Canonical and noncanonical intraflagellar transportregulates craniofacial skeletal developmentKazuo Nodaa,1,2, Megumi Kitamia,2, Kohei Kitamia,b, Masaru Kakuc, and Yoshihiro Komatsua,d,3

aDepartment of Pediatrics, The University of Texas Medical School at Houston, Houston, TX 77030; bDivision of Orthodontics, Niigata University GraduateSchool of Medical and Dental Sciences, Niigata 951-8514, Japan; cDivision of Bioprosthodontics, Niigata University Graduate School of Medical and DentalSciences, Niigata 951-8514, Japan; and dGraduate Program in Genes and Development, The University of Texas Graduate School of Biomedical Sciencesat Houston, Houston, TX 77030

Edited by Marianne Bronner, California Institute of Technology, Pasadena, CA, and approved March 30, 2016 (received for review September 30, 2015)

The primary cilium is a cellular organelle that coordinates signalingpathways critical for cell proliferation, differentiation, survival, andhomeostasis. Intraflagellar transport (IFT) plays a pivotal role inassembling primary cilia. Disruption and/or dysfunction of IFT com-ponents can cause multiple diseases, including skeletal dysplasia.However, the mechanism by which IFT regulates skeletogenesisremains elusive. Here, we show that a neural crest-specific deletionof intraflagellar transport 20 (Ift20) in mice compromises ciliogenesisand intracellular transport of collagen, which leads to osteopenia inthe facial region. Whereas platelet-derived growth factor receptoralpha (PDGFRα) was present on the surface of primary cilia in wild-type osteoblasts, disruption of Ift20 down-regulated PDGFRα pro-duction, which caused suppression of PDGF-Akt signaling, resultingin decreased osteogenic proliferation and increased cell death. Al-though osteogenic differentiation in cranial neural crest (CNC)-de-rived cells occurred normally in Ift20-mutant cells, the process ofmineralization was severely attenuated due to delayed secretion oftype I collagen. In control osteoblasts, procollagen was easily trans-ported from the endoplasmic reticulum (ER) to the Golgi apparatus.By contrast, despite having similar levels of collagen type 1 alpha 1(Col1a1) expression, Ift20mutants did not secrete procollagen becauseof dysfunctional ER-to-Golgi trafficking. These data suggest that in themultipotent stem cells of CNCs, IFT20 is indispensable for regulatingnot only ciliogenesis but also collagen intracellular trafficking. Ourstudy introduces a unique perspective on the canonical and non-canonical functions of IFT20 in craniofacial skeletal development.

cranial neural crest cells | intracellular trafficking | intraflagellar transport |PDGF signaling | primary cilia

Intraflagellar transport (IFT) proteins are required for the as-sembly of primary cilia (1, 2), which coordinate signaling pathways

that are critical for governing cell proliferation, differentiation,survival, and homeostasis (3, 4). Mutations in ciliary proteins resultin a diverse set of clinical disorders, termed ciliopathies, which in-clude a wide variety of skeletal phenotypes (5, 6). IFT proteinsorganize into two complexes, IFT-A and IFT-B (2, 7). IFT-B pro-teins mediate anterograde transport from the cell body to the ciliumtip, whereas IFT-A proteins regulate retrograde trafficking. Genesencoding IFT-A proteins are commonly altered in skeletal cil-iopathies. For example, IFT122, IFT140, and IFT144 are frequentlymutated in Sensenbrenner and Jeune syndromes, indicating that allsix IFT-A components are linked to skeletal ciliopathies (8–10).Conversely, with the exception of IFT80 and IFT172 in Jeune as-phyxiating thoracic dystrophy (11, 12), the function of IFT-B com-plexes in skeletal diseases is unknown.Intraflagellar transport 20 (IFT20) is the smallest IFT protein

in the IFT-B complex and interacts with IFT57/Hippi and thekinesin-II subunit KIF3B (13). Conventional Ift57 and Kif3a/Kif3b null mutations in mice lead to early embryonic lethality dueto left–right axis defects, accompanied by a loss of nodal cilia(14–16), suggesting that the IFT-B complex is critical for cilio-genesis. Supporting this idea, the deletion of Ift20 in mousekidney causes a lack of primary cilia and leads to polycystickidney disease (17). In addition to the traditional role of IFT in

cilium assembly, recent studies have found that an IFT-likeparticle organized by IFT20 is recruited to the immune synapsefor T-cell receptor recycling (18, 19). These studies suggest thatIFT proteins participate in intracellular membrane trafficking inimmune cells; however, it remains unclear how IFT20 governsdevelopmental processes, including skeletal formation, and howit functions in other cell types, such as multipotent stem cells.Several rare pleiotropic diseases that affect craniofacial skeletal

formation show disruption in primary cilia (20, 21), and oral-facial-digital (OFD) syndrome and Bardet–Biedl syndrome (BBS)are related to ciliary dysfunction (22, 23). Consistent with thefindings in humans, mutations in Ofd1 in mice and zebrafish causesevere craniofacial abnormalities, including cleft palate and short-ened Meckel’s cartilage (22, 24). Both patients and mice with amutant BBS6 gene display similar broad midfacial malformationand nasal hypoplasia (23). Disrupting ciliogenic components incranial neural crest cells (CNCCs) frequently results in embryonicand/or postnatal craniofacial abnormalities (25, 26), indicating atight relationship between primary cilia and craniofacial morphogen-esis. Because CNCCs are multipotent stem cells essential for formingcraniofacial skeletal components (27, 28), it is important to understandhow primary cilia function in CNCCc during facial development.The purpose of this study is to investigate the molecular

function of IFT20 in a multipotent stem cell population duringcraniofacial development. We found that IFT20 plays a vital rolein controlling not only ciliogenesis but also the intracellular

Significance

Intraflagellar transport (IFT) plays a critical role in assemblingprimary cilia that mediate growth factor signaling. The disrup-tion or dysfunction of IFT components can generate multiplediseases, including skeletal dysplasia. However, the mechanismby which IFT regulates skeletogenesis remains elusive. In thepresent study, we show that neural crest-specific deletion of thegene that encodes intraflagellar transport 20 (IFT20) in micecompromises ciliogenesis and the intracellular transport of col-lagen, leading to osteopenia in the face. Our findings highlight aunique function of IFT beyond its role in cilium assembly duringcraniofacial development, suggesting that IFT20 is indispens-able for the regulation of not only ciliogenesis but also theintracellular trafficking of collagen in the unique multipotentstem cell population of cranial neural crests.

Author contributions: Y.K. designed research; K.N., M. Kitami, K.K., and Y.K. performedresearch; K.N., M. Kitami, K.K., M. Kaku, and Y.K. analyzed data; and K.N. and Y.K. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1Present address: Department of Plastic and Reconstructive Surgery, Kyoto UniversityGraduate School of Medicine, Sakyo-ku, Kyoto 606-8507, Japan.

2K.N. and M. Kitami contributed equally to this work.3To whom correspondence should be addressed. Email: yoshihiro.komatsu@uth.tmc.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1519458113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1519458113 PNAS | Published online April 26, 2016 | E2589–E2597

DEV

ELOPM

ENTA

LBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

020

trafficking of collagen in CNCCs, demonstrating a function of IFTbeyond its role in ciliogenesis during facial skeletal formation.

