software TreeSAAP under the goodnessof-fit test (Woolley et al. 2003). All 31 structural and physiochemical properties were graded into eight magnitude categories, and amino acid substitutions with changes in magnitude from six to eight with P values ≤ 0.05 were generally considered radical property changing residues (McClellan et al. 2005). Amino acid residues with radical changes between each of the two echolocating branches were counted. Phylogenetic reconstruction Phylogenetic reconstruction was done using the maximum likelihood (ML) method in RAxML version 8 and Bayesian inference (BI) in MrBayes version 3.2 (Ronquist et al. 2012) based on entire nucleotide sequences and amino acid sequences of each gene separately. We chose the best fitting model for each coding region sequence in accordance with AIC criterion by ModelTest version 3.7 (Posada and Crandall 1998) for nucleotide evolution and ProtTest version 3 (Darriba et al. 2011) for amino acid sequence evolution. Trees were built based on the third codon position and the first plus second codon positions, separately, by partitioning each codon. After that, ML and BI trees were constructed independently based on nonsynonymous amino acid substitutions and synonymous substitutions to reveal the variation in evolutionary signals. Multilocus concatenated tree reconstruction was conducted in species with at least five genes. ML bootstrap support was obtained from a rapid bootstrap analysis, and the best-scoring ML tree was identified from 1000 replicates under the best-fitting evolutionary model for each dataset. As for the BI method, the best tree of each gene was derived from running at least 1,000,000 generations of Metropolis-coupled Markov chains with a burn-in of 2500 replications until the standard deviation of split frequencies was below 0.01. Three-dimensional prediction and distribution of sites on protein structures We mapped the amino acids that might affect molecular convergent evolution of echolocation onto three-dimensional protein structures by PyMOL (The PyMOL Molecular Graphics System, Version 1.4 Schrödinger, LLC.). To mark the amino acids on a structurally obtained protein, we searched for gene names on the RCSB Protein Data Bank (http://www. rcsb.org/) (Bernstein et al. 1977). If tertiary structures were not obtained, the most likely three-dimensional model was predicted using the I-TASSER web tool according to the given quality criteria (http://zhanglab. ccmb.med.umich.edu/I-TASSER/) (Zhang 2008; Roy et al. 2010; Yang et al. 2015). We used human sequences of each protein gathered from UniProt as templates (http://www.uniprot.org/) (Apweiler et al. 2004). RESULTS Sequence convergence of echolocation in the common shrew Of the 11 hearing-related genes examined in this study, Kcnq4 (in cats, bottlenose dolphins, and finless porpoises) and Gjb6 (in cats) were excluded from our study due to premature stop codons. Besides, there was no substitution saturation in any sequence used in this study. A total of 93 true positive convergent or parallel amino acid substitutions were identified between the common shrew and other sophisticated echolocators (Table S2). Of them, four substitutions were convergent, and the remaining (89 of 93) were parallel. Except for page 4 of 11Zoological Studies 59: 4 (2020) © 2020 Academia Sinica, Taiwan Tmc1, 10 other hearing-related genes within our study were separately detected, and at least one convergent or parallel amino acid substitution was shared between the common shrew and sophisticated echolocators. Two convergent substitutions existed in CDH23 (G/ S1057N in the common shrew and the ancestor of Vespertilionoidea, T/R1693S in the common shrew and Chaerephon plicatus), and two occurred in SLC4A11 (N/D453S in the common shrew and the ancestor of Yangochiroptera, N/Q522H and Myotis lucifugus). There were 45 parallel amino acid substitutions in the protein CDH23, four parallel substitutions in DFNB59, three in GJB6, three in KCNQ4, four in MYO6, 17 in OTOF, three in PCDH15, one in POU3F4, eight in PRESTIN, and five in SLC4A11. In addition, four parallel substitutions in CDH23 (T449A, F1420V, I2105V, D3060E), one parallel substitution in DFNB59 (I72V), and two amino acids in OTOF (T1041S, S1312G) were shared by the common shrew, echolocating bats, and toothed whales. It was inferred that the parallel substitutions of all echolocating lineages might add weight to the molecular evidence that echolocation underwent convergent