ARTICLE Received3Dec 2015 |Accepted 1 Jul2016 |Published 3Aug 2016 DOI:10.1038/ncomms12426 OPEN Cingulin and actin mediate midbody-dependent apical lumen formation during polarization of epithelial cells Anthony J. Mangan1, Daniel V. Sietsema1, Dongying Li1, Jeffrey K. Moore1, Sandra Citi2 & Rytis Prekeris1 Coordinated polarization of epithelial cells is a key step during morphogenesis that leads to theformationofanapicallumen.Rab11anditsinteractingproteinFIP5arenecessaryforthe targeting of apical endosomes to the midbody and apical membrane initiation site (AMIS) during lumenogenesis. However, the machinery that mediates AMIS establishment and FIP5-endosome targeting remains unknown. Here we identify a FIP5-interacting protein, Cingulin, which localizes to the AMIS and functions as a tether mediating FIP5-endosome targeting. We analysed the machinery mediating AMIS recruitment to the midbody and determinedthatbothbranchedactinandmicrotubulesarerequiredforestablishingthesiteof the nascent lumen. We demonstrate that the Rac1-WAVE/Scar complex mediates Cingulin recruitment to the AMIS by inducing branched actin formation, and that Cingulin directly binds to microtubule C-terminal tails through electrostatic interactions. We propose a new mechanism for apical endosome targeting and AMIS formation around the midbody during epithelial lumenogenesis. 1DepartmentofCellandDevelopmentalBiology,SchoolofMedicine,AnschutzMedicalCampus,UniversityofColoradoDenver,Aurora,Colorado80045, USA.2CellBiologyDepartment,UniversityofGeneva,CH-1211GENEVA4,Switzerland.Correspondenceandrequestsformaterialsshouldbeaddressedto R.P.(email:[email protected]). NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications 1 ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 T he formation of an apical lumen is a key step during AMIS.Toidentifythesetargetingfactorsweimmunoprecipitated epithelialtissuemorphogenesisandfunction,anditisnow FIP5 from polarized Madin–Darby canine kidney (MDCK) cells wellestablishedthatRab-dependentendosometransportis (SupplementaryFig.1a).Manyoftheidentifiedproteins(Fig.1b) responsible for driving individual cell polarity as well as de novo arealreadyknowntoregulateapicalvesicletransport,confirming lumen formation1–4. Specifically, the Rab11 family of GTPases the efficacy of the immunoprecipitation. In fact, SNX18, were shown to regulate the transport of vesicles carrying apical dynamin-2 and Arp2/3 are all known to form a complex that is cargo to the site of the forming lumen, known as the apical essential for the budding of FIP5 apical endosomes8.Myosin Vb membrane initiation site (AMIS)1,2,5–8. AMIS is a transient andFIP1areotherknowncomponentsofapicalendosomes11–15, structure that contains many proteins, including the Par3/Par6 and clathrin is a general vesicle coating factor16,17. Interestingly, polarity complex, the Exocyst complex and tight junction (TJ) we also identified CGN as a putative FIP5-binding protein proteins such as ZO-1 and Cingulin (CGN)1,2,5,7,8. De novo (Fig.1b,SupplementaryFig.1a).CGNisaknownAMISmarker formationofasingleAMISisanessentialcellularstepleadingto that is concentrated almost exclusively around the midbody theproperinitiationandexpansionofasingleapicallumen1,2,7,8. during late telophase7,8, thus making it a strong candidate to Recent work from our and other laboratories has shown that serveasatetheringfactormediatingRab11/FIP5apicalendosome midbody formation and midbody-dependent AMIS recruitment targeting. during telophase is the first symmetry-breaking event that To further test whether CGN binds to FIP5 we immuno- determines the time and site of apical lumen formation1,7. precipitated GFP-CGN from MDCK cells stably expressing However, the factors involved in AMIS recruitment to the GFP-CGN. Consistent with our proteomics data, FIP5 midbody are still unknown and are the focus of this study. co-immunoprecipitated with GFP-CGN (Fig. 1c). Since we have In addition to midbody-dependent AMIS formation, apical previously shown that FIP5-T276 phosphorylation inhibits