ResultsDisruption of Ift20 in Neural Crest Cells Results in CraniofacialMalformation. To characterize the function of IFT20 in facial de-velopment, we disrupted Ift20 in a neural crest-specific manner usingthe wingless-related MMTV integration site 1 (Wnt1)-Cre driverline (17, 29) (Fig. S1 A and B; Ift20:Wnt1-Cre or cKO hereafter).Neural crest-specific deletion of Ift20 led to severe craniofacialmalformation in mice (Fig. 1). Ift20:Wnt1-Cre embryos were bornat Mendelian ratios (Table S1), but had hypertelorism andfrontonasal dysplasia with 100% phenotypic penetrance (Fig.1A). Ift20:Wnt1-Cre mice were found dead with no milk spot,indicating that they died shortly after birth due to difficulties infeeding and breathing. Histological analysis revealed that Ift20:Wnt1-Cre embryos had an abnormal expansion of the facial mid-line and lacked major craniofacial components, including the pal-atal shelves and a tongue, at embryonic day (E) 18.5 (Fig. 1A andFig. S1C). Skeletal staining of Ift20:Wnt1-Cre mice found either aloss or a truncation of craniofacial bones (Fig. 1B). The maxillaryand mandibular bones and palatal process of the maxilla in Ift20:Wnt1-Cre mice were hypoplastic, whereas the palatal process ofthe palatine and pterygoid were absent (Fig. 1B). Ift20:Wnt1-Cre mice had malformed nasal and frontal bones, with anopening in the metopic suture (Fig. 1B), suggesting that cranial

neural crest (CNC)-derived facial bones were severely com-promised. Mice with neural crest-specific inactivation of Kif3a, amember of the kinesin superfamily required for ciliogenesis, alsoshow bifid nasal cartilage, severe cleft palate, and widely sepa-rated frontal bones (30, 31), similar to the features observed inour Ift20:Wnt1-Cre embryos. Because IFT20 appears to bridgethe kinesin-II complex (KIF3A/KIF3B) (13), the phenotypicanalysis of Ift20:Wnt1-Cre mice further reinforces the idea thatthere is a link between IFT20 and KIF3A, demonstrating thatciliary dysfunction due to IFT components contributes to a rangeof craniofacial abnormalities.Because epithelial–mesenchymal interactions are essential in

craniofacial morphogenesis (21, 27), we also deleted Ift20 in anepithelial cell-specific manner using the keratin 14 (K14)-Credriver line (32). Epithelial-specific deletion of Ift20 (Ift20:K14-Cre)did not induce overt abnormalities in the craniofacial regions (Fig.S2), but Ift20:K14-Cre mice displayed abnormal skin phenotypes,including fewer hair follicles than control mice (Fig. S3). Theseresults suggest that IFT20 is indispensable for facial developmentin CNC-derived mesenchymal cells, but not in epithelial cells. Inthis study, we focused on the analysis of the molecular mechanismsof IFT20 in CNC-derived bones during skull development.

IFT20 Assembles the Primary Cilia That Induce PDGF SignalingRequired for Osteogenic Proliferation and Survival During SkullFormation. Because IFT20 is important for primary cilia assembly

md

md

fb

fb

NB E18.5

NB

*

fb

fb

nb

nb

bo

bo

bs

bsp

pb

pb

tr

tr

pb

pb

ib

ib

ib

ib

ppmx

ppp ptgmx

pmx

eo

ppmx

mx pmx

nb

nb

eo

mx

mx

jb

jb

WT

cKO

WT

cKO

WT

cKO

A

B

Fig. 1. Neural crest cell-specific Ift20 deletion causes craniofacial abnormalities in mice. (A, Left andMiddle) Lateral and frontal views of the face of a control(WT) and a Ift20:Wnt1-Cre (cKO) mouse at birth (NB). (A, Right) H&E staining of sections fromWT and cKO embryos at E18.5. (Scale bar: 1 mm.) (B) Alcian blue-and Alizarin red-stained skulls of a WT and a cKO mouse at NB. Arrows indicate bifid nasal cartilage. Note that the frontal bones of Ift20:Wnt1-Cre mice arenot only separated by a large open space but are also less mineralized (asterisk). bo, basioccipital; bs, basisphenoid; eo, exoccipital; fb, frontal bone; ib,interparietal bone; jb, jugal bone; md, mandible; mx, maxilla; nb, nasal bone; p, palatine; pb, parietal bone; pmx, premaxilla; ppmx, palatal process of maxilla;ppp, palatal process of palatine; ptg, pterygoid; tr, tympanic ring. (Scale bar: 2 mm.)

E2590 | www.pnas.org/cgi/doi/10.1073/pnas.1519458113 Noda et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

020

(17, 33), we attempted to identify the primary cilia, using immu-nohistochemistry, in control and Ift20:Wnt1-Cre embryos. In controlembryos at E12.5, CNC-derived osteogenic cells, which were la-beled with RUNX2, a key regulator of osteoblasts, had primarycilia (Fig. 2A, Left). On the other hand, Ift20:Wnt1-Cre embryoshad no primary cilia in their CNC-derived osteogenic cells (Fig.2A, Right), demonstrating that IFT20 is essential for ciliogenesisin CNC derivatives.Ciliary defects lead to craniofacial abnormalities due to dis-

ruptions in Hedgehog signaling (8, 23, 30). However, it remainselusive whether ciliary-mediated Hedgehog is the sole signal axisto govern facial bone development. For instance, neural crest-specific Smoothened (Smo) mice display skull abnormalities (34);therefore, one might predict that Smo in primary cilia transducesthe Hedgehog signaling in skull formation. However, it is not

known if Smo is present on cilia in CNC derivatives. This notionraises the question of whether the phenotypes in Smo mutantsare attributed to the alterations of ciliary-dependent Hedgehogsignaling. Importantly, our analyzed Ift20 mutants showed a skullphenotype distinct from the phenotype in other cilia-deficientmice. Whereas neural crest-specific Kif3a mutants display ametopic craniosynostosis with the alteration of Hedgehog sig-naling (30, 35), Ift20 mutants did not show any premature suturefusions (Fig. 1B). These notions let us hypothesize that otherciliary-dependent signaling, rather than Hedgehog, would beresponsible for skull malformation in Ift20 mutants. To addressthis hypothesis, we examined platelet-derived growth factor(PDGF) signaling in Ift20:Wnt1-Cre embryos.PDGF receptor alpha (PDGFRα) is localized on the surface of

primary cilia in fibroblasts (36). The disruption of PDGFRα-PI3K

WT

cKO

MergePDGFRAc-Tu/ -TuB A

RU

NX

2A

c-Tu

-Tu

RU

NX

2 H

oech

st

WT cKOE12.5

D

250 PDGFR

IFT20

Tubulin

15

150

50

WT cKO (kD)

F

PDGFR

Tubulin

250 150

P-PDGFR

50

250 150

PDGF-AA - + - + WT cKO

(kD) G

P-Akt

Tubulin

Akt

75

50

75

PDGF-AA + (min) 0 2.5 5 7.5 10 12.5 0 2.5 5 7.5 10 12.5

WT cKO

(kD) E

Rel

ativ

e in

tens

ity

of P

DG

FR

0.0

0.4

0.8

1.2 *

WT cKO

C

WT cKO

Rel

ativ

e ex

pres

sion

leve

l

1.2 1.0 0.8 0.6 0.4 0.2 0.0

Ift20 Pdgfra

* *

JI KWT

cKO

TUNEL MergeOSX Ki-67

*

*

*

* *

*

*

*

WT

cKO

L

H

WT cKO0.0

0.2 Rel

ativ

e in

tens

ity

of P

-Akt

*

0.4

0.6

0.8

1.0

1.2

*

Ki-6

7+ /

OS

X+

cells

(%)

0

5

10 15 20 25 30

35 N.S.

N.S.