evolution. Figure 2 shows the distribution of convergent and parallel substitutions among echolocating mammals. In addition to the test of convergent and parallel amino acid substitutions, site-wise log-likelihood detected amino acids with a positive ΔSSLSi value, which suggests higher fitness to the topologies of echolocator-convergence. First, for the nested hypotheses of H0 and HA in most genes, there was no significant difference between the ΔSSLSi and ΔSSLSi-control in each hearing-related gene (Fig. S1). After ranking all the amino acid sites of each gene by ΔSSLSi, we observed that nearly half of the amino acid residues of each locus had positive ΔSSLSi, which supports the echolocator-converged hypotheses. To search for true positive amino acid sites supporting the echolocator-converged hypotheses, we screened out sites with positive ΔSSLSi-control that indicated support for non-echolocator-converged hypotheses. The true positive sites could offer more evidence for the molecular convergence of echolocation. Consequently, table S3 illustrates true positive amino acid sites with positive ΔSSLSi, supporting echolocators convergence in the top 5% within each locus. In the top 5% of amino acids ranked by ΔSSLSi of 11 hearing-related genes, 5.6% to 69.8% were regarded as true positive echolocator-convergence supporting sites. In this study, we focus not only on the convergent changes to the exact same amino acid, but also on the convergent amino acid physicochemical property changes. Changes in convergent amino acid physicochemical properties allow for detailed structural and functional contributions to the convergent evolution of echolocation. Analytical results of TreeSAAP implied a set of amino acids with convergent radical property changes in echolocating species. Twenty-four amino acid residues (13 in protein CDH23, one in GJB6, eight in OTOF, and two in PRESTIN) were found with radical property changes occurring only in the common shrew and either echolocating bats or dolphins Fig. 2. Radical property changes and convergent/parallel amino acid substitutions in echolocating mammals. Lineages in red, green, and blue indicate echolocating bats, toothed whales, and the common shrew, respectively. Convergent and parallel amino acid substitutions in common shrew and echolocating bat lineages are in pink boxes. Sites with radical property changes in the common shrew and echolocating bats are depicted in red boxes. Sites in light green indicate the convergent and parallel substitutions between common shrew and toothed whales, and sites in green boxes indicate radical property changes. 72 105 244 96 233 424 461 473 35 876 705 864 128 17 612 623 625 714 743 307 453 522 534 116 186 210 628 1041 1104 1235 1275 1312 1390 1478 1865 1873 15 449 498 728 943 1057 1069 1143 1301 1420 1569 1572 1655 1684 1693 1872 1907 2105 2124 2235 2339 2395 2484 2545 2778 2825 3060 3184 5 449 1011 1156 1320 1420 1504 2105 2121 3060 Bos taurus Erinaceus europaeus Homo sapiens Toothed whales Non-echolocating bats Canis familiaris Equus caballus Sorex araneus Baleen whales Mus musculus Rattus norvegicus Echolocating bats Cdh23 Dfnb59 Gjb6 Kcnq4 Myo6 Pcdh15 Otof Pou3f4 Slc4a11Prestin Tmc1 449 498 728 1338 1340 1710 1860 2038 2235 2358 3060 96 585 743 5 449 1710 3060 116 210 628 1104 1235 1260 1478 56 1235 72 1041 1312 page 5 of 11 Zoological Studies 59: 4 (2020) © 2020 Academia Sinica, Taiwan (Table S4). All 24 amino acids with radical property changes in echolocators showed various quantities of altered properties ranging from one to thirteen. Of these, residues 449, 1710, and 3060 in CDH23, and 1235 in OTOF possessed the radical property changes shared by all the ultrasound users, including the common shrew, echolocating bat, and toothed whale branches. Co-occurrences of the four amino acids in CDH23 and OTOF identified in echolocators might be particularly strong evidence for convergent evolution of echolocation. Figure 2 shows the distribution of amino acid sites with radical property changes among echolocating mammals. We also found that 13 amino acid sites—5, 449, 498, 728, 2235, 3060, and 3184 in protein CDH23; 116, 210, 1104, 1235, and 1478 in protein OTOF; and 743 in PRESTIN—overlapped in the detection of convergent molecular shifting in echolocators, which supported the common shrew clusters with other echolocating mammals (Table 1). Together, convergent/parallel amino acid substitutions in hearing-related genes characterized by radical property changes as