endosometargetingandfusionattheAMISisalsoanimportant FIP5 (ref. 7), we next co-immunoprecipitated CGN with either step in apical lumen formation. Previous studies have begun to FIP5-T276A-GFP or FIP5-T276D-GFP and found that FIP5 identify the mechanisms of apical endosome budding and phosphorylation mimetic FIP5-T276D did not bind to CGN targeting and have shown that apical endosome transport is (SupplementaryFig.1b).Finally,purifiedrecombinant6His-FIP5 governed by Rab11 GTPase bound to its effector protein known and 6His-CGN were also co-immunoprecipitated with anti-FIP5 as Rab11 family interacting protein-5 (FIP5)6–8. The sequential antibody(Fig.1d),furtherdemonstratingthatCGNdirectlybinds interactions of Rab11/FIP5 targeting complex with Sorting to FIP5. Nexin-18 (SNX18) and Kinesin-2 regulate apical endosome CGN contains three main domains: a large globular head, formation and transport along central spindle microtubules coiled-coil rod and short globular tail (Fig. 1e)18. The head duringtheinitialstepsoflumenogenesis6,8.Althoughitisknown domain is known to bind ZO1 and mediate CGN recruitment that these vesicles fuse with the plasma membrane at the AMIS, to TJs18,19 and the coiled-coil rod is required for CGN homo- thespecificmechanismsoftargetingandtetheringofRab11/FIP5 dimerization and mediates binding to GEF-H1 (Fig. 1e)18,20. To endosomes to the AMIS are not fully understood. While several maptheFIP5-bindingmotifweusedaGSTpull-downassayand proteins, such as synaptotagmin-like proteins Slp2 and Slp4 as showed that FIP5 binds to the amino-terminal portion of the well as the Exocyst complex, were shown to be required for coiled-coilrod(aminoacidregion355–579)(Fig.1e).SinceCGN single-lumen formation9, it is unlikely that they alone can target (355–579)islocatedjustafterthelargeglobularheaddomain,we endosome transport to the AMIS, since most of these factors refer to it as the CGN-coiled-coil-1 region (Fig. 1e). localizeandfunctionatothersubcellularlocationsinadditionto the AMIS and/or midbody, thus limiting their ability to serve as AMIS-specific tethers for incoming apical vesicles. CGN is required for single apical lumen formation. To assess Here, we investigate the machinery that mediates AMIS the role of CGN in lumen formation, we created a doxycycline formation at the midbody, as well as the targeting/tethering of (dox) inducible MDCK-shCGN cell line (Supplementary apical endosomes during lumenogenesis. We have identified Fig. 1c,d) and tested the effects of CGN knock-down (KD) on CGN10 as a FIP5-binding protein and have shown that CGN apical lumen formation. After embedding MDCK cells in serves as the tethering factor that ensures the fidelity of apical Matrigel for 4 days, dox(cid:2) cells produced single-lumen cysts endosometargetingtotheAMIS.WealsoshowthatCGNbinds (Fig.2a,d)withCGN,FIP5,gp135and actinoutliningtheapical to the carboxy-terminal tails of midbody microtubules, and that domain (Fig. 2a, Supplementary Fig. 2a,c,i). In contrast, CGN this CGN and microtubule interaction may play a major role in knock-down led to increased formation of cysts with multiple recruiting the AMIS to the midbody during late telophase. lumens (Fig. 2b,d, Supplementary Fig. 2b,d,e,j). These multiple Finally, we uncovered a novel and midbody-dependent role of lumens are likely the result of apical cargo mistargeting, since Rac1-WAVE/Scar-inducedactinpolymerizationduringtheinitial CGN depletion did not have any effect on the angle and steps of apical lumen formation. As the result of this data, we positioning of the mitotic spindle (Supplementary Fig. 3c–f). propose a new apical lumen formation model that explains how Furthermore,eventhedoxþ cellsthatcontainedasinglelumen polarized endocytic membrane transport, midbody microtubules had defects in apical surface formation. As shown in Fig. 2c, and branched actin cytoskeleton interact and function as a Supplementary Fig. 2e, MDCK-shCGN cells appear to have a coincidencedetectionsystemthatregulatesthetimingandfidelity largeapicalplasmamembranesurfacethatextendsandfoldsinto