Hearing Aid for Vertebrates via Multiple Episodic AdaptiveEvents on Prestin Genes

Zhen Liu12 Gong-Hua Li1 Jing-Fei Huang1 Robert W Murphy13 and Peng Shi1

1State Key Laboratory of Genetic Resources and Evolution Kunming Institute of Zoology Chinese Academy of Sciences KunmingChina2Graduate School of the Chinese Academy of Sciences Beijing China3Centre for Biodiversity and Conservation Biology Royal Ontario Museum Toronto Canada

Corresponding author E-mail shipmailkizaccn

Associate editor Naruya Saitou

Abstract

Auditory detection is essential for survival and reproduction of vertebrates yet the genetic changes underlying the evolutionand diversity of hearing are poorly documented Recent discoveries concerning prestin which is responsible for cochlearamplification by electromotility provide an opportunity to redress this situation We identify prestin genes from the genomesof 14 vertebrates including three fishes one amphibian one lizard one bird and eight mammals An evolutionary analysis ofthese sequences and 34 previously known prestin genes reveals for the first time that this hearing gene was under positiveselection in the most recent common ancestor (MRCA) of tetrapods This discovery might document the genetic basis ofenhanced high sound sensibility in tetrapods An investigation of the adaptive gain and evolution of electromotility animportant evolutionary innovation for the highest hearing ability of mammals detects evidence for positive selections on theMRCA of mammals therians and placentals respectively It is suggested that electromotility determined by prestin mightinitially appear in the MRCA of mammals and its functional improvements might occur in the MRCA of therian andplacental mammals Our patch clamp experiments further support this hypothesis revealing the functional divergence ofvoltage-dependent nonlinear capacitance of prestin from platypus opossum and gerbil Moreover structure-based cdockinganalyses detect positively selected amino acids in the MRCA of placental mammals that are key residues in sulfate aniontransport This study provides new insights into the adaptation and functional diversity of hearing sensitivity in vertebrates byevolutionary and functional analysis of the hearing gene prestin

Key words prestin adaptation selection electromotility tetrapods NLC

IntroductionThe vertebrate auditory system is responsible for thedetection of sound signals in external environments whichis essential for survival and reproduction The ability to hearis typically characterized by parameters that quantifyfrequency selectivity and sensitivity These parametersgenerally include the high-frequency hearing limit (Fmax)(Masterton et al 1969) Previous behavioral studies revealthat Fmax varies enormously across different vertebratelineages (Fay 1996) For example most fishes can detectsounds ranging from 044 to 074 kHz (Fay 1988Popper and Fay 1997) although a few species are ableto detect ultrasonic sounds (Popper 2000 Mann et al2001) In most species of frogs Fmax values are about threeto five times higher than those in most fishes but5-foldlower than those of most of birds (Fay 1988) In generalmammals detect frequencies above 20 kHz and this isan important characteristic distinguishing them from othertetrapods (Fay 1988 Manley 1990 2000)

High-frequency hearing detection tends to increasephylogenetically from fishes to amphibians to mammalsCertainly a few species in each lineage have higher frequencydetection than their cohorts (Popper 2000 Mann et al 2001Feng et al 2006) This trend is also documented in anatom-

ical and electrophysiological studies (Heffner and Heffner1991 Fay 1996 Coffin et al 2004) yet little is known aboutthe underlying genetic variation and evolutionary trajectoriesof the gene(s) involved in high-frequency detection

Prestin a membrane motor protein driving the elonga-tion of outer hair cells (OHCs) is a major determinant ofcochlear amplification for hearing-frequency hearing(Brownell et al 1985 Belyantseva et al 2000 Zhenget al 2000 Brownell et al 2001) Prestin also namedSLC26A5 is a member of the solute carrier 26 (SLC26) genefamily It encodes anion exchangers capable of transportinga wide variety of monovalent and divalent anions (Sindicet al 2007 Detro-Dassen et al 2008) Some 10ndash12 trans-membrane domains linked by intra- and extracellular loopsin prestin are flanked by cytoplasmic N- and C-termini(Navaratnam et al 2005 Zheng et al 2005) Prestin a motorprotein essential for electromotility (Zheng et al 2000Dallos and Fakler 2002 Liberman et al 2002) can bemeasured by its robust voltage-dependent nonlinearcapacitance (NLC) (Santos-Sacchi 1991) In mammalsNLC and high-frequency hearing increase to a greater ex-tent than in other vertebrates without electromotility (Fay1988 Manley 1990 2000 Tan et al 2011) In additionprestin-knockout mice show significantly reduced

copy The Author 2012 Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution All rights reserved For permissions pleasee-mail journalspermissionsoupcom

Mol Biol Evol

Research

article

29(9)2187ndash2198 2012 doi101093molbevmss087 Advance Access publication March 13 2012 2187

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

high-frequency sensitivity (Liberman et al 2002 Wu et al2004) Mutations in this gene lead to nonsyndromicaudi-tory loss in humans (Liu et al 2003) Phylogenetic analysisof prestin sequence data unite echolocating bats that havethe ability of high-frequency sound detection (Li et al2008) although these bats are not monophyletic in the spe-cies tree (Teeling et al 2005) Surprisingly echolocatingwhales and bats also cluster together in phylogeneticanalyses of amino acid sequences (Li et al 2010 Liu Cottonet al 2010) The number of amino acid replacements ispositively corrected with an increase in ability to detecthigh frequencies in whales (Liu Rossiter et al 2010) Takentogether these studies strongly suggest that prestin is animportant factor in high-frequency detection

Herein we examine the evolutionary dynamics andselection pressure on prestin genes in vertebrate lineagesWe focus on detecting selection pressure in the initialemergence of land vertebrates because these animals tendto detect higher frequencies of sounds than do fishes Thisis followed by an investigation of the unique gain ofsomatic electromotility in mammalian OHCs the basisfor cochlear amplification (Brownell et al 1985 Ashmore1987) This amplification system gives mammals thegreatest ability to detect the highest frequencies amongvertebrates Our analyses of 48 vertebrate prestin genesdetect multiple episodes of adaptive evolution amongvertebrates suggesting that its function might change asorganisms evolve Furthermore our results from whole-cellpatch clamp functional experiments on platypus opossumand gerbil prestin support this hypothesis This studyprovides insights into the adaptation and functionaldiversity of high frequency hearing in vertebrates

Materials and Methods

Identification of prestin Genes and Collection ofFmax Data in MammalsThirty-four prestin sequences obtained from GenBankincluded two birds two fishes and 30 mammals In addi-tion we searched for prestin sequences in Ensembl (httpwwwensemblorg) and NCBI (httpwwwncbinlmnih-gov) from 14 vertebrate genomes that have high genomecoverage (6) The taxa included three fishes (medakaOryzias latipes stickleback Gasterosteus aculeatus andfugu Takifugu rubripes) one amphibian (frog Xenopustropicalis) one nonavian reptile (lizard Anolis carolinensis)one bird (turkey Meleagris gallopavo) and eight mammals(chimpanzee Pan troglodytes gorilla Gorilla gorilla orang-utan Pongo pygmaeus common marmoset Callithrixjacchus elephant Loxodonta africana guinea pig Caviaporcellus panda Ailuropoda melanoleuca and horse Equuscaballus) (supplementary table 1 Supplementary Materialonline) We used our previous pipeline for identifying geneswith multiple exons (Yang Shi et al 2005 Liu et al 2011) asbriefly follows TBlastN was employed to search prestinsequences in the genome databases using previously knownprestin protein sequences and known protein sequences ofprestins and the best hit genomic sequences were used to

conduct a proteinndashDNA comparison using Wise2 (httpwwwebiacukToolsWise2indexhtml) which providedthe exonintron structures and the full-length proteinsequences and cDNA sequences of the putative genesWe did not search low-coverage mammalian genomes(about 2 coverage) from each major vertebrate lineagebecause sequence mining of multiexon genes was not feasi-ble without chromosomal assemblies