E14.5 E16.5 E18.5

WT cKO

E16.5 E18.5 0

2 4 6 8

10

12 *

N.S.

TUN

EL

posi

tive

cells

(%) WT

cKO

Fig. 2. IFT20 is indispensable for ciliogenesis, PDGFRα signaling, and the proliferation and survival of osteoblasts. (A) Immunostaining for RUNX2 (orange),acetylated tubulin (Ac-Tu, green), and gamma-tubulin (γ-Tu) (magenta) on sections fromWT and cKO embryos at E12.5. Arrows show primary cilia. (Scale bars:Upper, 200 μm; Lower, 5 μm.) (B–H) Primary osteoblasts were harvested from E18.5 calvaria (nasal and frontal bones) of WT and cKO embryos. For furtheranalysis, cells were cultured for 9 d in osteogenic medium. (B) Immunofluorescence images of primary osteoblasts labeled with Ac-Tu (green), γ-Tu (magenta),and PDGFRα (cyan). Arrows show primary cilia. (Scale bar: 5 μm.) (C) Quantitative RT-PCR analysis for Ift20 and Pdgfra. Expression levels were normalized toGapdh expression. (D) Immunoblot analysis of cell lysates. (E) Quantification of the relative protein levels of PDGFRα. (F) WT and cKO osteoblasts were treatedwith PDGF-ascorbic acid (AA) (50 ng/mL) and harvested to analyze the activation of PDGFRα (P-PDGFRα). (G) Akt phosphorylation (P-Akt) was examined byWestern blot after PDGF-AA stimulation. (H) Quantification of the relative protein levels of P-Akt in the cells treated with PDGF-AA for 10 min. (I) Coronalsections at E16.5 of WT and cKO frontal bones were double-labeled with OSX (green) and Ki-67 (magenta) to detect osteoblastic cells and proliferative cells,respectively. Asterisks show nonspecific staining of blood cells. (Scale bar: 50 μm.) (J) Quantification of Ki-67–positive cells in OSX-positive cell populations atthe indicated stages. (K) TUNEL staining of sections of WT and cKO frontal bones at E16.5. (Scale bar: 50 μm.) (L) Quantification of TUNEL-positive cells in thefrontal bone area at the indicated stage. Data in C, E, H, J, and L are represented as mean ± SD, n = 3 in each group. *P < 0.05. N.S., not significant.

Noda et al. PNAS | Published online April 26, 2016 | E2591

DEV

ELOPM

ENTA

LBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

020

signaling in CNCCs causes cleft palate and separated fron-tal bones (37). Thus, we hypothesized that PDGFRα on thesurface of primary cilia in CNCCs regulates osteogenic pro-liferation and cell survival during skull formation. To investigatethis idea, we established cultures of primary osteoblasts fromnasal and frontal bones, which are CNC-derived (27). Consistentwith a previous report (36), PDGFRα localized to the primarycilia in control osteoblasts (Fig. 2B); however, due to the absenceof primary cilia, it was not observed in Ift20:Wnt1-Cre osteoblastsbut was still present on the cell surface (Fig. 2B). QuantitativeRT-PCR and Western blot analysis confirmed that the tran-scriptional and translational levels of PDGFRα were reduced inIft20:Wnt1-Cre osteoblasts (Fig. 2 C–E). To determine whetherthe loss of PDGFRα in primary cilia attenuated PDGF-drivenintracellular signaling, we measured the activation of PDGFRαand the induction of Akt phosphorylation, one of the effectors ofPDGF signaling (38). Control osteoblasts stimulated with PDGFligand [PDGF-AA] robustly activated PDGFRα and inducedAkt phosphorylation, but Ift20:Wnt1-Cre osteoblasts did not (Fig.

2 F–H). These results suggest that PDGFRα, localized to primarycilia, regulates Akt phosphorylation through PDGFRα activation,which may govern multiple cascades of cellular events inosteoblasts.To investigate these cellular events, we first evaluated osteo-

genic differentiation by using preosteoblast markers RUNX2and Osterix (OSX). At E12.5, the RUNX2-positive area infrontal bone primordia was comparable between control andIft20:Wnt1-Cre embryos (Fig. S4A). During middle to late ges-tation, OSX-positive cells in the frontal bone were also equallydistributed between the two groups (Fig. S4B). Next, we exam-ined cell proliferation in embryonic skulls using Ki-67 immuno-histochemistry and costaining with OSX. Whereas the levels ofOSX were normal, Ift20:Wnt1-Cre embryos had reduced cellproliferation compared with controls at E16.5 (Fig. 2 I and J).We further analyzed osteogenic cell survival using a TUNELassay. Compared with controls, there was a significant increase inthe amount of cell death in Ift20:Wnt1-Cre embryos (Fig. 2 Kand L), suggesting that primary cilia-mediated PDGF signaling

A

B

C

D

E

von

Kos

saH

&E

C

ol1a

1O

SX

* *

* *

*

* *

*

1 w

eek

2 w

eeks

#1 #2 #3

#1 #2 #3

#1 #2 #3

#1 #2 #3

Nor

mal

l

ight

P

olar

ized

lig

ht * *

* *

Normal light Polarized light

Rel

ativ

e ex

pres

sion

leve

l

0.0

0.5

1.0

1.5

2.0

Runx2 Osx Ibsp Ocn

N.S. N.S. N.S.

N.S.

WT cKO

WT cKO

WT

cKO

WT

cKO

WT

cKO

WT cKO

E16.5

E16.5

Fig. 3. IFT20 is essential for osteogenic differentiation as well as for mineralization in vitro and in vivo. (A) H&E staining, von Kossa staining, immunoflu-orescent staining for OSX, and section in situ hybridization for Col1a1 for WT and cKO embryos at E16.5. Asterisks indicate the osteogenic front of frontalbones. Arrows show dotted calcium deposition. (Scale bars: 100 μm.) (B) Quantitative RT-PCR analysis of Runx2, Osx, Ibsp, and Ocn in WT and cKO calvaria(nasal and frontal bones) at E17.5 (data are represented as mean ± SD, n = 3 in each group). (C and D) Primary osteoblasts were harvested from E18.5 calvaria(nasal and frontal bones) of WT and cKO embryos. For advanced analysis, cells were cultured in osteogenic medium. (C) Bone nodule staining with Alizarinred. Three individual primary osteoblast cultures (#1, #2, and #3) from WT and cKO embryos were examined. (Scale bars: 100 μm.) (D) Collagen fiber or-ganization was analyzed by picrosirius red staining after 2 wk of preosteoblast culture. (Scale bars: 100 μm.) (E) Picrosirius red staining of sections from WTand cKO embryos at E16.5. Asterisks indicate the osteogenic front of frontal bones. Arrows show reduced development of collagen fibers. (Scale bars:100 μm.)