of single apical lumen formation. the luminal space. Significantly, the addition of shRNA-resistant GFP-CGNtodoxþ cellsrescuedthesingle-lumenphenotypeas wellastheintegrityoftheapicalsurface(Fig.2d,Supplementary Results Fig. 2f, t-test). CGN is a FIP5 binding protein concentrated at the AMIS. Our data suggest that CGN regulates formation of a single During de novo lumen formation the AMIS is established at the apicallumenbyaffectingtargetingofRab11/FIP5endosomes.To midbodyduringlatetelophase,markingthesiteofafutureapical test this we analysed MDCK-shCGN cells embedded in Matrigel lumen1,7(Fig.1a).FollowingAMISformation,Rab11/FIP5apical for 12h. Under these conditions,mostembedded cellsare either endosomesaretransportedtotheAMIS(Fig.1a)1,6.Whatisnot inlatetelophase,orjustcompletedtheirfirstdivisionandformed knownarethemechanismsthattargetRab11/FIP5vesiclestothe anascentmini-lumenbetweentwodaughtercells7.Todetermine 2 NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 a Telophase cell b Midbody Proteins identified in FIP5 immunoprecipitate AMIS Protein Spectral index FIP5 185 SNX18 31 Forms dynamin-2 24 complex Arp2/3 4 Rab11/FIP5 Myosin Vb 36 endosomes GEF-H1 10 Forms Cingulin 5 complex FIP1/RCP 11 clathrin 3 Nucleus TJ n e m u L Nucleus TJ Post-mitotic cell c ate e Cingulin ys P Head 346 Rod Tail K l GF 0– –1,203 kDa MDC IgG Anti- ZO1/2 FIP5 GEF-H1 1,155 17550–– --GFIFPP5-CGN mg) 406) 5–579) 1–794) 1–1,025) 5 (2 N(1– N(35 N(57 N(78 P G G G G 50– -Heavy chain FI C C C C s- T T- T- T- T- kDa 6Hi GS GS GS GS GS d 5 ng FIP 75– –6His-FIP5 kDa 100 IgG Anti- 50– 150– –6His-CGN 75– –6His-FIP5 25– Coomassie 50– -Heavy chain Figure1|CGNisanAMIS-associatedFIP5-bindingprotein.(a)Schematicmodelshowingtheformationofanascentapicallumen.Rab11/FIP5 endosomesaretransportedalongcentralspindlemicrotubulesandfusetotheAMISpresentatthemidbodyduringlatetelophase.(b)Listofproteins identifiedinFIP5immunoprecipitatebymassspectroscopyanalysis.Proteinsknowntobepresentinthesamecomplexarehighlightedtogether. (c)FIP5co-immunoprecipitationwithGFP-CGNfromMDCKlysates.Imagesshownaretheimmunoblotsafterprobingwithanti-CGN(topgel)oranti-FIP5 (bottomgel)antibodies.(d)Co-immunoprecipitationofpurified6His-CGNwith6His-FIP5.Imagesshownaretheimmunoblotsafterprobingwith anti-CGN(topgel)oranti-FIP5(bottomgel)antibodies.(e)Mapping6His-FIP5bindingdomainusingglutathionebeadpull-downassays.Imageshownis CoomassiestainedSDS–PAGEgel. whether CGN knock-down affects Rab11/FIP5 endosome contrast, in doxþ treated cells actin accumulation at the apical transport, we stained cells with anti-FIP5 antibodies. As plasma membrane is lost, and actin is instead present in distinct previously reported, FIP5 and gp135 are predominantly focialongtheentireplasmamembrane(Fig.2f–h,l).Insomecells concentrated at the AMIS (Fig. 2i,k, arrows; Supplementary we also observed large intracellular structures containing Fig. 2g) as marked by either tubulin or actin. In contrast, in the enrichedactin(Fig.2g,h,asterisks).Theseintracellularstructures doxþ treated cells FIP5 endosomes are distributed throughout are likely either enlarged apical endosomes or ectopic apical the daughter cells, indicating that CGN is required for the lumen-like structures, since they also contain FIP5 (Fig. 2j, targetingofFIP5(Fig.2j,l).Consistently,gp135,awell-established asterisk). CGN knock-down also changes the appearance of FIP5-endosome cargo protein, was also scattered throughout the actin in two-dimensional cultures, where stress fibre formation cytosol (Supplementary Fig. 2h). Additionally, in some cells we is induced at the basal side of the epithelial monolayer couldobservegp135accumulatinginenlargedcellularstructures (Supplementary Fig. 3a,b). that may represent ectopic nascent lumens (Supplementary Fig. 2h, arrow). Surprisingly, CGN knock-down also disrupted actin cytoske- TheWAVE/ScarcomplexispresentattheAMIS.Althoughwe leton localization (Fig. 2f–h,l). Typically, during lumenogenesis, have previously shown that CGN and the AMIS are recruited to