Two analyses were performed in order to exclude falseprestin homologies First we blasted the putative genes inGenBank to ensure the best hits were known prestin genesSecond we constructed a tree using the neighbor-joiningmethod (Saitou and Nei 1987) with protein Poisson distances(Nei and Kumar 2000) We employed SLC26A6 the closestrelated gene to prestin in the SLC26 gene family (Franchiniand Elgoyhen 2006) as the outgroup to root the tree(supplementary fig 1 Supplementary Material online)

We collected Fmax values which measured high-frequency hearing and defined the highest frequencyaudible at 60 dB sound pressure level (SPL) (Fay 1996) frommammals in order to detect differences among species Val-ues were collected from two monotremes the platypus(Gates et al 1974) and echidna (Mills and Shepherd 2001)threemarsupials the opossum (Reimer 1995) quolls (Dasyur-sus Aitkin 1995) and brushtail possum (Trichosurus Aitkin1995) and 23 placentals including the elephant (Heffnerand Heffner 1982) guinea pig (Heffner et al 1971) horseshoebat (Long 1977) rat (Gourevitch 1965) bottlenose dolphin(Herman and Arbeit 1973) mouse (Ehret 1976) chinchilla(Clack 1966) macaque (Clack 1966) cat (Nienhuys and Clark1979) human (Wier et al 1977) squirrel monkey (Green1975) yellow baboon (Hienz et al 1982) lemur (Mitchellet al 1971) bushbaby (Heffner et al 1969b) tree shrew(Heffner et al 1969a) dog (Heffner 1983) ferret (Kellyet al 1986) kangaroo rat (Heffner and Masterton 1980)gerbil (Ryan 1976) rabbit (Heffner and Masterton 1980)little brown bat (Dalland 1965) common harbor seal (Mohl1968) and frog-eating bat (Ryan et al 1983)

Evolutionary AnalysesThe prestin sequences were initially aligned using ClustalW(Chenna et al 2003) followed by manual adjustmentsPairwise comparisons of the numbers of synonymous sub-stitutions per synonymous site (dS) and nonsynonymoussubstitutions per nonsynonymous site (dN) were estimatedby the modified NeindashGojobori method (Zhang et al 1998)in MEGA5 (Tamura et al 2011) Tree-based selection testswere calculated by the branch-site likelihood methodimplemented in PAML4 (Yang Wong et al 2005 Zhanget al 2005) because this test is conservative and robustagainst violations of various model assumptions unlikethe previous version of the test (Yang and Nielsen2002) which has a relatively high rate of false positives(Zhang 2004) This method compared two models Onemodel defined four classes of sites in terms of x the ratioof dN to dS Codons conserved throughout the tree wereassigned to site class 0 (estimated 0 x0 1) Site class1 contained codons that were neutral throughout the tree

Liu et al middot doi101093molbevmss087 MBE

2188

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

(estimated x1 5 1) Site class 2a included codons conservedon the background versus foreground branches (0x0 1)These were assumed to fall under positive selection (x2 1)on the foreground branches Finally site class 2b was com-prised of codons assumed to be neutral (x15 1) on the back-ground branches but positive selection on the foregroundbranch The corresponding null model fixed x2 5 1 whichdiffered from the alternativemodel A likelihood ratio test wasused to compare the two models When the likelihood of thealternativemodel was significantly higher than that of the nullmodel it was assumed to indicate positive selection on theforeground branchesWhere tests indicated positive selectionwe recorded the sites under selection according to high pos-terior probabilities (095) following Bayes empirical Bayes(BEB) prediction (Yang Wong et al 2005)

Gene Synthesis Cell Culture and TransientTransfectionThe entire coding regions of platypus and opossum weresynthesized and cloned into the expression vectorpcDNA31() (Invitrogen Inc) Correct orientation andreading frames were verified by sequencing analysis Inaddition the expression vector of gerbilrsquos prestin was giftedfrom Dallosrsquo lab (Northwestern University)

HEK293 cells were grown in 35-mm dishes containingDulbeccorsquos modified Eaglersquos medium supplemented with10 bovine calf serum (Invitrogen Inc) When cell conflu-ence reached roughly 50ndash60 of the surface area of thedishes cotransfection of the expression vectors of theprestins and pEGFP-N1 (GFP) were accomplished using lip-ofectamine 2000 transfection reagent (Invitrogen Inc) Weused a ratio of 1 lg (GFP)3 lg (prestin) added to 10-lllipofectamine The pEGFP-N1 plasmid generated a cytoplas-mic EGFP protein as an independent marker for successfultransfection of cells After 24- to 48-h incubation the suc-cessfully transfected cells were used for NLC measurements

Electrophysiological Experiments for NLCMeasurementsNLC was measured using whole-cell patch-clamp record-ings that were performed by HEKA EPC 10 USB (HEKAInstruments Inc) at room temperature (22ndash26 C) Electro-des were pulled from borosilicate glass with resistances of25ndash4 MX and filled with the internal solution containing140 mM CsCl 2 mM MgCl2 10 mM EGTA and 10 mMHEPES The cells were bathed during the recordings inan external solution containing 120 mM NaCl 20 mMTEA-Cl 2 mM CoCl 2 mM MgCl2 10 mM HEPES and5 mM glucose Both solutions were adjusted to pH 72Osmolarities of the internal and external solutions wereadjusted to 300 and 320 mOsml1 with glucose respec-tively Voltage-dependent capacitance was measured usingthe stair-step voltage protocol to obtain the parameters ofcharge movement (Huang and Santos-Sacchi 1993)Voltage was stepped from140 to 100 mV in 10 mV incre-ments of 10 ms each The capacitive currents were sampledat 100 kHz and low-pass filtered at 5 kHz using PatchMastersoftware (HEKA Instruments Inc) For each voltage the

measured membrane capacitance (Cm) was plotted asa function of membrane voltage (Vm) and fitted withthe derivative of a two-state Boltzmann function

Cm 5Qmaxa

expfrac12aethVm V1=2THORNeth1 thorn expfrac12 aethVm V1=2THORNTHORN2thorn Clin

where Qmax was the maximum charge transfer V12 was thevoltage at which the maximum charge was equally distributedacross the membrane Clin was the linear capacitance and a5zekT was the slope factor of the voltage dependence ofcharge transfer where k was Boltzmannrsquos constant T was ab-solute temperature z was valence and e was electron charge

The Clin was proportional to the surface area of themembrane (cell size) To compare the magnitude ofNLC obtained from different cells with different levels ofprestin expression as a function of cell size we normalizedthe NLC by the linear capacitance of the cells Becausedifferences in Qmax could have been caused by cell sizethe charge movement was normalized to Clin This quantitydesignated as charge density had units of fCpF