E2592 | www.pnas.org/cgi/doi/10.1073/pnas.1519458113 Noda et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

020

in osteoblasts plays a critical role in regulating both osteogeniccell proliferation and cell survival. Together, these results in-dicate that IFT20 is essential for primary cilia formation andcontrols both osteoblast proliferation and survival throughPDGF signaling during skull formation.

Attenuation of IFT20 Does Not Influence Osteogenic Differentiation,but Leads to Abnormal Bone Mineralization in the Skull. BecauseIFT20 may regulate the onset of craniofacial skeletogenesis inmiddle to late gestation, we examined bone mineralization incontrol and Ift20:Wnt1-Cre embryos at E16.5. H&E stainingshowed continuous bone tissues in control skulls but not inIft20:Wnt1-Cre skulls (Fig. 3A). The von Kossa staining revealeddiscontinuous calcium deposition in Ift20:Wnt1-Cre calvaria (Fig.3A). OSX and collagen type 1 alpha 1 (Col1a1), both markers ofosteoblast differentiation, were also analyzed using immunohis-tochemistry and section in situ hybridization. Both OSX andCol1a1 were normally produced and expressed in osteogenictissues in control and Ift20:Wnt1-Cre skulls (Fig. 3A). Quantita-tive RT-PCR further confirmed that the expression of osteogenicdifferentiation genes, such as Runx2, Osx, Ibsp, and Ocn, wasunchanged in the nasal and frontal bones of Ift20:Wnt1-Creembryos (Fig. 3B), suggesting that IFT20 is not required forosteogenic differentiation but is necessary for the process ofmineralization during skull development.To address this hypothesis, we examined the process of min-

eralization using primary osteoblasts. After adding ascorbic acidand β-glycerol phosphate, an inducer of mineralization, bothcontrol and Ift20-mutant osteoblasts were examined. The controlosteoblasts formed white clusters at 1 wk, becoming solid nod-ules that positively stained with Alizarin red within 2 wk (Fig.3C). With Ift20:Wnt1-Cre osteoblasts, no cell clusters developedand few osteogenic nodules stained with Alizarin red (Fig. 3C).Because collagen fibrils are necessary for mineralization whenskull bones are formed (39), we visualized the collagen in thecultured osteoblasts. Picrosirius red staining revealed that theamount of collagen produced by Ift20:Wnt1-Cre osteoblasts wasless than the amount of collagen produced by control cells (Fig.3D), and discontinuous collagen fibrils were observed in the skullof Ift20:Wnt1-Cre mice (Fig. 3E). Taken together, these resultsdemonstrate that IFT20 is essential for normal mineralization inthe skull.

IFT20 Controls Procollagen Trafficking from the EndoplasmicReticulum to the Golgi Apparatus in CNC-Derived Osteoblasts. Al-though the transcriptional levels of Col1a1 were comparable be-tween control and Ift20:Wnt1-Cre embryos (Fig. 3A), the amount ofcollagen in the extracellular matrix was decreased in Ift20:Wnt1-Creosteoblasts (Fig. 3 D and E). These results suggest that Col1a1was transcribed normally and that procollagen was produced butwas not secreted appropriately into the external cellular environ-ment, thus leading to decreased collagen fibril formation. Thisfinding prompted us to hypothesize that IFT20 participates in theintracellular trafficking of procollagen.We first analyzed the localization of IFT20 in wild-type oste-

oblasts by costaining with the cis-Golgi marker GM130. Consis-tent with a previous report (40), IFT20 localized to the Golgiapparatus (Fig. S5). Staining of calnexin, the marker of endo-plasmic reticulum (ER), further confirmed that IFT20 was notpresent in the ER (Fig. S6). Next, we examined whether thesecretion of procollagen in Ift20-mutant osteoblasts was affectedby treatment with ascorbic acid. Before ascorbic acid stimulation,procollagen was initially located in the ER (Fig. S7). After 2 h ofascorbic acid stimulation, the procollagen was smoothly trans-ported into the Golgi in 80% of control osteoblasts. On the otherhand, only 40% of Ift20-mutant osteoblasts showed collagen Itransition to the Golgi, suggesting that procollagen secretion isslower than for the control cells. Procollagen secretion was still

hampered at 3 h in Ift20-mutant osteoblasts. Although therewere tendencies to delay procollagen secretion, Ift20-mutantosteoblasts could secrete procollagen to a similar extent ascontrols after 4 h of ascorbic acid stimulation (Fig. 4 A and B).These data suggest that procollagen transfer to the Golgi wasslow at an early phase of collagen secretion, but once the oste-oblasts started to secrete collagen I, there were not so manydifferences in intracellular collagen I localization between con-trol and Ift20-mutant osteoblasts. Western blot analysis furtherverified that the procollagen in Ift20-mutant osteoblasts re-mained intracellularly much longer than in controls (Fig. 4 C andD), indicating that procollagen secretion is hampered by thedysfunction of ER-to-Golgi trafficking.To monitor the dynamics of intracellular protein transport

from the ER to the Golgi, we used a plasmid encoding vesicularstomatitis virus G protein tagged with EGFP (VSVG-GFP) (41).VSVG-GFP is a powerful tool used in the study of membranetransport because of its reversible misfolding and retention in theER at 40 °C, and its ability to move out of the ER and into theGolgi upon reducing the temperature to 32 °C (41). We trans-fected control and Ift20-mutant osteoblasts with the VSVG-GFPplasmid and examined the movement of GFP using immunocy-tochemical costaining with GM130. After temperature reductionfrom 40 °C to 32 °C, GFP colocalized with GM130 within 5 minin the majority (84%) of control cells (Fig. 4 E and F). On theother hand, in the same time course, only 51% of Ift20-mutantosteoblasts showed colocalization (Fig. 4 E and F).To gain further insight into the potential cellular mechanisms

underlying the defective mineralization in Ift20:Wnt1-Cre embryos,we monitored the ER-to-Golgi transport of VSVG-GFP withtime-lapse imaging. Live imaging revealed that GFP accumulatedmore slowly in the Golgi apparatus of Ift20 mutants than in theGolgi apparatus of control osteoblasts (Movies S1 and S2). Fi-nally, we examined whether the transport of procollagen into theGolgi was affected in Ift20-mutant skulls in vivo. We noted thatonly osteoblasts near the osteogenic front of wild-type skulls wereenriched for procollagen in the Golgi. Therefore, we focused onthe area close to the osteogenic front. At E16.5, 55% of wild-typeosteoblasts contained procollagen in the Golgi, whereas only 8%of Ift20-mutant osteoblasts had procollagen (Fig. 4 G and H). AtE18.5, 50% of Ift20-mutant osteoblasts had procollagen in theGolgi compared with 70% of wild-type osteoblasts (Fig. 4 G andH), indicating that there is catching up, to some extent, of thedelayed ossification in Ift20 mutants at E18.5, which is consistentwith in vitro observations. These results suggest that IFT20 is re-quired for the ER-to-Golgi transport of procollagen, which is acritical first step of procollagen intracellular trafficking duringskull mineralization.Taken together, we conclude that IFT20 plays a pivotal role in

skull development through (i) the formation of primary cilia,which are required for mediating the PDGF signaling that con-trols both osteogenic proliferation and cell survival, and (ii) theregulation of the ER-to-Golgi transport critical for collagen se-cretion during skull development (Fig. 5).