actinclearlymarksthesurfaceoftheapicallumen(Fig.2e,k).In the midbody during late telophase7, the mechanisms mediating NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications 3 ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 a CGN CGN b CGN CGN Actin Actin 10 µm dox – 10 µm 10 µm dox + 10 µm c CGN CGN d 80 Actin er (% total) 4600 P<0.001 P<0.001 b m u n 20 st y dox + C 0 10 µm 10 µm dox – + + – + + – + + GFP-CGN – – + – – + – – + One lumen Two lumens Multi-lumens e Actin f Actin g Actin h Actin CGN CGN CGN CGN * * dox – dox + dox + dox + 10 µm 10 µm 10 µm 10 µm i FIP5 Tubulin j FIP5 Tubulin * 10 µm dox – 10 µm 10 µm dox + 10 µm k FIP5 Actin l FIP5 Actin AMIS AMIS 10 µm dox – 10 µm 10 µm dox + 10 µm Figure2|CGNisrequiredfortheformationofasingleapicallumen.MDCKcellsstablyexpressingCGNshRNAwereembeddedin3DMatrigeland allowedtogrowfor4days(a–d)or24h(e–l)inthepresence(doxþ)orabsence(dox(cid:2))ofdoxycycline.Cellswerethenfixedandstainedwith phalloidin-Alexa594(a–c,e–h,k,l;red),anti-CGN(a–cande–h;green),anti-tubulin(i,j;red)oranti-FIP5(i,j;green)antibodies.Asterisksin(g,h,j)mark ectopiclumenformationsites.Arrowsin(k)pointtotheAMIS.(d)showsquantificationofcystswithone,two,ormultiplelumens.Datashownaremeans ands.d.ofthreeindependentexperiments.Consecutiveimageswithoutanindividualletterlabelshowthesamecell. theestablishmentoftheAMISatthemidbodyremainunknown. embeddedinthree-dimensional(3D)matrix.AsshowninFig.3c, Thus, we next synchronized MDCK cells at telophase and Nap1co-localizesaroundthemidbodyduringlatetelophaseand immuno-precipitatedCGN.Isolatedproteinswerethenanalysed alsoformsaringstructurelocalizedtoTJsaftertheformationof by mass spectroscopy and compared with those pulled down in the nascent lumen (Fig. 3d,e), mirroring what we previously anIgGcontrol.Amongtheproteinsidentifiedonlyinanti-CGN observedwithCGN7.Likewise,Arp3becomesconcentratedatthe (Fig. 3a) we identified ZO1, a known CGN-binding protein. midbody and co-localizes with CGN during AMIS formation ConsistentwithourfindingthatCGNisanFIP5-bindingprotein, (Fig. 3f). We also analysed the actin cytoskeleton during AMIS we also identified FIP5 and SNX18 from anti-CGN antibody formationandlumenexpansion,anddetectedanincreaseinactin precipitates (Fig. 3a). polymerization around the midbody (Fig. 3g,h, Supplementary In addition to expected CGN-interacting proteins we also Fig. 4a). These actin structures emanate from the AMIS, identified Nap1 and Abi2, two components of the WAVE/Scar suggesting that they could be actively playing a role in the complex (Fig. 3a,b). The WAVE/Scar complex is known to be establishmentoftheapicallumen.Afterformationofthenascent activated by Rac1 and induces the polymerization of branched lumen,actinisstillcloselyalignedwithCGN,markingtheapical actin through the activation of Arp2/3 (Fig. 3b)21–23. surface and TJs (Fig. 3i). On the basis of these findings we To determine if the WAVE/Scar complex may play a role in hypothesized that the midbody-associated WAVE/Scar complex the recruitment of CGN to the midbody, we first tested the is helping to recruit the AMIS and may also drive apical lumen localization of Nap1 and Arp3 during division of MDCK cells formation. 4 NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 a b Proteins identified in CGN immunoprecipitate Core Protein Spectral index Rac1 complex CGN 201 ZO1 11 Forms Wave/scar FIP5 10 complex Sra1 SNX18 9 TJ-associated protein 1 (Tjap1) 5 Nap1 4 Forms Arp2/3 Abi2 4 complex Activation TBC1D1 10 Nap1 Abi2 CGNL1 20 c Nap1 Acetylated tubulin Nap1 Acetylated tubulin Midbody Midbody Midbody Telophase d Nap1 e Nap1 f Arp3-GFP Arp3-GFP CGN Midbody Midbody AMIS Late anaphase TJs Telophase g Actin h Actin i Actin CGN Acetylated tubulin CGN TJ Lumen Midbody AMIS 5 µm 5 µm 5 µm Figure3|ComponentsoftheWAVE/Scarcomplexandbranchedactinfilamentsarepresentatthemidbodyduringlumenformation.(a)Listof proteinsidentifiedinCGNimmunoprecipitatebymassspectroscopy.Proteinsknowntointeractasacomplexarehighlightedtogether.(b)Schematic representationoftheWAVE/Scarcomplex.