Predicting Prestin Protein Tertiary Structure andStructural SuperpositionThe tertiary structure of prestin was predicted online by theprofilendashprofile matching algorithms implemented in Phyre(httpwwwsbgbioicacukphyrehtmlcasp8html) Thequality of the predicted proteins was estimated by E valuewhere an E value 0001 corresponded to an estimatedprecision 95 (Kelley and Sternberg 2009)

Structural superposition was performed using the CEsoftware package (Shindyalov and Bourne 1998) Rootmean square deviation (RMSD) values between twoprotein structures were used to measure the degree ofstructural similarity for superposition RMSD values 4 Awere assumed to indicate similar overall structures (Tungand Yang 2007)

Binding Affinity of Prestin Protein and AnionsThe binding affinity between prestin and anions (ClHCO3

and SO42) was assessed using CDOCKER (Wu

et al 2003) in Accelrys Discovery Studio 21 (Accelrys SoftwareInc) a Charmm-based molecular docking tool for analyzingreceptor-ligand interactions The binding affinity was quanti-fied by CDOCKER interaction energy (CIE) the interactionenergy between the proteins and their binding ligands Gen-erally CIE 0 denoted the attraction of proteins and ligandsthe lower the value the higher the attraction In contrast CIE 0 indicated repulsion between ligands and proteins that isligands could not bind to acceptor proteins the higher theCIE value the stronger the repulsive force

Results and Discussion

Vertebrate prestin GenesHerein a total of 48 prestin sequences representing all majorvertebrate lineages except for turtles and crocodilians weresubjected to analyses (fig 1) Alignment of the amino acidsequences (supplementary fig 2 Supplementary Material

Adaptive Evolution of Vertebrate prestin Genes middot doi101093molbevmss087 MBE

2189

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

online) revealed that the N- and C-termini contained thehighest levels of variability We computed Poisson-correctedevolutionary distances for each region in each major lineageto further quantify regional sequence diversity Gene regionswere assigned according to the prestin membrane topologyproposed by several lines of functional evidence (Ludwiget al 2001 Matsuda et al 2004 Deak et al 2005 Rajagopalanet al 2006) The mean sequence distance in the N-terminalregion (NR) (fig 2) was significantly higher than that of thefull-length sequence in both teleosts (P 5 00047 FisherrsquosExact Test) and tetrapods (P 5 00027) The same was truefor the C-terminus (teleosts P 5 00005 tetrapods P 5

00001) In contrast the sequences of the transmembraneregion (TR) in both lineages were more conserved than thoseof other regions These comparisons revealed that NRs andC-terminal regions (CRs) evolved faster than TRs at theprotein sequence level in bothmajor vertebrate lineages Thisresult taken together with a previous result in mammals(Okoruwa et al 2008) suggested high levels of variationsin NRs and CRs might comprise a general pattern in verte-brate prestin genes

Adaptive Evolution of prestin Genes during theEmergence of TetrapodsTo further understand the evolutionary dynamics andselective pressure on prestin genes between fishes andtetrapods we compared the mean nonsynonymous nucle-

otide substitution distances between teleosts andtetrapods in NR CR TRs extracellular regions (ERs) andintracellular regions (IRs) using the modified NeindashGojoborimethod (Zhang et al 1998) As presented in figure 3 dNvalues were higher in NRs and CRs than in ER IR andTR This suggested either that positive selection had actedto favor amino acid replacements in the terminal regions orthat some of the amino acids in these regions enjoyed fewerfunctional constraints

To distinguish between positive selection and theabsence of functional constraints we compared dN anddS because the former was expected to exceed the latterin cases of positive selection We estimated the meansynonymous distance for the five regions between teleostsand tetrapods and found that the mean dN values weremuch lower than dS values in ER IR and TR In contrastthe dNdS ratio was larger than 10 in NR although themean dN value was not significantly higher than dS (fig3) The absence of significance between dN and dS mighthave owed to the saturation of synonymous and nonsy-nonymous substitutions (Tanaka and Nei 1989)

Positive selection between teleosts and tetrapods wasdifficult to detect by pairwise comparisons at this high levelof sequence divergence (05 dS 1) because subsequentsubstitutions may have hidden the signal of positive selec-tion especially if the emergence of tetrapods occurred ina relatively short evolutionary time span (Zhang et al 1998Shi et al 2003) To rectify this methodological limitation

FIG1 Phylogeny of vertebrates considered herein (Murphy et al 2004) Black blocks mark the species in which the prestin genes are newlyidentified in this study The species of bats and whales have been collapsed for illustration and detailed species information is presented insupplementary table 1 (Supplementary Material online) Positive selection tests are indicated by square labels

Liu et al middot doi101093molbevmss087 MBE

2190

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

we used an improved branch-site likelihood method(Zhang et al 2005) to look for selection signal while restrict-ing our signal test to the ancestral branch of tetrapods Thismethod more reliably and powerfully detected positive se-lection at this level of sequence divergence and saturationby computational simulation (Anisimova et al 2001 2002)

We assigned the ancestral branch of tetrapods as theforeground branch and all others as background branches(fig 1) This approach obtained a significantly higher likeli-hood for the alternative model than that of the null model(P 0001 v2 test) suggesting positive selection on prestinin the ancestor of tetrapods Conservative BEB (YangWong et al 2005 Zhang et al 2005) identified 20 sitesunder positive selection and with posterior probabilities 095 (table 1) In addition almost half of these sites werelocated in NRs and CRs This could have explained thehigher dN values found in these two regions

Functional and morphological audition innovations intetrapods are believed to be adaptations for processingairborne sound (Fritzsch 1991) For example when theancestral tetrapods moved onto land the spiracular pouchin fishes transformed into a tympanic middle ear thehyomandibular bone transformed into the stapes (Gaupp1898 1913 Werner 1960 Thomson 1966 Lombard andBolt 1988) and the basilar papilla formed as a uniquesensory adaptation for airborne sound detection (Retzius1881 1884) Consistent with these morphological innova-tions the rate of amino acid replacements in prestin geneswhich has been shown to be highly related to high-frequency sensitivity dramatically changed at the sametime The positive selection tests indicated an acceleratedrate of amino acid changes in the most recent commonancestor (MRCA) of tetrapods suggesting that prestingenes might have been involved in the functional shiftfrom low-frequency hearing in fishes to higher frequencyaudition in tetrapods To our knowledge this analysis pro-vided the first evidence that positive selection on prestin

FIG 2 Poisson distance of different regions of the prestin gene inteleosts and tetrapods The dashed lines show the average Poissondistance for the full length of the protein in teleosts (light gray) andin tetrapods (dark) An asterisk indicates the significant difference ofPoisson distance between different regions and the full-lengthprotein

FIG 3 Pairwise synonymous (dS) (filled circles) and nonsynonymous (dN) nucleotide distances (open triangles) and dNdS ratio (filled columns)for different regions of the prestin gene between teleosts and tetrapods

Adaptive Evolution of Vertebrate prestin Genes middot doi101093molbevmss087 MBE

2191

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

genes might have enhanced the high-frequency hearing oftetrapods

Multiple Episodic Adaptive Events on Mammalianprestin GenesSound amplification via electromotility gives mammals asuperior ability to hear high-frequency sound (Fay 1988Manley 1990 2000) This represents a major evolutionary ad-vancement Therefore it is important to evaluate the evolu-tionary tempo and mode of prestin in mammals because theprotein is responsible for somatic electromotility (Brownellet al 1985 Ashmore 1987 Zheng et al 2000)

