© 2013 Nature America, Inc. All rights reserved. University, Portland, Oregon, USA. 6 University of Virginia, Charlottesville, USA. Oregon, USA. 1 and bundles vestibular chick from 1,100 over detected we function bundle for critical are that several including proteins, 59 detected trometry bundle’s the quantify and detect complement. to accuracy and sensitivity the spectrometry mass quantitative scarce, are bundles Although essential. are strategies biochemical proteins, additional these cover To dis paralogs. by for compensated be can cases, some or, in ment may have escaped detection because they are essential during develop function bundle for essential proteins many identified studies have genetics Although possible. seems operation and assembly well as their concentrations and interactions—understanding bundles’ present—as are proteins which know we once task, specialized a out controlling release. potential, neurotransmitter membrane cell’s hair the modulates turn external in as channels the links of closing and tip opening the by bundle; the oscillate gated forces mechanically are which channels, kinocilium axonemal an develop ment, during least at and, stereocilia -filled 50–100 of sists Protruding con system. typically bundle a cell, hair sensory a of nervous surface apical the from central the to propagate that signals electrical to movement head and sound ear, converting inner the for signals mechanical transduces bundle hair vertebrate the purpose, An outstanding example of devoted to a a specialized single establishes a framework for future of characterization dynamic processes that shape bundle structure and function. functions essential for bundle structure and function. The quantitative mass spectrometry of bundle proteins reported here hub proteins, including RDX (radixin) and SLC9A3R2 (NHERF2), which interact with many bundle proteins and may perform membrane connectors; these values compared well to those obtained from mass spectrometry. Network analysis revealed several imaging Three-dimensional using electron tomography allowed us to count the number of actin-actin and cross-linkers actin- have been shown to regulate structure and dynamics, energy metabolism, phospholipid synthesis and cell signaling. from stereocilium, purified chick bundles; 336 of these were enriched significantly in bundles. Bundle proteins that we detected stimulation. Using mass spectrometry, we identified and quantified >1,100 proteins, present from a few to 400,000 copies per Hair bundles of the inner ear have a specialized structure and protein composition that underlies their sensitivity to mechanical Dongseok Choi James M Pagana Jung-Bum Shin bundle Molecular architecture of the chick vestibular hair nature nature Received 20 August 2012; accepted 17 December 2012; published online 20 January 2013; Department Department of Public Health & Preventive Medicine, Oregon Health & Science University, Portland, Oregon, USA. Department of University Neuroscience, of Virginia, Charlottesville, USA. Our previous analysis of hair-bundle proteins using mass spec mass using proteins hair-bundle of analysis previous Our Because hair bundles have a reduced protein complement and carry NEUR 3 Life Life Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California, USA. OSCI 4 . Here, using a more sensitive mass spectrometer, spectrometer, mass sensitive more a using Here, . 1 6 EN 1 , , Larry L David , Larry 2 , 2 , C 8 , , Nicholas E Sherman E , , Jocelyn F Krey

advance online publication online advance 8 These These authors contributed equally to this work. should Correspondence be addressed to P.G.B.-G. ( 1 . A bundle enlists ~100 transduction transduction ~100 enlists . A bundle 5 , , Manfred Auer 2 5 , 8 Department Department of Biochemistry & Molecular Biology, Oregon Health & Science University, Portland, Oregon, USA. , Ahmed , HassanAhmed 4 , , Erin D Jeffery

3 2 & Peter G Barr-Gillespie , others others , 3 has has 3 , , Zoltan Metlagel - - - ­ ­

4 2 , , Kateri J Spinelli Oregon Oregon Hearing Research Center, Oregon Health & Science University, Portland, are functional utricles age, this at chicks; 20–21 day embryonic from acceleration, ( utricles of epithelia (LC- and bundles hair from proteins spectrometry identified we mass MS/MS), tandem liquid-chromatographic Using Mass spectrometry of purified hair bundles RESULTS mechanisms. alternative out rules also families, protein key of absence the of basis the on and, bundles hair in pathways functional reveals spectrometry mass quantitative by offered view prehensive proteins transmembrane many to RDX couples that tein pro adaptor PDZ-domain a 2), regulator 3 member 9 family carrier important in hair-bundle function teins: the ezrin-radixin-moesin (ERM) family member RDX (radixin), Moreover, CALM (calmodulin). we two identified other key hub pro (PtdIns(4,5)P phosphatidylinositol-4,5-bisphosphate actin, by played function hair-bundle in roles central that the map highlights interaction an assembled we interactions, structural and proteins. To the bundle’splace proteome into a of network functional orably to mass-spectrometric estimates for cross-linker and connector counts those compared favconnectors; and linkers actin-membrane cross- tomography using and electron counted actin-actin stereocilia , ear. We imaged inner also the in function deafness-associated essential their confirming from expressed are proteins enriched ­identified those proteins selectively targeted to bundles. Many bundle- doi:10.1038/nn.331 Supplementary Fig. 1 Fig. Supplementary 7 . . Bundles (BUN) were enriched 40-fold, to ~80% purity 3 4 W.M. Keck Biomedical Mass Spectrometry Laboratory, , , Andrew N Tauscher 2 2 , , Hongyu Zhao , 7 7 2 Vollum Institute, Oregon Health & Science ), vestibular organs that detect linear linear detect that organs vestibular ), 5 , , and SLC9A3R2 (NHERF2; solute 2 , , Phillip A Wilmarth [email protected] 3 ,

e r O S 2 , Ca ), 6 ). . The com The . r u 2+ and and c 5 e ,  - - - ­

© 2013 Nature America, Inc. All rights reserved. by detecting proteins diluted in an an in diluted proteins human detecting by mixtures protein complex with experiments We mimicked protein? ous validation: does riBAQ accurately report the mole fraction of each of iBAQthe the linearity approach strated demon report previous a Although protein. each for value (riBAQ) iBAQa relative iBAQ generating to of sum iBAQ the all value values, peptides observable theoretically of number the by intensities precursor-peptide all of sum the divides To quantify hair-bundle proteins, we used the iBAQ algorithm, which Quantification using relative iBAQ values experiments. four all in 2,147 including experiments, more or two in proteins 2,753 identified we epithelia, utricular In peptide. unique single a with identified were (<2%) proteins 20 Only experiments. four all in identified were 728 ( bundles from proteins 1,125 identified we experiments, more or two in found proteins sidering con only by stringency Increasing UTR. and BUN of union the in ing member. We a of proteins or total 2,944 protein groups identified combined into protein groups, which were denoted by their best scor were peptides detected their of 20% than more shared that Proteins MaxQuant and engine search Andromeda the with data lyzing and hair-cell the peel 1b Fig. to eyelash an supporting-cell used layer from we the underlying stroma (UTR), layer ( technique epithelia twist-off the using below), (see bundles. in only detected proteins right, Far protein. each for relative points) represents Hue connecting right. fit at of bundles in coefficient enriched (power highly most enrichment those left, at indicated are epithelium the in enriched highly most Proteins enrichment. bundle-to-epithelium represents them connecting line the of slope the (right); bundle and (left) epithelium ( indicated. are epithelium and bundles in proteins abundant most The lowest. to highest from abundance, ( fit. Gaussian single epithelium; and bundles in detected ( fits. Gaussian single proteins; epithelium and ( unity. ref. from ( mean are points data standards; as proteins purified with immunoblots quantitative using measured fraction mole to values riBAQ from determined fraction mole relating curve ( origin the through constrained line a with fit were 10 of fractions mole (mean point data each for indicated is proteins identified of number The value. riBAQ to into spiked standards protein human of fraction mole relating ( summed. class each in proteins of fractions mole the with except left as same right, Bottom frequency). protein (by epithelium-enriched and unenriched bundle-enriched, as proteins bundle of representation left, Bottom experiments). more or (two epithelium and bundles in identified ( proteins. hair-bundle 1 Figure e r  proteins human abundant more the Only background. protein a as e R

) Enrichment distribution of proteins proteins of distribution Enrichment ) = 0.97). Data for CKB and GAPDH were were GAPDH and CKB for Data 0.97). = We identified proteins using an Orbitrap mass spectrometer, ana spectrometer, mass Orbitrap an using proteins We identified O S ± ± d y s.e.m. and are fit by by fit are and s.e.m. ). Four experiments each of BUN and UTR were analyzed. were UTR and of BUN each experiments Four ). = 1.02 = s.d.). The points corresponding to to corresponding points The s.d.). ) Abundance distribution of bundle bundle of distribution Abundance )

4 Quantitative analysis of chick chick of analysis Quantitative . Dotted lines in in lines Dotted . r u f ) Cumulative protein molar molar protein Cumulative ) g x ) Mole fractions of proteins in in proteins of fractions Mole ) ; ; c R = 0.999). ( 0.999). = −2 e , 10 , a ) Top, proteins Top, proteins ) b ) Calibration curve curve Calibration ) −3 a , and 10 and y b E. coli E. = 0.98 = correspond to to correspond c

Fig. 1 Fig. ) Calibration Calibration )

extract extract −4 x

1

a

1

and and . We normalized each protein’s each We. normalized

1 Supplementary Table1 Supplementary 1

, , we a sought more rigor 4 Escherichia coli Escherichia , 8 . To obtain utricular utricular obtain To . a f Cumulative protein abundance 0.25 0.50 0.75 1.00 Supplementary Proteins detected 527 HIST1H4 ACTG1 H2AFJ TUBB4 H3 AGR3 CKB GAPDH (1,125 proteins) ACTG1 CKB OCM TUBA5 TUBB4 GAPDH FSCN2 HSPA8 ENO1 2 143 B Bundles E B 500 extract extract Ranked proteins(byabundance) 455 0.3–3 1,095 proteins 2,753 proteins 9 <0.3 , 1,000 Epithelium 1 >3 Bundles 0 ); ); - - - - - 156 proteins(6% 2,559 proteins(93% 1,090 proteins(97%oftotal) 9 proteins(0.3% 6 proteins(0.5% . 29 proteins(1% 1 protein(0.1% 28 proteins(2%

enrichment Fold is nearly exact, at least above to a mole fraction of ~10 of fraction mole a to above at least exact, nearly is riBAQ fraction and mole between correspondence the that conclude fall on the regression line, only two of eight proteins were detected. We ( points data fraction 1.02 We low mole out fractions. (log a regression carried linear at proteins in detecting limitations the demonstrating were detected, similarly detected in both preparations. both in detected similarly lium proteins ( epithe and hair-bundle for similar was values abundance protein of ( stereocilium per ecules mol into protein bundle hair each for values fraction mole convert to value fraction actin’smole and estimate this used we below), see tomography; electron (by molecules actin filamentous 400,000 has present at minimal levels in stereocilia mol of number are monomers actin the for Because any protein. stereocilium per estimate ecules to used be can values fraction protein, mole measured accurately one of stereocilium per molecules of or stereocilium, per molecules of number total of knowledge With Quantification of hair-bundle proteins accurately. fraction ( 1:1 to close very was fit The protein. of amounts known with curves standard accurate generate to us allowed which standards, protein ( blotting concentrations of five hair-bundle proteins using quantitative immuno 1,500 y = 0.98 = To independently verify our riBAQ calibration, we measured the the measured we calibration, riBAQ our verify Toindependently Protein molefraction (2,753 proteins) 30 bundle–only ± Epithelium 0.01 × log × 0.01 2,000 47% ) ) ) ) ) ) x Fig. 1 Fig. advance online publication publication online advance ; ; ) 46% R 2,500 = 0.97), confirming that riBAQ values reported mole mole reported values riBAQ that confirming 0.97), = 7% Fig. Fig. 1 c ). We only used proteins for which we had purified purified had we which for proteins used We only ). 10 mole fraction) with the 10 the with fraction) mole d b d Fig. 1 Fig.