(c–i)MDCKcellswereembeddedin3DMatrigelandallowedtogrowfor24h.Cellswerethenfixedand stainedwithphalloidin-Alexa594(g–i;red),anti-CGN(f,g,i),anti-acetylatedtubulin(c;redandh;green)oranti-Nap1(c–e;green)antibodies.In(f), cellisexpressingArp3-GFP.Whitearrowsidentifystructuresaslabelled.Consecutiveimageswithoutanindividualletterlabelshowthesamecell. Active Rac1 is required for AMIS formation at the midbody. subcellular Rac1 localization can provide clues about its possible TheWAVE/Scarcomplexisactivatedbythebindingofthesmall function, ultimately GTP-bound Rac1 is what activates WAVE/ GTPase Rac1 (refs 21–23). Our finding that WAVE/Scar and Scarandinducesactinpolymerization.Totestthelocalizationof therefore Rac1 may be involved in regulating AMIS formation activeRac1weusedanRac1biosensor28.AsshowninFig.4e,in during late telophase is somewhat unexpected, since it is well metaphase,activatedRac1appearstobeequallydistributedalong established that RhoA rather than Rac1 drives cleavage furrow the entire plasma membrane. In contrast, once the cell enters formation and ingression during cytokinesis24–26. Importantly, telophase, Rac1 becomes activated at the AMIS (Fig. 4f,g). itwasdemonstratedthatRhoAisinactivatedatlatetelophaseto AlthoughtheseexperimentsdirectlylinkRac1activationtothe allow for the disassembly of the contractile actomyosin ring27. establishment of the apical lumen, it does not explain the This suggests that during late telophase RhoA inactivation is mechanism governing the recruitment and activation of Rac1 followedbyRac1activation,thusallowingatransitiontoArp2/3- during late telophase. Previous studies have shown that Arf6 is dependentbranchedactincytoskeleton.Tothatendweanalysed recruited to the midbody during telophase29,30. It was suggested the localization of Rac1 during lumen formation (Fig. 4a–d). that Arf6 binds and recruits Tiam1, a known Rac1 GEF31. During anaphase, Rac1 is evenly distributed throughout the Consistent with this hypothesis, Arf6 can be observed at the cytosolofdividingcells,consistentwithRac1notbeinginvolved midbody of the MDCK cells during lumenogenesis (Fig. 4h). in the formation of the actomyosin ring (Fig. 4a). During late WenextwantedtodetermineifRac1playsafunctionalrolein telophase, Rac1 becomes concentrated at the AMIS (Fig. 4b) the establishment of the apical lumen. It has been previously whereitco-localizeswithfilamentousactinandCGN(Fig.4b,c). shownthatRac1knock-downleadstoformationofinvertedcysts Upon formation of the nascent mini-lumen, Rac1 remains with the apical pole of the cells facing the ECM, thus lacking a enriched at the apical plasma membrane (Fig. 4d), indicating its well-defined apical lumen32–34. However, the use of cell lines possible involvement in maintenance of apical polarity. While expressingRac1shRNAisdifficulttointerpret,sinceRac1isone NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications 5 ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 a CFP-Rac1 Actin b CFP-Rac1 Actin AMIS AMIS 10 µm 10 µm 10 µm 10 µm c CFP-Rac1 CGN d CFP-Rac1 Actin AMIS AMIS Lumen Lumen 10 µm 10 µm 10 µm 10 µm e YFP CFP FRET g Area #1 Area #2 Area #2 20 2m) 15 5 µm Metaphase 5 µm 5 µm Rac1 biosensor m f YFP CFP FRET er p u. 10 T (a. P<0.0001 E 5 R AMIS AMIS AMIS F 0 5 µm Telophase 5 µm 5 µm Rac1 biosensor Area #1 Area #2 h Acetylated tubulin Arf6-CFP Arf6-CFP Arf6-CFP Figure4|Rac1isactivatedattheAMIS.(a–d)MDCKcellstransientlyexpressingCFP-Rac1(green)wereembeddedin3DMatrigelandallowedto growfor24h.Cellswerethenfixedandstainedwithphalloidin-Alexa594(a,b,d;red)oranti-CGN(c;red)antibodies.(e–g)MDCKcellstransiently expressingFRET-basedRac1biosensorwereembeddedin3DMatrigelandimagedeitheratmetaphase(e)ortelophase(f).(g)showsquantificationof FRETsignalattheAMIS(area#1)andcellperiphery(area#2)duringtelophase.Datashownarethemeansands.d.derivedfromsixcells.