Using placental and marsupial prestin sequencesFranchini and Elgoyhen (2006) report positive selection sig-nals in the MRCA of mammals However because mam-mals consist of three major clades of monotremesmarsupials and placentals the absence of monotremeprestin sequences in their study precludes understandingsof the evolution of this gene and its function in mammals

The available platypus genome (Warren et al 2008) anda clone of platypus prestin (Okoruwa et al 2008) providean opportunity to reexamine selective pressures on prestingenes for the MRCA of all mammals The test assignsthe ancestral branch that leading to all mammals to

the foreground and all other amniotes to backgroundbranches (fig 1) The results (table 1) show a significantlyhigher likelihood of the alternative model than thatof the null model (after multiple testing correction P

005 v2 test) indicating that the MRCA of mammals expe-rienced positive selection This result supports the conclu-sion that the origin of electromotility in mammals likelyhappened in the MRCA of all mammals (Franchini andElgoyhen 2006)

Furthermore analyses of all available audiogram datafrom monotremes marsupials and placentals reveals thatthe average upper hearing limit in monotremes (12 kHz) issignificantly lower than that of therians (38 kHz P 001t-test ) and placentals (619 kHz P 001) These obser-vations lead to the hypothesis that prestin experienced ad-ditional adaptive selection for detecting high-frequencysound shortly after the origin of electromotility duringthe evolution of mammals

To test this hypothesis we examined selection pressureson prestin genes on the ancestral branches that lead totherians and placentals respectively The two brancheswere separately assigned as foreground branches and allother amniotes as background branches (fig 1) Significantsignals of positive selection were detected on the ancestralbranches of therians and placentals respectively (table 1)

Table 1 Detection of Positive Selection in the Different Lineages of Prestin Orthologous Genes

Foreground Branches 2DLa P Valueb

Estimates of theParameters in theModified Model Ac Positively Selected Sitesd

Ancestral branchof tetrapods

3456 P 5 165 3 1028 p0 5 080027p1 5 009229p2a 5 009633p2b 5 001111v0 5 008133v2 5 30677

30E 49A 73A 155D169E 184L 192C 273L292L 305A 312S 368Q449K 460F 584N 594K602E 606K 608E 627E

Ancestral branchof mammals

527 P 5 0022 p0 5 068286p1 5 007127p2a 5 022263p2b 5 002324v0 5 006303v2 5 202367

44D 47K 50F 59N 68T73A 75N 76F 151L 247T257L 338L 415C 493I 540I588A 598E 599V 618P 631R

634P 662G 690N

Ancestral branchof therian mammals

2668 P 5 720 3 1026 p0 5 087403p1 5 009010p2a 5 003252p2b 5 000335v0 5 006793v2 5 3266711

124C 225M 260C 330N460F 521I 583G 617P

Ancestral branchof placental mammals

1013 P 5 0003 p0 5 086350p1 5 008660p2a 5 004535p2b 5 000455v0 5 006629v2 5 643691

68T 76F 196C 240I 268V

a Twice the difference between the log likelihood of the alternative model and that of the null model The modified model A with x2 fixed at 1 is the null model Themodified model A is used as the alternative modelb Multiple testing corrections are performedc x values are the nonsynonymoussynonymous rate ratios p0 is the proportion of codons that have x0 in all branches p1 is the proportion of codons that have x1 5 1 inall branches p2a is the proportion of codons that have x0 in the background branches but x2 in the foreground branches and p2b is the proportion of codons that have x1

in the background branches but x2 in the foreground branches Note that as long as x2 significantly exceeds 1 (as indicated by the likelihood ratio test) its exact value haslittle biological meaning due to the large estimation errord Sites with the Bayes empirical Bayes posterior probabilities higher than 95 are shown The sites are indexed by the amino acids at the site in the gerbil prestin The sitesin different regions are shown as followings italic in NR and CR single underlined in TR double underlined in ER and boxed in IR

Liu et al middot doi101093molbevmss087 MBE

2192

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

We also identified 23 8 and 6 positively selected sites in theMRCA of mammals therians and placentals respectively(table 1) The inferred positively selected amino acid sub-stitutions on these three branches might have provided ev-idence for functional divergence

Functional Variations of prestin in PlatypusOpossum and GerbilIf the ancestral branches of mammals therians and placen-tals are indeed under positive selection as described abovewe would expect that functional changes of prestin mightbe observed in monotreme marsupial and placental mam-mals Voltage-dependent NLC is one of the uniquecharacteristics of prestin and it is often used to measureprestin function (Santos-Sacchi 1991) NLC exhibits a bell-shaped dependence on membrane potential and it can befitted with the first derivative of a two-state Boltzmannfunction (Santos-Sacchi 1991 Oliver et al 2001) Conse-quently we measure the NLC of prestin genes from theplatypus opossum and gerbil the representative mono-treme marsupial and placental mammals respectively

As shown in figure 4 functional variations occur in threemammalian prestin genes For platypus prestin the param-eters of NLC from 12 cells after fitting two-state Boltzmannfunction are as follows QmaxClin 5 15 plusmn 23fCpF V12 5384 plusmn53 mV 1a5 4057 plusmn 24 mV (mean plusmn SE) Thesevalues are consistent with previous reports (Tan et al2011) Compared with the platypusrsquo prestin 1a value ofopossum normalized from 25 cells is significantly lower(P 5 001 Studentrsquos t-test fig 4A) suggesting changesin the reactivity of prestin to the membrane charge trans-fer Other parameters are also well fitted a two-state Boltz-mann function curve with following values of QmaxClin 597plusmn 16 fCpF V125301plusmn 37 mV and a5 5898plusmn 43mV

In comparison with platypus and opossum prestin thepeak voltage of NLC (V12) of gerbil prestin is significantlyshifted toward the hyperpolarizing direction (P 001Studentrsquos t-test fig 4B) with a value of 679 mv (n 5

20) Furthermore the curve-fitting parameters of chargedensity (QmaxClin 5 197 plusmn 27 fCpF) and 1a value(3582 plusmn 27 mV) also differ significantly from that ofopossum prestin (P 001 Studentrsquos t-test fig 4C)

In addition to functional variation among the threemammalian lineages functional changes occur betweennonmammalian vertebrates and mammals and betweenfish and tetrapods For example whereas the prestin geneof all three mammals exhibits a robust bell-shaped voltage-dependent NLC those of the zebrafish and chicken do notFurthermore the magnitude of NLC in chicken prestin isconsiderably larger than that of zebrafish (Tan et al 2011)

The generation of robust bell-shaped NLC in mamma-lian prestins from platypus and opossum to gerbil as well asthe functional improvement of NLC in nonmammalianvertebrates (eg zebrafish and chicken) might owe tomultiple positive selection events on the ancestralbranches of tetrapods all mammals therians andplacentals This possibility requires that positively selectedsites involve functional changes and site-directed mutagen-esis studies provide strong support for this For examplethe NLC experiment of chimera gerbil prestin constructedby exchanging 225M a positively selected site on theancestral branch of therians into the corresponding siteof gerbil prestin reveals functional changes of NLC andthe motility of prestin-expressing cells (Kumano et al2009) Another site 260C on the same branch and site196C on the ancestral branch of placentals also play animportant role in functional changes of prestin (Rajagopalanet al 2006 Kumano et al 2009 McGuire et al 2010) Themutagenesis of 415C a positively selected site in the MRCAof mammals can significantly decrease the magnitudes ofNLC suggesting 415C is required for the increase of NLCin mammals (McGuire et al 2010) Amino acid 192C is in-ferred to be positively selected on the ancestral branch oftetrapods and it plays an important role in charge move-ment of prestin (McGuire et al 2010) All of these sitesare in the list of positively selected sites (table 1)