DiscussionIFT plays a crucial role in the assembly of primary cilia, whichare highly conserved organelles that project from the surface ofmany cells. However, the details of how IFT governs craniofacialskeletal development remain elusive. In this study, we showedthat IFT20 is required for assembling the primary cilia thatcontrol skeletogenic proliferation and cell survival through theprecise regulation of PDGF signaling during craniofacial for-mation. We also found a potential link between IFT20 andprocollagen trafficking, which may be important for the secretionof collagen. Our study introduces a unique perspective on thecanonical and noncanonical functions of IFT20 in craniofacialskeletal formation.

Noda et al. PNAS | Published online April 26, 2016 | E2593

DEV

ELOPM

ENTA

LBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

020

IFT20 Is Essential for Assembling Primary Cilia That Transduce PDGFSignaling During Craniofacial Skull Formation. Growth factor sig-naling in CNCCs regulates the patterning and growth of facialprimordia (21, 27, 28). Although it has been widely recognizedthat primary cilia are projected like antennas and serve as eitherchemosensory or mechanosensory organelles, their role inCNCCs remains unclear. Because neural crest-specific loss ofPkd2, which encodes a mechanoreceptor in cilia, leads to mul-tiple craniofacial anomalies at postnatal ages, but not in miceembryos (42), one might speculate that primary cilia pre-dominantly function as chemosensors that mediate numerousgrowth factor signals during embryonic development (36, 43, 44).Consistent with a previous report analyzing the disruption ofIFT20 in kidney development (17), IFT20 was required for theformation of primary cilia in CNC-derived osteogenic cells (Fig.2 A and B), demonstrating that IFT20 plays an important role in

assembling primary cilia that may transduce growth factor sig-naling in facial development.During skull development, primary cilia might predominantly

participate in the regulation of PDGF signaling, rather than inHedgehog signaling. Supporting this idea, we found that one ofthe PDGF receptors, PDGFRα, clearly localized to the ciliumsurface in primary osteoblasts (Fig. 2B). In addition, we foundthat PDGF-Akt signaling through primary cilia is required forosteoblast proliferation and survival (Fig. 2 B–L). In agreementwith the phenotype of mouse embryos with a CNCC-specificinactivation of Pdgfra (45), the control and Ift20-mutant embryoshad a similar number of proliferating CNCC-derived osteogeniccells during early-middle gestation, but a much lower numberduring middle-late gestation in Ift20 mutants (Fig. 2J), indicatingthat PDGF signaling via primary cilia in CNCCs may not beessential at the early stage of osteoblastogenesis, but is required

B

0 20 40 60 80

100

-AA +AA

1 2 3 4 5 (hrs)

* *

WT cKO

% o

f cel

ls w

ith

Col

lage

n I i

n G

olgi

C

olla

gen

I G

M13

0 M

erge

C

olla

gen

I G

M13

0 M

erge

-AA +AA

1 2 3 4 5 (hrs)

WT

cKO

A

WT cKO

H

0

20

40

60

80

100

E16.5 E18.5

% o

f cel

ls w

ith

Col

lage

n I i

n G

olgi

**

F

Time (min)

0

20

40

60

80

100

0 2.5 5 10

% o

f cel

ls w

ith V

SV

G in

Gol

gi

*

WT cKO

E

VS

VG

-GFP

G

M13

0 M

erge

WT cKO

G

Col

lage

n I

GM

130

Mer

ge

E16.5 E18.5 WT cKO WT cKO

C

Collagen I 150

50 Tubulin

WT WT WT cKO cKO cKO0 1+ AA (h) 2

(kD)

D

010 20 30 40 50

% o

f col

lage

n I

in ly

sate

WT cKO

*

Fig. 4. IFT20 is critical for the secretion of intracellular type I collagen by regulating its ER-to-Golgi trafficking in osteoblasts. (A and B) WT and cKO os-teoblast cultures were treated with AA (50 μg/mL) for the indicated time and compared with untreated cells. (A) Coimmunostaining of type I collagen(collagen I) and GM130. Arrows show localization of intracellular collagen I in the Golgi. (Scale bar: 5 μm.) (B) Quantification of the percentage of the cellswith collagen I in the Golgi. (C) Immunoblot analysis of intracellular collagen I. (D) Quantification of the relative protein levels of intracellular collagen I. Proteinlevels were normalized to α-tubulin. (E and F) Control and Ift20:Wnt1-Cre osteoblasts were transfected with the plasmid for VSVG-GFP. After incubation at40 °C for 24 h, cells were shifted to 32 °C to allow VSVG-GFP transport from the ER into the Golgi. Cells were fixed at 0, 2.5, 5, and 10 min after shifting to 32 °C,followed by immunostaining for GFP and GM130. (E) Representative images of immunostaining at 5 min. Arrows show localization of VSVG-GFP to the Golgi.(Scale bar: 20 μm.) (F) Quantification of the percentage of coimmunostained cells in VSVG-GFP–positive cell populations. (G) Coimmunostaining of collagen Iand GM130 in a WT skull and a cKO skull at E16.5 and E18.5. Arrows show the localization of intracellular collagen I in the Golgi. Asterisks show the osteogenicfront. (Scale bar: 5 μm.) (H) Quantification of the percentage of cells with collagen I in the Golgi at the indicated stage. Data in B, D, F, and H are represented asmean ± SD, n = 3 in each group. *P < 0.05.

E2594 | www.pnas.org/cgi/doi/10.1073/pnas.1519458113 Noda et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

020

for cell proliferation and survival during late-stage skull devel-opment. However, we cannot exclude the possibility thatHedgehog signaling may also be involved in the molecularpathogenesis, because dysregulation of Hedgehog signaling hasbeen frequently observed in ciliary mutants displaying skeletaldysmorphogenesis (46, 47). Although the localization of Smo toprimary cilia has been identified in nodal cilia (44), a key ques-tion would be whether Hedgehog signaling mediators, includingSmo, are present in the primary cilia of CNC-derived osteoblasts.This alternative possibility will need to be addressed to un-derstand the synergistic regulatory mechanisms of growth factorsignaling (e.g., PDGF and Hedgehog signaling) in primarycilia during skull formation.

IFT20 Is Critical for Regulating the Trafficking of Procollagen in CNC-Derived Osteoblasts. Compared with other IFT-B proteins, IFT20has an unusual distribution. In addition to being found in bothprimary cilia and the centrosome pool, it is present in the Golgiapparatus (40). This finding raises the possibility that IFT20might participate in intracellular membrane trafficking, in ad-dition to its role in ciliary assembly. Supporting this hypothesis,recent studies have demonstrated that IFT20 is associated withthe trafficking of ciliary membrane proteins from the Golgi tothe cilium, opsin trafficking, and membrane trafficking in lym-phoid and myeloid cells (18, 19, 33, 48); however, it remainsunclear whether IFT20 functions as a noncanonical IFT systemduring craniofacial skeletal development.Ift20:Wnt1-Cre embryos showed immature bone mineralization