), indicating that low-abundance proteins were that ), low-abundance indicating Log riBAQ Number 10 g mass spectrometry Log mole fraction 100 200 300 400 500 Supplementary Table 1 Table Supplementary –6 –3 –2 –1 10 –5 –4 –4 –3 –2 –1 –0 –6 –5 0 0 –8 –6 b Log ). Although the 10 the Although ). –5 SNRNP200 spiked proteins SMARCC1 SMARCA2 DNAJC1 DNAJC1 COL14A 2/8 Log –6 GOLGA4 LRPPRC CTNND1 CTNNA1 PLXNA4 COL8A SLC1A3 NOMO1 LIMCH1 PITRM1 10 ZNF326 OTOGL CYFIP CLINT1 UGGT CNOT MYEF NCBP1 PLOD2 TOP2 AKAP9 PRPF DHX3 SF3B MYH9 –4 GPD2 CHD4 GLG1 ANK3 RPL 6/8 DMD molefraction GAK CPD VILL TNC 10 B 3 0 1 2 2 2 1 1 8 0 1 4 Epithelium –4 –3 (riBAQ) Log 8/8 –3 Epithelium –2 8/8 Bundles 10 8 –2 and because each stereocilium –2 bundlepurity –1 0 0 –1 nature nature Bundles c e −2 –0 Log10 mole fraction

−5 Number , 10 , 200 100 150 immunoblot –6 –3 –2 –1 50 –5 –4 data point did not not did point data ). The distribution distribution The ). 0 0 FYCO1 PLS2 PSMB7 LT SLC9A3R2 ADH1 CLIC CRABP1 YWHAE TPI1 FSCN2 OCM ACTG1 MYO1 MYO15A USH1 ATIC INTS SET ATP8B1 SRM ESPN CEP290 GSTO1 XIRP2 PLS1 RDX CAL PCDH1 UNC119B NUDC FBXO4 SPUNK PTPRF PFN2 PPI –6 –6 F −3 D Log mass spectrometry M 5 7 Log NEUR B C H –4 –5 4 and 10 and 5 Bundles only Contamination −5 –0.8 10 0.8 2 –2 level –4 (enrichment) MYO1 molefraction ( CALM coefficient Power 10 Fig. 1 Fig. –3 OSCI GAPD 2 0 FSCN2 C (UTR) riBAQ = TRIM2 WDR1 HYI EIF3 CAB39L PDZD7 ARL13B SERPINB7 ESPNL GRXCR1 CHAD MIM1 MRC2 PSMD1 SERPINF ALDOA CA7 –2 ANXA5 −4 H J 6 3 FSCN2 mole mole 1 –1 (BUN) EN 4 b 1 CK ). B 0 C E - - - ­

© 2013 Nature America, Inc. All rights reserved. rtis eetd n h 10 the in detected proteins log Table 1 in reported abundance and enrichment protein in with the error in the calibration slope measurement, to estimate errors ( proteins by immunoblotting bundle 14 and immunocytochemistry by proteins bundle 16 of tion distribu appropriate observed we analysis, our in detected proteins of enrichment bundle the of confirmation further As spectrometry. of fraction 0.30 a contamination of we measured BUN from 6 samples, and sets UTR ( bundles hair in trated FSCN2 by immunoblotting ( ( bundles hair from absent are proteins these that confirmed we istry, system of cells the vestibular supporting and cells hair of bodies cell in found protein filament intermediate surfaces apical cell supporting on present antigen), cell supporting the as known also J, receptor phosphatase ponent of the mitochondrial citric acid cycle; PTPRJ (protein tyrosine a 2), com dehydrogenase (malate MDH2 marker; reticulum plasmic cells supporting and cells hair of membranes basolateral the on found polypeptide), Na (ATPase, ATP1A1 bundles: hair from absent be to known proteins five of sample BUN the in presence the ( immunoblotting tive approxi purified 40-fold. mately thus were bundles bundles; hair ~80% contains 0.20 was proteins 81 these from estimated fraction contamination The 0. of value enrichment an assigned were bundles in detected not were that proteins experiments; utricle or more three in detected proteins are absent from stereocilia which nuclei, and mitochondria from proteins of abundances molar ratio for proteins known not to be in bystereocilia, measuring relative ( proteins of types three all of presence the reflected values enrichment binned of histogram broad The nants. contami cell-body contain also will preparation bundle the Finally, ~1. of enrichment an with bundles, in present be also will proteins stereocilia are not closed compartments, however, diffusible cell-body as large Because as 50-fold. (enrichment) ratio bundle-to-epithelium a have could proteins targeted Methods), (Online epithelium the of ~2% constitute bundles that Given function. for important ticularly symbols. italic by indicated are analysis) statistical to subject not (hence runs epithelium fewer or two in detected those symbols, boldface by indicated are fraction contamination the over ( enriched significantly proteins deafness-linked indicated; are genes deafness-associated by encoded Proteins abundance. indicates size symbol class; protein indicates Color enrichment. of order in ranked ( stoichiometry. 1:1 in be to known proteins highlight callouts magenta and Blue symbols. boldface by indicated are fraction ( enriched significantly proteins proteins; actin-associated abundant most the indicate callouts Red enrichment. panel in (key class protein indicates color Data-point abundance. of order in ranked proteins 2 Figure nature nature data, the through fit a generate to model mixed-effects linear a Supplementary Fig. Supplementary We used riBAQ measurement errors, propagated in combination combination in propagated errors, measurement riBAQ Weused To validate the estimated contamination fraction, we used quantita We the contamination determined the average fraction, BUN/UTR par be may bundles hair to targeted selectively are that Proteins 10 ± mole fraction (reversed axes from from axes (reversed fraction mole 0.25 (mean (mean 0.25 . . To error,the slope determine we log plotted

NEUR Protein composition of chick hair bundles. ( bundles. hair chick of composition Protein OSCI ± 1 0.14 (mean 3 b ; HSPA5 (heat shock protein 5; GRP78), an endo an GRP78), 5; protein shock HSPA5(heat ; ); symbol size represents bundle-to-epithelium bundle-to-epithelium represents size symbol ); ± EN s.d.), which suggests that the BUN preparation preparation BUN the that suggests which s.d.),

C 2c Supplementary Figs. 2a Figs. Supplementary Supplementary Fig. 2c Fig. Supplementary E

). As controls, we also examined actin and and actin examined also we controls, As ). advance online publication online advance 1 2 . . We chose 81 nuclear and mitochondrial Supplementary Fig. Supplementary 2a −2 ± Supplementary Figs. 4 Figs. Supplementary s.d.), to similar that estimated by mass –10 −4 mole fraction range. We used used We range. fraction mole P < 0.05) over the contamination contamination the over < 0.05) 1 5 1 . . Using immunocytochem + 4 Fig. 1 Fig. /K ; and VIM (), an (vimentin), VIM and ; i. 1 Fig. + ). In 14 measurements In). measurements 14 transporting, alpha 1 alpha transporting, , b b e b a ). ) Bundle proteins proteins ) Bundle and ) Hair bundle bundle ) Hair b ); ); each is concen ) for all human human all for ) 10 Supplementary Supplementary and 3 riBAQ against ) to measure measure to ) 5 P

< 0.05) < 0.05) ). ------

USH2A (usherin), all of which are especially large, may be too low. too be may large, especially are which of all (usherin), USH2A and 98) receptor coupled G-protein (VLGR1; GPR98 23), (cadherin CDH23 for values enrichment and Consequently, concentrations of slices. estimates UTR our corresponding the from than efficient less been have may slices these from proteins of recovery Methods), (Online wells sample to adjacent were that slices gel BUN for only purification (SCX) exchange cation strong out carried we Because hair-bundle riBAQ values for high-molecular-mass proteins, however. We of is tion limit the the below detection. may have underestimated domain (PDZ PDZD7 7) example, containing bundles—for in identified only ( purity bundle approximate against for in samples both ment, proteins all mole plotting detected fraction ( bundles in abundance molar proteome total the of 50% for accounted (0.6%) proteins seven Only families). protein shock heat kDa 90 and 70 the enzymes were abundant, as were HSPA8( 2) and HSP90AA1 (fascin (representingFSCN2 and ) beta and alpha all (representing TUBB4B and TUBA5 CPV3), (parvalbumin OCM kinase), creatine (brain CKB ), all representing actin, (gamma proteins. bundle-enriched for reliable more are values corrected much these tainty in the fraction, contamination ( ­bundles the we corrected samples, and epithelium hair-bundle in protein of each 1.00 the the through slope of was origin; constrained curve the calibration b Bundle a We highlighted proteins with high bundle or epithelium enrich epithelium or bundle high with proteins Wehighlighted ACTG1 included bundles hair in proteins abundant most The BUN/UTR enrichment

only 10 10 Molecules per stereocilium abundance of each protein to reflect its actual concentration in in concentration actual its reflect to protein each of abundance 10 10 10 10 10 10 10 10 ± –2 –1 0 1 2 1 2 3 4 5 6 0.03. Using the contamination fraction and the mole fraction fraction mole the and fraction contamination the Using 0.03. TUBB4B TUBA5 Supplementary Table 1 Supplementary Fig. 1 Fig. ACTG1 Average contaminantenrichment FSCN2 GRXCR1 LDH LDH MYO6 PDZD7 Stoichiometric (1:1)complexes RDX Membrane Traffi Metabolism Chaperon Nucleu /axoneme Energy Actin cytoskeleton A B ATP5A1 PCDH1 ATP5B PLS1 USH1 1 XIRP2 6 CLIC5 SPTAN SPTBN FSCN2 f 200 200 5 —presumably because their epithelium concentra epithelium their because —presumably c ). SPTBN1 C MYO15A SPTAN1 s PLS1 ATP6V1B2 ATP6V1A SLC9A3R2 ESPN e 1 1 DSTN ATP8B1 ACTR1 RDX FSCN1 GNB1 GNA Ranked bundleproteins(byabundance Ranked bundleproteins(byenrichment) ACTG1 ACTN4 A ACTN1 CLIC CAPZA1/2 TWF2 CAPZB2 MYO1 SHOOTIN1 5 PTPRQ 400 400 WDR1 ACTR3 ACTR2 OC9 C CAPZB2 s CKB ESPN 0

CDH2 CAPZA2 PRKAR2A PRKAC MYO7 MYO3 MYH1 USH2 3 ARPC1 ARPC ARPC Unknown Antioxidant Adhesion Ion homeostasis Lipid synthesis/transpor Signaling Otolithic membrane/ECM Protein synthesis Proteolysis TECTB A A ACTR3 ‘Deafness 0 A GPR9 proteins’ PLS2 B associated proteins ). Because of the substantial uncer of substantial the ). Because COCH A ACTR2 4 2 PCDH1 CDH2 600 600 ATB2B1 MYO3 8 Abundant actin– GSN MYO6 PLS3 OTOL B 3 TPM1 Fig. 1 Fig. 5 TECTA CAPZA1 1 MYO7 ARPC4 ARPC2 PFN2 A FLNB EPS8L2 g CLIC 800 800 RAC3 ). Some proteins were were proteins Some ). t CORO2 4 GRXCR1 e r B Fig. 1 Fig. ) O S 10 10 10 10 10 0.01 0.1 1 10 >100 1,000 1,000 SERPINB6 2 3 4 5 f OTOF ). Glycolytic Glycolytic ). OTOA stereocilium Molecules per r u COL11A enrichment epithelium Bundle/ MYH9 1 1,200 1,200 c e  - - -

© 2013 Nature America, Inc. All rights reserved. enrichment. Actincross-linking proteinsareindicatedinbold; thepredominantactin-to-membraneconnector familyisindicatedinunderlined bold. ratio indicatesthebundle/utricle enrichment;correctedmolecules perSCistheestimatedmolecules stereocilium,determinedwithriBAQquantification andcorrectedfor proteins inbundlesthataresummedtogether inanentry(alphabeticalorderafterprincipalentry);paralogBUNindicates whetheraparalogispresentinbundles;BUN/UTR principal entry, theproteinwithmostmass-spectrometricevidence;symbolisofficial proteinsymbol(basedonhumangenes);groupmemberslistssymbolsforall Identifier orproteingroupcolumngivesthe EnsemblorNCBIidentifier, orexperimentallyassignedproteingroupname;description columngivesthecommonnameof ENSGALP0000000764 ENSGALP0000000994 ENSGALP0000001236 ACTL6A ENSGALP E ENSGALP0000000104 ENSGALP0000003939 TPM1/3 ENSGALP0000001983 ENSGALP0000000952 ENSGALP0000003866 ENSGALP0000001563 E E ENSGALP0000001527 ENSGALP0000000641 XP_417532. ACTN1/2 ENSGAL ENSGA TWF2 ENSG ENS EN E NP_00117160 SPTAN1 ENSGALP0000000898 ENSGALP000000308 ENSGALP00000017 ENSGALP0000000416 ENSGALP0000002557 ERM NP_00117120 ENSGALP000000273 ENSGALP00000017 ENSGALP0000001 ENSGALP000000 EBP41/L3 ENSGALP00000 ENSGALP0000 ENSGALP000 ENSGALP00 ENSGALP0 ENSGALP ENSGAL ENSGA ENSGALP ENSGAL RAC ENSGA ENSG ENS EN ACT Identifier orproteingroup Table 1 in two or more experiments each of bundles and epithelium, 336 had a detected proteins 1,095 the Of bundles. hair in present reliably were e r  NSGALP0000001080 NSGALP0000002739 NSGALP0000000745 NSGALP0000001310

We used the contamination fraction to determine which proteins which determine to fraction contamination the Weused SGALP0000000219 SGALP0000001409 GALP0000001324 GALP0000003813 O S ALP0000002308 ALP0000001865 LP0000001505 LP0000003667 LP0000003975 P0000003648 P0000001303 P0000000191

0000000793 0000001010 0000000179 Quantification of Quantification actin and proteins actin-binding in chick stereocilia 000000795 r u 00000821 3 0000821 3 9 000136 03962 2293 c 963 76 26 9 0 e 4 6 1 4 0 2 6 9 2 1 1 7 4 0 3 5 0 7 9 8 2 5 4 3 4 8 1 5 9 5 2 4 9 1 6 6 6 4 4 5 2 0 7 0 Myosin phosphatase Rho interacting protein LIM domain and actin binding 1 Myosin Myosin IIIA Actin-like 6A (ARP4) Actin related protein 2/3 complex, Myosin VIIA Capping protein, alpha 1 1 ARP3 protein actin-related 3 homolog Plastin 3 ARP2 protein actin-related 2 homolog Myosin IIIB Plastin 2 Actin related protein 2/3 complex, Capping protein, alpha 2 Capping protein, beta 2 , alpha 1 WD repeat domain 1 Shootin-1 Twinfilin-2 Actinin, alpha 4 , beta, 1 non-erythrocytic (actin factor) depolymerizing ARP1 protein actin-related 1 homolog A, Fascin 1 Spectrin, alpha, 1 non-erythrocytic Solute carrier family 9 member 3 regulator 2 Chloride channel intracellular 5 Xin protein repeat-containing actin-binding 2 Plastin 1 Myosin VI Radixin Fascin 2 LIM LIM domain 7 Mediator of cell motility 1 protein Actin-related 10 homolog (ARP11) Coronin, actin binding protein, 1C Erythrocyte band 4.1 Huntingtin interacting protein 1 Myosin IH Myosin IC Myosin XVA Synapsin I Coronin, actin binding protein, 2B cysteine Glutaredoxin, rich 1 Espin-like Myosin, heavy chain 10, non-muscle Chloride channel intracellular 4 Ras-related C3 botulinum toxin substrate 3 EPS8-like 2 Actin related protein 2/3 complex, 2 Actin gamma 1 subunit subunit 4, 20 kDa subunit 1A, 41 kDa centractin alpha subunit subunit 2, 34 kDa Description