Thus our sequence analysis is consistent with experimen-tal results and it may help discover more key functional sites

platypus

opossum

gerbil

20

30

40

50

60

70

1α(mV)

80

70

50

30

10

platypus

opossum

gerbil

)V

m(V

2 1

5

10

15

20

25

platypus

opossum

gerbil

ytisnedegrahc (Q

C fCpF)

nilxa

m

0

A

C

B

FIG 4 Voltage-dependent membrane capacitance (NLC) of HEKcells transiently transfected with prestin orthologs of platypusopossum and gerbil respectively The capacitancendashvoltage plots arefitted with two-state Boltzmann function (A) 1a values of prestinin three species are analyzed platypus 4057 plusmn 24 mV (n 5 12)opossum 5898 plusmn 43 mV (n 5 25) and gerbil 3582 plusmn 27 mV (n 5

20) (B) Comparison of V12 for three mammalian species platypus384 plusmn 53 mV (n 5 12) opossum 301 plusmn 37 mV (n 5 25) andgerbil 679 plusmn 42 mV (n 5 20) (C) Charge density for all threespecies is as follows platypus 15 plusmn 23 fCpF (n 5 12) opossum 97plusmn 16 fCpF (n 5 25) and gerbil 197 plusmn 27 fCpF (n 5 20) Allvalues are mean plusmn SE P 005 P 001

Adaptive Evolution of Vertebrate prestin Genes middot doi101093molbevmss087 MBE

2193

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

of prestin which is associated with the gain and subsequentevolution of NLC and electromotility More importantly ourevolutionary analysis of prestin genes taken together withbehavioral and functional analyses suggests that the geneexperienced at least three adaptive selection events inmammals alone The functional improvement of prestinmight be a very complex stepwise process in mammals

Positive Selection in the MRCA of PlacentalMammals for Transporting SulfateIn addition to changing NLC of electromotility the func-tion of prestin differs substantially among vertebrates withrespect to its ability to transport sulfate For example in thezebrafish and chicken prestin acts as an electrogenic anti-porter exchanging SO4

2 for Cl with a 11 stoichiometry(Schaechinger and Oliver 2007) This plesiomorphicfunction is not known to occur in mammals including ger-bils (Oliver et al 2001) The vertebrate lineage that expe-rienced this functional change remains to be identifiedRegardless functional data suggest that this transformationmight be associated with structural conformation changesof prestin (Schaechinger and Oliver 2007)

The 3D structure of prestin facilitates an evaluation ofwhether or not structural conformation changes are in-volved in functional sulfate transport Whereas the 3Dstructure of the C-terminus is known (Pasqualetto et al2010) it remains unknown for TRs Because these areimportant functional domains for anion transport (Baiet al 2009 McGuire et al 2010) our understanding ofthe changes in sulfate transport mechanisms are limitedUpon using Phyre to predict the 3D structure of TRs ofprestin the best hit of the gerbilrsquos prestin is a chloride chan-nel (PDB ID 5 1ots) (Dutzler et al 2003 E value 5 31 105) All other vertebrate prestins hit the same model

(1ots) with a predicted high accuracy (95) SignificantE values are found to range from 103 to 105 although thesimilarity between the template and query sequences is notso high (alignments in supplementary fig 3 SupplementaryMaterial online)

Technological limitations of 3D modeling and therelatively few available crystallographic structures of mem-brane proteins require us to evaluate the reliability of thepredicted 3D structure of prestin The predicted structureagrees with the key properties of prestin First thepredicted structure has 12-transmembrane helixes and thisis consistent with the secondary topology demonstrated bymost functional and modeling assays (Oliver et al 2001Deak et al 2005 Rajagopalan et al 2006) Second bothtemplate and prestin are anion channels that functionto conduct Cl across cell membranes in all vertebrates(Oliver et al 2001 Dutzler et al 2003 Schaechinger andOliver 2007) Third the accuracy of our 3D model canbe validated by comparisons of the predicted functionsby molecular docking and functional assays Whereas a pos-itive CIE value implies little or no affinity between prestinand anions a negative value suggests that prestin can driveanions freely through the membrane If our predicted pres-tin structure holds true then all of the prestins in non-mammals and mammals should have a high affinity forCl and HCO3

and negative CIE values Here Cl andHCO3

should be freely transported by prestins in all ver-tebrates as evidenced by functional experiments on thezebrafish chicken and gerbil (Oliver et al 2001Schaechinger and Oliver 2007) As expected the CIEs be-tween prestins and ClHCO3

in the zebrafish chickenand gerbil are all negative (data not shown)

For SO42 functional assays (Oliver et al 2001

Schaechinger and Oliver 2007) predict that the zebrafishand chicken prestins will have high-binding affinities and

BA

FIG 5 (A) Distribution of CIE values between prestins and SO42 in representative vertebrate species Asterisks indicate genes that provide

functional evidence (B) CIE values before and after artificial mutations of positively selected sites in the opossum and gerbil

Liu et al middot doi101093molbevmss087 MBE

2194

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

the gerbilrsquos prestin should have a lower affinity Consistentwith the functional assays CIE values of prestin-SO4

2 in thezebrafish and chicken are10968 and10068 respectivelysuggesting that their prestins can easily bind and transportSO4

2 In contrast the CIE value of prestin-SO42 in the ger-

bil is 519 indicating that this mammal cannot as easily trans-port SO4

2 (fig 5A) Therefore our modeled structure forprestin is congruent with the key aspects of transportingSO4

2 and ClHCO3 The structure appears to provide

a reliable means for determining the functional changesfor permeability to SO4

2 and for inferring positively selectedamino acid sites associated with changes in permeability inmammals

CIE values of prestin-SO42 can be used to evaluate when

functional change might have occurred in the vertebratesThe values for the zebrafish frog lizard chicken platypusand opossum are negative ranging from 1126 to10939 In contrast placental mammals have positiveCIE values ranging from 436 to 5951 (fig 5A) Thus thefunctional change in sulfate transport most likely has its or-igin in the MRCA of placental mammals and the positivelyselected amino acids in the MRCA of these animals might beassociated with changes in the ability to transport sulfate

We tested whether the functional change depended onthe structural conformation caused by positive selection ornot by performing structural analyses after artificially ex-changing positively selected sites identified on the branchof placental MRCA First we created two chimera prestinschimera opossum prestin (CP1) and chimera gerbil prestin(CP2) CP1 was constructed by inserting the positively se-lected sites of placental mammals into the correspondingsites of opossum prestin CP2 was established by exchang-ing the positively selected sites of gerbil prestin with thecorresponding sites of opossum prestin Second the 3Dstructures of CP1 and CP2 were modeled and reliabilityof the modeling was validated Both chimera prestins alsohit the same model (1ots) with highly predicted accuracyand significant E values (supplementary table 2 and align-ments in supplementary fig 3 Supplementary Material on-line) When inferred positively selected sites in the prestinsequence of opossum were replaced by those from the ger-bil prestin the CIE value changed from 985 to 4389 (fig5B) suggesting these positively selected amino acidsaffected the ability to transport sulfate This might haveowed to positively selected amino acids changing the localstructural conformation of the pore region of the channel

