miR-34/449 control apical actin network formation during multiciliogenesis through small GTPase ...

14 Pages·2015·2.35 MB·English

by  Benoît Chevalier

ARTICLE
Received7Jul 2015 |Accepted 17 Aug 2015 |Published 18 Sep2015 DOI:10.1038/ncomms9386 OPEN
miR-34/449 control apical actin network formation
during multiciliogenesis through small GTPase
pathways
Benoˆıt Chevalier1,2,*, Anna Adamiok3,*, Olivier Mercey1,2,*, Diego R. Revinski3, Laure-Emmanuelle Zaragosi1,2,
Andrea Pasini3, Laurent Kodjabachian3, Pascal Barbry1,2 & Brice Marcet1,2
Vertebrate multiciliated cells (MCCs) contribute to fluid propulsion in several biological
processes. We previously showed that microRNAs of the miR-34/449 family trigger MCC
differentiationbyrepressingcellcyclegenesandtheNotchpathway.Here,usinghumanand
Xenopus MCCs, we show that beyond this initial step, miR-34/449 later promote the
assemblyofanapicalactinnetwork,requiredforproperbasalbodiesanchoring.Identification
of miR-34/449 targets related to small GTPase pathways led us to characterize R-Ras as a
key regulator of this process. Protection of RRAS messenger RNA against miR-34/449
binding impairs actin cap formation and multiciliogenesis, despite a still active RhoA. We
proposethatmiR-34/449alsopromoterelocalizationoftheactinbindingproteinFilamin-A,
a knownRRAS interactor, near basal bodies in MCCs. Our study illustrates the intricate role
playedbymiR-34/449incoordinatingseveralstepsofacomplexdifferentiationprogramme
by regulating distinct signalling pathways.
1CNRS,InstitutdePharmacologieMole´culaireetCellulaire(IPMC),UMR-7275,660routedesLucioles,06560Sophia-Antipolis,France.2Universityof
Nice-Sophia-Antipolis(UNS),InstitutdePharmacologieMole´culaireetCellulaire,660routedesLucioles,Valbonne,06560Sophia-Antipolis,France.
3Aix-MarseilleUniversite´,CNRS,UMR7288,InstitutdeBiologieduDe´veloppementdeMarseille(IBDM),13288Marseille,France.*Theseauthors
contributedequallytothiswork.CorrespondenceandrequestsformaterialsshouldbeaddressedtoP.B.(email:[email protected])ortoB.M.
(email:[email protected]).
NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications 1
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
M
ulticiliated cells (MCCs), characterized by the presence Songetal.48also showedthatthecentriolarprotein Cp110isan
ofmultiplemotileciliaattheirapicalsurface,havebeen additionalimportantmiR-34/449targetthatmustberepressedto
described in many vertebrates1,2. Coordinated ciliary allow the maturation of basal bodies. Thus, beyond their initial
beating allows efficient fluid movement and is required for effectoncellcycleexitandentryintodifferentiation,miR-34/449
physiological processes such as elimination of mucus from the probably regulate later steps of the complex multiciliogenesis
respiratory tract, circulation of the cerebrospinal fluid or process. As the development of functional motile cilia appears
migration of the embryo in the fallopian tubes1. The exquisitely sensitive to the reorganization of the actin
physiological importance of MCCs is highlighted by the ever cytoskeleton, we reasoned that miR-34/449 may also regulate
growing number of human disorders associated with defects of one or more molecules associated with actin dynamics or small
the motile cilia1,3–5. Multiciliogenesis, which occurs during GTPase pathways. Here we show that miR-34/449 indeed
normal development and during regeneration of damaged contribute to the establishment of the apical actin cytoskeleton,
tissues, can be studied in experimental setups, such as primary via a mechanism involving the direct repression of the small
cultures of human airway epithelium6 and Xenopus embryonic GTPase R-Ras. This further establishes miR-34/449 as a central
epidermis7. Several characteristic steps are observed as follows: controlsystemofmulticiliogenesisactingatseveraldistinctlevels
(i) exit from the cell cycle of MCC precursors, (ii) massive of this complex physiological process.
postmitotic multiplication of centrioles (centriologenesis),
(iii) reorganization of the apical actin cytoskeleton into a dense
cortical meshwork of actin, (iv) migration of the newly Results and Discussion
synthesized centrioles towards the apical pole of the cell, where TheapicalactinnetworkofhumanandXenopusMCCs.Apical
they anchor to the actin meshwork and mature into ciliary actincytoskeletonformationwasexaminedatseveraltimepoints
organizing centres known as basal bodies, and (v) elongation of during differentiation of primary cultures of human airway
one cilium from each basal body8–15. Several key regulators epithelial cells (HAECs) grown at an air–liquid interface (ALI)
of multiciliogenesis have been identified, such as Notch and in Xenopus embryonic epidermis9. Formation of the apical
and bone morphogenetic protein (BMP) pathways16,17, the meshwork of filamentous actin (F-actin) was monitored directly
transcription factors FOXJ1, MYB and RFXs (regulatory factor bystainingwithfluorescentphalloidinandindirectlybystaining
X)18–24, and the geminin-related nuclear protein Multicilin25. for ezrin or phospho-ERM. In human and Xenopus, the
During multiciliogenesis, the reorganization of the apical actin acetylated tubulin-positive MCCs displayed a strong enrichment
cytoskeleton is controlled by several factors including FOXJ1, of apical F-actin (human: Supplementary Fig. 1a–d; Xenopus:
Multicilin, the ERK7 mitogen-activated protein kinase and small Supplementary Fig. 1f). Basal bodies, which are positive for
GTPases such as RhoA14,19,20,25–28. Following FOXJ1- and g-tubulin labelling, are embedded within an apical F-actin and
RhoA-pathway-dependent phosphorylation, proteins of the ezrinmeshwork(SupplementaryFig.1b).Reorganizationofactin
ezrin-radixin-moesin (ERM) family, which link actin to filaments involves cofilin, a ubiquitous G-actin-binding factor50.
the cell membrane, can interact with cortical actin29,30. The As unphosphorylated cofilin depolymerizes actin filaments,
subcellularlocalizationofezrinanditsinteractingproteinEBP50 cofilin phosphorylation appears essential for cytoskeletal
at the apical membrane of airway MCCs also appears to be reorganization50. Phosphorylation of cofilin-1 also contributes
mediated by a FOXJ1-dependent mechanism14,19,31,32. Focal toactinnetworkstabilizationandformationoffocaladhesions51.