MPRIP LIMA1 MYO3A ACTN1 WDR1 SHOOTIN1 TWF2 ACTN4 SPTBN1 DSTN ACTR1A FSCN1 SPTAN1 SLC9A3R2 CLIC5 XIRP2 PLS1 MYO6 RDX FSCN2 ACTL6A ARPC4 MYO7A CAPZA1 TPM1 ACTR3 PLS3 ACTR2 MYO3B PLS2 ARPC1A CAPZA2 CAPZB2 ESPN ACTG1 LMO7 MEMO1 ACTR10 CORO1C EPB41 VCL HIP1 MYO1H MYO1C MYO15A SYN1 CORO2B GRXCR1 ESPNL MYH10 CLIC4 RAC3 EPS8L2 ARPC2 PFN2 GSN symbol Protein

the rate of incorrect assignments among enriched proteins enriched among assignments incorrect of rate the a withtion bundle-to-epithelium enrichment higher than the contamination frac

MPRIP LIMA1 ACTN1, ACTN1, ACTN2 WDR1 SHOOTIN1 TWF2, TWF2/WDR82 ACTN4 SPTBN1 DSTN ACTR1A FSCN1 SPTAN1 SLC9A3R2 CLIC5 XIRP2 PLS1 MYO6 RDX, EZR, MSN FSCN2 MYO3A ACTL6A ARPC4 MYO7A CAPZA1 TPM1, TPM3 ACTR3, ACTR3B PLS3 ACTR2 MYO3B PLS2 ARPC1A, ARPC1B CAPZA2 CAPZB2, CAPZB1 ESPN ACTBL2, ACTBL2, ACTC1, ACTG2 ACTG1, ACTA1, ACTA2, ACTB, LMO7 MEMO1 ACTR10 CORO1C EPB41, EPB41L3 VCL HIP1 MYO1H MYO1C MYO15A SYN1 CORO2B GRXCR1 ESPNL MYH10, MYH1 CLIC4, CLIC6 RAC3, RAC1, RAC2, RHOG EPS8L2 ARPC2 PFN2 GSN advance online publication publication online advance P value <0.05, adjustedrate<0.05,(FDR),false-discoveryvalue forthe members Group

Paralog in BUN? Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes No No No No No No No No No No No No No No No No No No No No No No No No

BUN/UTR BUN BUN only BUN BUN only nature nature 17 13 12 40 21 10 ratio 16 34 0.9 3 5 4 1.4 0.4 1.3 1.7 4 0.5 6 0.8 9 0.3 2 4 0.7 3 2 4 1.2 1.3 2 7 0.2 1.0 1.8 0.97 1.7 5 0.4 1.1 0.2 0.3 2.7 0.2 1.3 0.5 0.9 1.3 0.7 0.4 0.4 6 1.0 0.9

NEUR molecules perSC Corrected 400,000 40,000 OSCI 1,100 1,300 1,300 1,300 1,300 1,400 2,000 2,400 4,600 5,500 6,600 6,800 1 7 100 130 130 180 180 190 210 250 260 300 400 400 430 430 460 470 540 690 710 790 870 910 950 . Many . 18 19 24 31 36 46 47 50 65 69 81 91 95

1 3 4 5 6 9 EN

C E -

© 2013 Nature America, Inc. All rights reserved. in the chick Ensembl database, so cannot be detected with our mass mass our with detected be cannot so database, Ensembl chick the in utricles present not is 20–21 STRC and protein, day the of absence the embryonic for accounting in undetectable are transcripts not were USH1G detect: to (stereocilin) expect not did we proteins STRC Two these of detected. and (Sans) USH1G 5), partner-like (whirlin), CLRN1 (clarin 1), LHFPL5 (TMHS; lipoma HMGIC fusion successfully identified functionally important proteins. Only DFNB31 29 of enrichment bundle-to-epithelium average an had proteins 22 These stereocilia in expressed be to known proteins mouse 27 of deafness-associated 22 detected experiments spectrometry mass our esis, Table ( proteins in deafness known other Adding expressed are stereocilia. that dysfunction vestibular or deafness with ated function. demon vestibular and auditory with for significance proteins strated for reference rich a thus is epithelium over (MGD) Database Genome ( significantly also were >2-fold enriched Proteins database. Ensembl to in genes of 0.7% the human the ~23,500 corresponding , or deafness with annotated entries 163 has database OMIM ( <2-fold enriched proteins 2,667 database OMIM the in associated were deafness with >2-fold enriched proteins 277 the of 4% bundles; “deafness” “Usher syndrome.”and data: OMIM the ( groups tein from BUN and identified UTR identify samples, including redundant proteins proteins present in pro of to list the used in was proteins 2012) ­deafness-associated October (downloaded IDs accession marker MGI and genes human to mapping their and terms (OMIM) ( pro teins those encoded by deafness- or labeled vestibular-dysfunction-associated genes and enrichment bundle by proteins ranked We ‘Deafness proteins’ are enriched in hair bundles dances ( relativeabunexpected thestoichiometric presentatcomplexeswere ( cilium actin-associatedproteins were present at100 or more copies perstereo Chr isthehumanchromosome;descriptiongivesdescriptivenameforprotein;BUN/UTRratiobundle/utricleenrichmentinchick. protein mappingtoahumandeafness-associatedgene;homologousgeneistheEnsemblidentifierforwhichgivenchickmaps; The humandeafnesslocuscolumngivestheunmappedidentifier;chickproteinidentifierischickenEnsemblforgeneencoding DFNX5 DFNX5 DFNB85 DFNB57 DFNB57 DFNB57 DFNB55 DFNB55 DFNA32 DFNA18 DFNA16 AUNA1 AUNA1 Human deafnesslocus Table 2 nature nature detectable known, of 10% about only Thus approach. spectrometric P i. 2 Fig. 10 < The OMIM and MGI databases do not include all genes associ genes all include not do databases MGI and OMIM The hair in enriched were detected proteins deafness-associated Most ±

1 12 (mean (mean 12 b

), including those implicated by means of targeted mutagen targeted of means by implicated those including ), Fig. 2 Fig. 2 Fig. −2 NEUR ). A list of 7,112 Online Mendelian Inheritance in Man Man in Inheritance Mendelian Online 7,112 of list A ). Candidates Candidates for mapped but genes uncloned deafness-associated ) associated with mouse deafness entries in the Mouse Mouse the in entries deafness mouse with associated ) Supplementary Table 1 Supplementary a a OSCI ), which independently corroborated our quantification. and ± s.e.m.), confirming that the enrichment analysis analysis enrichment the that confirming s.e.m.), EN Table C ENSGALP00000031 ENSGALP000 ENSGA ENSGALP0000003255 ENSGALP00000015 ENSGALP000 ENSGA ENSGALP0000002226 ENSGALP00000039 ENSGALP000 ENSGA ENSGALP0000002741 ENSGALP00000027 Chicken proteinidentifier E

advance online publication online advance 1 1 9 ). Moreover, proteins known to be in 1:1 in be to Moreover,known proteins). . The list of proteins enriched . in The list of bundles proteins enriched LP0000000782 LP0000002232 LP0000001563 P 0000952 0001387 0000416 < 10 < ). We used two terms to search search to terms two We ). used 1 8 , compared to only 0.7% of of 0.7% only to compared , 74 05 75 39 −4 , Fisher’s exact test). The The test). exact Fisher’s , 6 5 0 6 4 8 3 3 9 5 2 7 1 Homologous humangene ENSG0000016570 ENSG0000010 ENSG00 E ENSG0000018716 ENSG0000010 ENSG00 E ENSG0000017710 ENSG0000012 ENSG00 E ENSG0000013616 NSG0000018686 NSG0000012805 NSG0000010254 Supplementary Supplementary 00017566 00010926 00007190

202 803 075 2 4 4 2 2 4 9 5 0 6 6 9 7 7 0 2 ------­ , .

because they are in auditory but not vestibular stereocilia. vestibular not but auditory in are they because or abundance low of their because either spectrometry, by mass tion observa escaped 25) of (3 proteins deafness-associated stereocilia or ~400,000 actin monomers in filaments (~3 mM). (~3 filaments in monomers actin ~400,000 or that there were ~210 actin filaments in a stereocilium of that diameter, in length and hence ~0.2 fl in volume. Electron tomography indicated ~5 diameter, in nm ~250 was microscopy electron transmission by and visualized fluorescence prototypical chick stereocilium utricle cross- ( connectors of actin-membrane number and actin-actin linkers of number spacing, actin-actin length, filament maps. rendered (low estimate) of our high-threshold building model i staining possible and noise reconstruction by complicated was which maps, density thresholding objectively in difficulties addressed numbers, cross-linker of estimates minimum and maximum generated which olds to build The models. simple use of ball-and-stick two thresholds, thresh density two tomography, used we electron by obtained maps plane jumping from one actin filament to another. To interpret density are but views complicated by these the section cross-linkers, of-plane ultrathin sections. Oblique views allow more reliable detection of out- that links can is be of relatively detected because actin-actin small the of number the but orientations, all in distances actin-actin of ment measure ready allow views Transverse anisotropy. data tilt-related limited to owing detected reliably less are cross-linkers out-of-plane but distances, long for filaments actin individual follow to us allows view longitudinal The quantification. for advantages distinct has set 6 Fig. longitudinally ( ( utricles electron t used we data, mass-spectrometric the validate further To Actin cytoskeleton structure from electron tomography ( loci deafness 8 for candidates new 13 we loci, identified but deafness uncloned identified with locations homologs for proteins bundle-enriched and human correlating genomic map identifying By deafness. with associated be plausibly might human loci for ~400 given and, bundles, in expressed are genes deafness-associated of all one-third At least genes. deafness-associated yet-undiscovered o Chr mography 17 10 10 10 11 13 13 More proteins enriched in hair bundles may be encoded by as- by encoded be may bundles hair in enriched proteins More We measured in several independent subvolumes the total actin actin total the subvolumes independent several in Wemeasured 4 4 3 2 X X ) and obliquely ( obliquely and ) Hypoxanthine-guanine Hypoxanthine-guanine Plastin 3 TOM1-like protein 2 PDZ domain containing 7 Shootin-1 Xaa-Pro 1 aminopeptidase protein Uncharacterized KIAA1211 protein Multifunctional ADE2 Epidermal growth factor receptor Plastin 1 Myosin Myosin IIIB Calcium binding protein 39-like Plastin 2 Fig. Fig. phosphoribosyl-transferase kinase substrate 8-like protein 2 n homogeneity. We omitted out-of-plane cross-linkers in in cross-linkers out-of-plane omitted We homogeneity. 2 2 to countto cytoskeletal structures stereociliain fromchick 3 ). We generated tomograms from stereocilia oriented oriented stereocilia from tomograms We ). generated Fig. 3a Description – c Fig. 3i Fig. ), transversely (

– k ) to the plane of section. Each data data Each section. of plane the to ) Table 2 1 Fig. 3f Protein symbol , another 100 bundle proteins HPRT PLS3 TOM1L2 PDZD7 SHOOTIN1 XPNPEP1 KIAA1211 PAICS EPS8L2 PLS1 MYO3B CAB39L PLS2 2 ). – h e r and O S Supplementary Fig. 3d Fig. BUN/UTR ratio

Bundle Bundle only Bundle only r u , 21 e 4 3 1 5 3 3 1 1 4 3 , l ,

s c ). A ). µ m e  - - - © 2013 Nature America, Inc. All rights reserved. Focusing on the actin cytoskeleton, we identified potential inter potential identified we cytoskeleton, actin the on Focusing Stereociliary protein network 5,800 1,100 connectors per with stereocilium the low-density threshold and 7,300 Wecounted connectors. contain actin-membrane 7,200 could as many as stereocilium each filament, actin each along nm 36 every appear should site binding a as membrane; plasma the to cent adja filaments actin ~52 has stereocilium prototypical The graphy. filament actin ~87,000. is maximum of nm 36 every for cross-linkers 30,000 cross- were out-of-plane linkers, consider not do also which threshold, density (mean stereocilium prototypical per cross-linkers tomogram) orientation 91,000 and tomogram) (longitudinal-orientation 62,000 estimated we tomogram, per subvolumes six from ( equal with (red), Gaussians three of sum the with fit were data The cross-bridges. of sites at distances actin-actin ( alone. ( map. density the on overlaid ( membrane. the to adjacent region a in map density ( ( ( orientation. stereocilia longitudinal from region; mid-shaft in model ( map. density three-dimensional filtered bilaterally of slice ( reconstruction. dimensional ( plane. sectioning to respect with ~18° of angle an at is axis longitudinal Stereocilia reconstruction. tomographic for used image oblique of projection tilt 0° dimensional, ( distances. actin-actin of measurements precise allowing orientation, transverse of map density ( reconstruction. three-dimensional unfiltered ( reconstruction. tomographic for recorded image transverse ( ( density. membrane segmented and alone ( thick). nm (6.4 map density surface-rendered the on overlaid ( density. membrane depicts green light and connectors actin-membrane represent lines orange cross-linkers, actin-actin represent lines blue actin, represent lines Red tomogram. ( map. density filtered bilaterally of slice ( reconstruction. three-dimensional unfiltered of slice ~0.8-nm ( reconstruction. tomographic for recorded image projection tilt 0° dimensional, ( section. of plane the to respect with longitudinally oriented stereocilium ( stereocilia. 3 Figure e r  proteins hair-bundle and them using between graph actions depicted s n m j i h f f d