Superposition of the opossum prestin and CP1 struc-tures detected a mismatch between one region in opossumprestin (256ndash260) and the corresponding part in the CP1(256ndash260) The side chain of LYS256 in the CP1 projectedinto the pore and this might have blocked the entryway ofthe anion channel for the sulfate (fig 6A) The ability of thegerbil prestin to transport sulfate was fully rescued byreplacing positively selected sites with the amino acidsof opossum prestin (fig 5B) Superposition of the structuresof the gerbil prestin and CP2 showed that the helix turnregion (431ndash436) in the CP2 was similar to that of theopossum prestin (fig 6B) It might have swung out of

the channelrsquos pore to allow sulfate penetration and henceobtained a negative CIE value (fig 5B)

Results from the cdocking analysis and artificialmutagenesis of positively selected sites on the ancestralbranch of placental mammals support the above findingsto some extent However exactly how the positivelyselected residues influence the changes of SO4

2 transportability remains to be detailed One possible scenario is thatthe positively selected residues are located on the pore ofthe channel and they directly bind to SO4

2and Cl An-other scenario is that the positively selected sites surroundthe pore and indirectly bind the anions this would changethe conformation and thus result in defective sulfatetransport The absence of functional data precludes the

FIG 6 Structural superposition of the prestins (A) Structuralsuperposition of the opossum prestin (green color) and its chimeraCP1 (tan color) Amino acids under positive selection are marked bya red ball and stick The lsquolsquorsquorsquo denotes positively selected aminoacids in the chimera The side chain of the LYS256 in opossumprestin (purple ball and stick) and the corresponding residue (alsoLYS256) in CP1 (blue ball and stick) are shown (B) Structuralsuperposition of the gerbil prestin (yellow color) and the chimeraCP2 (cyan color) Only those amino acids under positive selectionare marked by red ball and stick Thelsquolsquorsquorsquo denotes positivelyselected amino acids in the CP2 The helix turn of gerbil prestin(431ndash436) and the CP2 (431ndash436) are colored by purple and bluerespectively Note that the PRO240 is not labeled because thissite in the structure prediction of CP2 was not available

Adaptive Evolution of Vertebrate prestin Genes middot doi101093molbevmss087 MBE

2195

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

unambiguous selection of one scenario Regardlessour cdocking results more strongly support the latterpossibility

The loss of SO42 transport ability seems to have hap-

pened in the MRCA of placental mammals The positivelyselected amino acids in this MRCA appear to be relatedto changes in sulfate transport ability The prestin of theMRCA of placental mammals as well as that in the majorityof if not all placental mammals has a novel anion transportfunction that might further enhance the ability to detect highfrequencies Interestingly our audiogram analysis supportsthis conjecture Placental mammals generally possess a supe-rior ability to detect higher frequencies (average upper hear-ing limit 619 kHz) than domarsupials (average upper hearinglimit 38 kHz)

ConclusionOur evolutionary analysis of prestin genes from 48 verte-brates provides evidence for multiple instances of positiveselection and functional divergence events during verte-brate evolution Prestin appears to have undergonepositive selection during the emergence of tetrapodsand for the first time adapted hearing for a terrestriallifestyle Moreover our analyses indicate three indepen-dent adaptive events in the evolution of mammalianprestin genes The first is predicted to have occurredin the MRCA of mammals and this possibly resulted fromthe gain of NLC and electromotility The second adaptiveevent seems to have occurred in the MRCA of therianmammals and this might be related with the functionalimprovement of electromotility The third appeared inthe MRCA of placentals which is associated with theability of prestin to transport solutes further enhanceshigh-frequency detection Our functional experimentssupport sequentially functional enhancements of prestinin monotremes marsupials and placentals respectivelyCombined with other results on bats and whales (Liet al 2008 2010 Liu Cotton et al 2010 Liu Rossiteret al 2010) these findings suggest that prestin genes un-derwent at least six positive selection events during theevolution of vertebrates This discovery represents an un-usually detailed understanding of how adaptation leadsto functional diversity for the perception of high-frequency sound

Supplementary MaterialSupplementary figures 1ndash3 and tables 1ndash2 are available atMolecular Biology and Evolution online (httpwwwmbeoxfordjournalsorg)

AcknowledgmentsWe thank Professor Peter Dallos (Northwestern UniversityIllinois USA) for the expression clone of the gerbil prestinWe also thank Professor Chen Zhang for consultationregarding electrophysiological techniques Valuable com-

ments were made by the members of the Shi lab This workwas supported by a start-up fund of lsquolsquoHundreds-TalentProgramrsquorsquo from Chinese Academy of Sciences and by grantsfrom Key Project from National Natural Science Founda-tion of China (30930015) to PS and by a Visiting Professor-ship for Senior International Scientists and Discovery GrantA3148 from the Natural Sciences and Engineering ResearchCouncil (Canada) to RWM

ReferencesAitkin L 1995 The auditory neurobiology of marsupials a review

Hear Res 82257ndash266Anisimova M Bielawski JP Yang Z 2001 Accuracy and power of the

likelihood ratio test in detecting adaptive molecular evolutionMol Biol Evol 181585ndash1592

Anisimova M Bielawski JP Yang Z 2002 Accuracy and power ofBayes prediction of amino acid sites under positive selectionMol Biol Evol 19950ndash958

Ashmore JF 1987 A fast motile response in guinea-pig outer haircells the cellular basis of the cochlear amplifier J Physiol388323ndash347

Bai JP Surguchev A Montoya S Aronson PS Santos-Sacchi JNavaratnam D 2009 Prestinrsquos anion transport and voltage-sensing capabilities are independent Biophys J 963179ndash3186

Belyantseva IA Adler HJ Curi R Frolenkov GI Kachar B 2000Expression and localization of prestin and the sugar transporterGLUT-5 during development of electromotility in cochlear outerhair cells J Neurosci 20RC116

Brownell WE Bader CR Bertrand D de Ribaupierre Y 1985 Evokedmechanical responses of isolated cochlear outer hair cellsScience 227194ndash196

Brownell WE Spector AA Raphael RM Popel AS 2001 Micro- andnanomechanics of the cochlear outer hair cell Annu Rev BiomedEng 3169ndash194

Chenna R Sugawara H Koike T Lopez R Gibson TJ Higgins DGThompson JD 2003 Multiple sequence alignment with theClustal series of programs Nucleic Acids Res 313497ndash3500

Clack TD 1966 Effect of signal duration on the auditory sensitivityof humans and monkeys (Macaca mulatta) J Acoust Soc Am401140ndash1146

Coffin A Kelley M Manley GA Popper AN 2004 Evolution ofsensory hair cells In Manley GA Fay RR Popper AN editorsEvolution of the vertebrate auditory system New York Springer-Verlag p 55ndash94

Dalland JI 1965 Hearing sensitivity in bats Science 1501185ndash1186Dallos P Fakler B 2002 Prestin a new type of motor protein Nat

Rev Mol Cell Biol 3104ndash111Deak L Zheng J Orem A Du GG Aguinaga S Matsuda K Dallos P

2005 Effects of cyclic nucleotides on the function of prestin JPhysiol 563483ndash496

Detro-Dassen S Schanzler M Lauks H Martin I zuBerstenhorst SMNothmann D Torres-Salazar D Hidalgo P Schmalzing G Fahlke C2008 Conserved dimeric subunit stoichiometry of SLC26multifunctional anion exchangers J Biol Chem 2834177ndash4188

Dutzler R Campbell EB MacKinnon R 2003 Gating the selectivityfilter in ClC chloride channels Science 300108ndash112

Ehret G 1976 Critical bands and filter characteristics in the ear ofthe housemouse (Mus musculus) Biol Cybern 2435ndash42

Fay R 1988 Hearing in vertebrates a psychophysics databookWinnetka (IL) Hill-Fay Associates

Fay R 1996 Structure and function in sound discrimination amongvertebrates In Webster DB Fay RR Popper AN editors Theevolutionary biology of hearing New York Springer-Verlagp 246ndash247

Liu et al middot doi101093molbevmss087 MBE

2196

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

Feng AS Narins PM Xu CH Lin WY Yu ZL Qiu Q Xu ZM Shen JX2006 Ultrasonic communication in frogs Nature 440333ndash336

Franchini LF Elgoyhen AB 2006 Adaptive evolution in mammalianproteins involved in cochlear outer hair cell electromotility MolPhylogenet Evol 41622ndash635

Fritzsch B 1991 The water-to-land transition evolution of thetetrapod basilar papilla middle ear and auditory nuclei InWebster DB Fay RR Popper AN editors The evolutionarybiology of hearing New York Springer-Verlag p 18

Gates GR Saunders JC Bock GR Aitkin LM Elliott MA 1974Peripheral auditory function in the platypus Ornithorhynchusanatinus J Acoust Soc Am 56152ndash156

Gaupp E 1898 Ontogenese und phylogenese des schalleitendenapparates bei den wirbeltieren Ergeb Anat Entwickl 8990ndash1149

Gaupp E 1913 Die reichertsche theorie (Hammer Amboss undKieferfrage) Arch Anat Physiol Anat Abt Suppl 1ndash416

Gourevitch G 1965 Auditory masking in the rat J Acoust Soc Am37439ndash443

Green S 1975 Auditory sensitivity and equal loudness in thesquirrel monkey (Saimiris ciureus) J Exp Anal Behav 23255ndash264

Heffner H Masterton B 1980 Hearing in glires domestic rabbitcotton rat feral house mouse and kangaroo rat J Acoust Soc Am681584ndash1599

Heffner HE 1983 Hearing in large and small dogs absolute thresholdsand size of the tympanic membrane Behav Neurosci 97310ndash318

Heffner HE Ravizza R Masterton B 1969a Hearing in primitivemammals III tree shrew (Tupaia glis) J Aud Res 912ndash18

Heffner HE Ravizza RJ Masterton B 1969b Hearing in primitivemammals IV bushbaby (Galago senegalensis) J Aud Res 919ndash23

Heffner R Heffner H Masterton B 1971 Behavioral measurementsof absolute and frequency-difference thresholds in guinea pigJ Acoust Soc Am 491888ndash1895

Heffner RS Heffner HE 1982 Hearing in the elephant (Elephasmaximus) absolute sensitivity frequency discrimination andsound localization J Comp Physiol Psychol 96926ndash944

Heffner RS Heffner HE 1991 Evolution of sound localization inmammals In Webster DB Fay RR Popper AN editors Theevolutionary biology of hearing New York Springer-Verlagp 691ndash711

Herman LM Arbeit WR 1973 Stimulus control and auditorydiscrimination learning sets in the bottlenose dolphin J Exp AnalBehav 19379ndash394

Hienz RD Turkkan JS Harris AH 1982 Pure tone thresholds in theyellow baboon (Papio cynocephalus) Hear Res 871ndash75

Huang G Santos-Sacchi J 1993 Mapping the distribution of theouter hair cell motility voltage sensor by electrical amputationBiophys J 652228ndash2236

Kelley LA Sternberg MJ 2009 Protein structure prediction on theWeb a case study using the Phyre server Nat Protoc 4363ndash371

Kelly JB Kavanagh GL Dalton JC 1986 Hearing in the ferret(Mustela putorius) thresholds for pure tone detection Hear Res24269ndash275

Kumano S Tan X He DZ Iida K Murakoshi M Wada H 2009Mutation-induced reinforcement of prestin-expressing cellsBiochem Biophys Res Commun 389569ndash574

Li G Wang J Rossiter SJ Jones G Cotton JA Zhang S 2008 Thehearing gene Prestin reunites echolocating bats Proc Natl AcadSci U S A 10513959ndash13964

Li Y Liu Z Shi P Zhang J 2010 The hearing gene Prestin unitesecholocating bats and whales Curr Biol 20R55ndashR56

Liberman MC Gao J He DZ Wu X Jia S Zuo J 2002 Prestin isrequired for electromotility of the outer hair cell and for thecochlear amplifier Nature 419300ndash304

Liu XZ Ouyang XM Xia XJ et al (17 co-authors) 2003 Prestina cochlear motor protein is defective in non-syndromic hearingloss Hum Mol Genet 121155ndash1162

Liu Y Cotton JA Shen B Han X Rossiter SJ Zhang S 2010Convergent sequence evolution between echolocating bats anddolphins Curr Biol 20R53ndashR54

Liu Y Rossiter SJ Han X Cotton JA Zhang S 2010 Cetaceans ona molecular fast track to ultrasonic hearing Curr Biol 201834ndash1839

Liu Z Li S Wang W Xu D Murphy RW Shi P 2011Parallel evolution of KCNQ4 in echolocating bats PLoS One6e26618

Lombard RE Bolt JR 1988 The evolution of the stapes in Paleozoictetrapods In Fritzsch B Ryan M Wilczynski W Hetherington TWalkowiak W editors The evolution of the amphibian auditorysystem New York Wiley and Sons p 37ndash67

Long GR 1977 Masked auditory thresholds from the batRhinolophus ferrumequinum J Comp Physiol A 116247ndash255

Ludwig J Oliver D Frank G Klocker N Gummer AW Fakler B 2001Reciprocal electromechanical properties of rat prestin themotor molecule from rat outer hair cells Proc Natl Acad Sci U SA 984178ndash4183

Manley GA 1990 Peripheral hearing mechanisms in reptiles andbirds New York Springer-Verlag

Manley GA 2000 Cochlear mechanisms from a phylogeneticviewpoint Proc Natl Acad Sci U S A 9711736ndash11743

Mann DA Higgs DM Tavolga WN Souza MJ Popper AN 2001Ultrasound detection by clupeiform fishes J Acoust Soc Am1093048ndash3054

Masterton B Heffner H Ravizza R 1969 The evolution of humanhearing J Acoust Soc Am 45966ndash985

Matsuda K Zheng J Du GG Klocker N Madison LD Dallos P 2004N-linked glycosylation sites of the motor protein prestin effectson membrane targeting and electrophysiological functionJ Neurochem 89928ndash938

McGuire RM Liu H Pereira FA Raphael RM 2010 Cysteinemutagenesis reveals transmembrane residues associated withcharge translocation in prestin J Biol Chem 2853103ndash3113

Mills DM Shepherd RK 2001 Distortion product otoacousticemission and auditory brainstem responses in the echidna(Tachyglossus aculeatus) J Assoc Res Otolaryngol 2130ndash146