(Fig. 1), but this phenotype was not simply explained by thedown-regulation of osteogenic differentiation activities (Fig. 3).Osteoblastic differentiation genes, including Col1a1, wereexpressed normally; however, Ift20-mutant osteoblasts were notcapable of mineralization to the same degree as controls. Ourstudy suggests that this defect is because Ift20-mutant cells can-not secrete procollagen molecules into the external cellular en-vironment within the same time course as controls (Fig. 4),demonstrating that IFT20 may be involved in procollagen traf-ficking of osteoblasts. Meanwhile, noncollagen factors secretedby bones may also be affected in Ift20:Wnt1-Cre mutants. Forexample, bioactivity of both bone morphogenetic protein-15(BMP-15) and growth differentiation factor-9 (GDF-9) is tightlyregulated by the Golgi-mediated secretory pathway (49, 50).Therefore, IFT20 might participate in the regulation of the se-creting pathway of those growth factors in skull formation; thispossibility needs to be clarified in the future.In humans, mutations in SEC23A, a gene encoding a compo-

nent of the coat protein complex II (COPII), cause craniolenti-culosutural dysplasia (CLSD) (51, 52). CLSD is an autosomalrecessive disorder characterized by skeletal defects and sepa-

rated frontal bones, along with hypertelorism (53), resemblingthe features of Ift20:Wnt1-Cre embryos. COPII-coated vesicleformation is critical in anterograde protein trafficking from theER to the Golgi apparatus (54). Whereas procollagen bundlesexiting from the ER have a length of 300 nm, a much larger sizethan COPII can generally transport (39), studies suggest thatlarge procollagen cargo might be transported to the Golgi by aspecialized ER-Golgi intermediate compartment, known as thevesicular tubular cluster (VTC) complex, in a COPII-dependentmanner (55). Because ER export complexes are composed ofarrays of budding vesicles and cytoplasmic VTCs (56), thisfinding prompted us to hypothesize that IFT20 participates inthe composition of VTCs and plays a role in procollagen traf-ficking from the ER to the Golgi, which is essential for the initialprocess of procollagen transport. Correspondingly, we found thatprocollagen transport from the ER to the Golgi was delayed inIft20-mutant osteoblasts (Fig. 4). VSVG-GFP time-lapse imagingfurther revealed that the transport machinery was affected inIft20 mutants (Movies S1 and S2), suggesting that IFT20 may beinvolved in ER export complexes.The golgin GMAP210/TRIP11 anchors IFT20 to the Golgi

complex, and both molecules function together in the traffickingof ciliary membrane proteins (40). An initial report suggests thatloss of GMAP210/TRIP11 in mice alters glycosylation in theGolgi, leading to the intracellular accumulation of Perlecan inchondrocytes and resulting in a neonatal lethal form of skeletaldysplasia (57). However, recent studies suggest that GMAP210/TRIP11 is required for efficient membrane trafficking because itregulates vesicle tethering from the ER (56, 58, 59). Therefore,the IFT20–GMAP210 complex in CNC-derived osteoblasts mayact in ER-to-Golgi trafficking.In summary, our study introduces a unique perspective for ca-

nonical and noncanonical IFT function in craniofacial skeletalformation. Our findings may also contribute to our understandingof the molecular pathogenesis of skeletal ciliopathies.

Experimental ProceduresAnimals. Ift20-floxed mice (17), Wnt1-Cre mice (29), and K14-Cre mice (32)were obtained from The Jackson Laboratory. All mice were maintained inthe animal facility of The University of Texas Medical School at Houston. Theexperimental protocol was reviewed and approved by the Animal WelfareCommittee and the Institutional Animal Care and Use Committee of TheUniversity of Texas Medical School at Houston.

Skeletal Preparations, Histological Analysis, Immunofluorescent Staining, andTUNEL Assays. Staining of bone and cartilage of embryos with Alizarin red/Alcian blue was carried out as described previously (60). H&E staining, im-munofluorescent staining, and TUNEL assays of paraffin sections were per-formed as described previously (61). The von Kossa staining was performedusing a von Kossa staining kit (Abcam; ab150687). Picrosirius red staining

Fig. 5. Canonical and noncanonical IFT20 regulates craniofacial skeletal development. IFT20 is critical for regulating (i) the formation of primary cilia re-quired for mediating PDGF signaling, which controls both osteogenic proliferation and cell survival, and (ii) ER-to-Golgi transport in the process of collagensecretion during skull development.

Noda et al. PNAS | Published online April 26, 2016 | E2595

DEV

ELOPM

ENTA

LBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

020

was performed using 1% picrosirius red solution (Sigma-Aldrich; 365548 andP6744). Section in situ hybridization was performed as previously described(62). Primary antibodies used in immunofluorescent staining were as follows:RUNX2 (1:400, Cell Signaling; no. 12556), acetylated tubulin (1:1,000, Sigma-Aldrich; T6793), gamma-tubulin (1:1,000, Sigma-Aldrich; T5326), OSX (1:200,Abcam; ab22552), Ki-67 (1:100, BD Biosciences; 550609), GM130 (1:200, BDBiosciences; 610822), calnexin (1:100, EMD Millipore; MAB3126), and colla-gen type I (1:200, Cedarlane; CL50151AP). Slides were viewed with anOlympus FluoView FV1000 laser scanning confocal microscope using thesoftware FV10-ASW Viewer (version 3.1).

Isolation and Culture of Primary Osteoblasts.We collected calvarial osteoblastsfrom control and Ift20:Wnt1-Cre embryos at E18.5. Nasal and frontal boneswere subjected to five sequential digestions with an enzyme mixture con-taining 1 mg/mL collagenase type I (Sigma-Aldrich; C0130) and 1 mg/mLcollagenase type II (Sigma-Aldrich; C6885). Cell fractions (from two to five ofthe sequential digestions) were collected and cultured in growth medium[α-MEM supplemented with 10% (vol/vol) FBS, 100 U/mL penicillin, and100 mg/mL streptomycin] and in osteogenic medium (growth medium supple-mented with 50 μg/mL ascorbic acid and 2 mM β-glycerol phosphate). For theanalysis of PDGFRα signaling, cells were treated with 50 ng/mL PDGF-ascorbicacid (PeproTech; 315-17).

Western Blot Analysis. We prepared the osteoblast lysates in radio-immunoprecipitation assay buffer. After centrifugation, the supernatantswere separated by SDS/PAGE, blotted onto a PVDF membrane, analyzed withspecific antibodies, and visualized with enhanced chemiluminescence. Theantibodies used were as follows: α-tubulin (Abcam; ab7291), IFT20 (Pro-teintech; 13615-1-AP), PDGFRα (R&D Systems; AF1062), P-PDGFRα (Sigma-Aldrich; P8246-1VL), Akt (Cell Signaling; no. 9272), P-Akt (Cell Signaling; no.4060), and collagen type I (Cedarlane; CL50151AP). The intensities of theimmunostained bands were quantified with Image Lab Version 5.0 (Bio-Rad).