adhesion proteins are also required for the interaction between Focal adhesion proteins indeed participate to ciliary adhesion
basal bodies and apical actin network during multiciliogenesis33. complexes in MCCsand allow interactions betweenbasal bodies
The action of small GTPases on actin cytoskeletal dynamics is and the apical actin network33. In HAECs, the levels of
regulated by a complex network of interactions with additional phosphorylated cofilin-1 and ezrin increased during MCC
GTPases, such as the Ras family member R-Ras34–39, and other differentiation (Supplementary Fig. 1e). We noticed a parallel
regulatoryfactorsincludingguaninenucleotideexchangefactors, increase in the expression of the ERM-binding protein EBP50
GTPase-activating proteins (GAPs), GDP-dissociation inhibitors (Supplementary Fig. 1e), an adapter protein required for the
(GDIs)40,41 and microRNAs (miRNAs)42. Recent work has also maintenance of active ERM proteins at the apical membrane of
highlighted the importance of interactions between the Rho polarizedepithelia14,32.Interestingly,wealsoobservedapunctate
GTPase signalling and the planar cell polarity pathway in labelling of phosphorylated cofilin-1, which was localized
controlling the assembly of apical actin filaments, as well as the sub-apically close to basal bodies in MCCs, but was absent
dockingandplanarpolarizationofthebasalbodiesinMCCs43,44. from non-ciliated cells (Supplementary Fig. 1c,d).
miRNAsormiRsareaclassofsmallsingle-strandedandnon-
coding regulatory RNAs that control many biological processes
by limiting the stability and the translation of their target miR-34/449 control apical actin network assembly in MCCs.
mRNAs45,46.AbnormalmiRNAactivityhasbeenassociatedwith We examined whether the miR-34/449 family can control the
awidevarietyofhumanpathologiesincludingairwaydiseases47. formation of the apical actin network, a prerequisite for basal
We have previously demonstrated that the miR-34/449 family is bodyanchoringandciliumelongation.ToinvalidatemiR-449or
important for the initiation of human and Xenopus MCC miR-34b/c activity in human, we transfected HAECs with a
differentiation. Members of this family share high sequence cholesterol-conjugated antagomiR directed against miR-449
homology and miR-449a/b/c, which are located on the same (Antago-449) or against miR-34 (Antago-34) and assessed MCC
genomic locusas Multicilin, were identified as the most strongly differentiation.Antago-449aswellasantago-34stronglyblocked
induced miRNA species in human and Xenopus during MCC miR-449a/b, whereas antago-34 blocked miR-34a/b/c more
differentiation.WeshowedinthesetwospeciesthatmiR-34/449 efficiently than antago-449 (Supplementary Fig. 2b). Antago-449
promotecellcycleexitandentryintodifferentiationbyrepressing andantago-34blockedtothesameextenttheformationofMCCs
severalcomponentsofthecellcyclecontrolmachineryandofthe andnoadditiveeffectwasobserved(SupplementaryFig.2c).The
Notch signalling pathway9. Their inactivation was sufficient to hugeinductionofmiR-449expressionatearlyciliogenesis(stage
blockcentrioleamplificationandmultiplemotileciliaformation9. EC) suggests that early effects were mainly mediated through
TworecentstudiesconfirmedourfindingsbyshowingthatmiR- miR-449, without excluding a later role for miR-34
34/449-deficient mice exhibited impaired multiciliogenesis48,49. (Supplementary Fig. 2a). This could also explain the lack of
2 NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
compensationofmiR-34inourantago-449conditions,whichdo MCC differentiation. Of note, in Xenopus, the percentage of
notaltertheexpressionofmiR-34.InHAECs,miR-449silencing cells exhibiting defective apical actin meshwork was higher in
caused a decrease of 51%±3.5 in the number of acetylated miR-34/449 morphants (Fig. 2b) than in PO-Dll1 morphants
tubulin-positive cells and of 33%±7 in the number of ezrin- (Fig. 2d), suggesting that miR-34/449 may affect additional
positivecells(Fig.1b).InXenopusepidermis,miR-449aandmiR- targets. We also noticed a reduction of miR-449 levels when
34bwerebothspecificallyexpressedinMCCs,andtheirlevelsin preventing miR-34/449 binding on Notch1 in differentiating
the developing epidermis showed a similar evolution HAECs (Fig. 1f) and on Dll1 in frog epidermis (Fig. 2f).
(Supplementary Fig. 2d, see also ref. 9). Previously, we knocked Conversely,miR-449expressionwasincreasedaftertreatmentof
down miR-449 in Xenopus MCCs by injecting a cocktail of differentiatedHAECswithDAPT(Fig.1f).Altogether,thesedata
morpholino antisense oligonucleotides against miR-449a/b/c reveal the existence of a double-negative feedback loop between
(449-MOs) into the prospective epidermis at the eight-cell miR-449 and the Notch pathway. We hypothesize that once this
stage9. In Xenopus, miR-34b was detected in MCCs loopislockedinastateofhighmiR-449expression,interactions
(Supplementary Fig. 2d–f), and in situ hybridization (ISH) ofmiR-449withadditionaltargetsexpressedatsubsequentsteps
experiments also revealed that 449-MOs not only blocked the ofmulticiliogenesis remainpossible(see Fig.8d).Accordingtoa
expression of miR-449 but also blocked the expression of recent work, CP110 would represent one such target. Its
miR-34b (Supplementary Fig. 2f), suggesting that 449-MOs repression by miR-34/449appearsto affectcentriolematuration,
collectively inhibit miR-34/449 miRNAs (Supplementary butnotapicalactinnetworkassembly48.Wethuslookedforother
Fig. 2e,f). MiR-449 knockdown suppressed multiciliogenesis and possible targets of miR-34/449 that would be directly related to
apical actin web formation in both HAECs (Fig. 1a,b) and actin dynamics.