Mitchell C Vernon J Herman P 1971 What does the lemur reallyhear J Acoust Soc Am 50710ndash711

Mohl B 1968 Auditory sensitivity of the common seal in air andwater J Aud Res 827ndash38

Murphy WJ Pevzner PA OrsquoBrien SJ 2004 Mammalian phyloge-nomics comes of age Trends Genet 20631ndash639

Navaratnam D Bai JP Samaranayake H Santos-Sacchi J 2005 N-terminal-mediated homomultimerization of prestin the outerhair cell motor protein Biophys J 893345ndash3352

Nei M Kumar S 2000 Molecular evolution and phylogenetics NewYork Oxford University Press

Nienhuys TG Clark GM 1979 Critical bands following the selectivedestruction of cochlear inner and outer hair cells Acta Oto-laryngol 88350ndash358

Okoruwa OE Weston MD Sanjeevi DC Millemon AR Fritzsch BHallworth R Beisel KW 2008 Evolutionary insights into theunique electromotility motor of mammalian outer hair cellsEvol Dev 10300ndash315

Oliver D He DZ Klocker N Ludwig J Schulte U Waldegger SRuppersberg JP Dallos P Fakler B 2001 Intracellular anions asthe voltage sensor of prestin the outer hair cell motor proteinScience 2922340ndash2343

Pasqualetto E Aiello R Gesiot L Bonetto G Bellanda MBattistutta R 2010 Structure of the cytosolic portion of themotor protein prestin and functional role of the STAS domainin SLC26SulP anion transporters J Mol Biol 400448ndash462

Popper AN 2000 Hair cell heterogeneity and ultrasonic hearingrecent advances in understanding fish hearing Philos Trans RSoc Lond B Biol Sci 3551277ndash1280

Adaptive Evolution of Vertebrate prestin Genes middot doi101093molbevmss087 MBE

2197

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from

Popper AN Fay RR 1997 Evolution of the ear and hearing issuesand questions Brain Behav Evol 50213ndash221

Rajagopalan L Patel N Madabushi S et al (12 co-authors) 2006Essential helix interactions in the anion transporter domain ofprestin revealed by evolutionary trace analysis J Neurosci2612727ndash12734

Reimer K 1995 Ontogeny of hearing in the marsupial Monodelphisdomestica as revealed by brainstem auditory evoked potentialsHear Res 92143ndash150

Retzius G 1881 Das gehororgan der wirbeltiere I Das gehororgander fische und amphibien Stockholm (Sweden) Samson andWallin

Retzius G 1884 Das gehororgan der wirbeltiere II Das gehororgander amphibien Stockholm (Sweden) Samson and Wallin

Ryan A 1976 Hearing sensitivity of the Mongolian gerbil Merionesunguiculatis J Acoust Soc Am 591222ndash1226

Ryan MJ Tuttle MD Barclay MR 1983 Behavioral responses of thefrog-eating bat Trachops cirrhosus to sonic frequencies J CompPhysiol A 150413ndash418

Saitou N Nei M 1987 The neighbor-joining method a newmethod for reconstructing phylogenetic trees Mol Biol Evol4406ndash425

Santos-Sacchi J 1991 Reversible inhibition of voltage-dependentouter hair cell motility and capacitance J Neurosci 113096ndash3110

Schaechinger TJ Oliver D 2007 Nonmammalian orthologs ofprestin (SLC26A5) are electrogenic divalentchloride anionexchangers Proc Natl Acad Sci U S A 1047693ndash7698

Shi P Zhang J Yang H Zhang YP 2003 Adaptive diversification ofbitter taste receptor genes in mammalian evolution Mol BiolEvol 20805ndash814

Shindyalov IN Bourne PE 1998 Protein structure alignment byincremental combinatorial extension (CE) of the optimal pathProtein Eng 11739ndash747

Sindic A Chang MH Mount DB Romero MF 2007 Renalphysiology of SLC26 anion exchangers Curr Opin NephrolHypertens 16484ndash490

Tamura K Peterson D Peterson N Stecher G Nei M Kumar S 2011MEGA5 molecular evolutionary genetics analysis using maxi-mum likelihood evolutionary distance and maximum parsi-mony methods Mol Biol Evol 282731ndash2739

Tan X Pecka JL Tang J Okoruwa OE Zhang Q Beisel KW He DZ2011 From zebrafish to mammal functional evolution ofprestin the motor protein of cochlear outer hair cells JNeurophysiol 10536ndash44

Tanaka T Nei M 1989 Positive Darwinian selection observed atthe variable-region genes of immunoglobulins Mol Biol Evol6447ndash459

Teeling EC Springer MS Madsen O Bates P OrsquoBrien SJ Murphy WJ2005 A molecular phylogeny for bats illuminates biogeographyand the fossil record Science 307580ndash584

Thomson KS 1966 The evolution of the tetrapod middle ear in therhipidistian-amphibian transition Am Zool 6379ndash397

Tung CH Yang JM 2007 fastSCOP a fast web server for recognizingprotein structural domains and SCOP superfamilies NucleicAcids Res 35W438ndashW443

Warren WC Hillier LW Marshall Graves JA et al (106 co-authors)2008 Genome analysis of the platypus reveals unique signaturesof evolution Nature 453175ndash183

Werner G 1960 Das labyrinth der wirbeltiere Jena (Germany)Fischer Verlagp309

Wier CC Jesteadt W Green DM 1977 Frequency discrimination asa function of frequency and sensation level J Acoust Soc Am61178ndash184

Wu G Robertson DH Brooks CL 3rd Vieth M 2003 Detailedanalysis of grid-based molecular docking a case study ofCDOCKER-A CHARMm-based MD docking algorithm J ComputChem 241549ndash1562

Wu X Gao J Guo Y Zuo J 2004 Hearing threshold elevationprecedes hair-cell loss in prestin knockout mice Brain Res MolBrain Res 12630ndash37

Yang H Shi P Zhang YP Zhang J 2005 Composition and evolutionof the V2r vomeronasal receptor gene repertoire in mice andrats Genomics 86306ndash315

Yang Z Nielsen R 2002 Codon-substitution models for detectingmolecular adaptation at individual sites along specific lineagesMol Biol Evol 19908ndash917

Yang Z Wong WS Nielsen R 2005 Bayes empirical Bayes inferenceof amino acid sites under positive selection Mol Biol Evol221107ndash1118

Zhang J 2004 Frequent false detection of positive selection by thelikelihood method with branch-site models Mol Biol Evol211332ndash1339

Zhang J Nielsen R Yang Z 2005 Evaluation of an improved branch-site likelihood method for detecting positive selection at themolecular level Mol Biol Evol 222472ndash2479

Zhang J Rosenberg HF Nei M 1998 Positive Darwinian selectionafter gene duplication in primate ribonuclease genes Proc NatlAcad Sci U S A 953708ndash3713

Zheng J Du GG Matsuda K Orem A Aguinaga S Deak LNavarrete E Madison LD Dallos P 2005 The C-terminus ofprestin influences nonlinear capacitance and plasma membranetargeting J Cell Sci 1182987ndash2996

Zheng J Shen W He DZ Long KB Madison LD Dallos P 2000Prestin is the motor protein of cochlear outer hair cells Nature405149ndash155

Liu et al middot doi101093molbevmss087 MBE

2198

at University of T

oronto Library on A

ugust 30 2012httpm

beoxfordjournalsorgD

ownloaded from