well as echolocatorconverged topology support are speculated to contribute to the convergent evolution between echolocators. Phylogenetic reconstruction The ML and BI trees based on complete nucleotide alignments of each gene basically agreed with the well-accepted species tree. Neither the ML nor BI trees based on entire amino acid sequences, nor the trees based on different codon position alignments, could group the three subsets of ultrasonic users together. The trees based on synonymous or nonsynonymous amino acid substitutions did not cluster the common shrew with other sophisticated echolocators. The multilocus concatenated tree based on amino acid alignment clustered echolocating bats of two suborders and grouped toothed whales with baleen whales, which was the same as in the species tree (Fig. 3B). Surprisingly, the BI tree based on the combined datasets of sites possessing the radical property changes, true positive convergent and parallel substitutions, and true positive echolocator-converged hypotheses supporting amino acids of 11 genes grouped the common shrew and echolocating bats with a notably high posterior probability (Fig. 3A). DISCUSSION Convergent evolution of echolocation acts as a complex and specialized adaptation, and studies on bats and dolphins have given molecular evidence for natural selection in hearing-related genes (Li et al. 2008; Liu et al. 2010 2011; Davies et al. 2012; Ru et al. 2012; Shen et al. 2012; Kirwan et al. 2013). For instance, the motor protein PRESTIN expressed in OHCs was speculated to undergo sequence convergence, and therefore the lineages of echolocating bats were clustered with dolphins in gene trees based on amino acid alignment (Liu et al. 2010; Li et al. 2010). Similar to the environmental challenges that bats and dolphins face, such as low luminosity and dysphotia, the common shrew is characterized as a nocturnal vocal mammal with poor eyesight that lives exclusively in dense Table 1. Amino acids show evidence for hypotheses that echolocation is convergent. Sites overlapped in analyses of convergent/parallel substitutions, radical property changes, and site-wise log-likelihood support Gene Site* Convergent/parallel substitution Radical property changes SSLSi H1H2H3 Cdh23 5√√√ 449 √ √ √ 498 √ √ √ √ 728 √ √ √ √ 2235 √ √ √ √ 3060 √ √ √ √ 3184 √√√ Otof 116 √ √ √ √ 210 √ √ √ √ 1104 √√√ 1235 √√√ 1478 √ √ √ √ √ Prestin 743 √√√ Site*: With human protein sequence as a reference. page 6 of 11Zoological Studies 59: 4 (2020) © 2020 Academia Sinica, Taiwan vegetation (Merritt et al. 1994). Although the laryngealemitted broadband ultrasonic clicks of the common shrew do not emit as much high-sound pressure as the sophisticated echolocation clicks from bats or toothed whales, they are necessary and effective in probing protective cover and avoiding predators (Forsman and Malmquist 1988). However, echolocation in the common shrew has so far not been addressed through molecular evidence. This study used a series of molecular evolutionary analyses of hearing-related genes and provided clues regarding echolocation in the common shrew and convergent evolution between the common shrew and sophisticated echolocating bats and dolphins. Convergent or parallel substitutions, in combination with radical amino acid property changes and sitewise log-likelihood support (SSLS) analysis, identified 13 amino acids as strong candidate convergent sites between different mammalian echolocating groups (Table 1). These convergent sites are usually located within or near important functional domains of hearingrelated genes. For example, CDH23 is expressed in the cochlea and critical for proper morphogenesis and functioning of mechanosensitive hair bundles of the inner ear’s neurosensory cells (Frolenkov et al. 2004). Great quantities of mutations in CDH23 have been associated with hearing loss phenotypes and Usher syndrome type 1D, which is a rare inherited disorder characterized by the sensory neural hearing loss (Frolenkov et al. 2004; Miyagawa et al. 2012). The convergent residue 449 identified in this study is located in cadherin domain 4, where three calcium ions are usually bound at the interface to rigidify the connections by imparting a strong curvature to the full-length ectodomain, whereas residues 498, 728, and 2235 are located in cadherin domains 5, 7, and 21, respectively, and play a role in mediating adhesive specificity in the process Fig. 3. Convergence of echolocation in Sorex araneus and sophisticated echolocators. (A) Convergence of the sophisticated echolocating bats and simple echolocator Sorex araneus shown by the multilocus concatenated gene tree based on 481 potential convergence contributing amino acid sites in BI method. (B) Multilocus concatenated BI tree based on the whole multilocus concatenated amino acid alignment. (Numbers at internal nodes indicate posterior probability). 