) Single 0.8-nm slice of unfiltered three- unfiltered of slice 0.8-nm Single ) – ) Two-dimensional, 0° tilt projection of of projection tilt 0° Two-dimensional, ) – ) Close-up view of model. Scale bars: 100 nm in in nm 100 bars: Scale model. of view Close-up ) ) Model alone. ( alone. Model ) filtered bilaterally of slice 0.8-nm Single ) , We also counted actin-to-membrane connectors with electron tomo data the averaging and limit) (upper threshold low the Using k ) Model overlaid on the density map. map. density the on overlaid Model ) h e ) Oblique stereocilia tomogram. ( tomogram. stereocilia Oblique ) ) Stereocilia model from longitudinal longitudinal from model Stereocilia ) ) Transverse stereocilia tomogram. tomogram. stereocilia Transverse ) O S ±

± r 900 using the high threshold ( threshold high the using 900 ) Histogram showing distribution of of distribution showing Histogram )

d Electron tomography of chick chick of tomography Electron s.e.m.). The conservative lower-limit estimates with a high- with estimates lower-limit conservative The s.e.m.). ) Overview of stereocilia model model stereocilia of Overview ) r u a o – ) Density map only. ( only. map Density ) l c , m ) Tomogram from a from Tomogram ) g c c ) Scaled views of density and and density of views Scaled ) ) Single ~0.8-nm ~0.8-nm Single ) ) Single 0.8-nm slice of of slice 0.8-nm Single ) o – q e ) Close-up views of of views Close-up ) e ) Overview of model model of Overview ) ± l ) Density map only. only. map Density ) 1,000 and 36,000 36,000 and 1,000 k q ) Single 0.8-nm 0.8-nm Single ) ) Model Model ) a b ) Two- ) ) Single Single ) p

) Model Model ) i ) Two- )

Fig. 3g Fig. ± 2,000. Assuming three three Assuming 2,000. – i ). a – a p o l i f 2 k ± 3 , 20 nm in in nm 20 , , the theoretical theoretical the , 2,000 (oblique- 2,000 ± l 1,000 1,000 – q σ , 50 nm in in nm 50 , (width) for each. Individual fits for 8-, 11- and 15-nm peaks are shown in gray. gray. in shown are peaks 15-nm and 11- 8-, for fits Individual each. for (width) ± - ­ ­ q m j g b

clustering coefficient coefficient clustering law power a data were distribution fit well the byprotein-interaction interactions, the hold together links nodes which linked sparsely few many, a hubs—have only connected have others—highly nodes but most networks, molecular most ions. and In molecules signaling as well as proteins, hair-bundle tant Table 3 Supplementary in given are identified interactions All ones. additional identify and each protein us in interactions to allowing the network, validate both ( list protein our interaction curated manually therefore We databases. interaction protein-protein or other are in these interactions ever, known not all org ( the STRING ( bundles hair in enriched significantly were that proteins cytoskeletal major the of most chose cally defined, undirected graph that illustrates these relationships. We theory The network ( The network / s ) databases identified some interactions for these proteins; how proteins; for these interactions some identified ) databases . r

2 Number 4 20 40 60 with spring-electrical modeling spring-electrical with 0 advance online publication publication online advance 9 6 3 0 Actin-actin distance(nm 2 6 8 nm n k h c http://string-db , with with , Fig. Fig. Supplementary Table 2 Table Supplementary P 4 ( 12 11 nm C and k ) ) i . = 2 = 15 nm 15 ∝

Supplementary Fig. 7 Fig. Supplementary Table k 18 n –1.3 ) .org i / 2 k d s 6 , , i . For the 69 nodes with at least two two least at with nodes 69 the For . ( R / 1 k ) and BioGRID ) ( and BioGRID = 0.79 and and 0.79 = ) to seed the network. Searching Searching network. the seed to ) i – 1), where where 1), – 2 ) by searching PubMed for for PubMed searching by ) 5 , generating a mathemati , generating nature nature P < 10 < n ) highlights impor ) highlights e i is the number of number the is http://thebiogrid. NEUR −3 . The average average The . OSCI EN C E - - -

© 2013 Nature America, Inc. All rights reserved. (arrowheads). RHOA is also substantially enriched in hair cells over over cells hair in enriched substantially also is RHOA (arrowheads). bundle the of stereocilia tallest the or kinocilium the and (arrow) ( (*). region taper the above a band in and tips stereociliary toward concentrated ( RDX. is than tapers, the including panel), bottom in arrows (double bases stereociliary toward concentrated more is SLC9A3R2 but bundle, the throughout overlap ( in bar Scale (SC). stereocilia in as (MV), microvilli of bases from absent is immunoreactivity pERM cells. hair and supporting of surfaces apical of magnification Inset, panel). bottom in arrows (double region taper the in except colocalize pERM ( immunoblotting. by detected was RDX associated and immunoprecipitated, was SLC9A3R2 Tagged combinations. indicated the in cells HEK in expressed were SLC9A3R2 and T564D-RDX RDX, wild-type chick epitope-tagged (HA) ( (*). kDa ~80 at and kDa) (~70 size expected the at both bands detected moesin) and radixin ezrin, phosphorylated recognizes (which anti-pERM and (~10 ears 4 chick from epithelium sensory utricle represents lane epithelium sample; bundles the in that to equivalent agarose represents lane (~0.6 ears chick 40 from bundles represents ( 5 Figure activated to binds SLC9A3R2 RDX; with immunoprecipitated SLC9A3R2 expressed in a human embryo cell line (HEK293T) ( immuno by blotting bundles utricle chick in detected each were network, protein hair-bundle the in hubs as identified SLC9A3R2, and RDX RDX and SLC9A3R2 Ca and RDX (8) (9), CALM links were actin (33 interactions), PtdIns(4,5)P work indicates strong clustering how interconnect nodes the connecting links nature nature cells. supporting d a ) Protein immunoblotting with purified hair bundles. Hair bundle lane lane bundle Hair bundles. hair purified with immunoblotting ) Protein ) RDX and SLC9A3R2 immunocytochemistry. RDX and SLC9A3R2 SLC9A3R2 and RDX immunocytochemistry. SLC9A3R2 and ) RDX µ g protein). used are indicated at left. Both anti-RDX anti-RDX Both left. at indicated are used Antibodies g protein).

NEUR Identification of RDX and SLC9A3R2 in hair bundles. bundles. hair in SLC9A3R2 and RDX of Identification Fig. 5 Fig. f ) RHOA immunocytochemistry. Staining is seen in stereocilia stereocilia in seen is Staining immunocytochemistry. ) RHOA b OSCI ) RDX-SLC9A3R2 interaction. Myc or hemagglutinin hemagglutinin or Myc interaction. ) RDX-SLC9A3R2 a and and PIP4K2B c EN ) RDX and pERM immunocytochemistry. RDX and and RDX immunocytochemistry. pERM and ) RDX k C i Supplementary Fig. 8 Fig. Supplementary neighbors of node node of neighbors e E 2 ) MYO7A immunocytochemistry. MYO7A is MYO7A immunocytochemistry. ) MYO7A

6 . . The advance online publication online advance 2+ (8). C 2 c 7 i is 10 10 is value of 0.24 for measured our net . . Nodes with the largest numbers of Ptdlns(4,5)P µ m and applies to to applies m and µ i g total protein); agarose agarose protein); g total to each other, represents represents other, each to Hypothetical Bundle proteinparalog Known interaction 2 2 ). We also found that that found also We ). (20), SLC9A3R2 (12), PLS1 c ANXA6 – f

.

Molecules per stereociliu

10 10 10 10 10 - - m 2 3 4 5 in a narrow band above basal tapers, at the site of the ankle links ankle the of site the at tapers, basal above band narrow a in RDXcilia, is activated by PtdIns(4,5)P phos ( phorylation activating the mimics that threonine-to-aspartate proteins ERM SLC9A3R c a MYO7A CDC42 RHOA MYO6 pER RDX MV M 2 SC * *

Hair bundles Fig. 5 Fig. 2 8 , and binding was indeed stronger to RDX with a with RDX to stronger indeed was binding and , Agarose b e and and MYO7

* Epithelium represents protein abundance in bundles. Ca in bundles. abundance protein represents as in key (same classification functional indicate colors Node proteins. deafness-associated indicate labels Underlined CALB2. and of OCM exception the with plotted, are interactions or more two with nodes only messengers; or second proteins bundle represent (nodes) Symbols proteins. 4 Figure reference supporting the interaction. the supporting reference a to PubMed hyperlinked link each with figure 7 Figure Supplementary nodes. less-connected other many and (hubs) nodes connected highly a few contains plot the that indicates law, a by which power well fit is plot the in links and nodes of distribution The nodes. nearby between links of density the by controlled is plot the of layout The from interactions SLC9A3R2 example, (for interactions hypothetical represent links dashed proteins; of bundle paralogs involving to interactions correspond links Dotted evidence. Table 3 Supplementary citations; literature with validated interactions represent links Solid symbols. node diamond PtdIns(4,5)P Supplementary Fig. 8 Fig. Supplementary * A 260 160 260 20 20 40 50 80 80 pERM pERM Actin Actin RDX RDX MYO7

Actin Interaction network for hair-bundle hair-bundle for network Interaction Myc-T564D-RDX A HA-SLC9A3R 2 SLC9A3R SLC9A3R and cyclic AMP are indicated by indicated are AMP cyclic and d b 2 Fig. Fig. and is phosphorylation found f Myc-RDX RHO RDX RDX 110 50 60 80 2 2 2 A b 2 lists all interactions and interactions all lists ); node symbol size size symbol ); node Actin reproduces this this reproduces – – + – – e r ). In bullfrog stereo bullfrog In ). RHO Actin + – A O S + + – + – SLC9A3R2 SLC9A3R2 r u 5% load 5% load HA IP HA IP Table Actin RDX RDX

c

HA blot HA Myc blot Myc 2+ 3 ). ). , e 5  - ­ .

© 2013 Nature America, Inc. All rights reserved. As expected given the structure of the stereocilium cytoskeleton, actin Actin cytoskeleton transduction. out carry and bundle the make of to how proteins a cooperate bundle understanding comprehensive gain will we interactions, protein-protein cataloging and abundance protein measuring localization, protein dynamic and static tigating relationships functional determining bundle between proteins, inves ment, and composition and assembly ofdevelop the transduction complex. during By cytoskeleton bundle of the of biology construction cell cells: the hair for topics crucial two address systematically to ( sets data expression inner-ear of website SHIELD the on browsing for available of sive hair compendium the bundles, proteins of vestibular exten an establish to spectrometry mass quantitative used we Here DISCUSSION and USH2A USH1C, ( GPR98 PCDH15, CDH23, including partners, ing ligands SLC9A3R2 for sequences consensus of basis the on binding; for able ( cytoskeleton actin the control which CDC42, and RAC3 RHOA, GTPases Rho-family ( SLC9A3R2 and tapers basal above ( microvilli Likewise, pERM was concentrated in the upper half of ( tapers stereociliary above found only was (pERM) protein ( stereocilium each of half Similarly, and SLC9A3R2 total RDX were concentrated in the bottom form ofATP2B2 inhairbundlesis‘a’. a acid), XTXF, GVGL,ESDL,STHMandTLGA,allwhichbindtoSLC9A3R2(ref. We searchedtheC-terminalfouraminoacidsofallchickbundleproteinsforinstancesXTXL(whereXisany ENSGALP0000000568 ENSGALP00000023 ENSGALP0000001 ENSGALP000000 ENSGALP00000 ENSGALP0000 ENSGALP000 ENSGALP00 ENSGALP0 ENSGALP ENSGAL ENSGA ENSG ENS EN E EN ENSGA ENSG ENS EN E ENSGALP0000000091 ENSGALP0000003871 ENSGALP0000000431 Identifier Table 3 e r  and actin-associated proteins were abundant in the mass spectr ATP2B1 andATP2B2 areinthesameproteingroup.Only‘b’spliceformsofthese proteinsbindPDZdomains;thesplice NSGALP0000000740 NSGALP0000001993

When bound to RDX, the PDZ domains of SLC9A3R2 are avail are SLC9A3R2 of domains PDZ the RDX, to bound When SGALP0000004023 SGALP0000001540 SGALP0000000450 GALP0000000348 GALP0000001570 O S ALP0000000957 ALP0000002357 LP0000001596 LP0000001835 P0000000966 3

0 0000002433 Table Candidate SLC9A3R2-binding Candidate proteins SLC9A3R2-binding , we identified 24 hair-bundle proteins as candidate bind candidate as proteins hair-bundle 24 identified we , 000000748 https://shield. r u 00003928 Fig. 5 Fig. 0001940 001757 02301 3 1999 ). c 281 c Fig. 5 Fig. 30 2 , inset). Although MYO7A also appears in a band band a in appears also MYO7A Although inset). , 9 e Supplementary Table 1 Supplementary , its distribution was distinct from those of RDXof those from distinct was distribution its , 4 9 9 0 6 0 4 2 6 1 5 7 9 7 2 6 7 2 6 7 1 5 4 2 2 e Last 4 residues hms.harvard.edu ETSL ESDL STCL GTSL PTTL QTEL DTDL QTEF VTLL DTEL KTSL NTFF STAL ATVL PTGF LTLL ESDL ETSL DTHL DTHL CTVF ITEL ETKL MTFF STAL ). Finally, we detected in hair bundles the the bundles hair in detected we Finally, ). Fig. 5c Fig. b ACTN1 andACTN2areinthesameproteingroup.