Xenopus embryonic epidermis (Fig. 2a,b). In embryos injected
with 449-MOs, the number of acetylated tubulin-positive cells miR-34/449targetscomponentsofthesmallGTPasepathways.
went down to 18%±13 of the control and the number of apical Several small GTPase proteins such as RhoA or Rac1 act as
actin cap-positive cells decreased to 9%± 8 of the control key regulators of multiciliogenesis14,27,55,56. We assessed the
(Fig. 2a/b). These results establish that the miR-34/449 family functional impact of miR-449 on the activity of RhoA and
interferes with MCC apical actin meshwork formation in both Rac1,2,3 by transfecting proliferating HAECs with miR-449 and
models.AsmiR-34andmiR-449miRNAssharethesametargets, differentiated HAECs with antago-449. In proliferating primary
we only used miR-449 in the rest of this study. The impact of HAECs, miR-449 overexpression caused a 50% increase in the
miR-449 on the actin cytoskeleton was further investigated by level of active RhoA-GTP, similar to the effect of the Rho
monitoring the formation of focal adhesion and stress fibres, activator calpeptin (Fig. 3a), while it decreased by 35±14%
which are thick and relatively stable actin filaments involved in Rac1,2,3 activity (Supplementary Fig. 3a),as previouslyobserved
cell adhesion and morphogenesis52,53. We found that in in another cellular context with miR-34a (ref. 57). In contrast,
proliferating A549 cells, a human lung cell line devoid of Notch pathway inhibition by DAPT had no impact on RhoA
miR-449andmiR-34b/c,miR-449overexpressionincreasedactin activity in proliferating HAECs (Fig. 3a), indicating that
stress fibres and focal adhesion formation (Fig. 1c,d, see also exogenous miR-449 modulated the RhoA pathway in a Notch-
Supplementary Fig. 3d,e). In addition, western blot analysis independent manner in this assay. MiR-449 silencing caused a
revealed increased ERM phosphorylation following miR-449 modest but significant reduction of RhoA activity in
transfection in proliferating HAECs (Fig. 1e), consistent with differentiating HAECs (Fig. 3a). Conversely, DAPT caused a
the regulatory role of phospho-ERM during actin cytoskeleton significant increase in RhoA activity in differentiating HAECs
dynamics50,54. Thus, our data show that the miR-34/449 family (Fig. 3a), consistent with the concomitant upregulation of
clearly contributes to actin cytoskeleton remodelling in several miR-449 expression (Fig. 1f)9. RhoA activation was also
independent models. Next, we addressed the precise mode of examined in Xenopus MCCs by injecting embryos with an
action of miR-34/449 in the construction of the apical actin RNA encoding the Rhotekin rGBD-GFP, a sensor of activated
network in MCCs. RhoA56. The rGBD-GFP signal was detected in MCCs from
controlembryosbutnotinembryosinjectedwithNotch-ICD,in
Mutual repression between miR-449 and the Notch pathway. which MCC differentiation was abolished (Fig. 3b). In control
As the miR-34/449 family represses the Notch pathway during embryos and in miR-34/449 morphants, almost all rGBD-GFP-
MCC differentiation9, we assessed the contribution of the Notch positive cells expressed the early MCC differentiation marker
signal to the actin web reorganization. We treated proliferating a-tubulin (Fig. 3c). By contrast, in miR-34/449 morphants only
HAECs or human lung A549 cells (which both are devoid 13% of rGBD-GFP-positive cells displayed acetylated tubulin
ofendogenousmiR-449)eitherwithmiR-449orwithN-[N-(3,5- ciliary staining, compared with 96% in the control situation
Difluorophenacetyl)-L-alanyl]-S-phenylglycinet-butylester (DAPT), (Fig. 3b,c). Thus, following the experimental inhibition of
ag-secretaseinhibitorthatblocksNotchactivation.Asexpected, miR-34/449 in Xenopus, RhoA activation can still be detected
both ectopic expression of miR-449 in proliferating HAECs and in MCCs unable to grow cilia. However, we cannot rule out
DAPT repressed the expression of the Notch target gene HES1 discretechangesinthesub-cellularlocalizationofactivatedRhoA
(Fig.1g).InproliferatingA549cells,NotchinhibitionwithDAPT in miR-34/449-deficient embryos.
alone had neither impact on the formation of actin stress fibres Collectively,theseresultssuggestthat,althoughthemodulation
and focal adhesions (Fig. 1c,d) nor on ERM phosphorylation in ofRhoAactivitybymiR-34/449mayplayaroleinthecontrolof
proliferatingHAECs(Fig.1e).ThissuggeststhatmiR-449causea apical actin polymerization, other miR-34/449 targets contribute
rise of levels of phosphorylated ERM independently of Notch to the profound disruption of the actin cap observed after
repression. miR-34/449 inactivation.
We observed that preventing the binding of miR-34/449 on In a bid to identify such additional factors, we applied several
Notch1(PO-Notch1inHAECs,Fig.1a,b)orontheNotchligand miRNA targetpredictiontools58 toidentify putative miR-34/449
Dll1 (PO-Dll1 in Xenopus, Fig. 2c–f) with protector oligo- targetsamongthesmallGTPasepathways.Specifically,welooked
nucleotides coordinately blocked multiciliogenesis and apical for relevant miR-34/449 mRNA targets that were repressed
actin network formation. This is consistent with the need for an during MCC differentiation and after overexpression of
early repression of the Notch pathway by miR-34/449, to allow miR-34/449 in proliferating HAECs (Gene Expression Omnibus
NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications 3
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
a d
Human airway differentiated epithelium (LC) A549 cells
er 5
CTR-Neg Antago-449 PO-Notch1 mbR) 4 ***
n 1 5 9 of F.A. nuorm. / CT 23
F-acti Ratio (n 01
20 μ2m 20 μ6m 20 1μ0m miR-NmeigR-4 4m9ia+R -DNAePgT
e
Prolif. HAECs
n
zri 1.0 2.8 0.7 kDa
E
P-ERM 80
20 μm 20 μm 20 μm HSP60 60
ub. 3 7 11 miR-N emigmiR-R4-4N9eag + D A P T
c. t
A f Diff. HAECs (LC)
20 μm 20 μm 20 μm **
4 8 12 n 200
o
ge pressiCTR) 150
Mer 9 exm. / 100
20 μm 20 μm 20 μm miR-44(nor 500 ***
b Human airway differentiated epithelium (LC) PCOT-RN otch 1 D A P T
Ezrin g
R) MCCs Prolif. HAECs
% of MCC or ezrin(+)cell number (norm./ CT 1075205050 *** ** *** ** Hes1 expression(norm./CTR) 1025705050 *** **
CTR-Neg Antago-449 PO-Notch1 CTR/miR-Neg + – +
miR-449a – + –
DAPT – – +
c
A549 cells
miR-Neg miR-449a miR-Neg+DAPT
PI 1 2 3
A
D
n
cti
A
n
xilli
a
P
10 μm 10 μm 10 μm
Figure1|MiR-449affectsactinnetworkremodellingandmulticiliogenesisinHAECs.(a–c)EffectofatreatmentbycontrolantagomiR(CTR-Neg),
anti-miR-449a/b(Antago-449)andmiR-449::Notch1protector(PO-Notch1)ondifferentiatingHAECs.(a)StainingforF-actin(a1,5,9),ezrin(a2,6,10)and
acetylatedtubulin(a3,7,11),atLCstage.(b)ThehistogramindicatestheaveragepercentageofMCCs(inmagenta)andapicalezrin-positive(ingreen)cell
numberrelativetocontrol(means±s.d.fromnineandthreedonorsforMCCandezrinquantifications,respectively.***Po0.001,**Po0.01;Student’s
t-test).(c)ImmunostainingoffocaladhesionsproteinPaxillin(ingreen),F-actin(inred)andnuclei(inblue)inA549epithelialcellstransfectedfor72h
withcontrolmiRNA(miR-Neg),miR-449aormiR-Negplus10mMDAPT.(d)Ratiooffocaladhesionnumberpercell,normalizedtocontrol(n¼5fieldsin
threeindependentexperiments;***Po0.001;Student’st-test).(e)EffectofmiR-449overexpressionandDAPT(10mM)onERMphosphorylationin
proliferatingHAECs.Phosphorylatedproteinlevelswerenormalizedwithnon-phosphorylatedERMandwithanantibodyagainstHSP60asaloading
control.Normalizedfoldchangesareindicatedonthecorrespondingbands.Experimentswererepresentativeofthreedonors.(f)EffectofPO-Notch1and
DAPT(10mM)indifferentiatingHAECsatLCstageonmiR-449expression,normalizedwithRNU44.(g)Real-timeRT–PCRofHES1transcriptsincontrol,
DAPT(10mM,48h)-treatedormiR-449-overexpressingproliferatingHAECs.TranscriptlevelsofHES1werenormalizedagainstUBCtranscriptasan
internalcontrol.(f,g)Datarepresentthemeanands.d.ofthreeindependentexperiments(***Po0.001,**Po0.01;Student’st-test).
4 NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
Xenopus epidermis embryo St.25
a c
CTR-Neg 449-MOs CTR-Neg PO-Dll1
1 5 1 5
n n
cti cti
a a
F- F-
10 μm 10 μm 20 μm 20 μm
2 6 2 6
X X
A A
A A
C C
P- P-
F F
G G
10 μm 10 μm 20 μm 20 μm
3 7 3 7
b. b.
u u
c. t c. t
A A
10 μm 10 μm 20 μm 20 μm
4 8 4 8
e e
g g
er er
M M
10 μm 10 μm 20 µm 20 µm
b d
Normal
% Injected cells withmotile cilia (MCCs)or with actin cap 12340000 MAcCtiCn* s*cap ** % Injected cells withactin cap 1200 * Defecti*v*e*
0 0
C T R-N e4g4 9-M O s C T R-N e4g4 9-M O s C T R-N e gP O-Dll1 C T R-N e gP O-Dll1
e f
800 *** 100
NAonTR) 600 9onTR)
Dll1 mRexpressinorm. / C 240000 miR-44expressinorm. / C 50
( (
0 0
C T R-N e g P O-Dll1 C T R-N e g P O-Dll1
Figure2|MiR-449affectsactinnetworkremodellingandmulticiliogenesisinXenopusepidermis.(a–d)Theepidermisprecursorblastomeres
(eight-cellstageXenopusembryos)wereinjectedwithnegativecontrolmorpholinos(CTR-Neg)(a,c),withmorpholinosagainstmiR-449(449-MOs)
(a,b)orwithprotectormorpholinosofDll1(PO-Dll1)(c,d).Stainingatstage25forF-Actin(inred,a1,5andc1,5)andmotilecilia(Ac.Tub.inmagenta,
a3,7andc3,7).InjectedcellsweredetectedbytheexpressionofasyntheticmRNAcodingformembrane-boundGFP(GFP-CAAXingreen,a2,6andc2,6).
(b)ThehistogramindicatesthepercentageofGFP-CAAX-positiveinjectedcellsthatdevelopmotileciliaorapicalactincapincontrols(Stage24þ25:
n¼5fields/583injectedcells)andinmiR-449morphants(Stage24þ25:n¼8fields/625injectedcells;P-valuest.24þ25¼0.0087;Mann–Whitneytest
withtwo-tailedP-value).(d)PercentageofinjectedcellspositiveforGFPfluorescencewithnormalordefectiveactincapincontrol(n¼30fieldsper1,345
injectedcells)versusPO-Dll1morphants(n¼32fieldsper1,268injectedcells;P-value(normalversusdefectiveincontrol)st.24þ25o0.0001,Pvalue
(normalversusdefectiveinPO-Dll1-injectedst.24þ25¼0.0033;Mann–Whitneytestwithtwo-tailedP-value).(e)EffectofprotectingtheDll1mRNAfrom
theinteractionwithmiR-449(PO-Dll1)inXenopusepidermisonDll1expression(normalizedwithornithinedecarboxylase(ODC)).(f)EffectofPO-Dll1in
XenopusepidermisonmiR-449expression,normalizedwithU6.Dataaremeans±s.d.oftwoindependentexperiments.
NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications 5
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
a c
Prolif. HAECs Diff. HAECs Xenopus
ns epidermis St.25
** ** ** ** α-tub.
% RhoA activity (norm. /CTR)m211iR0505-00000Neg + + – + – * GBD-GFP(+) cellber (norm. / CTR) 11505000 Ac. tub.ns
Rhom aiRct-i4va4t9oar –– +– +– –– +– % rnum **
0
AntagoD-ANPeTg – – – + + +– –– ++ C T R-N e g 4 4 9-M O s
Antago-449 – + –
b Xenopus embryonic epidermis St.25
CTR-Neg NICD CTR-Neg 449-MOs
1 3 5 7
P P
F F
G G
D- D-
B B
G G
r r
30 μm 30 μm 5 μm 5 μm
2 4 6 8
P
F
mR ub.
ub. c.T
T A
c.
A
30 μm 30 μm 5 μm 5 μm
Figure3|miR-449controlssmallGTPasepathwaysduringMCCdifferentiation.(a)Left(Prolif.HAECs):RhoAactivityinproliferatingHAECs
transfectedfor72hwithmiR-Neg,miR-449aand/orincubatedwithDAPT(10mM)oraRhoactivator(calpeptin,1Uml(cid:2)1,2h).Right(Diff.HAECs):
RhoAactivityindifferentiatingHAECsatPostagetreatedfor72hwithantago-Neg,antago-449orDAPT(10mM).RhoAactivityisexpressedasa
percentagerelativetocontrol.Dataaremean±s.d.fromatleastthreeindependentexperiments(*Po0.05,**Po0.01and***Po0.001;Student’st-test).
(b)RhoAactivityinXenopusepidermisatstage25,assessedbymeasuringthefluorescenceofrhotekinrGBD-GFP,anactiveRhoAsensor(green,upper
panels).Injectedcellsareidentifiedbythemembrane-boundRFP(mRFPinred,lowerpanels).Cellswereeitheruntreated(CTR-Neg,b1,2,5,6),injected
withmiR-449morpholinos(449-MOs,b7,8)orwithNotchintracellulardomainNICD(b3–4).Motilecilia’sstainingisinmagenta(lowerpanels).
(c)ThehistogramindicatesthepercentageofrGBD-GFP-positivecellsstainedfora-tubulinmRNA(a-Tub.inblack)oracetylatedtubulin(Ac.Tub.in
magenta)inmiR-449morphants(449-MOs,n¼68)relativetonegativecontrol(n¼110).Dataaremean±s.e.m.(**Po0.005,ns,notsignificant,
one-wayanalysisofvariancewithDunnett’stest).
(GEO) data set GSE22147). This survey led us to identify ARHGAP1, DAAM1 and NDRG1 decreased at late ciliogenesis
several candidates related to small GTPase signalling and actin (LC).TheexpressionofARHGDIBtranscriptslightlydecreasedat
cytoskeleton remodelling: (1) ARHGAP1, a member of the theonsetofdifferentiation(thatis,polarizationstep,Po)butrose
Rho-GAP family59; (2) ARHGDIB, also called Rho-GDI2 again during the phase of multiciliogenesis (Fig. 4a). The RRAS
(ref. 40); (3) DAAM1, the diaphanous-related formin transcript level decreased throughout the whole time course of
Dishevelled-associated activator of morphogenesis60; (4) NDRG1, MCC differentiation (Fig. 4a). In proliferating HAECs, miR-449
N-myc downstreamregulated gene 1, an iron-regulated metastasis overexpression strongly reduced the transcript levels of
suppressor61; (5) R-Ras, a member of the superfamily of small ARHGDIB, ARHGAP1 and RRAS, whereas the expression of
GTPases, related toRas38. DAAM1andNDRG1transcriptswasslightlydecreased(Fig.4b).
By looking more precisely at the three major families of These putative targets were further investigated using a dual
regulators of small GTPases expressed during HAECs differentia- luciferase reporter assay in HEK293 cells. MiR-449a and
tion or after miR-34/449 transfection, we detected 23 distinct miR-449b reduced the relative luciferase activity of chimeric
ARHGAPs, including ARHGAP1; we also detected 22 distinct constructs containing the wild-type 30-untranslated regions
ARHGEF transcripts but none of them were predicted as direct (30-UTRs) of ARHGAP1, ARHGDIB, NDRG1 and RRAS, but
miR-34/449 targets; regarding the three known mammalian not of DAAM1 30-UTR (Fig. 4c). Next, we focused on the three
ARHGDI transcripts40, only ARHGDIB was further analysed, as most significantly regulated miR-34/449 targets ARHGAP1,
ARHGDIA expression levels did not change during HAECs ARHGDIB and RRAS. MiR-449-mediated silencing of either
differentiation or after miR-34/449 transfection and ARHGDIG ARHGAP1,ARHGDIBorRRASwasrespectivelyabolishedwhen
wasnotdetected(SupplementaryFig.3b,seealsoGEOGSE22147). miR-34/449-predicted binding sites were mutated (Fig. 4c).
The transcripts ARHGAP1, ARHGDIB, DAAM1, NDRG1 and Among the four miR-449-binding sites in the 30-UTR of RRAS,
RRASareallmodulatedduringHAECdifferentiation(Fig.4a,see the strongest effect was observed for the most 30-site, which also
alsoGEOGSE22147).Figure4ashowsthatthetranscriptlevelof corresponds to the unique conserved site between human and
6 NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
a b
RRAS ARHGAP1 ARHGDIB DAAM1 NDRG1
pression ofm. / Pr) (/UBC) 112...050 **** ****** expressionR) (/UBC)1680000 CTR-Neg * miR-449a
Relative extranscripts (nor 00..05 Pr*P*o**E*C**L*C*PrPoESCt*aL*gC*e Porf *dP*iof*fEeCreLnCtiatPiornPoECLC PrPoECLC % transcript (norm. /CT 24A000R H G DAIR*B*H*G A P*D*1A*A M 1N D R G*1R R A*S**
c CTR-Neg miR-449a miR-449b d
1.0 0.1 kDa
100
R-Ras 24
y
ctivitR) 80 * ARHGAP1 1.0 0.2 50
aT *
e C 60 1.0 0.2
% luciferas(norm. / 2400 ****** ****** ****** ARHHSGPDmi6IRB0-N emgiR-4 4 9 a 6203
0
wt Mutant wt Mutant DAAM1 NDRG1 wt Mut.1 Mut.2 Mut.3 Mut.4 Full mut.
ARHGDIB ARHGAP1 RRAS
Figure4|ARHGAP1,ARHGDIBandRRASaretargetedbymiR-449inHAECs.(a,b)TranscriptexpressionlevelsofARHGAP1,ARHGDIB,DAAM1,NDRG1
orRRASwereanalysedusingreal-timeRT–PCRduringHAECdifferentiation(a)orfollowingmiR-449aoverexpression(48h)inproliferatingHAECs(b),
andnormalizedwithUBCtranscriptasaninternalcontrol.(c)SpecificinteractionbetweenmiR-449a/bandthe30-UTRsofARHGAP1,ARHGDIB,DAAM1,
NDRG1andRRASmRNAswasassessedusingluciferasereporterassayonconstructscarryingeitherthewild-type(wt)ormutants(mut.)30-UTR-binding
sitesformiR-449.ValueswerenormalizedwiththeinternalRenillaluciferasecontrol.(d)DetectionofARHGAP1,ARHGDIBandR-Rasproteinsafter
miR-449overexpressioninproliferatingHAECsfor72h.HSP60isusedasaninternalcontrol.Quantificationofproteinlevelsareindicatedaboveeach
correspondingband.Alldataaremeans±s.d.ofatleastthreeindependentexperiments(***Po0.001,**Po0.01and*Po0.05;Student’st-test).
Xenopus (Fig. 4c and Supplementary Fig. 4a). The protein levels arhgap1 expression was very faint at similar stages and did not
of ARHGAP1, ARHGDIB and R-Ras proteins were also show any specific pattern of distribution (Fig. 5e). No arhgdib
dramatically decreased after transfection of proliferating HAECs signal was detected in the Xenopus epidermis at any
with miR-449 (Fig. 4d). These results establish ARHGAP1, developmental stage (Supplementary Fig. 3i).
ARHGDIB and RRAS transcripts as bona fide targets of The silencing of ARHGDIB, ARHGAP1 and RRAS using
miR-34/449. Consistent with this conclusion, both ARHGDIB specificsmallinterferingRNAs(siRNAs)inproliferatingHAECs
and R-Ras proteins were excluded from acetylated tubulin- strongly reduced the level of expression of the corresponding
positive MCCs, while being enriched in non-ciliated CD151- proteins (Supplementary Fig. 3c). In proliferating human A549
positive basal HAECs9,62 (Fig. 5a,b). This is consistent with a cells, the silencing of RRAS or ARHGAP1 but not ARHGDIB
recent gene expression profiling study performed in mouse increased actin stress fibres and focal adhesion formation
trachea, which reported a higher level of expression of RRAS (Supplementary Fig. 3d,e), and mimicked the effects observed
transcripts in non-ciliated cells than in ciliated cells63 after miR-449 overexpression (Fig. 1c,d and Supplementary
(GSE42500). We could not address this possibility for Fig. 3d,e). These effects were abolished when cells were treated
ARHGAP1 as none of the antibodies that we tested worked in with an inhibitor of Rock (Y27632), an important RhoA
immunofluorescence. In differentiating HAECs, both R-Ras and effector55 (Supplementary Fig. 3e). In proliferating HAECs, we
ARHGAP1 protein level strongly and continuously decreased also found that RhoA activity increased after silencing of RRAS
from Po to LC, concomitantly with the increase in miR-449 but not of ARHGAP1 or ARHGDIB. This effect remained
expression (Fig. 5c,d). Conversely, ARHGDIB protein level however smaller than the one induced by miR-449 expression
increased during human HAEC differentiation (Fig. 5d). As (SupplementaryFig.3f).Rac1,2,3activitywasslightlyreducedby
ARHGDIB protein was mainly excluded from human MCCs miR-449expression,whereasitwasnotaffectedbyasilencingof
(Fig. 5a,b), the global protein increase observed during HAEC RRAS (Supplementary Fig. 3a). These observations point to the
differentiation in western blotting is probably explained by participation of miR-34/449 and R-Ras to actin network
its stronger expression in non-ciliated HAECs. In Xenopus reorganization and their capacity to alter the RhoA activity.
epidermis, rras expression measured by quantitative reverse They arein agreement witha previous work showingthatRRAS
transcriptase–PCR (qRT–PCR) became detectable at neurula deficiencyisassociatedwithcorticalactinreorganizationinadult
stage 16, then dramatically increased until stage 20, before haematopoietic progenitor cells38. Our data suggest that the
multiciliogenesis and subsequently dropped (Fig. 5f,g). induction of RhoA activity by miR-449 may at least in part
Interestingly, rras transcript levels were anti-correlated with involve the silencing of RRAS, notwithstanding possible
miR-449a levels during the course of MCC differentiation contributions by additional regulators. The lack of effect on
(Fig. 5g). When analysed by ISH, rras transcripts were RhoA activity after ARHGAP1 and ARHGDIB silencing is
primarily detected in inner-layer cells that were negative for the probably in line with the existence of redundant or
MCCmarkera-tubulin,atstages16and19(Fig.5e).Incontrast, compensatory mechanisms controlling RhoA activity in the
NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications 7
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
a b
Isolated differentiated HAECs Cultured HAECs (LC)
1 2 b.
u
T
b. Ac.
u
Ac.t Tub. DIB
c. G
1 A H
5 R
D1 A
C n
20 μm 5 μm cti 20 μm
A
3 4 c Differentiating HAECs
s s 25
a a
R-R151 Tub.R-R miR-449a expression 1120505
CD Ac. 0
20 μm 20 μm Pr Po EC LC
d kDa
5 6
B B ARHGAP1 50
DI DI
G G
H H ARHGDIB 23
R R
A A
R-Ras 24
1 b.
15 Tu HSP60 60
CD 20 μm Ac. 20 μm Pr Po EC LC
Stage of HAEC differentiation
e Xenopus epidermis f Xenopus epidermis
2.0
St.16 1 St.17 3 of
α-tubulin α-tubulin xpression anscripts 11..05
rras 20 μm arhgap1 20 μm Relative e rras tr 00..05 1416181920232425
Stage
α-tubulin St.19 25 α-tubulin St. 20 4 expression gm. / St. 13) 11268 mrraiRs-*4*49*a ** **
rras 20 μm arhgap120 μm rras (nor 04 St.1 3 St.1 9* St.2 3 S*t*.*2 5
Figure5|CellularspecificityofexpressionofARHGAP1,ARHGDIBandR-RasduringMCCdifferentiation.(a,b)Fluorescenceimmunocytochemistry
experimentsoncytospinsfromdissociatedHAECs(a1–6)andprimaryHAECculturesatLCstep(b)illustratethecell-specificexpressionofhuman
ARHGDIB(ingreen,a5–6,b)andR-Rasproteins(ingreen,a3,4).BasalandciliatedcellsareCD151þ (inred,a1,3,5)oracetylatedtubulinþ (inmagenta
a2,4,6andb),respectively.Panela2isamagnificationofanisolatedacetylatedtubulin-positiveMCC.HAECsarestainedfornucleiwith4,6-diamidino-2-
phenylindole(DAPI;inblue,a1–6),F-Actinwithphalloidin(inred,b).ExpressionlevelsofmiR-449a(c)andofARHGAP1,ARHGDIBandR-Rasproteins
(d)duringHAECdifferentiation.MiR-449levelsarenormalizedwithRNU44(c)andHSP60wasusedasaloadingcontrol(d).(e)FISHanalysisofrras
(e1,2)andarhgap1(e3,4)mRNAonsectionsofXenopusembryonicepidermisatstages16and19(e1,2)oratstages17and20(e3,4).MCCprecursorsare
positivefora-tubulin(magenta;e1–4).DAPIstainingisinblueandwhitedottedlinesindicatethesurfaceoftheouterlayer.(f)Real-timeRT–PCRofrras
transcriptsinXenopusepidermisperformedateachcorrespondingstagesindictedinthefigure.(g)Transcriptlevelsofrrasdecreaseconcomitantlywiththe
inductionofmiR-449expression.TranscriptlevelsofrrasandmiR-449werenormalizedandcomparedwiththeirrespectivevaluesobtainedinstage13.
TranscriptlevelsofrraswerenormalizedagainstOdctranscriptasaninternalcontrolandmiR-449expressionwasnormalizedwithU2asaninternal
control.Datarepresentthemeanands.d.ofthreeindependentexperiments(***Po0.001,**Po0.01and*Po0.05;Student’st-test).
contextofmulticiliogenesis.Asthepatternofexpressionofboth We designed target protection assays in which cholesterol-
arhgap1 and arhgdib in Xenopus epidermis was not consistent conjugated modified oligonucleotides were transfected in
with an association to MCC precursors, arhgap1 and arhgdib differentiating HAECs to compete with the binding of
were not further analysed in Xenopus. miR-34/449 on the site identified within the human 30-UTR of
8 NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
a c miR-34/449controlapicalactinassemblybyrepressingR-Ras.
Diff. HAECs (LC) Diff. HAECs (LC)
We used target protection assays (cholesterol-conjugated mod-
CTR-Neg PO-RRAS si-RRAS
ified oligonucleotides in HAECs or morpholino oligonucleotides
1 4 1
infrogepidermis)tocompetewiththebindingofmiR-34/449on
sitesidentifiedwithinthehumanandXenopus30-UTRsofRRAS
n n mRNA. In human cells, the RRAS protector oligonucleotide
cti cti
A A (PO-RRAS) strategy effectively blocked the action of ectopic
miR-449, as evidenced on RRAS 30-UTR in luciferase assays
20 μm 20 μm 20 μm (SupplementaryFig.4b).Inaddition,miR-34/449knockdownor
protection of RRAS mRNA from miR-34/449 in frog epidermis
2 5 2
ledtoanincreaseinrrastranscriptlevels(SupplementaryFig.4c).
We checked that in both species the expression of RRAS2, an
Ezrin c. tub. RRRRAASS2-reelaxtperdessgioenne,redmidainneodtaitntaervfeerrey lwoiwthleRve-lRadsursiniggnaMlliCngC.
A
differentiation and in response to miR-449 overexpression, and
20 μm 20 μm 20 μm was not altered by miR-34/449 knockdown or PO-RRAS protec-
tion (HAECs, see GEO GSE22147; Xenopus, Supplementary
3 6 3
Fig. 4c). In both models, the PO-RRAS strategy was also able
to increase endogenous R-Ras protein level (Supplementary
ub. ge Fig.4d,e).Collectively,theseassaysunambiguouslyestablishthat
c. t Mer RRAS transcripts were specifically targeted by miR-34/449 in
A
human and Xenopus MCCs.
20 μm 20 μm 20 μm Inhuman(Fig.6a,b)andinXenopus(Fig. 7a,d),protectionof
the RRAS transcript from miR-34/449 binding led to a strong
b reduction in apical actin meshwork and motile cilia formation.
% of MCCs or zrin(+) cell number(norm. / CTR)11055000 ***EMzCriCn*s** *** StbsianiulilollogoenwcgnwkehcaesiliadtnscpohgtihcmsRiatatrRsultolAalntReicSgcv-tilRieiylnnilaoasogchffeafupenpemcleaxtsfeyapiodsnsrremaasa(pstaFiikotaicigennao.ylneha6raacaobrstnll,iyecndt)os.immtnebTepuesmhlhtoteiaiwfcsgtieouhHlirrtoAklarygteEeifsofionCunrnelmtsedsio-siatf.ftufsieInotMrnrneeodnCnfa,trCgniotalsdoyg-,
e 0
CTR-Neg PO-RRAS si-RRAS epidermis, both apical and sub-apical phalloidin staining were
altered in PO-rras-injected embryos (Fig. 7a,b). In contrast,
Figure6|ThedirectrepressionofRRASbymiR-449affects
F-actinstaining remaineddetectableatcellularjunctionsinboth
multiciliogenesisandapicalactincytoskeletonreorganizationinHAECs.
models. Importantly, actin cap formation and multiciliogenesis
(a)DifferentiatingHAECswerechronicallytreatedwithnegative
wererescuedinXenopusepidermiswhenamorpholinodesigned
oligonucleotide(CTR-Neg,a1–3)orwithanoligonucleotideprotectingthe
to block R-Ras translation (MO-ATG-rras) was co-injected with
miR-449-bindingsiteonRRAS(PO-RRAS,a4–6)andstainedatLCstagefor
PO-rras (Fig. 7a,d). At the dose used for this assay, injection of
F-actin(a1,4),ezrin(a2,5)andacetylatedtubulin(Ac.Tub.,a3,6).
MO-ATG-rras alone had no significant effect on either apical
(b)Alternatively,differentiatingHAECsweretransfectedatseedingtime
actin meshwork formation or multiciliogenesis (Fig. 7a–d),
withsi-RRASandstainedatLCstageforF-actin(inred,c1)andmotilecilia
suggesting that rras expression may be already repressed by the
(Ac.Tub.inmagenta,c2).(c)Thehistogramindicatesthenumberofapical
presence of endogenous miR-34/449 in those MO-ATG-rras
ezrin-positivecells(ingreen)andMCCs(inmagenta)normalizedin
maturing MCCs.
percentagefromcontrol.ProtectingtheRRASmRNAfrominteractionwith
Our data unambiguously indicate that the repression of RRAS
miR-449leadstodefectsinapicalactinreorganizationtogetherwitha
at a late step of MCC differentiation by miR-34/449 is required
decreaseinMCCdifferentiationsimilartothatobservedafterinhibitionof
forapicalactinnetworkassemblyandmulticiliogenesisinhuman
miR-449activity.Alldataaremeans±s.d.fromatleastthreeindependent
(Fig.6)aswellasinfrog(Fig.7).ConsideringthatR-Rasactivity
experiments(***Po0.001;Student’st-test). can be increased by Notch pathway activation, as previously
observed in another cellular model39, and considering the early
ARHGDIB mRNA (PO-ARHGDIB). No effect was observed repression of Notch signalling by miR-34/449 during vertebrate
either on apical actin web or multiciliogenesis after protection multiciliogenesis9, miR-34/449 is therefore able to control R-Ras
withPO-ARHGDIBorafterARHGDIBsilencingwithsiRNAsin function at two distinct levels: (1) directly, via the inhibition of
differentiating HAECs (Supplementary Fig. 3g,h). Incidentally, RRAS expression, and (2) indirectly, via the inhibition of R-Ras
the existence of five miR-34/449-binding sites in the 30-UTR activity through the repression of the Notch pathway.
of ARHGAP1 made elusive the assessment of a target protection AsobservedformiR-449morphantswithrhotekinstainingin
of ARHGAP1 against a miR-34/449 action in primary Xenopus epidermis (Fig. 3b), PO-rras-injected MCCs displayed
HAEC cultures. Intriguingly, silencing ARHGAP1 in human at reducedactincapandmotileciliastaining,butmaintainedRhoA
an early step of HAEC differentiation strongly affected apical activationattheapicalsurface(Fig.7c,e).ThissuggeststhatRRAS
actin meshwork formation and blocked multiciliogenesis mayinteractwithdownstreameffectorsofRhoAduringactincap
(Supplementary Fig. 3g,h), suggesting that ARHGAP1 does play formation.Inparticular,previousworksdescribedaninteraction
a rolein multiciliogenesis, in whichits level of expression has to between R-Ras and Filamin A (FLNA), a non-muscle F-actin
be finely controlled during maturation of MCCs. cross-linking protein involved in epithelial cell shape, actin
These data posit the absence of R-Ras in miR-34/449- cytoskeleton remodelling and primary cilia formation36,37,64–66.
expressing MCCs as a conserved feature across tetrapods. We FLNA can interact with the RhoA and Shroom3 signalling
thereforefocusedonthefunctionalimpactofmiR-449-mediated pathways67–69, and has also been involved in ciliogenesis and
repression of RRAS on apical actin reorganization in MCCs. basal body positioning through its interaction with Meckelin64.
NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications 9
&2015MacmillanPublishersLimited.Allrightsreserved.
ARTICLE
NATURECOMMUNICATIONS|DOI:10.1038/ncomms9386
a b
Xenopus epidermis st. 25 Xenopus epidermis St. 25
PO-rras
CTR-Neg PO-rras MO-ATG-rras +MO-ATG-rras CTR-Neg PO-rras
1 5 9 13 1 3
ctin calctin
A piA
F- AF-
20 μm 20 μm 20 μm 20 μm 10 μm 10 μm
2 6 10 14 2 4
GFP-CAAX SubapicalF-Actin
20 μm 20 μm 20 μm 20 μm 10 μm 10 μm
c
3 7 11 15 1 3
P
Ac.Tub. ApicalGBD-GF
20 μm 20 μm 20 μm 20 μm r 5 μm 5 μm
4 8 12 16 2 4
Merge ApicalAc.Tub.
20 μm 20 μm 20 μm 20 μm 5 μm 5 μm
d e
***
20 ** Actin α-Tub.
% of injected cellswith motile cilia (MCCs)or apical actin cap 11505 *** ** **** MCCs % rGBD-GFP(+) cellnumber (norm. / CTR)110550000 Ac. Tubn.s **
0 CTR-Neg PO-rras
CTR-Neg + – – – + – – –
PO-rras – + – + – + – +
MO-ATG-rras – – + + – – + +
Figure7|ThedirectrepressionofRRASbymiR-449affectsmulticiliogenesisandapicalactincapformationinXenopus.(a,b)Eight-cellstageXenopus
embryoswereinjectedintheepidermisprecursorblastomereswithamixtureofsyntheticmRNAcodingforGFP-CAAXtolabeltheinjectedcells(ingreen,
a2,6,10,14)andnegativecontrolMO(CTR-Neg,a1–4,b1,2),oramorpholinoprotectingrrasagainstbindingbymiR-449(PO-rras,a5–8,b3,4)ora
morpholinoblockingthetranslationofrras(MO-ATG-rras,a9–12),oracombinationofPO-rrasandMO-ATG-rras(a13–16).MCCsarestainedwithan
anti-acetylatedtubulinantibody(inmagenta,a3,7,11,15andc2,4),F-actinwithphalloidin(red,a1,5,9,13andb1–4).(b)TheimpactofPo-rrasinMCCson
apical(upperb5)andsub-apicalactin(lowerb6)wasobservedwithF-actinstaining.(c)Inindependentexperiments,weusedrGBD-GFPtoexamineRhoA
activityinXenopusMCCs(c1–4).Eight-cellstageXenopusembryoswereinjectedintheepidermisprecursorblastomereswithareporterofRhoAactivity
(rGBD-GFPingreen,c1–4)andwithnegativecontrolMO(CTR-Neg,c1–2)orwithPO-rras(c3–4).MCCsarestainedinmagenta(c2,4).Protectingtherras
mRNAfrominteractionwithmiR-449resultsindefectsinactincapformation(F-Actin,a5andb3,4)togetherwithalossofMCCs(Ac.Tub.,a7,c4)
withoutaffectingapicalRhoAactivity(rGBD-GFP,c1,3).ThisphenotypeisrescuedwhenthetranslationoftheprotectedrrasmRNAisblockedby
coinjectionofMO-ATG-rras(a13–16).(d)Thehistogramindicatesthepercentageofinjectedcells(positiveformGFP)thatdevelopproperapicalactincap
(inred)andmotilecilia(inmagenta)inXenopusepidermisatstage25.CTR-Neg,n¼10embryosper413injectedcells;PO-rras,n¼8embryosper350
injectedcells;MO-ATG-rras,n¼8embryosper290injectedcells;MO-PO-rrasþMO-ATG-rrasn¼9embryosper395injectedcells(***P¼0.009and
Po0.0001,and**P¼0.0016;Mann–Whitneytest).Dataaremean±s.e.m.(e)ThehistogramindicatesthepercentageofrGBD-GFP-positivecellsstained
fora-tubulin(a-tub.inblack)oracetylatedtubulin(Ac.Tub.inmagenta)inPO-rrasmorphants(PO-rras,n¼150)incomparisonwiththenegativecontrol
(CTR-Neg,n¼110).(**Po0.005;ns,nosignificant;one-way-analysisofvariancewithDunnett’stest).Dataaremean±s.e.m.
WedetectedincontrolfullydifferentiatedHAECs(Fig.8a)FLNA miR-449, R-Ras and FLNA were enriched in the membrane
in MCCs near apically docked basal bodies. By contrast, it was fraction (Fig. 8b). Following miR-449 overexpression, R-Ras
more homogeneously distributed in non-ciliated cells. The same disappeared from the membrane fraction, while FLNA was
apical enrichment of FLNA was observed in Xenopus MCCs redistributed from the membrane to the cytoskeletal fraction
(Fig. 8c). We finally noticed a miR-449-dependent subcellular (Fig. 8b). In a tentative model (Fig. 8d), we propose that the
redistribution of FLNA, after quantifying R-Ras and FLNA silencingofR-RasbymiR-34/449inMCCsaffectstheinteraction
protein levels in different cellular fractions of proliferating betweenR-RasandFLNA,andfavoursaredistributionofFLNA
HAECs that overexpress or not miR-449. In the absence of inacytoskeletalcomponentsinvolvedintotheanchoringofbasal
10 NATURECOMMUNICATIONS|6:8386|DOI:10.1038/ncomms9386|www.nature.com/naturecommunications
&2015MacmillanPublishersLimited.Allrightsreserved.