(h)MDCKcells transientlyexpressingCFP-Arf6(red)wereembeddedin3DMatrigelandallowedtogrowfor24h.Cellswerethenfixedandstainedwith anti-acetylatedtubulin(red)antibodies.Arrowsmarkthemidbody.Consecutiveimageswithoutanindividualletterlabelshowthesamecell. ofthemainregulatorsofactincytoskeleton.Asaresult,insteadof nuclei). Occasionally we could also observe a small CGN and shRNA, we used an inhibitor that specifically targets the Rac1 actin ring (Fig. 5c) that forms on the neck of a small bud-like GEFs (Tiam1 and Trio). To ensure that we are inhibiting Rac1 structure on one of the daughter cells. Finally, in some cases only during the first mitotic division, we embedded MDCK cells the cells divide, but fail to polarize and form any TJs of nascent into Matrigel for 4h, followed by a 12-h incubation with Rac1 mini-lumens (Fig. 5d). inhibitor. After 16h, most of the cells successfully complete the To further analyse the role of Rac1 in regulating AMIS first division, and the CGN-enriched TJ ring surrounds the formation we followed GFP-CGN dynamics during MDCK cell nascent mini-lumen, which is always found between the two division by time-lapse microscopy. Consistent with previously daughter cells (Fig. 5a). In contrast, cells treated with Rac1 published work7, CGN accumulated at the midbody during inhibitorfailedto formtheCGNring andapicallumen between telophase(Fig.5e,arrow).Uponcompletionofcelldivision,CGN thetwodaughtercells(Fig.5b).Instead,CGNformsaTJringat formedaringbetweentwodaughtercells,anindicationofapical thesideofoneofthedaughtercells(Fig.5b),despitethefactthat lumen and TJ formation (Fig. 5e). In contrast, treatment of cells cells successfully completed the first mitotic division (Fig. 5b; with Rac1 inhibitor blocked accumulation of CGN at the arrows mark the actin-rich plasma membrane between two midbody and consequently prevented apical lumen formation 6 NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 a Actin CGN Lumen Lumen Control 10 µm 10 µm 10 µm b Actin CGN Rac1 inh 10 µm 10 µm 10 µm c Actin CGN Rac1 inh 10 µm 10 µm 10 µm d Actin CGN Rac1 inh 10 µm 10 µm 10 µm e 0 h 0.5 h 1.5 h 8 h 11 h f 0 h Rac1 inh 1.5 h 8 h 13 h 15 h Figure5|Rac1inhibitionaffectsmidbody-associatedAMISformation.(a–d)MDCKcellswereembeddedin3DMatrigelandallowedtogrowfor24h. Whereindicated,cellsweretreatedwith200mMRac1inhibitor(b–d).Cellswerethenfixedandstainedwithphalloidin-Alexa594(red)andanti-CGN (green)antibodies.Whiteboxesindicateareazoomedinandshowninfarrightbox.Whitearrowspointtonascentlumen,blackarrowspointtoareasof actinconcentration.Consecutiveimageswithoutanindividualletterlabelshowthesamecell.(e,f)MDCKcellsstablyexpressingGFP-CGNwere embeddedin3DMatrigel.Untreated(e)orRac1inhibitortreated(f)cellswerethenimagedusingtime-lapsemicroscopy.Inallcases,cellprogressionfrom metaphasetoabscissionwasanalysed.Arrowin(e)pointstothemidbodyassociatedGFP-CGN. and CGN accumulation at the TJs (Fig. 5f, Supplementary formation of branched actin filaments at the midbody may also Fig. 5a). Since Rac1 activates the WAVE/Scar protein complex play a role in the correct placement of TJs around the midbody wenextcheckedthelocalizationofAbi2,thecoresubunit ofthe and nascent apical mini-lumen. complex. Similar to Nap1, GFP-Abi2 also localized to the Since Rac1 is required for a recruitment of the AMIS to the midbody during telophase (Fig. 6a), as well as to TJs after midbody, Rac1 inhibition should lead to defects in targeting of completion of the first division and the establishment of the apicalproteins.Totestthis,weanalysedlocalizationofgp135,an nascent apical lumen (Fig. 6c). Incubation of dividing cells with apicalplasmamembraneproteindeliveredtotheapicallumenvia Rac1 inhibitor blocked GFP-Abi2 recruitment to the midbody Rab11andFIP5endosomes.Aspreviouslyreported7,8incontrol (Fig. 6b). Interestingly, even recruitment of GFP-Abi2 to the TJs cellsgp135isdeliveredtothemidbody-associatedAMIS(Fig.6e). was affected, although not completely inhibited (Fig. 6d). Taken These gp135-containing endosomes eventually fuse to form together,thesedataindicatethatRac1andWAVE/Scar-dependent nascentmini-lumen(Fig.6f).InhibitionofRac1ledtoafailureof NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications 7 ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 a Acetylated tubulin Acetylated tubulin b Acetylated tubulin Acetylated tubulin GFP-Abi2 GFP-Abi2 10 µm Control 10 µm 10 µm Rac1 inh 10 µm c GFP-Abi2 d GFP-Abi2 10 µm Control 10 µm Rac1 inh e CGN gp135 f CGN gp135 CGN CGN Control Control 10 µm 10 µm g CGN gp135 h CGN gp135 CGN CGN 10 µm Rac1 inh 10 µm 10 µm Rac1 inh 10 µm i CGN gp135 j CGN gp135 CGN CGN 10 µm Rac1 inh 10 µm 10 µm Rac1 inh 10 µm Figure6|Rac1isrequiredforthegp135targetingduringapicallumenformation.