Protein Transport Assays and Immunocytochemistry. Primary osteoblasts weretransfected with the plasmid containing VSVG-GFP (Addgene; no. 11912),using Lipofectamine 3000 (Thermo Fisher Scientific), and incubated at 40 °Cfor 24 h. To shift the temperature from 40 °C to 32 °C, we used a ThermoPlate (Tokai Hit) for immunocytochemistry and a Stage Top Incubator (TokaiHit) for time-lapse imaging. For quantification, 10 images were captured

from one coverslip with at least 43 GFP-positive cells. This experiment wasperformed independently in triplicate. To analyze intracellular Col1a1, cellswere cultured for 12 h after seeding and then treated or not treated with50 μg/mL ascorbic acid. Cells were fixed with 4% (wt/vol) paraformaldehydefor 10 min at room temperature (RT) and permeabilized with 0.1% TritonX-100 in PBS (PBST) for 15 min at RT. They were then blocked in 2% (vol/vol)sheep serum in PBST for 30 min at RT before staining with primary anti-bodies. The primary antibodies used were as follows: acetylated tubulin(1:1,000, Sigma-Aldrich; T6793), gamma-tubulin (1:1,000, Sigma-Aldrich;T5326), PDGFRα (1:200, R&D Systems; AF1062), IFT20 (1:500, Proteintech;13615-1-AP), GM130 (1:200, BD Biosciences; 610822), collagen type I (1:200,Cedarlane; CL50151AP), and GFP (1:500, Thermo Fisher Scientific; A-11122).Images were captured with an Olympus FluoView FV1000 laser scanningconfocal microscope equipped with the software FV10-ASW Viewer(version 3.1).

Quantitative RT-PCR. Using TRIzol Reagent (Thermo Fisher Scientific), totalRNA was extracted from the nasal and frontal bones of control and Ift20:Wnt1-Cre embryos at E17.5. The total RNA was treated with DNase I(Roche) before cDNA synthesis. From the cultured osteoblasts, total RNAwas purified using the RNeasy Plus Mini Kit (Qiagen). cDNA was syn-thesized using iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad). Quantitative RT-PCR was carried out using SsoAdvanced UniversalSYBR Green Supermix (Bio-Rad). The sequences of the specific primersets are listed in Table S2.

Statistical Analysis. The Student’s t test was used for statistical analysis. AP value of less than 0.05 was considered statistically significant.

ACKNOWLEDGMENTS. We thank Dr. Gregory J. Pazour and Dr. AndrewP. McMahon for Ift20-floxed, Wnt1-Cre, and K14-Cre mice; Dr. Yingzi Yang,Dr. Ernestina Schipani, and Dr. Wei Hsu for in situ plasmids; Dr. JenniferLippincott-Schwartz for sharing the pEGFP-VSVG plasmid; Dr. Sunny Wongfor skin analysis; Dr. Trisha Castranio for the critical reading of this manu-script; and Patricia Fonseca for editorial assistance. This study was supportedby Grant R00DE021054 from the National Institute of Dental and Craniofa-cial Research/NIH (to Y.K.) and by a fellowship from the Uehara MemorialFoundation (to K.N.).

1. Kozminski KG, Johnson KA, Forscher P, Rosenbaum JL (1993) A motility in the eukaryoticflagellum unrelated to flagellar beating. Proc Natl Acad Sci USA 90(12):5519–5523.

2. Rosenbaum JL, Witman GB (2002) Intraflagellar transport. Nat Rev Mol Cell Biol 3(11):813–825.

3. Sharma N, Berbari NF, Yoder BK (2008) Ciliary dysfunction in developmental abnor-malities and diseases. Curr Top Dev Biol 85:371–427.

4. Eggenschwiler JT, Anderson KV (2007) Cilia and developmental signaling. Annu RevCell Dev Biol 23:345–373.

5. Haycraft CJ, Serra R (2008) Cilia involvement in patterning and maintenance of theskeleton. Curr Top Dev Biol 85:303–332.

6. Yuan X, Serra RA, Yang S (2015) Function and regulation of primary cilia and intra-flagellar transport proteins in the skeleton. Ann N Y Acad Sci 1335:78–99.

7. Pazour GJ, Rosenbaum JL (2002) Intraflagellar transport and cilia-dependent diseases.Trends Cell Biol 12(12):551–555.

8. Ashe A, et al. (2012) Mutations in mouse Ift144 model the craniofacial, limb and ribdefects in skeletal ciliopathies. Hum Mol Genet 21(8):1808–1823.

9. Walczak-Sztulpa J, et al. (2010) Cranioectodermal Dysplasia, Sensenbrenner syn-drome, is a ciliopathy caused by mutations in the IFT122 gene. Am J Hum Genet 86(6):949–956.

10. Miller KA, et al. (2013) Cauli: A mouse strain with an Ift140 mutation that results in askeletal ciliopathy modelling Jeune syndrome. PLoS Genet 9(8):e1003746.

11. Beales PL, et al. (2007) IFT80, which encodes a conserved intraflagellar transportprotein, is mutated in Jeune asphyxiating thoracic dystrophy. Nat Genet 39(6):727–729.

12. Halbritter J, et al.; UK10K Consortium (2013) Defects in the IFT-B component IFT172cause Jeune and Mainzer-Saldino syndromes in humans. Am J Hum Genet 93(5):915–925.

13. Baker SA, Freeman K, Luby-Phelps K, Pazour GJ, Besharse JC (2003) IFT20 links kinesin IIwith a mammalian intraflagellar transport complex that is conserved in motile flagellaand sensory cilia. J Biol Chem 278(36):34211–34218.

14. Houde C, et al. (2006) Hippi is essential for node cilia assembly and Sonic hedgehogsignaling. Dev Biol 300(2):523–533.

15. Nonaka S, et al. (1998) Randomization of left-right asymmetry due to loss of nodalcilia generating leftward flow of extraembryonic fluid in mice lacking KIF3B motorprotein. Cell 95(6):829–837.

16. Takeda S, et al. (1999) Left-right asymmetry and kinesin superfamily protein KIF3A:New insights in determination of laterality and mesoderm induction by kif3A-/- miceanalysis. J Cell Biol 145(4):825–836.

17. Jonassen JA, San Agustin J, Follit JA, Pazour GJ (2008) Deletion of IFT20 in the mouse

kidney causes misorientation of the mitotic spindle and cystic kidney disease. J Cell

Biol 183(3):377–384.18. Finetti F, et al. (2009) Intraflagellar transport is required for polarized recycling of the

TCR/CD3 complex to the immune synapse. Nat Cell Biol 11(11):1332–1339.19. Finetti F, et al. (2014) Specific recycling receptors are targeted to the immune synapse

by the intraflagellar transport system. J Cell Sci 127(Pt 9):1924–1937.20. Zaghloul NA, Brugmann SA (2011) The emerging face of primary cilia. Genesis 49(4):

231–246.21. Szabo-Rogers HL, Smithers LE, Yakob W, Liu KJ (2010) New directions in craniofacial

morphogenesis. Dev Biol 341(1):84–94.22. Ferrante MI, et al. (2006) Oral-facial-digital type I protein is required for primary cilia

formation and left-right axis specification. Nat Genet 38(1):112–117.23. Tobin JL, et al. (2008) Inhibition of neural crest migration underlies craniofacial dys-

morphology and Hirschsprung’s disease in Bardet-Biedl syndrome. Proc Natl Acad Sci

USA 105(18):6714–6719.24. Ferrante MI, et al. (2009) Convergent extension movements and ciliary function are

mediated by ofd1, a zebrafish orthologue of the human oral-facial-digital type 1

syndrome gene. Hum Mol Genet 18(2):289–303.25. Goetz SC, Anderson KV (2010) The primary cilium: A signalling centre during verte-

brate development. Nat Rev Genet 11(5):331–344.26. Chang CF, Schock EN, Attia AC, Stottmann RW, Brugmann SA (2015) The ciliary baton:

Orchestrating neural crest cell development. Curr Top Dev Biol 111:97–134.27. Noden DM, Trainor PA (2005) Relations and interactions between cranial mesoderm

and neural crest populations. J Anat 207(5):575–601.28. Bronner ME, LeDouarin NM (2012) Development and evolution of the neural crest: An

overview. Dev Biol 366(1):2–9.29. Danielian PS, Muccino D, Rowitch DH, Michael SK, McMahon AP (1998) Modification

of gene activity in mouse embryos in utero by a tamoxifen-inducible form of Cre

recombinase. Curr Biol 8(24):1323–1326.30. Brugmann SA, et al. (2010) A primary cilia-dependent etiology for midline facial

disorders. Hum Mol Genet 19(8):1577–1592.31. Kolpakova-Hart E, Jinnin M, Hou B, Fukai N, Olsen BR (2007) Kinesin-2 controls de-

velopment and patterning of the vertebrate skeleton by Hedgehog- and Gli3-dependent

mechanisms. Dev Biol 309(2):273–284.32. Dassule HR, Lewis P, Bei M, Maas R, McMahon AP (2000) Sonic hedgehog regulates

growth and morphogenesis of the tooth. Development 127(22):4775–4785.