ATPase, Ca Actinin, alpha 2 Laminin, beta 1 Golgin A4 LIM and calponin homology domains 1 Anterior gradient 3 homolog 88 kDa protein), beta (cadherin-associated 1, Family with sequence similarity 129, member B CSE1 segregation 1-like Tight junction protein 2 (zona occludens 2) Coatomer protein complex, subunit beta 1 Ribosomal protein L18a Rho GTPase activating protein 17 Dmx-like 1 Casein kinase 1, alpha 1 Cold inducible RNA binding protein Actinin, alpha 1 ATPase, Ca G receptor protein-coupled 98 Usherin (Usher syndrome 2A) Ras-related C3 botulinum toxin substrate 3 Cadherin-23 Espin Harmonin (Usher syndrome 1C) Protocadherin-15 , d ), and phosphorylated total ERM ERM total phosphorylated and ), 2+ 2+ / ). These data will allow us us allow will data These ). transporting, plasma transporting, membrane 2 plasma transporting, membrane 1 and and b b Description Fig. 5a Fig.

supporting-cell supporting-cell , f ). Fig. 5 Fig. o 3 metry metry 0 ). NA,notapplicable. c ). ). - - - - -

a a ity of stereociliary actin filaments suggests that their barbed ends, ends, barbed their that suggests filaments actin stereociliary of ity mechanisms. myosin-mediated only involves ment in late that suggest develop polymerization control of actin-filament ment value suggested that it was a contaminant. Together these results one detected formin (DIAPH2) in one its bundle experiment, enrich we although and members, Ena/VASP any family detect Wenot did stereocilia. in inactive is likely complex ARP2/3 the however, bers, WASP/WAVE any we activating not did detect Because mem family neutrophils human in that to similar concentration at present is at 3 complex, which mediates polymerization of branched actin networks; networks connectors. to-membrane myosin ~7,500 counted we molecules per Altogether stereocilium, which could account for remaining actin- system. vestibular the in MYO15A with partner might the EPS8L2 binds complex, MYO15A-DFNB31 8) substrate pathway receptor EPS8 factor paralog growth its (epidermal because 2); (EPS8-like EPS8L2 of molecules 130 found we DFNB31, partner binding MYO15A the detect not did we ESPN. to binds that paralog profilin the PFN2, of that and ESPN partner binding its of that to close very was MYO3B cilium, nearly all of which was MYO3B. Notably, the concentration of myosin XV families development late through MYO1H. related closely the and MYO1C of each ~50 and MYO7A 250 MYO6, chick 6,600 MYO6, of of >400 MYO7A and 100 of MYO1Cstereocilium per molecules >700 have bundles bullfrog LAMB1 GOLGA4 LIMCH1 AGR3 CTNNB1 FAM129B CSE1L TJP2 COPB1 RPL18A ARHGAP17 DMXL1 CSNK1A1 CIRBP ACTN1 ATP2B1 GPR98 USH2A RAC3 CDH23 ESPN USH1C PCDH15 ATP2B2 ACTN2 symbol Protein Although actin may not treadmill from tip to taper to tip from treadmill not may actin Although actin of polymerization control also proteins other Several of MYO15A. Although Westereocilium per 50 molecules detected least at dynamic is which stereocilia, in polymerization Actin µ M (~340 molecules per stereocilium), M the complex per ARP2-ARP3 (~340 stereocilium), molecules

3 advance online publication publication online advance 3 . We detected five of seven subunits of the ARP2/3 ARP2/3 the of subunits seven of five detected We . BUN/UTR 10 22 35 ratio 0.01 0.02 0.04 0.05 0.08 0.1 0.1 0.2 0.2 0.3 0.4 0.8 0.9 0.9 0.9 1 1.3 2 2 3 NA NA 1 . We detected 430 myosin III molecules per stereo

estimated by tomography (5,800–7,300). by tomography estimated for the majority; this value is within the range stere (moesin) together totaled 6,800 molecules per spectr mass By connectors. most actin-membrane the of contributes likely at intervals, actin ~36-nm to complexes ­membrane-protein the (33,000–77,000). estimates tomography with agreement good in cross-linkers, ~48,000 has stereocilium each that estimates previously found (710 espins and PLS3) 400 PLS2; 460 fimbrin; as known also PLS1, (5,500 plastins FSCN1), 1,300 FSCN2, of stereocilium per molecules (40,000 fascins of cross-linkers: classes three ( method riBAQ the using spectrometry mass by measured tors connec actin-membrane and cross-linkers actin-actin known of abundances esti to mates tomography the compared We data. surements. By quantitative immunoblotting, immunoblotting, By quantitative surements. mea independent with that well corresponded abundance myosin of estimates gave family super myosin the of members be may tors The ERM family, which cross-links cross-links which family, ERM The eann atnt-ebae connec actin-to-membrane Remaining 3 2

o ESPN and 90 ESPNL), similar to what we , is controlled by the myosin III and and III myosin the by controlled is , cilium ( cilium o 1 . Mass-spectrometric quantification quantification Mass-spectrometric . metry, EZR (ezrin), RDX and MSN MSN and RDX (ezrin), EZR metry, 2 9 Table ; mass spectrometry estimated estimated ; mass spectrometry 3 1 . Mass spectrometry thus thus spectrometry Mass . 1 nature nature ), with RDX accountingRDXwith ), Table

1 NEUR ). We detected detected We ). 3 5 , the stabil the , OSCI EN C 3 4 E ­ ------­ .

© 2013 Nature America, Inc. All rights reserved. PCDH15-containing transient lateral links required for forming a forming for required links lateral transient PCDH15-containing and SLC9A3R2 SLC9A3R2. RDX might anchor also the and CDH23- links ankle the at located are GPR98 and USH2A function. and structure importance bundle known for of are SLC9A3R2 for ligands likely Moreover, may complexes in that be stereocilia. dynamic SLC9A3R2 suggesting microvilli in dynamic highly is SLC9A3R1 stereocilia. assembling microvilli assembling in SLC9A3R1 paralog the of function critical the Given SLC9A3R2. interactions. protein-protein other for scaffold a viding RDXpro activate membranes, and target its receptor to stereociliary PtdIns(4,5)P by preactivation membrane plasma the of to of a CLIC5, CLIC4, paralog translocation induces in stereocilia, could bind RDX molecules not bound by SLC9A3R2. RHOA, CLIC5 detected stereocilium, per molecules 2,400 At forms. transmembrane intra (chloride CLIC5 cellular be to thought is RDX for partner binding major A interaction phosphorylation. RDX by direct facilitated is a that them including between interactions, 12 and 8 ­respectively mechanotransduction to CALM-dependent enzyme activity. tions; Ca Also prominent were CALM and Ca stereocilia. in proteins membrane-associated some concentrate may PtdIns(4,5)Ptobind do actin with PtdIns(4,5)P in the network map ( membranetargeting stereocilia prominence in tions; PtdIns(4,5)P interac 33 with actin, was hubhair-bundle largest The proteins. etal actin. bind nevertheless may XIRP2 that (ref. LIMA1 protein binding ( domains these contain not do XIRP2 bundle rat and chick repeats, Although most species’ XIRP2 proteins contain >30 actin-binding Xin is the most abundant protein in hair without bundles an obvious role. rootlet. the to ends filament cross-link protein) and binding F-actin (TRIO TRIOBP component rootlet the latter, the if or cilium, Q) type, receptor phosphatase, tyrosine (protein PTPRQ protein membrane the to ends pointed the material rootlet into central the membrane taper the on terminate either filaments actin capper pointed-end candidate capper pointed-end characterized best- the , detect not We did there. depolymerization or capping systematic a suggesting tapers, stereociliary form to nate stereocilia. shorter and longer in differentially occur could which ends, filament for compete they that suggests filaments actin over cappers of excess The molecules. (twinfilin-2) TWF2 950 and stereocilium per heterodimers protein) (capping CAPZ ~700 pers: We two main stereocilium. cap per be cappers ~210 should detected there of the run length and the stereocilium, if filament per filaments present is capper barbed-end one If capped. are tips, stereociliary at nature nature bundle cohesive Supplementary Fig. 4b Fig. Supplementary Other membrane proteins will interact with RDX through through RDX with interact will proteins membrane Other had stereocilium) per molecules (2,000 SLC9A3R2 and RDX cytoskel other several of role the highlighted analysis Network At 4,600 per XIRP2 molecules stereocilium, (Xin-related protein 2) termi progressively filaments actin stereociliary of ends Pointed

channel 5) channel NEUR 2+ entering stereocilia through transduction channels couples 2 ; moreover, several bundle proteins not known to interact OSCI 1 . 2 4 1 EN had 20 interactions, given which was expected its 4 ; RDX may anchor ankle links there through through there links ankle anchor may RDX ; 4 Fig. , which can interconvert between cytosolic and and cytosolic between interconvert can which , 3 4 . As is clear from co-clustering of the two hubs two the ofco-clustering from clear is As . C 7 4 E , we propose that SLC9A3R2 is necessary for necessary is SLC9A3R2 that propose we , 5 4

). However, XIRP2 is a paralog of the actin- the of paralog a is However,). XIRP2 4 ; moreover, RHOA can activate RDX after after RDX activate can RHOA moreover, ; 5 0 advance online publication online advance ), many bundle proteins bind both actin and and actions in actin polymerization actin in actions and (detected in one experiment only) could could only) experiment one in (detected 4 2 3 1 (ref. (ref. 3 3 7 9 ), also known as EPLIN, suggesting suggesting EPLIN, as known also ), 8 . The pointed ends of stereociliary stereociliary of ends pointed The . , present at 1,500 copies per stereo per copies 1,500 at present , ; ; if former,the MYO6 anchor could 2 2+ , suggesting that the phospholipid the thatsuggesting , 3 6 , with respectively 9 and 8 interac 4 nr i w dtc tprn a taperin, detect we did nor , 6 ). RHOA may therefore both both therefore may RHOA ). 2 3

or gather gather or 4 2 and 4 8 ­ ------­ ,

be required to detect it above the background of actin and other other and actin of proteins. background cytoskeletal the above it detect to required be be may channel 10 every for present transduction the of molecules 1–10 only because ~10 matched we run, spectrometry mass each In sensitivity. of lack than rather spectrometry, mass of range from arises the dynamic limited channel—likely elusive transduction kinocilial ( links in and bundles developing of links lateral the in present also are they cadherins, the of each two contain only should cilium link tip a of anchor end upper to the cluster which molecules, (harmonin) USH1C 60 as well as 15), (protocadherin PCDH15 and CDH23 of cadherins, each link tip the stereocilium per molecules ~20 detected we Nevertheless, stereocilium per channels transduction active two and link tip one Mechanotransduction molecules are rare in hair bundles; there is only Other hair bundle proteins mal small GTPase ARL13B. GTPase small mal and RSPH6A axone the and RSHL1, IFT172 molecule transport intraflagellar the molecules RSPH9, head and spoke DNAH5 radial the axonemal DNAH9, the detected we , besides California Berkeley Electron MicroscopyBerkeley California Laboratory for assistance with Facility. We thank K. McDonald, andR. Zalpuri G. Min of the University of Laboratory and Spectrometry The University of Virginia Biomedical Research Mass was out carried spectrometry by the W.M. Keck Biomedical Mass Note: Supplementary information is available in the identifier set data with tory reposi partner PRIDE the via Consortium ProteomeXchange to the codes. Accession the in v available are references associated any and Methods M bundle. hair the of function and maintenance to contributes that any protein identify principle in can approach ric by roles; the mass spectromet contrast, proteins developmental with Indeed, genetic screens for deafness likely miss ubiquitously expressed phenotype. lethal have and in an an could tions embryonic organism func multiple out carry that proteins as such genetics, by identified identify functionally important bundle to proteins us that haveallows not yet been spectrometry Mass future. the in deafness to linked be deafness- by genes, associated suggesting that proteinsother bundle-enriched may encoded are bundles hair in enriched proteins Many Conclusions experiments. LC-MS/MS in detected well not are transmembrane peptides because perhaps ATP2B2), (mostly ~1,700 fewer, Ca 8,000 have as in bullfrog (2,000 molecules ~4 noblotting the 70 (~60 concentration CALM estimated The cilium, together corresponding to a total of ~2 mM Ca stereo per (calmodulin) CALM and 7,300 (calretinin) CALB2 8,000 Acknowledgmen ersion of the pape the of ersion ethods Axonemal kinocilia are present in the hair-bundle preparation hair-bundle the in present are kinocilia Axonemal the as proteins—such transduction other detect to inability Our Diffusible Diffusible Ca µ m Supplementary Fig. 4 Fig. Supplementary µ 2 ; if the density of the Ca of the density ; the if M estimated for bullfrog hair bundles by quantitative immu by quantitative hair bundles for M bullfrog estimated 4 9 . The membrane area of the prototypical stereocilium is is stereocilium prototypical the of area membrane The . 2+ 2+ pumps. Mass spectrometry estimated significantly significantly estimated spectrometry Mass pumps. buffers were prominent; we estimated 63,000 OCM, Mass spectrometry data have been deposited deposited been have data spectrometry Mass r ts . 6 bundle molecules, substantial enrichment may enrichment substantial molecules, bundle 1 . Although the tip link of a single stereo single a of link tip the Although . PXD00010 ) 1 µ . 2+ m pump in chick bundles is the same same the is bundles chick pump in −2 ; ref. 4 . 5 4 0 online version of the pape spectra to chick peptides; peptides; chick to spectra µ ), each stereocilium would M) is nearly identical to to identical nearly is M) e r O S 2+ binding sites. r u

online online r . c e 4 1  ------­ . ;

© 2013 Nature America, Inc. All rights reserved. 18. 17. 6. 5. 4. 3. 2. 1. reprints/ at online available is information permissions and Reprints Published online at The authors declare no competing interests.financial The manuscript was written by P.G.B.-G. out data mass analysis, spectrometry with assistance from P.A.W. and L.L.D. experiments of experiments of out experiments. K.J.S.mass spectrometry outcarried immunocytochemistry electron tomography and analyzed tomography data. N.E.S. and E.D.J. carried samples for mass analysis. spectrometric A.H., Z.M., A.N.T. and M.A. outcarried immunoblots of 5 out immunoblotting experiments of and immunocytochemistry J.-B.S. and P.G.B.-G. the designed overall approach and analysis. J.F.K. carried (L.L.D.), P30 EY10572 (L.L.D.) and P01 GM051487 (M.A.). (P.G.B.-G.), (P.G.B.-G.),R01 DC011034 (P.G.B.-G.),P30 DC005983 R01 EY007755 NIH grants (J.B.S.),K99/R00 DC009412 (J.F.K.),F32 DC012455 R01 DC002368 Center for Research Resources (NIH). Work here described was supported by instrumentation grants S10 RR023432 and S10 RR025440 from the National The Jungers Center (Oregon Health & University),Science supported by shared assistance, we acknowledge A. Snyder of the Advanced Light Microscopy Core at (NIH) National Center for Research Resources grant RR017573. For technical Automated Molecular Microscopy, supported by US National Institutes of Health for help with electron microscopy data atcollection the National Resource for pressure Wefreezing. would like to thank A. Cheng, B. andCarragher C. Potter high-pressure freezing and imaging; D. Jorgens provided mentoring in high- e r 1 16. 15. 14. 13. 12. 11. 10. 9. 8. 7. COM AUTH , , as well as 0

McKusick, V.A. Mendelian Inheritance in Man and its online version, OMIM. version, online its and Man in Inheritance Mendelian V.A. McKusick, and practical a rate: discovery false the Controlling Y. Hochberg, & Y. Benjamini, M. Donowitz, P.G.Gillespie, membrane & Large Brugger,B. J.B., Shin, D.E., Williams, H., Zhao, J.B. Shin, Domon, B. & Aebersold, R. Options and considerations when selecting a quantitative and biophysical the Integrating A.J. Ricci, & B. Pan, F.T., Salles, A.W., Peng, hearing. of biology cell The U. Müller, & B. Kachar, M., Schwander, oe, .. Jns TA Otgn o vsiua cmon ato ptnil in potentials action compound vestibular of Ontogeny T.A. Jones, & S.M. Jones, Grati, M. Grati, the in vimentin of localization Immunohistochemical J.C. Adams, & B.A. Schulte, R.P. Kruger, the in sites pump Na-K of localization Quantitative C.E. Stirling, & J.A. Burnham, Tanaka,chicken’sstudy.the TEM of and Structure SEM C.A. ear: Smith, inner & K. B. Schwanhäusser, Cox, J. & Mann, M. MaxQuant enables high peptide identification rates, individualized J. Cox, and bundles hair bullfrog of isolation High-purity A.J. Hudspeth, P.G.& Gillespie, J. Hum. Genet. Hum. J. 57 testing. multiple to approach powerful network. protein (2012). syndrome 14288–14293 Usher the with protein scaffolding ear. inner gerbil 19 ear. inner avian the of epithelia sensory the in expressed phosphatase sacculus. frog Anat. J. Am. control. quantification. protein proteome-wide and Biotechnol. accuracies mass p.p.b.-range environment. (1991). 625–640 proteins. constituent of localization topological and subcellular chicken. domestic the (2005). 3–11 32 activation. radixin constricted spatially specify bundles hair in domains strategy. proteomics 2 mechanotransduction. cell hair auditory of mechanisms molecular 190 , 523 (2011). 523 , P O S , 289–300 (1995). 289–300 , (1999). 4815–4827 , (2012). 4600–4609 , O E , 9–20 (2010). 9–20 , index.htm TI R R

CON t al. et NG 53 Nature Supplementary Supplementary Figures 2 r u et al. et , 371–386 (2007). 371–386 , et al. et

FI 26 Figure Supplementary Figure 4

nrmd: ppie erh nie nertd no h MaxQuant the into integrated engine search peptide a Andromeda: Figure TRIBUTI 153 t al. et J. Proteome Res. Proteome J. et al. et J. Neurocytol. J. , 1367–1372 (2008). 1367–1372 , N l

Localization of PDZD7 to the stereocilia ankle-link associates this associates ankle-link stereocilia the to PDZD7 of Localization

. 473 Hair bundles are specialized for ATP delivery via creatine kinase. creatine via ATPdelivery for specialized are bundles Hair 80 http://www.nature.c A c J. Histochem. Cytochem. Histochem. J. , 251–271 (1978). 251–271 , NC , 588–604 (2007). 588–604 , h spotn-el nie: rcpo-ie rti tyrosine protein receptor-like a antigen: supporting-cell The NHERF family and NHE3 regulation. NHE3 and family NHERF 5 , 337–342 (2011). 337–342 , e t al. et Nat. Biotechnol. Nat. 1 . . D.C. outcarried the statistical analyses. P.G.B.-G. carried . . J.-B.S. and J.M.P. prepared hair-bundle and epithelium IA J. Assoc. Res. Otolaryngol. Res. Assoc. J. ON L Global quantification of mammalian gene mammalian of quantification Global I N S

13 T

10 E , 617–638 (1984). 617–638 , R , 1794–1805 (2011). 1794–1805 , E and om/doifinder/10.1038 STS

28 . . H.Z. outcarried immunoprecipitation . . tt Sc Sre B tt Methodol. Stat. B Series Soc. Stat. R. J. 4 , 710–721 (2010). 710–721 ,

; ; she outalso carried quantitative 37 , 1787–1797 (1989). 1787–1797 ,

1 , 232–242 (2000). 232–242 , http://www. J. Physiol. (Lond.) Physiol. J. /nn.331 Figures Figures 2 . el Biol. Cell J. . Neurosci. J. 2 a. Commun. Nat. . .

nature.com/ J. Cell Biol. Cell J. J. Neurosci. J. J. Neurosci. J.

and

112 567

Nat. Am. 32 , , ,

32. 31. 30. 29. 28. 27. 26. 25. 24. 23. 50. 49. 48. 47. 46. 45. 44. 43. 42. 41. 40. 39. 38. 37. 36. 35. 34. 33. 22. 21. 20. 19.

Rzadzinska, A.K., Schneider, M.E., Davies, C., Riordan, G.P.actin Riordan, An Kachar,C., B. & Davies, Schneider, M.E., A.K., Rzadzinska, J.B. Shin, Tonikian,R. Hasson, T. role the cortex: cell the Organizing A. Bretscher, & A.I. McClatchey, R.G., Fehon, networks. ‘small-world’ of dynamics Collective S.H. Strogatz, & D.J. Watts, functional cell’s the understanding biology: Network Z.N. Oltvai, & A.L. Barabási, placement. force-directed by drawing Graph E.M. Reingold, T.M.J.& Fruchterman, using networks regulatory and biochemical of organization the into Insights A. Ma’ayan, stereocilia, filaments, Actin D.J. DeRosier, & M.S. Tilney, L.G., Tilney, uot R.A. calmodulin- Dumont, and Calmodulin P.G. Gillespie, & A.J. Hudspeth, R.G., Walker, Garbett, D. & Bretscher, A. PDZ interactions regulate rapid turnover of the scaffolding D.P.LaLonde, D., Garbett, Bretscher,regulates & EBP50 protein scaffolding The A. phosphorylation RhoA-dependent T. Jacks, & F. Solomon, M., Henry, R.J., Shaw, B. Ponsioen, L.H. Gagnon, W.D. Heo, cytoskeleton. actin the of regulation Phosphoinositide P.A. Janmey, & H.L. Yin, R.S. Maul, S. Kitajiri, H. Sakaguchi, C.M. Hackney, & R. Fettiplace, M., Ohashi, S., Mahendrasingam, D.N., Furness, A.U. Rehman, end. slow the at V.M.Fowler,Tropomodulins:life & R.S. Fischer, D.S. Zhang, Wiskott- of terminus C the of T.D.Influence Pollard, & L. Blanchoin, H.N., Higgs, factors assembly matching actin: of faces many The H.N. Higgs, & E.S. Chhabra, Auer, M. Auer, in impairment. hearing on influences environmental K.P.and Steel, Genetic K.J. Spinelli, Eppig, J.T. J. Neurosci. J. crosslinker in hair-cell stereocilia, underlies early-onset hearing loss of DBA/2J mice. (2008). 137 proteins. ERM of 393 organization. Exp. Pract. Softw. analyses. theory graph measure. and count (1992). cells how cells: hair (2008). tomography. electron-microscopic 2007). UK, Cambridge, Press, University (Cambridge 505–515 N.D.) intensity.Medicine Modern in Genetics Diseases: MS2 Common and using spectrometry mass quantitative USA Sci. Acad. Natl. Proc. by revealed epithelia hair bundles. hair (1993). bundles. hair in proteins binding microvilli. in EBP50 protein (2010). 397–413 manner. phosphorylation-dependent a in assembly microvillar in protrusions membrane/actin apical into proteins fibroblasts. ERM of relocalization and RhoA. by CLIC4 protein Neurosci. J. function. ear inner normal for essential is and stereocilia cell hair in levels high at membrane. plasma the to clusters Physiol. Rev. Annu. filaments. hearing. for essential stereocilia VI-based myosin of implication stereocilia: of end transport. proximal the at Q receptor mammalian and cells in low-frequency rootlets and (2008). high- stereociliary membrane. lateral the between to connection of a comparison for evidence cells: composition hair and cochlear dimensions The Genet. Hum. J. Am. C9orf75, encoding taperin, as the mutated gene in nonsyndromic deafness DFNB79. 13 stereocilia. hair-cell in turnover Biochemistry polymerization. actin on complex Arp2/3 the and (WASp) protein syndrome Aldrich structures. cellular with and self-renewal. architecture functional stereocilia maintain and treadmill molecular resourcemousebiology.for , 593–601 (2003). 593–601 , , 1287–1307 (1997). 1287–1307 , (1998). 440–442 , advance online publication publication online advance et al. et Cell Motil. Cytoskeleton Motil. Cell et al. J. Cell Biol. Cell J. et al. et et al. et al. et t al. et Mol. Biol. Cell Biol. Mol. et al. et

et al. et

30 26 t al. et J. Cell Biol. Cell J. Nat. Rev. Genet. Rev. Nat. 38 et al. et J. Neurosci. J. et al. et t al. et et al. et Three-dimensional architecture of hair-bundle linkages revealed by revealed linkages hair-bundle of architecture Three-dimensional t al. et The Mouse Genome Database (MGD): from genes to mice–a community t al. et , 9683–9694 (2010). 9683–9694 , , 10188–10198 (2006). 10188–10198 , The R109H variant of fascin-2, a developmentally regulated actin regulated developmentally a fascin-2, of variant R109H The Unconventionalmyosinsininner-ear sensoryepithelia. Actin-bundling protein TRIOBP forms resilient rootlets of hair cell hair of rootlets resilient forms TRIOBP protein Actin-bundling , 15212–15222 (1999). 15212–15222 , PI(3,4,5)P EPLIN regulates actin dynamics by cross-linking and stabilizing and cross-linking by dynamics actin regulates EPLIN A specificity map for the PDZ domain family. domain PDZ the for map specificity A Nat. Rev. Mol. Cell Biol. Cell Mol. Rev. Nat.

ut-stp iaig as pcrmty eel so protein slow reveals spectrometry mass imaging Multi-isotope 21 Distinct energy metabolism of auditory and vestibular sensory vestibular and auditory of metabolism energy Distinct The chloride intracellular channel protein CLIC5 is expressed is CLIC5 protein channel intracellular chloride The

Dynamic compartmentalization of protein tyrosine phosphatase tyrosine protein of compartmentalization Dynamic ptoeprl euain f hoie nrclua channel intracellular chloride of regulation Spatiotemporal

lsa ebae Ca membrane Plasma 65 agtd atr ad etgnrto sqecn identifies sequencing next-generation and capture Targeted 86 , 1129–1164 (1991). 1129–1164 ,

J. Biol. Chem. Biol. J. , 761–789 (2003). 761–789 , 160 , 378–388 (2010). 378–388 , Nat. Cell Biol. Cell Nat. Mol. Biol. Cell Biol. Mol.

164

21 9

NucleicAcidsRes. , 399–407 (2003). 399–407 , 109 3 , 403–419 (1998). 403–419 , J. Cell Biol. Cell J. n PI(4,5)P and , 5066–5078 (2001). 5066–5078 ,

, 887–897 (2004). 887–897 , 5 Nature , E268–E277 (2012). E268–E277 , , 101–113 (2004). 101–113 , Cell

65 Science rc Nt. cd Si USA Sci. Acad. Natl. Proc. . so. e. Otolaryngol. Res. Assoc. J.

141

, 528–538 (2008). 528–538 , 284

481 9 20

, 1110–1121 (2007). 1110–1121 , , 786–798 (2010). 786–798 , 198

, 5451–5455 (2009). 5451–5455 , , 4664–4672 (2009). 4664–4672 ,

11 2+ 2 , 520–524 (2012). 520–524 , 314 iis agt rtis ih polybasic with proteins target lipids nu Rv Cl Biol. Cell Rev. Annu.

APs ioom a s h PC of PMCA the is 2a isoform -ATPase , 195–203 (2012). 195–203 , , 276–287 (2010). 276–287 , 33 , 1458–1461 (2006). 1458–1461 , , D471–D475,(2005). nature nature J. Neurosci. J. (eds. Wright, A.F. & Hastie, A.F.& Wright, (eds. NEUR PLoS Biol. PLoS . el Biol. Cell J.

90 28 Trends Cell Biol. TrendsCell

2807–2811 , , 6342–6353 , 9 OSCI 8 215–224 , 257–274 , J. CellBiol.

6 , e239 , Nature EN Genes

191 C E ,

© 2013 Nature America, Inc. All rights reserved. acetic acid gradient at a flow rate of 0.5 ofrate flow a atgradient acid acetic (7.5 packed with Phenomenex C18 Jupiter of 10 reversed-phase capillary column of 8 cm length × 75 a to interfaced was which source,nanosprayProtana ion a with spectrometer ofconfidentbers protein assignments. Peptides were introduced into massthe resolution of peptide peak exceptionaltem;the Orbitrap ofthe mass accuracy instrument allowsfor high sys spectrometer mass VelosETD Orbitrap Electron Thermo a of consisted M analyzed by six LC-MS/MS runs, corresponding to the six gel pieces. experiment’sepitheliumsingle wasor bundlesA hair acid. of acetic worth 3% tube. The eluate was vacuum dried, then the sample reconstituted with 15 elutedthen with25 CapTrap50 withwashed was from Bruker-Michrom (TR1/25109/35; size 0.5 × 2 mm, bed volume 0.5 peptides were first purified away from interfering polymers on a SCX CapTrap analysis. For the gel slice immediately adjacent to the agarose in the sample well, formic5% acid.wereextracts These combined and evaporated to15 30- two in polyacrylamide the from extracted added. The sample was digested overnight at 37 °C and the peptides formed excesswere enzyme solution was removed and 20 pancreas, dimethylated) in 50 mM ammonium bicarbonate dratedon icein 20for ng/ 10 min. Any the pieces completely dried by vacuum centrifugation. The gel pieces were rehy The pieces were dehydrated in 100 removed and the gel pieces rehydrated in 100 removed and the gel pieces dehydrated in 100 30 removed and the sample alkylated at room temperature (20–25 °C) for 0.5 h andreducedat room temperature solutionwith DTT wasThe for h. °C)(20–250.5 rehydrated30 in 200 The 1 cm of gel with separated proteins was manually sliced into six pieces. rinsed with water, then stained with Imperial Protein Stain (Thermo Scientific). bundlesample);Bis-Tris lanesper gelsweretwooneorwell; 10 × mm (1.5gel sample buffer. Proteins were separated by running ~1 cm into NuPAGEa with solubilized EpithelialNuPAGE proteinsalso weremin. LDS 15 for °C 70 4–12% final volume of 80 sample buffer (Invitrogen) with 50 mM dithiothreitol was added to a combined interfering polymers from the agarose used for bundle isolation. NuPAGE LDS comparisondetectionin othertomethods reproducibilityandsensitivity substantiallydigestionincreasedof trypsin and gested that bundles make up ~2% of the total protein in the utricle. the total protein in the utricle. Given the uncertainties in each, the methods sug 1.8 up make bundlesthatestimated method interest,thisregions of ( Fiji Usingmicroscopy. light by examined utricles osmium-stainedfixed, of sections plastic ofimages in bodies cell andbundles utricle’s total protein. In the second method, we estimated the areas taken up by 2.4 athere recovery (~40%); this value was then divided by the protein per utricleutricle (estimated(16 ng; ref. independent methods. In the first, we divided the amount of bundle protein per estimated the fraction of epithelium protein accounted for by bundlesmethod twist-off using the two using chicks 20–21 day embryonic from purified spectrometry. mass for samples of Preparation Panel on Euthanasia. was done using methods approved by American Veterinary Medical Association approval number was A684. All experiments began by killing the animal, which Health & Science University Institutional Animal Care and Use Committee; the Animals. METHODS ONLINE doi:10.1038/nn.3312 source was operated at 2.5 kV. The digest was analyzed using the data-dependent ass spectrometry data acquisition and analysis. analysis. and acquisition data spectrometry ass Gel pieces were transferred to siliconized tubes and washed and destained in Separation of proteins by a short SDS-PAGE run before reduction, alkylation µ µ µ l of 50 mM iodoacetamide in 0.1 M ammonium bicarbonate. The reagent was l) was injected and peptides eluted from the column by an acetonitrile/0.1 M l 50% methanol overnight. The gel pieces were dehydrated in acetonitrile, Animal experiments reported here were approved by the Oregon the by approved were here reported experiments Animal ± 0.2 µ l trypsin (Sigma-Aldrich T6567 proteomics grade, from porcine µ µ 4 g). This approach suggested bundle protein was 1.7% of the of 1.7% approachproteinwasbundle Thissuggested g). µ ), measured with a fluorescence protein assay, by the estimated l of 10 mM dithiothreitol in 0.1 M ammoniumbicarbonateM 0.1 dithiothreitolin mM 10 of l µ l per 100 utricles’ worth of bundles; samples were heated to l 1 M ammoniumM 1 l acetate into separatea microcentrifuge m / z (mass-to-charge ratio), which leads to high num µ l of 1% acetic acid (voidvolumeacidand collected) acetic 1% of l µ l acetonitrile, the acetonitrile removed and µ l/min over 1.2 h. The nanosprayionThe h. 1.2 over l/min 3 µ 1 µ µ l of 50 mM ammonium bicarbonate µ , in part because we couldweremovebecause part in , m particle size. An extract aliquot l of 0.1 M ammonium bicarbonate. l acetonitrile. The acetonitrile was µ l aliquots of 50% acetonitrile, 50% of aliquots l Utricle hair bundles were bundles hair Utricle µ http://fiji.sc The LC-MS/MS system LC-MS/MS The m internal diameter, self- / ) to measure to ) µ ± 0.6% of 0.6% l forl MS µ 4 l). The , 8 µ . We. l of - - - -

with a single unique peptide are flagged in flagged arepeptide unique single a with espin ( ( several sequences with longer or full-length sequences, including actin gamma 1 Andromeda using 2012) spectrometry data were searched against Ensembl version 66 Mass (releasedcells. hair Februarycomponentsof as dismissed fully be not should they that ing components). to contamination white from red blood cells egg hemoglobins, Nevertheless, alpha (, and beta hemoglobins, workflow which appear inpurification the preparation owing bundle- enteredimpuritiesadditionalthe that add to and example,(for actin) bundles hair in present be to known entries remove to edited was download quantification amino acid sequence. tide molecular weights, and 20 product-ion spectra in the ion trap to determine mass in spectrum the Orbitrap detector at 60,000 resolution to determine pep ­capability of the instrument, acquiring—in sequential scans—a single full-scan were further grouped in our analysis (‘shared-peptideanalysisour with in groups’);grouped entry thefurther were group. Redundant groups that shared more than 20% of their identified peptides the entry with the largest number of peptides was used to identify the redundant another protein, MaxQuant groups those proteins togetherof that withincompletely contained(‘redundantor to identical wasprotein a groups’);for peptides ase A ( ( are particularly poorly recovered, leading to reduced detection of this class of class this of detection reduced to recovered,leading poorly particularly are peptides derived from transmembrane segments of integral membrane proteins accuracy when standards for each protein are lacking. For example, hydrophobic injectedintoliquidchromatographthe limitsquantificationwhichvaries, also between intensity me liquid following recovery variable of because peptide to peptide from widely varies amount of peptide delivered to the detector, although of the efficiency delivery to-chargeratio withresolution. high Intensity depends onboth chargethe and instrument we used measures their intensities, as well as measuring their mass- Orbitrap the spectrometer,mass the to delivered are and ionization undergo chromatograph, liquid the from elute peptides As spectrometer. mass the in peptide’sdetected each reliessignal ion-current that method label-free a used Intensity-based mass-spectrometric quantification. Supplementary Table 1 enriched at as well as which were present in the group of all proteins that were significantly protein or protein group were identified in the combinedmatching BUNto a andprotein UTRor dataprotein sets,group. We also determined which paralogs proteinfor entry a or group, each calculating for the genomeaverage chicken sequence the identityin for paralogs all paralogs all identified also we tool, BioMart Chicken Ensembl database ( Paralog identification. assigned to it ( All proteins in the BUN preparation had one (and only one) of these ontologies consistent set of ontology annotations to apply to all bundlewas not proteinsuseful for (seeannotation of bundle proteins. Accordingly, we chose a simple, determined by the Organization, for a protein’s symbol. andsymbol. Whenever possible, we chose thehuman ortholog’s gene name, as species, particularly mouse and human, to determine an appropriate description Ensembl-identifiedandotheroforthologs chicken Ensembl entry the of tions proteins all identified in the BUN preparation, we manually examined annota via the PRIDE partner repository ProteomeXchange Consortium ( output files and modified Ensembl FASTA database, have been deposited to the group.proteomicsmass spectrometry All data,including rawdata,MaxQuant thegreatest intensity associated with it towas used identify theshared-peptide NP_ NP_ MaxQuant version 1.2.2.5 software was used for protein identification andidentification proteinfor used was software 1.2.2.5 versionMaxQuant Protein identifications were reported with an FDR of 1%. Proteins identified annotation of the chicken genome is poor and, in many cases, Annotation of the chicken genome is incomplete and occasionally wrong. For 001007825. 00117120

chromatography and differing degrees of ionization. Thus the relationship NP_001159798. XP_417532. P < 0.05 over the contamination fraction. These data are reported in 1 Supplementary Table 1 9 0 . The default contaminants file associated with the MaxQuantthe with associatedcontaminants default file The . ), ATP synthase beta ( 1 ), actin beta ( 3 ) and CACNA2D2 ( a sured in the mass spectrometer and the amount of a peptide 1 ), calbindin 2 ( . Using data downloaded during October 2012 from the 9 te neb FSA ie a eie b replacing by edited was FASTA file Ensembl the ; http:// NP_99084 http://proteomecent 5 www.ensembl.org/Gall 1 with the data set identifier PXD000104. NP_990647. 4 NP_0010265 , are expressed in chick utricle ). XP_427707. 9. 1 SupplementaryTable 1 ), fascin 1 ( 1 ), PDZD7 ( 62. nature nature 3 ral.proteomexchange. ). 1 To quantify proteins, we NP_00117160 ), peptidylprolyl isomer us_gallus/Info/Index NEUR XP_00364 OSCI 2 . If a set of set a If . 3 0 ), fascin 2 , suggest Fig. 2 1537. EN org C b 1 / / ). ), E - - - - ) )

© 2013 Nature America, Inc. All rights reserved. pressure freezing and freeze substitution freeze and freezing pressure E using the exactRankTests package in the R statistical computing environment. tiple test corrections by the false discovery rate nation fraction of log riBAQ/UTR riBAQ) were significantly greater than the corresponding contami Permutation tests were employed to test whether the mean values of log the increased numbers of deafness-associated proteins in the Statisticalhair analysis.bundle samples. bundles, for 56 kDa kDa for epithelium). (46 spectrometry mass by estimated fraction mole the by the average molecular mass for all proteins in each sample, which was weighted the amount of protein per ear’s worth of bundles or epithelium, then dividing by moles of total protein per ear. We estimated moles of protein per ear by dividing convertedwasearof ngper ear,moltoper dividedwasthenbynumber the of estimate the protein, each of fraction mole the Toestimateprotein. standard each for out carried were experiments Three samples.epithelium and bundle siswas carriedout to determine theamount of each protein (ngear) per inthe intensities were measured using Fiji imaging software andpreviously described as essentiallyimmunoblottinglinearout carried were regression analy MYO1C and four dilutions between 1.25 and 10 ng for FSCN2. SDS-PAGE and for CALM and ANXA5 standard curves, four dilutions between 0.01 and 1 fromng for human; USCN Life Sciences). Five dilutions between Coluccio,BiomedicalInstitute)Research1BostonL. (recombinant, FSCN2and and 200 ng were used ImmunoToolschicken;from MYO1CofGmbH),(recombinant, gift rat;from standardswere CALM(bovine brain; Millipore),EMD (recombinant, ANXA5 with dilutions of purified protein standards, and separated byples wereSDS-PAGE. loaded into 10- or 15-well Novex Protein NuPAGE Bis-Tris 4–12% gels, along MaxQuant analysis with riBAQ determination. and trypsin proteolysis; LC-MS/MS using the Orbitrap mass spectrometer; and alkylation reduction, pieces; six into gel the slicing by followed was that tion SDS-PAGE1-cm separa a includingepithelia, utricular and bundles hair for We carried out four independent experiments using methods identical to those proteins, eight each at 10 pmol, 1 pmol, 100 fmol, 10 fmol, 1 fmol and 100 amol. (~500 pmol total) of the standard proteins (Universal Proteomics Standard 2; Sigma-Aldrich) into 25 iBAQ values, generating an riBAQ value for each protein: divided each remaining protein’s iBAQ value by the sum of all non-contaminant (from blood), egg white proteins (for example, ovalbumin) and trypsin. We then preparation workflow, which include keratins (from human skin), hem our enteredcontaminantproteinsthat all analysis the fromremoved first We(riBAQ). iBAQ relative abundance, molar of measure normalized a its summed intensity and its iBAQ value. the MaxQuant (intensity) into one that is proportional to molar amount (iBAQ). The release of acidsin length. This operation converts ameasure that is proportional to mass peptides, which is considered to be all tryptic peptides between 6 and 30 amino each protein are summed and divided by the number of theoretically observable is averaged out and quantification accuracy is improved. withlarger proteins—which generate many peptides—inter-peptide variability surement of protein abundance uses the sum of many peptide measurements, so quantified less accurately than those with an average peptide. Nevertheless,are peptides tryptic mealongor overabundance shortan proteinsof with so acids, amino 6–30 typically range,relativelymass narrow a in peptides detect mally proteins nature nature saline chick in dissected were epithelia utricle Chick samples. sensitive lectron microscopy sample preparation and tomography.and preparation sample microscopy lectron For iBAQ validation experiments, we spiked one-fifth of a vial of UPS2 UPS2 of vial a of one-fifth spiked we experiments, validation iBAQ For generatedTowe stereocilia, protein absolutein amountseach determineof In the iBAQ algorithm For quantitative immunoblot validation of riBAQ values, BUN and UTR sam 5 2 NEUR . Moreover, mass spectrometers like the one we used are tuned to opti 1 0 we used (version 1.2.2.5 OSCI Fisher’s exact test was used to determine the significance of 2 (0.20). Exact EN E. coli 1 C 1 , the intensities of the precursor peptides that map to E extract. Each experiment thus included 48 human riBA P Q values were computed and adjusted for mul = i= ∑ ) n reports for each identified protein both iBAQ 1 5 iBAQ 3 to optimally preserve optimally to 1 7 i . All computations were done by i We used high- We used