(a–d)MDCKcellstransientlyexpressingGFP-Abi2wereembeddedin 3DMatrigelandallowedtogrowfor24h.Whereindicated,cellsweretreatedwith200mMRac1inhibitor(b,d).Cells(a,b)werethenfixedandstained withanti-acetylatedtubulin(red)antibodies.Whitearrows(a,b)pointtothemidbody.(e–h)MDCKcellswereembeddedin3DMatrigelandallowed togrowfor24h.Whereindicated,cellsweretreatedwith200mMRac1inhibitor(g,h).Cellswerethenfixedandstainedwithanti-gp135(green)and anti-CGN(red)antibodies.Whitearrowin(e)pointstothemidbody.Whitearrowin(g)pointstomisplacedCGN.(i,j)MDCKcellswereembeddedin 3DMatrigelandallowedtogrowfor2days.Cellsweretreatedwith200mMRac1inhibitorforthefirst12hoftheincubation.Cellswerethenfixedand stainedwithanti-gp135(green)andanti-CGN(red)antibodies.Whitearrowin(j)pointstomisplacedCGN.Consecutiveimageswithoutanindividual letterlabelshowthesamecell. gp135 targeting to nascent lumen. Instead gp135-containing withthegp135-containingplasmamembranefacingtheMatrigel vesicles accumulated in multiple clusters within the cytosol (Fig.6j).Theseinvertedcystsweresimilartotheonesobservedin (Fig.6g).Significantly,insomecellsweobserveaCGNringthat cells expressing dominant-negative Rac1 mutants32,34. forms at the neck of small bud-like structures, and gp135- If Rac1 is required for correct placement of the apical lumen containingvesiclescouldalsobeobservedatthosenecks(Fig.6g, andTJs,onewouldpredictthatinhibitionofRac1duringthefirst arrows),confirmingthatCGNdoesfunctionasafactormediating mitotic division should lead to the formation of mature cysts apicalendosometargeting/tethering.Insomecases,after16hthe containing multiple lumens. To test this prediction, we treated cells underwent two rounds of division, thus generating small embedded cells with Rac1 inhibitor for 12h, then changed the four-cell cysts. Typically, at this stage gp135 could already be media and allowed the cells to form mature cysts (4 days) observed enriched within small mini-lumen (Supplementary (Fig. 7a). As reported previously, untreated cells developed into Fig. 2f). Rac1 inhibition led to the formation of multiple well-structured,single-lumencysts(Fig.7b,c)withCGNmarking mini-lumens surrounded by CGN-containing TJs (Fig. 6i). TJsandactinoutliningtheapicaldomain(Fig.7c).Themajority However, B45% of all the cysts contained an inverted polarity, of cells treated with Rac1 inhibitor (Fig. 7b) formed cysts 8 NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 a b 80 al) ot 60 Incubate with % t Rac1 inhibitor Image sts ( 40 y c K DC 20 1 h 13 h 96 h M First cell Single Multiple No division lumen lumens lumen Rac1 inh. – + – + – + c Actin CGN Actin CGN Lumen 10 µm 10 µm 10 µm d Actin CGN Actin CGN Rac1 inhibitor 10 µm 10 µm 10 µm e Actin CGN Actin CGN 10 µm Rac1 inhibitor 10 µm 10 µm Figure7|Rac1isrequiredfortheformationofasingleapicallumen.(a)Schematicrepresentationshowingtimingof12htreatmentwithRac1 inhibitor.(b–e)MDCKcellswereembeddedin3DMatrigelandallowedtogrowfor4daysintheabsence(c)orpresence(d,e)ofRac1inhibitor. Thetimingofinhibitoradditionisshownin(a).Cellswerethenfixedandstainedwithphalloidin-Alexa594(red)andanti-CGN(green)antibodies. Consecutiveimageswithoutanindividualletterlabelshowthesamecell.