E2596 | www.pnas.org/cgi/doi/10.1073/pnas.1519458113 Noda et al.

Dow

nloa

ded

by g

uest

on

June

1, 2

020

33. Follit JA, Tuft RA, Fogarty KE, Pazour GJ (2006) The intraflagellar transport proteinIFT20 is associated with the Golgi complex and is required for cilia assembly. Mol BiolCell 17(9):3781–3792.

34. Jeong J, Mao J, Tenzen T, Kottmann AH, McMahon AP (2004) Hedgehog signaling inthe neural crest cells regulates the patterning and growth of facial primordia. GenesDev 18(8):937–951.

35. Liu B, Chen S, Johnson C, Helms JA (2014) A ciliopathy with hydrocephalus, isolatedcraniosynostosis, hypertelorism, and clefting caused by deletion of Kif3a. ReprodToxicol 48:88–97.

36. Schneider L, et al. (2005) PDGFRalphaalpha signaling is regulated through the pri-mary cilium in fibroblasts. Curr Biol 15(20):1861–1866.

37. Fantauzzo KA, Soriano P (2014) PI3K-mediated PDGFRα signaling regulates survivaland proliferation in skeletal development through p53-dependent intracellularpathways. Genes Dev 28(9):1005–1017.

38. Klinghoffer RA, Hamilton TG, Hoch R, Soriano P (2002) An allelic series at thePDGFalphaR locus indicates unequal contributions of distinct signaling pathwaysduring development. Dev Cell 2(1):103–113.

39. Canty EG, Kadler KE (2005) Procollagen trafficking, processing and fibrillogenesis.J Cell Sci 118(Pt 7):1341–1353.

40. Follit JA, et al. (2008) The Golgin GMAP210/TRIP11 anchors IFT20 to the Golgi com-plex. PLoS Genet 4(12):e1000315.

41. Presley JF, et al. (1997) ER-to-Golgi transport visualized in living cells. Nature389(6646):81–85.

42. Khonsari RH, et al. (2013) Multiple postnatal craniofacial anomalies are characterizedby conditional loss of polycystic kidney disease 2 (Pkd2). Hum Mol Genet 22(9):1873–1885.

43. Corbit KC, et al. (2008) Kif3a constrains beta-catenin-dependent Wnt signallingthrough dual ciliary and non-ciliary mechanisms. Nat Cell Biol 10(1):70–76.

44. Corbit KC, et al. (2005) Vertebrate Smoothened functions at the primary cilium.Nature 437(7061):1018–1021.

45. Tallquist MD, Soriano P (2003) Cell autonomous requirement for PDGFRalpha inpopulations of cranial and cardiac neural crest cells. Development 130(3):507–518.

46. Haycraft CJ, et al. (2007) Intraflagellar transport is essential for endochondral boneformation. Development 134(2):307–316.

47. Koyama E, et al. (2007) Conditional Kif3a ablation causes abnormal hedgehog sig-naling topography, growth plate dysfunction, and excessive bone and cartilage for-mation during mouse skeletogenesis. Development 134(11):2159–2169.

48. Keady BT, Le YZ, Pazour GJ (2011) IFT20 is required for opsin trafficking and pho-toreceptor outer segment development. Mol Biol Cell 22(7):921–930.

49. Tibaldi E, et al. (2010) Golgi apparatus casein kinase phosphorylates bioactive Ser-6 ofbone morphogenetic protein 15 and growth and differentiation factor 9. FEBS Lett584(4):801–805.

50. Lasa-Benito M, Marin O, Meggio F, Pinna LA (1996) Golgi apparatus mammary glandcasein kinase: Monitoring by a specific peptide substrate and definition of specificitydeterminants. FEBS Lett 382(1-2):149–152.

51. Boyadjiev SA, et al. (2006) Cranio-lenticulo-sutural dysplasia is caused by a SEC23Amutation leading to abnormal endoplasmic-reticulum-to-Golgi trafficking. Nat Genet38(10):1192–1197.

52. Boyadjiev SA, et al. (2011) Cranio-lenticulo-sutural dysplasia associated with defects incollagen secretion. Clin Genet 80(2):169–176.

53. Boyadjiev SA, et al. (2003) A novel dysmorphic syndrome with open calvarial suturesand sutural cataracts maps to chromosome 14q13-q21. Hum Genet 113(1):1–9.

54. Lang MR, Lapierre LA, Frotscher M, Goldenring JR, Knapik EW (2006) Secretory COPIIcoat component Sec23a is essential for craniofacial chondrocyte maturation. NatGenet 38(10):1198–1203.

55. Stephens DJ, Pepperkok R (2002) Imaging of procollagen transport reveals COPI-dependent cargo sorting during ER-to-Golgi transport in mammalian cells. J CellSci 115(Pt 6):1149–1160.

56. Roboti P, Sato K, Lowe M (2015) The golgin GMAP-210 is required for efficientmembrane trafficking in the early secretory pathway. J Cell Sci 128(8):1595–1606.

57. Smits P, et al. (2010) Lethal skeletal dysplasia in mice and humans lacking the golginGMAP-210. N Engl J Med 362(3):206–216.

58. WongM, Munro S (2014) Membrane trafficking. The specificity of vesicle traffic to theGolgi is encoded in the golgin coiled-coil proteins. Science 346(6209):1256898.

59. Sato K, Roboti P, Mironov AA, Lowe M (2015) Coupling of vesicle tethering and Rabbinding is required for in vivo functionality of the golgin GMAP-210. Mol Biol Cell26(3):537–553.

60. Kamiya N, et al. (2015) SHP2-Deficiency in Chondrocytes Deforms Orofacial Cartilageand Ciliogenesis in Mice. J Bone Miner Res 30(11):2028–2032.

61. Noda K, Mishina Y, Komatsu Y (2015) Constitutively active mutation of ACVR1in oral epithelium causes submucous cleft palate in mice. Dev Biol, 10.1016/j.ydbio.2015.06.014.

62. Komatsu Y, Kishigami S, Mishina Y (2014) In situ hybridization methods for mousewhole mounts and tissue sections with and without additional β-galactosidasestaining. Methods Mol Biol 1092:1–15.

Noda et al. PNAS | Published online April 26, 2016 | E2597

DEV

ELOPM

ENTA

LBIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

June

1, 2

020