osmotically osmotically 4 ,

o 3 sample- 1 globins 2 . Band . (BUN (1) µ g ­ ------

Microscopy Sciences), 0.2% tannic acid, 100 mM KCl, 5 mM MgCl was fixed for 2 h at room temperature (20–25 °C) in 3% glutaraldehyde (Electron electrical embedding uses two forces. The ‘attractiveforces.The twoforce’ uses embedding electrical dimensions.Spring-two in graph the of modelphysical a of energy the mizes a straight-edge drawing algorithm, spring-electrical embedding keptwerewasrelativelyand nodesminimizedspacing between Weeven. used the mathematicalTosense). representinteractions in the aesthetically, crossingsgraph,link (a drawing a in displayed are they interactions, these between drawn with line segments rather than arrows. To visualize the inte because in our data teininteractions. Two vertices nodes Table2 Stereociliary protein network. to allow adequate display of the cross-linkers and connectors. deliberatelydiametersmallerincylinders used than theirrespective structures connectors and cross-linkers filaments, actin showing model simplified The an FEI Tecnai T12 LaB6 electron microscope operated at 120 kV. Tecnai F20 electron microscopes, whereas analyzed tilt series waswere primarilycollected doneusing with JEOL1200-EX (TEM only), Philips CM200 FEGtions and usingFEI a Denton Vacuum system (DV-502). Initialminimize charging, screeninga thin film ofand carbon tomographywas deposited on grids containing sec fiducial markers for tomography. For greater stability upon with beam0.6% exposureFormvar andcoating tofor imaging and decorated with 10 nm or 15 nm gold120 nm in thickness, were placed on 2 × 1 mm oval-hole copper-rhodium grids 120 nm for screening or for tomography, respectively. Thin sections, nominally described outas infiltrationcarriedand dingwas mix containing 0.1% uranyl acetate in 100% acetone. Epon-araldite resin embed freezer. Thereafter, freeze substitution was performed using a freeze substitution 20% glycerol cryoprotectant and frozen using a Bal-Tec HPM 010 high pressure water.deionizedwith immediatelyrinses threeby Samplesa werewith coated with 2% aqueous uranyl acetate atout room carried temperaturewas staining water.Subsequent deionized (20–25with °C) rinses formore three 1 h, followed and bufferphosphate sodium M 0.1 with rinses three by followed ice, on h 1 glutaraldehyde and tannic acid. times for 5 min each at room temperature (20–25 °C)3-( in the same solution without piperazineethanesulfonic(HEPES))containingacid CaCl mM 0.1 (155 a5; for ATP2B2, F2a (ref. ATP1A1,forDMA1;Sigma-Aldrich Developmental StudiesHybridoma Bank anti-chick ANXA5 from G. Richardson (University of Sussex); for alpha-tubulin, methods. which we used to generate the network figure. (WolframResearch; 8 Mathematica roles in bundles. The spring-electrical embedding algorithm is implemented in similar out carry may which proteins,related clustering thus ones, unrelated These modeling rules draw together nodes with similar interactions and disperse determined by minimizing the energy function described by these two forces nodes: between distance ‘electrical force’ (repulsive) is global and inversely proportional to the Euclidian spring-likea constant that maintains optimalthe distance The nodes. between adjacent nodes is proportional to the Euclidian distance between them ( Chimera UCSF using done was building model simplified interactive and programClipIMOD.filteringmedianusing theinordiffusion, Segmentation tions of a bilateral filter using PRIISM warepackageIMOD were aligned and reconstructed into a three-dimensional volume with the soft axis tilt series were recorded at 1° intervals for angles of up to magnification, corresponding to a pixel size of 0.8 nm atquantification ~0.8using werecollected the specimen level. Dual- (F20)orGatan a 2,048-pixel10002,048× camera CCD (T12). Tomograms for CDD 4K TietzF415 a (CM200), Gatan CCD First(JEOL1200-EX),a 4K Light Images were recorded with a Gatan MegaScan Model 794/20 2K CCD CCD 2K 794/20 Model MegaScan Gatan a with recorded were Images OsO 1% with stained Sampleswere N

-morpholino)propanesulfonicthreewashed then 6.8, pH(MOPS) atacid mM NaCl, 5 mM KCl, 5 mM N can be represented in the form of a graphrepresenteda formof be the canin correspond to bundle proteins and links  Primary antibodies were as follows: for ANXA5, polyclonal n 5 4 , m . Resultingmaps. were iteraprocessedthree eitherwith  5 0

f ∈ r ); for ATP5A1, BD Transduction 612516; for CALM, = =

L implies n K and The protein interaction data in 2 / d d ij -glucose and 10 mM 4-(2-hydroxyethyl)-1- . The layout of the graph vertices is then then is vertices graph the of layout The . m http://www.wol 5 5  form anedge of thegraph if 4 or with successive rounds of smoothing, m in 0.1 M sodium phosphate buffer forphosphate buffer sodium M 0.1 in µ , n m underfocus at a nominal13,500× a atunderfocus m 

∈

L , the graph is undirected and is L 2 2 G correspond to specific pro . Sections were cut at 70 or 70at werecut Sections . = fram.com/mathematica  N , doi:10.1038/nn.3312 L f , a ± = where the set of set thewhere 65°. Projections Supplementary 2 d 5 r , which mini ij relationships 2 2 2 and 20 mM / . The tissueThe .  K n between between , m d ij  );

∈ K

5 2 / is L 6 5 ), ------; . .

© 2013 Nature America, Inc. All rights reserved. buffer. SDS-PAGE and immunoblotting were carried out essentially as above. sample SDS with eluted were proteins and washed Immunoprecipitateswere (20 anti-HA-agarose with at16,000 centrifuged was extract inhibitors;proteasethe and NP40 Triton0.5% X-100, probesonicatorcontainingausing PBSweretransfection.lysed with 1%Cells (Qiagen) Effectene using cells HEK293T in h 24 for expressed were Proteins utricle cDNA into expression vectors with respectively Myc and HA epitope tags. diluted 1:10,000. were used at 1:1,000, except anti-CALM (1:500). All secondary antibodies were ting were carried out essentially as described previously branes were removed without protease treatment. SDS-PAGE and immunoblot in cold, oxygenated chicken saline containing 4 mM CaCl University) for immunoblotting. Hopkins Johns Donowitz; M. (from 2331B (anti-NHERF2) anti-SLC9A3R2 fromJ. Kendrick-Jones (MRC Laboratory of Molecular Biology).We used also obtained were MYO6 to Antibodies CQRHSHSFSSHSSRKDLNGQKE). and (CHSDLQSPGKESEDGDSEK peptides SLC9A3R2 chicken two of mixture a against We sera polyclonal8D4. raised Sigma-Aldrich , for and (67B9); 2117 Signaling RHOA, forCell (41A3); 3149 Signaling Cell pERM, for (1D9); PTPRJ, monoclonal antibody D37 (ref. for PCDH15, G19 anti–tip link antigen Synthesis, Inc.); for MYO7A, 138-1 (Developmental Studies Hybridoma Bank); rabbit anti-GDWIRKPLYGLFQYNSSMIGLESLC (produced for us by Genemed 22 amino acids (from B. Burnside,anti– QHF UniversityMYO3A, for Inc.); of Synthesis, California, Berkeley); for MYO3B, (ref. Genemed by us for (produced kDa 15 2652 C-terminal MYO1Hanti-chick G5991 antibody MYO1C, Sigma-Aldrich for MDH2, anti-CVSTEPAWLALAKRKAKAWSD for HPA019714; Inc.); Synthesis, rabbit Genemed by us for KIAA1211, (produced for ab32618; anti-CYTLEFKAGKLAFKD (ref. FSCN2, for (NIH); Kachar B. from pan-espin and University) (Stanford Heller S. from anti-espin ESPN, for H-300; Cruz Santa ENO1, for University); State (Colorado Bamberg J. from ADF/cofilin anti- DSTN, for 610154;Transduction BD Labs CTNNB1, for Zürich); (ETH Wallimann T. from anti-chBCK rabbit CKB, for Inc.); Synthesis, Genemed by CATRPAPPDRERQ against (raised junction EC15/16 anti-CDH23 goat Millipore 05-173; for CDC42, SignalingCell 2466 (11A11); for CDH23, C2367 doi:10.1038/nn.3312 For immunoprecipitation,For Forimmunoblotting, utricles were dissectedfrom day 20–21 chick embryos 3 1 ; o GPH Ceio MB7; o HP5 GP8, Abcam (GRP78), HSPA5 for MAB374; Chemicon GAPDH, for ); g for 15 min. The supernatant was removed and 300 µ l 50% slurry; Sigma A2095) overnight at 4 °C. °C. 4 at overnight A2095) Sigma slurry; 50% l RDX and 5 1 8 SLC9A3R2 4 ; for PRKAR2A, Santa Cruz M-20; for ); for RDX, Abnova H00005962-M06 Xenopus laevis Xenopus were cloned from chicken from cloned were 5 7 ); for MYO1H, rabbit rabbit MYO1H, for ); 4 , 3 MYO3A C-terminal C-terminal MYO3A 2 1 , and otolithic mem . Primary antibodies µ lwas incubated - -

58. 52. 51. the stack. using Fiji ( before saving the images as 24-bit merged TIFF files. All formostpositivethe image andother appliedimagesthe for consistencyall to objective using an iterative algorithm of ten iterations.imagesweredeconvolved opticaltransferoptimizedthe function forwiththat The histogram was adjusted 0.2- and zoom 3× with objective a Apo Plan NA 1.42 60× using software, acquisition 3.0 AF10-ASW with system microscope focal 0.12 mm deep, Invitrogen). Vectashieldin (Vector Laboratories) spacer(eightusing wells,one Secure-Seal slides on Invitrogen). mounted and each min Probes, 20 for times (Molecular three rinsed then were Phalloidin Organs 488 Fluor Alexa U/ml 0.4 and bated for 3–4 h in blocking solution with 1:1,000both at Alexa1:500), then rinsed three Fluortimes for 10 min each. secondaryOrgans were then incu antibodies (1:250 dilution for all primary antibodies except anti-SLC9A3R2 and anti-RDX, incubated overnight at 4 °C with primary antibodies diluted in blocking solution for 2 h in 2% bovine serum albumin, 5% normal goat serum in PBS. Organs rinsedwere in PBS, permeabilized for 10 min in 0.5% TritonwereOrgans saline. chicken inX-100 Sciences) Microscopy formaldehyde(Electron in PBS and blocked 57. 56. 55. 54. 53. Images were acquired on an Olympus Fluoview FV1000 laser scanning con 4% in min 25 for fixed were utricles dissected immunocytochemistry,For

Goodyear, R.J. & Richardson, G.P. A novel antigen sensitive to calcium chelation calcium to sensitive antigen novel G.P.A Richardson, & R.J. Goodyear, arr, .. Rbno, .. dacs n h ms setoer o membrane of spectrometry mass the in Advances C.V. Robinson, & N.P. Barrera, J.A. Vizcaíno, Dumont, R.A., Zhao, Y.D., Holt, J.R., Bahler, M. & Gillespie, P.G. Myosin-I isozymes E.F.Pettersen, system integrated an Priism: D.A. Agard, & J.W. Sedat, W.K., Clyborne, H., Chen, three- of visualization Computer J.R. McIntosh, & D.N. Mastronarde, J.R., Kremer, G.E. Sosinsky, rtis fo idvda poen t itc complexes. intact to proteins individual from proteins: Res. Acids ht s soitd ih h tp ik ad ioiil ik o snoy ar bundles. hair sensory of links Neurosci. kinocilial J. and links tip the with associated is that (2002). 375–389 epithelia. vestibular and auditory rodent neonatal in analysis. and (1992). images. microscope 3-D of analysis and display for IMOD. using data image dimensional tissue labile highly preserves (2008). substitution freeze applications. tomography electron for and ultrastructure freezing pressure high (2011). 247–271 http://fiji.sc

38

23 J. Comput. Chem. Comput. J. et al. et , D736–D742 (2010). D736–D742 , et al. et , 4878–4887 (2003). 4878–4887 , t al. et The Proteomics Identifications database: 2010 update. 2010 database: Identifications Proteomics The / UCSF Chimera–a visualization system for exploratory research exploratory for system visualization Chimera–a UCSF ); the Reslice function was used to generate an The combination of chemical fixation procedures with procedures fixation chemical of combination The

25 , 1605–1612 (2004). 1605–1612 , J. Struct. Biol. Struct. J. nature nature J. Struct. Biol. Struct. J. J. Assoc. Res. Otolaryngol. Res. Assoc. J.

rc SPIE Proc. 116 nu Rv Biochem. Rev. Annu. z , 71–76 (1996). 71–76 , -stacks were processed µ m NEUR z -steps. Confocal Confocal -steps.

1660

161 OSCI 784–790 , x , 359–371 , - z Nucleic slab of EN

80 C

3 E - - , ,