0.08 Balaena mysticetus Equus caballus Eonycteris spelaea Rattus norvegicus Erinaceus europaeus Canis lupus familiaris Homo sapiens Rousettus leschenaultii Chaerephon plicatus Hipposideros armiger Mus musculus Neophocaena asiaeorientalis Physeter macrocephalus Sorex araneus Balaenoptera acutorostrata Tursiops truncatus Myotis altarium Lipotes vexillifer Hipposideros larvatus Bos taurus Orcinus orca 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 98 100 0.08 Equus caballus Orcinus orca Balaenoptera acutorostrata Eonycteris spelaea Mus musculus Erinaceus europaeus Rousettus leschenaultii Neophocaena asiaeorientalis Sorex araneus Physeter macrocephalus Lipotes vexillifer Myotis altarium Balaena mysticetus Rattus norvegicus Hipposideros larvatus Canis lupus familiaris Homo sapiens Chaerephon plicatus Hipposideros armiger Bos taurus Tursiops truncatus 100 100 79 100 100 100 86 100 100 97 100 100 100 100 100 100 100 98 100 (A) (B) page 7 of 11 Zoological Studies 59: 4 (2020) © 2020 Academia Sinica, Taiwan of forming a coherent hair cell bundle and initiating contact between contiguous stereocilia (Boëda et al. 2002; Goodwin 2004). Residue 3184 is within a 20-residue peptide fragment corresponding to amino acid residues 3181–3200, which are crucial to form a complex structure binding to harmonin. This structure is necessary for the functional tip link of hair cells through multidentate binding modality (Pan et al. 2009; Caberlotto et al. 2011) (Fig. 4A and 4B). For the OTOF, five convergent amino acids were found in the β-strand of the C2A domain or the adjoining regions of domains C2B, C2D, and C2E, which may affect OTOF’s capacity to bind to the trigger Ca2+ and consequently release neurotransmitters in IHCs (Roux et al. 2006; Helfmann et al. 2011). It has been reported that mutations in position 1235 might be involved in the pathogenesis of auditory neuropathy (Zadro et al. 2010) (Fig. 4C). PRESTIN is an important motor protein of OHCs (Zheng et al. 2000) and is required for electro motility and cochlear amplification (Rodríguez-Ballesteros et al. 2008). Convergent residue 743 is perched in a coil structure of the C-terminus, which is a less conserved intracellular region, with E743D as a unique substitution in the common shrew and some echolocating bats (Fig. 4D); this suggests that it may function with other sequences or individual amino acids in the C-terminus to locate PRESTIN or sense voltage as a motor protein in cochlear OHCs (Zheng et al. 2001 2005; Pasqualetto et al. 2008). The convergent amino acid sites and their potential impacts on gene functions could provide novel insights into the evolution of echolocation in common shrews. Convergence can also be supported by discrepant phylogenetic clustering against the accepted species tree topology. Liu et al. (2010) found that trees based on PRESTIN alignment had good support for dolphins and echolocating bats. Coincidently, Li et al. (2010) reported bat-dolphin convergence based on protein PRESTIN sequence concurrently. Our phylogenetic reconstruction based on each coding region sequences generated topologies congruent with generally accepted species trees; however, the BI tree of combined Fig. 4. Predicted three-dimensional distribution of amino acids overlapped in convergent evolutionary analyses. (A), (B) CDH23 longer than 1500 amino acids were predicted into separate parts. Overlapped sites in CDH23 are indicated in blue. (C) Overlapped sites in OTOF are indicated in blue. (D) Overlapped sites in PRESTIN are indicated in blue. (A) (B) (D) (C) page 8 of 11Zoological Studies 59: 4 (2020) © 2020 Academia Sinica, Taiwan convergent amino acids (Fig. 3) had some unexpected results, such as a more significant convergence between the common shrew and echolocating bats. It is inferred that this kind of grouping might be due to the significant difference between habitat media for sound and body size. Foremost, both bats and the common shrew are mammals that hear in air, while dolphins use an entirely different medium. These differences in acoustic media lead to differences in sound speed