(b)showsquantificationofcellswithsingleormultiplelumens.Datashownare themeansands.d.derivedfromthreeindependentexperiments. containing multiple lumens (Fig. 7d). It is likely that this multi- To inhibit Arp2/3 only during the first cell division, we treated luminalphenotypeiscausedbyincorrectplacementoftheAMIS newly embedded cells with Arp2/3 inhibitor for 12h, then (ormultipleAMIS-likestructures)sinceinhibitionofRac1during changed the media and allowed the cells to form mature cysts the first cell division does affect mitotic spindle positioning and (SupplementaryFig.4b).ConsistentwiththeinvolvementofRac1 angleduringsubsequentcelldivisions(SupplementaryFig.3c–f). in mediating lumenogenesis, Arp2/3 inhibition led to the Surprisingly, after 4-day incubation we did not observe any formation of cysts containing multiple lumens (Supplementary inverted cysts, suggesting that re-activation of Rac1 (after Fig.4c–e).Weagainobservedcystswithectopicaccumulationsof inhibitor wash-out) was sufficient to correct polarity inversion CGN (Supplementary Fig. 4d,e, arrows). but not the multi-luminal phenotype. In contrast, some cells did To test whether Rac1 is required for apical lumen initiation not form any apical lumen and had no clear CGN-rich TJs only during the first mitotic division, we next treated MDCK (Fig. 7b,e). cells with Rac1 inhibitor 24h post embedding (Supplementary To further confirm that Rac1 pathway regulates apical lumen Fig. 5b). Surprisingly, treatment with Rac1 inhibitor after the formation we treated cells with Arp2/3 inhibitor CK-666. initiation of a lumen during the first cell division still led to a NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications 9 ARTICLE NATURECOMMUNICATIONS|DOI:10.1038/ncomms12426 a b d d CTT ar ar CTT d CGN (1–406) dCGN (1,015–1,203) kDa Stan – + Stan – + Tubulin α β 75– -CGN (1–406) 50– -CGN (1,015–1,203) Anti-GST 75– -CGN (1–406) 50– -Tubulin -CGN (1,015–1,203) Coomassie % similarity c to HsTUBB2C Organism Gene aa1–430 CTT amino acid sequence 2 B U dkDa Standard –– +– +– ++ STuubbutilliisnin e 7kD5–a Standard No TUB2 wt-TUB2 Tailless-T -CGN (1–406) 100– 75– -CGN (1–406) 50– −α-Tubulin 1.0 50– -Tubulin ng 0.75 35– N bindi(a.u.) 0.5 Coomassie G 0.25 C 0 f g 1.5 P<0.01 Human g DMPyootguhsoen N bindin(a.u.) 1.0 G 0.5 Eagle C Frog 0 – + – + – + Tubulin CGN CGN-M1 CGN-M2 Figure8|ElectrostaticinteractionsmediateCGNbindingtomicrotubuleCTTs.(a)Microtubulebindingassayshowingamountofpelleted GST-CGN(1–406)andGST-CGN(1015–1203)whenincubatedwith(þ)andwithout((cid:2))taxol-stabilizedmicrotubules(tubulin).Anti-GSTwesternblot (topgel),andCoomassiestainedSDS–PAGEgel(bottomgel)areshown.(b)Diagramdepictingpositioningofa-andb-tubulinCTTswithinpolymerized microtubule.(c)AminoacidsequencealignmentcomparingCTTsofhumanandyeastb-tubulins.(d)Microtubulebindingassaycomparingamountof CGN(1–406)thatpelletswithwild-typetubulinandsubtilisin-digestedtubulin.CoomassiestainedSDS–PAGEgelisshown.(e)Microtubulebindingassay comparingCGN(1–406)pelletingwithwild-typeyeasttubulinandmutantyeasttubulincontainingnob-CTT.Anti-tubulinandanti-CGNwesternblotis shown.(f)AminoacidsequencealignmentcomparingCGNbasicpatchindifferentvertebratespecies.Boxesmarkthesitesmutatedtoalanineforbinding analysisshownin(g).(g)Quantificationcomparingmicrotubulebindingofwild-typeCGN(1–406),CGN-M1(R36/37A)andCGN-M2(R40/41A). Datashownarethemeansands.d.derivedfromfourindependentbindingexperiments. multi-luminal defect (Supplementary Fig. 5c). However, these stable Rac1 knock-down leads to very dramatic effects on multi-luminal cysts were different from the ones formed by multiple stages of epithelia polarization32. inhibiting Rac1. Typically, these cysts contained a single central primary lumen with one or more small mini-lumens located at the cyst periphery (Supplementary Fig. 5d). This indicates that CGN binding to tubulin mediates targeting to the midbody. evenaftertheformationofaprimarylumenRac1stillplaysarole WhilewehavenowshownthatCGNisrequiredforsingle-lumen in subsequent polarized cell divisions, and inhibiting Rac1 leads formation, the mechanism ensuring the fidelity of CGN totheformationofmultiplesmallsecondarylumens.Thisisalso recruitment to the midbody is still unknown. The midbody is consistent with previously published studies demonstrating that composed of a dense network of microtubules that have been 10 NATURECOMMUNICATIONS|7:12426|DOI:10.1038/ncomms12426|www.nature.com/naturecommunications
Description: