案例二Alternative Splice Forms Influence Functions.pdf

ReportAlternative Splice Forms Influence Functions ofWhirlin in Mechanosensory Hair Cell StereociliaGraphical AbstractHighlightsdMajor WHRN isoforms WHRN-S and WHRN-L have distinctlocalizations within stereociliadLack of WHRN-S and WHRN-L causes short stereociliabundles and profound deafnessdIn absence of WHRN-L,WHRN-S can preserve stereocilialength in certain hair cellsdDifferential isoform expression underlies distinct phenotypesof known Whrn mutationsAuthorsSeham Ebrahim,Neil J.Ingham,Morag A.Lewis,.,Bechara Kachar,Johanna C.Pass,Karen P.SteelCorrespondencekaren.steelkcl.ac.ukIn BriefEbrahim et al.show that two majorisoforms of the WHRN gene have distinctlocalizations and functions within andacross mechanosensory hair cells in theinner ear,and isoform-specific mutationsare the likely cause of different auditorypathophysiologies associated withWHRN in mouse and humans.Accession NumbersAY739121.1NM_001008795.1Ebrahim et al.,2016,Cell Reports 15,935943May 3,2016 2016 The Authorshttp:/dx.doi.org/10.1016/j.celrep.2016.03.081Cell ReportsReportAlternative Splice Forms Influence Functionsof Whirlin in Mechanosensory Hair Cell StereociliaSeham Ebrahim,1Neil J.Ingham,1,2Morag A.Lewis,1Michael J.C.Rogers,3Runjia Cui,4Bechara Kachar,4Johanna C.Pass,1,2and Karen P.Steel1,2,3,*1Wolfson Centre for Age-Related Diseases,Kings College London,Guys Campus,London SE1 1UL,UK2Wellcome Trust Sanger Institute,Hinxton,Cambridge CB10 1SA,UK3MRC Institute of Hearing Research,Nottingham NG7 2RD,UK4National Institute on Deafness and Other Communications Disorders,NIH,Bethesda,MD 20892,USA*Correspondence:karen.steelkcl.ac.ukhttp:/dx.doi.org/10.1016/j.celrep.2016.03.081SUMMARYWHRN(DFNB31)mutations cause diverse hearingdisorders:profound deafness(DFNB31)or variablehearing loss in Usher syndrome type II.The knownrole of WHRN in stereocilia elongation does notexplain these different pathophysiologies.Usingspontaneous and targeted Whrn mutants,we showthat the major long(WHRN-L)and short(WHRN-S)isoforms of WHRN have distinct localizations withinstereocilia and also across hair cell types.Lack ofboth isoforms causes abnormally short stereociliaand profound deafness and vestibular dysfunction.WHRN-S expression,however,is sufficient to main-tain stereocilia bundle morphology and function ina subset of hair cells,resulting in some audi-tory response and no overt vestibular dysfunction.WHRN-S interacts with EPS8,and both are requiredat stereocilia tips for normal length regulation.WHRN-L localizes midway along the shorter stereo-cilia,at the level of inter-stereociliary links.We pro-pose that differential isoform expression underliesthe variable auditory and vestibular phenotypesassociated with WHRN mutations.INTRODUCTIONInner ear sensory hair cells(HCs)transduce sound and headmotion to electrical impulses via their mechanosensory hair bun-dles.Each hair bundle comprises dozens of specialized actin-filled protrusions,called stereocilia,organized in rows of gradedheights.Stereocilia heights and organization have importanteffects on the HCs operating range,sensitivity,and frequencyselectivity(Aranyosi and Freeman,2004),and a number of pro-teins involved in regulating stereocilia morphology are essentialfor normal hearing(Dror and Avraham,2009).The PDZ domain-containing protein whirlin(WHRN)has beenshown to localize to the tips of stereocilia,where it is involved inlength-regulation(Delprat et al.,2005;Holme et al.,2002;Manoret al.,2011),and also to the stereocilia base,where it is thoughttoplayaroleinthebundleorganizationduringdevelopment(Del-pratetal.,2005).Whrnconsistsof13exonswithtwomajorsplicevariants:a long isoform(referred to here as WHRN-L)encodedbyexons113,composedoftwoPDZdomainsattheNterminusfollowed by a proline-rich domain and a third PDZ at the C termi-nus;and a short form(WHRN-S),encoded by exons 613,whichlacks PDZ1 and PDZ2 of the N terminus(Mburu et al.,2003).Mutations in DFNB31(encoding WHRN)cause profound non-syndromic deafness,DFNB31(Mburu et al.,2003;Mustaphaet al.,2002;Tlili et al.,2005).DFNB31 has also been associatedwith Usher syndrome II,involving retinal degeneration and vari-able hearing loss(Aller et al.,2010;Audo et al.,2011;Ebermannet al.,2007).The underlying molecular pathogenic mechanismsare not known.We addressed this outstanding question in the current studyusing the Whrnwi/wimouse,in which both major WHRN iso-formsareablated(Mburuetal.,2003),andamousemutant,Whrntm1b(KOMP)Wtsi(referred to here as Whrntm1b),whichexpresses WHRN-S but not WHRN-L.While the Whrnwi/wimutant is profoundly deaf and exhibits circling and headbob-bing behavior,we found that the Whrntm1bmutant shows onlymoderate to severe hearing loss,suggesting that WHRN-Sis sufficient to prevent complete loss of auditory function aswell as vestibular dysfunction.We show that normal ster-eocilia are maintained in cochlear inner hair cells(IHCs)ofthe Whrntm1bmutant,but outer hair cell(OHC)stereociliamorphology,organization,and function are affected.Similarly,while a subset of vestibular HCs in the Whrntm1bmutant haveabnormally short stereocilia,the remainder have stereociliabundles with close-to-normal morphology.We use immunofluo-rescence with super-resolution,structured illumination micro-scopy(SIM)to determine the spatiotemporal localization ofboth WHRN isoforms along stereocilia.We show that WHRN-Slocalizes to the tips of stereocilia of IHCs from birth to adult-hood,colocalizes with the actin regulatory protein EPS8,andis required for normal stereocilia length regulation.We proposethat the localization of WHRN-L along the stereocilia shaftcoincides with inter-stereociliary links,such as lateral links orhorizontal top connectors,and its absence leads to disorga-nized bundles.Thus,WHRN isoforms are expressed differen-tially across HC types and within stereocilia,where they playdistinct roles in organization and elongation.Cell Reports 15,935943,May 3,2016 2016 The Authors935This is an open access article under the CC BY license(http:/creativecommons.org/licenses/by/4.0/).RESULTSWhrnwi/wiand Whrntm1b/tm1bMice Express DifferentWHRN IsoformsIn this study,we analyzed the whirler(Whrnwi/wi)mouse(Mburuet al.,2003;Mogensen et al.,2007),which has a spontaneousdeletion encompassing the majority of exons 610 of theWhrn gene,resulting in the ablation of both major isoforms ofWHRN,WHRN-L and WHRN-S(Figures 1A and 1B).We alsoused a mutant,Whrntm1b(KOMP)Wtsi(referred to from here asWhrntm1b),in which exon 4 of the Whrn gene is deleted and acassette including a b-galactosidase reporter(lacZ),is insertedinto intron 3(Figures S1A and S1B).Ensembl predicts five main WHRN isoforms:two short N-ter-minal isoforms(here referred to collectively as WHRN-N),amidsize isoform(WHRN-M),and several splice variants of bothFigure 1.Whrn Isoform Expression(A)Diagram of Whrn exons showing deletions inWhrnwi/wi(orange)and Whrntm1b/tm1b(blue),withregions coding for WHRN domains colored(green,PDZ1;blue,PDZ2;red,PDZ3;purple,proline-richdomain).Vertical arrows mark target region ofantibodies PB584 and PB595.Horizontal arrowsmark locations of primers.(B)Ensembl predictions of Whrn isoforms.Narrowboxes indicate UTRs and wider boxes protein-coding regions.Left:isoform names used inthispaper.Right:EnsembltranscriptIDs.ENSMUST00000155058 is classified as a retainedintron transcript but was recently found in theinner ear(Mathur et al.,2015b)and results inaproteinsequenceidenticaltotranscriptENSMUST00000119294,one of the WHRN-Sisoforms,when translated.(GenBank accessionnumbers provided in Table S1).Primers used todetect each isoform are shown on the relevantisoform.(CE)Transcriptsdetectedinwild-type(C),Whrntm1b/tm1b,(D)and Whrnwi/wi,and(E)innerears.Blacklinesindicatesplicejunctionsobserved by sequencing.Grey lines indicatesplice junctions inferred by presence of the iso-form,but not observed by sequencing.See also Figure S1.the long isoform(WHRN-L)and the shortC-terminal isoform(WHRN-S)(Figure 1B).We examined the transcripts present ininner ear tissue of each mutant by reversetranscriptase(RT)PCR followed by capil-lary sequencing(Figure 1B).In the wild-type inner ear,we foundbothWHRN-Nisoforms,WHRN-M,WHRN-L,and two variants of WHRN-S(Figure 1C).The two WHRN-N isoformswere also found in both WhrnwiandWhrntm1bhomozygotes(Figures 1D and1E).We were unable to test for the pres-ence of WHRN-M in Whrnwihomozygotesbecause the only specific primer set fell inthe deletion;if this isoform is transcribed,it is predicted to betruncated.Wedid confirm the presence of WHRN-M in Whrntm1bhomozygotes but could not observe the lack of exon 4 in thisisoform because exon 4 was not part of the sequence amplifiedby the specific primer set(Figure 1D).WHRN-L was detected in both Whrnwiand Whrntm1bhomozy-gotes(Figures 1D and 1E),and the deletion in each mutant allelewas observed by sequencing.In the Whrnwisequence,there is aclean break from exon 6 to exon 10,which introduces a frame-shift.The Whrntm1bsequence lacks exon 4 but has a sectionof 115 bp between exon 3 and exon 5 that is an exact matchto the region around the splice junction of the second exon ofEn2.It is likely that this is because the trapping cassette usedin making the original tm1a allele includes the mouse En2 spliceacceptor(Skarnes et al.,2011).If either the Whrnwior theWhrntm1bWHRN-L transcripts are translated,the mutations are936Cell Reports 15,935943,May 3,2016predicted to introduce a stop codon and result in a truncatedprotein.WHRN-S was also detected in both Whrnwiand Whrntm1bhomozygotes(Figures 1D and 1E).The Whrntm1bproduct wasidentical to the wild-type product.The deletion in the Whrnwiallele was clear,just as observed in the WHRN-L transcript.Inthe WHRN-S isoform,the Whrnwideletion starts in the 50UTRand removes the start codon,so it is unlikely that any protein isproduced from the mutant transcript.Whrnwi/wiand Whrntm1b/tm1bMice Show DistinctCochlear Physiology,Vestibular Phenotypes,and WhrnExpression PatternPhenotypic differences between the two mutants were immedi-ately apparent:Whrnwi/wimice are profoundly deaf and exhibitheadbobbing and circling behavior characteristic of severevestibular dysfunction,while Whrntm1b/tm1bmice intriguinglyshowed no overt vestibular abnormalities.We tested the contactrighting reflex,which uses a combination of visual,vestibular,and somatosensory inputs to make postural adjustments afterdisplacement,in adult mice.When Whrntm1b/tm1bmice wereFigure 2.Auditory Electrophysiology andWhrnExpressioninWhrnwi/wiandWhrntm1b/tm1bMice(A)Mean CAP threshold(SD)in P20 Whrnwimice.(B)Mean ABR threshold(SD)in P20 Whrnwimice.(C)Mean ABR threshold(SD)in P98 Whrnwimice.(D)Mean ABR threshold(SD)in P98 Whrntm1bmice.(E)Mean ABR masked tuning curves(SD)in P98Whrntm1bmice.Opensymbolsindicatemean(SD)12 kHz probe tone threshold.Filled symbols indi-cate mean(SD)masked tuning thresholds.(F)Mean 2F1-F2 DPOAE threshold(SD)in P98Whrntm1bmice.Red arrows on symbols indicatemaximum sound pressure level tested.(G)WaveformsfromWhrn+/wimouse(P17)showingrange of responses.CAP and SP responses areindicated.SP can be either negative(e.g.,12 kHzat 70 dB SPL)or positive(high frequencies,highintensities).(H)Waveforms from a Whrnwi/wimutant(P17),showing positive and negative SPs but no CAP.(I)Cochlear microphonic amplitudes plotted as afunction of stimulus intensity for a 6 kHz tone.(J)LacZstaininginorganofCortifromWhrntm1b/tm1bmice atP5(left:whole cochlea,right top:magnified)and P28(bottom right).See also Figure S2.placed on their backs,they immediatelyturned over to rest on all four feet,whileWhrnwi/wimice took several seconds toright themselves.Wenextevaluatedcochlearphysiology.In Whrnwi/wimice,recordings from theround window showed no compound ac-tion potentials(CAPs)even at the higheststimulusintensitiesused(Figure2A).Whrnwi/wimutants also showed no audi-tory brainstem response(ABR)to the highest sound stimulusat P20 and P98(Figures 2B and 2C).Whrn+/wimice had compa-rable ABR thresholds to Whrn+/+mice,suggesting the mutationwas completely recessive(Figure 2B).In contrast,14-week-oldmice homozygous for the Whrntm1b/tm1ballele showed a pro-found impairment only at higher frequencies(2430 kHz)andonly moderate impairment at lower frequencies(Figure 2D).Hearing loss in Whrntm1b/tm1bmice did not progress between4 and 14 weeks(data not shown).We assessed frequency tuning in Whrntm1bmice by measuringforward masked frequency tuning curves using a 12 kHz probetone presented at 20 dB above threshold(Figure 2E).Whrn+/tm1bmice produced a mean tuning curve with a sensitive tip locatedclosetothe12kHzprobetone.Incontrast,Whrntm1b/tm1bmutantsproduced a mean tuning curve that was flat across frequencies(Figure2E),suggestingimpairedfunctionofOHCs.OHCdysfunc-tion in Whrntm1b/tm1bmutants was further indicated by measure-ments of 2F1-F2distortion product otoacoustic emission(DPOAE)thresholds(Figure 2F).Whrn+/tm1bmice had sensitiveDPOAE thresholds for F2 frequencies,reflecting the shape ofthe ABR audiogram;Whrntm1b/tm1bmutants displayed moderateCell Reports 15,935943,May 3,2016937elevations in DPOAE threshold for F2 frequencies of 618 kHz,while DPOAEs from higher F2 frequencies were often not evokedeven at the highest stimulus levels tested(Figure 2F).HCs of Whrnwi/wimutant mice degenerate from around P21(Holme et al.,2002),so we analyzed mice younger than thisage to investigate HC function before the onset of degeneration.We recorded cochlear microphonics(an alternating voltage withthe same frequency as the stimulus)that reflect OHC functionand found that responses from Whrnwi/wimutants aged P13 toP20 were below detection limits,within the noise floor(Figure 2I).We also recorded summating potentials(Figures 2G and 2H),which are a sustained shift in potential for the duration ofthe 15-ms tone burst,and represent asymmetry of the receptorcurrent between positive and negative phases of the acousticstimulus(Dallos et al.,1972;Harvey and Steel,1992).Summat-ing potentials were detected in 27 of the 28 Whrnwi/wimutantsstudied,albeit at high stimulus intensities,suggesting thatsome HCs can depolarize in response to sound(example tracesin Figure 2H).These data are consistent with previous reportswhere mechanotransduction currents were recorded from earlypostnatal HCs in Whrnwi/wimutants(Stepanyan et al.,2006).Finally,we used lacZ staining to determine the endogenousexpression patternof WHRN-Lin theinner ear.Since thelacZ re-porter within the Whrntm1ballele is in the third intron(Figure S1A),it will be transcribed wherever WHRN-L is transcribed.The cod-ing sequence of WHRN-S begins partway through exon 6,so thelacZ will not report WHRN-S expression.At P5,when stereociliabundles are still developing,lacZ expression was detected inboth IHCs and OHCs of the organ of Corti(Figure 2I).In thevestibular organs lacZ expression was very strong in a subpop-ulation of HCs:the extrastriolar region of the utricular macula(Figures S2A and S2C)and the peripheral zone of the crista am-pularis(Figures S2B and S2D