and atmospheric absorption, as well as a different physical basis for auditory systems. Second, the high-frequency sounds emitted by echolocators are relative to the size of their transmitting aperture (Wilson et al. 2013; Madsen and Surlykke 2014). The body and aperture sizes of the common shrew are more similar to those of bats than toothed whales. CONCLUSIONS Evolutionary analyses of auditory genes on extended datasets including the common shrew and additional cetaceans give novel insights into the molecular convergence of echolocation in common shrews. Convergently evolving amino acids in the hearing-related proteins CDH23, OTOF, and PRESTIN were inferred to play a vital function in high-frequency hearing. Molecular evidence could further extend our understanding of echolocation evolution in mammals with diverse levels of echolocating capacity. To summarize, this study provides molecular evidence for the evolution of echolocation in the common shrew and potential convergence with other mammalian echolocators. Further studies are essential to clarify the inherent molecular mechanisms by which echolocation converged in these mammals, from comparative genomics to functional validation experiments. List of abbreviations BI, Bayesian Inference. IHCs, Inner Hair Cells. ML, Maximum Likelihood. OHCs, Outer Hair Cells. SSLS, Site-wise log-likelihood Support. Acknowledgments: We express our great gratitude to Kaiya Zhou, Xinrong Xu, and some students for their sample collecting. We also thank Zhengfei Wang, Di Sun, and other coworkers for their help and support, and Noah Last of Third Draft Editing for editing the English language. Authors’ contributions: BC, GY, WR, SC, and SX conceived this study. GL, SC, and XR collected and analyzed the data. SC was a major contributor in writing manuscript. GY, SX, and RT refined manuscript. All authors read and approved the final manuscript. Competing interests: BC, GY, WR, SC, RT, SX, GL, and XR declare that they have no conflict of interest. Availability of data and materials: The datasets supporting the conclusions of this article are included within the article and its additional files. Consent for publication: Not applicable. Ethics approval consent to participate: Not applicable. REFERENCES Apweiler R, Bairoch A, Wu CH, Barker WC, Boeckmann B, Ferro S, Gasteiger E, Huang H, Lopez R, Magrane M, Martin MJ, Natale DA, O’Donovan C, Redaschi N, Yeh LS. 2004. UniProt: the universal protein knowledgebase. Nucleic Acids Res 32:D115– D119. doi:10.1093/nar/gkh131. Bernstein FC, Koetzle TF, Williams GJ, Meyer EF, Brice MD, Rodgers JR, Kennard O, Shimanouchi T, Tasumi M. 1977. The protein data bank. FEBS J 80:319–324. doi:10.1111/j.14321033.1977.tb11885.x. Boëda B, El‐Amraoui A, Bahloul A, Goodyear R, Daviet L, Blanchard S, Perfettini I, Fath KR, Shorte S, Reiners J, Houdusse A, Legrain P, Wolfrum U, Richardson G, Petit C. 2002. Myosin VIIa, harmonin and cadherin 23, three Usher I gene products that cooperate to shape the sensory hair cell bundle. EMBO J 21:6689–6699. doi:10.1093/emboj/cdf689. Buchler ER. 1976. The use of echolocation by the wandering shrew (Sorex vagrans). Anim Behav 24:858–873. Caberlotto E, Michel V, Foucher I, Bahloul A, Goodyear RJ, Pepermans E, Michalski N, Perfettini I, Alegria-Prévot O, Chardenoux S, Do Cruzeiro M, Hardelin JP, Richardson GP, Avan P, Weil D, Petit C. 2011. Usher type 1G protein sans is a critical component of the tip-link complex, a structure controlling actin polymerization in stereocilia. P Natl Acad Sci USA 108:5825–5830. doi:10.1073/pnas.1017114108. Darriba D, Taboada GL, Doallo R, Posada D. 2011. ProtTest 3: fast selection of best-fit models of protein evolution. Bioinformatics 27:1164–1165. doi:10.1093/bioinformatics/btr088. Davies KTJ, Cotton JA, Kirwan JD, Teeling EC, Rossiter SJ. 2012. Parallel signatures of sequence evolution among hearing genes in echolocating mammals: an emerging model of genetic convergence. Heredity 108:480–489. doi:10.1038/hdy.2011.119. Edgar RC. 2004. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res 32:1792–1797. doi:10.1186/1471-2105-5-113. Eick GN, Jacobs DS, Matthee CA. 2005. A nuclear DNA phylogenetic perspective on the evolution of echolocation and historical biogeography of extant bats (Chiroptera). Mol Biol Evol 22:1869–1886. doi:10.1093/molbev/msi180. Fenton MB, Bell GP. 1981. Recognition of species of insectivorous bats by their echolocation calls. J Mammal 62:233–243. page 9 of 11 Zoological Studies 59: 4 (2020)