KIF16B/Rab14 Molecular Motor Complex Is Critical for Early Embryonic Development by Transporting FGF Receptor

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Developmental Cell Article

Hitoshi Ueno,1 Xiao Huang,1 Yosuke Tanaka,1 and Nobutaka Hirokawa1,* 1Department of Cell Biology and Anatomy, Graduate School of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan *Correspondence: [email protected] DOI 10.1016/j.devcel.2010.11.008

SUMMARY receptors are translated in the endoplasmic reticulum (ER), modified in the Golgi apparatus, budded off with membrane Kinesin-mediated membrane trafficking is a funda- vesicles, and transported to the apical surface by the biosyn- mental cellular process, but its developmental thetic pathway of membrane trafficking (Nakata et al., 1998). relevance is little understood. Here we show that They then undergo endocytosis into endosomes and are sorted the kinesin-3 motor KIF16B/Rab14 complex acts in to be degraded or recycled back to the plasma membrane biosynthetic Golgi-to-endosome traffic of the fibro- (Di Fiore and De Camilli, 2001; Seto et al., 2002). Endosomes blast growth factor receptor (FGFR) during early can also serve as intermediates of a post-Golgi biosynthetic –/– pathway for the apical membrane targeting regulated by a small embryonic development. Kif16b mouse embryos molecular G protein, Rab14 (Junutula et al., 2004; Kitt et al., failed in developing epiblast and primitive endoderm 2008). lineages and died in the peri-implantation stage, FGF signaling plays a fundamental role in embryonic develop- similar to previously reported FGFR2 knockout ment (Arman et al., 1998). Especially the development of epiblast embryos. KIF16B associated directly with the which is the primordium of three germ layers is highly dependent Rab14-GTP adaptor on FGFR-containing vesicles on FGF signaling (Chazaud et al., 2006). In the early postimplan- and transported them toward the plasma membrane. tation stage embryo, the inner cell mass of blastocyst acquires To examine whether the nucleotide state of Rab14 the first dorsoventral polarity and differentiates into primitive serves as a switch for transport, we performed endoderm in the outer surface; and primitive ectoderm (epiblast) Rab14-GDP overexpression. This dominant negative in the inner surface of the egg cylinder. These two tissues repre- approach reproduced the whole putative sequence sent the ventral and dorsal layers, respectively. Due to FGF signaling, cells of the primitive endoderm lineage appeared to of KIF16B or FGFR2 deficiency: impairment in secrete a layer of basement membrane containing laminin-a1 FGFR transport, FGF signaling, basement membrane and fibronectin (George et al., 1993; Li et al., 2004; Li et al., assembly by the primitive endoderm lineage, and 2001). This basement membrane is crucial for epiblast cells for epiblast development. These data provide one of surviving and developing a columnar epithelium that lines the first pieces of genetic evidence that microtu- proamnionic cavity inside the egg cylinder (Coucouvanis and bule-based membrane trafficking directly promotes Martin, 1995). Fgfr2 knockout embryos have been reported to early development. abolish these events in egg cylinder formation, and are lethal within few hours after implantation (Arman et al., 1998). Embryoid bodies expressing dominant negative FGFR2 failed INTRODUCTION in specification of primitive endoderm, assembly of basement membrane, and differentiation of epiblast (Li et al., 2001, 2004). Molecular motors, kinesin superfamily proteins (KIFs), play Epiblast cells that could not approach to the basement essential roles in intracellular transport of various cargos such membrane undergo apoptosis, and leave the proamnionic cavity as membranous organelle, protein complex, and mRNA within surrounded by the epiblastic epithelium at the egg cylinder cells (Hirokawa et al., 1998; Hirokawa et al., 2009; Vale, 2003). stage. Because the epiblast will later give rise to three germ KIF16B is a kinesin-3 family motor protein that contains a PX layers through gastrulation, elucidation of the molecular mecha- domain in its C terminus suggesting that it is associated with nism of epiblast differentiation will unravel how development of PI(3)P on some vesicle. Its endosome association and its role the fetus is initially set forth (Ang and Constam, 2004). in regulation of the expression level of EGF receptor have been To understand the physiological relevance of the KIF16B previously reported (Hoepfner et al., 2005). However, the precise protein within cells and whole body, we have performed gene characteristics of the cargo organelles, mechanistic link to them, targeting in mouse embryonic stem cells (ESCs) and directly and in vivo role of KIF16B have been largely elusive. Developing generated KO embryos and embryoid bodies from Kif16bÀ/À cells receive morphogens to determine their fate and behavior, ESCs. Significantly, these genotypes are lethal with a complete mainly via cell surface receptors (Seto et al., 2002). These loss of the basement membrane and deficiencies in epiblast

60 Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. Developmental Cell A Kinesin Transporting FGFR for Early Development

and primitive endoderm lineages during early postimplantation whereas the signal in KO ICM region was only at the background development, in parallel to the phenotype of previously reported level (n = 2 and 6; p < 0.05 and 0.01, c2 test; Figures 1C–1H). Fgfr2 KO embryos and FGFR2-dominant-negative embryoid Finally, immunoblotting of blastocysts consistently showed bodies. We have further identified that KIF16B is associated downregulation of pluripotency markers Oct4 and Sox2 in addi- with and essential for transport of FGFR2-containing vesicles tion to GATA6 by KIF16B deficiency (Figure 1I). Because it has from the Golgi toward the plasma membrane via a direct binding been reported that a basement membrane of the egg cylinder, partner, Rab14. Furthermore we found that the nucleotide state secreted by primitive endoderm, is essential for the survival of Rab14 critically regulates trafficking of FGFR2-carrying and differentiation fate of epiblast lineage (Coucouvanis and vesicles powered by KIF16B. Overexpression of Rab14-GDP Martin, 1995)(Figure 1J), the failure in differentiation of primitive reproduced the defects in the FGF-dependent germ layer forma- endoderm lineage could affect the overall development of the tion of embryos, suggesting that the KIF16B/Rab14 molecular embryo. Impairment in the developing potential of the KO motor complex critically regulates the biosynthetic pathway of embryo ICM was further examined using an explant culture FGFR2 and FGF signaling within developing cells. These results assay of these blastocysts (Figure S1D). Although the blasto- are one of the first pieces of evidence that kinesin-mediated cysts with WT ESCs could give rise to ES-cell-like colonies after membrane trafficking directly promotes early development of 5 days of culture, those with KO cells only developed fragmented the mouse embryo, through a newly discovered mechanism colonies with low viability. Because KIF16B was expressed regulating FGFR transport. within the ICM (Figure 1K), problems in housekeeping signal transduction pathways in the ICM were highly suspected. These RESULTS data suggest that KIF16B is essential for the correct differentiation of primitive endoderm that supports the development of Kif16b–/– Embryos Fail in Early Postimplantation peri-implantation mouse embryos. Development To examine the physiological relevance of KIF16B in vivo, Kif16b–/– Primitive Endoderm Fails to Assemble Kif16b–/– (KO) ESCs were generated by gene targeting (see a Basement Membrane Supporting the Epiblast Figures S1A–S1C available online). Because these ESCs prolif- Lineage of Embryoid Bodies erated normally but showed apparent differentiation problems To further examine the relevance of KIF16B in peri-implantation in teratoma formation assay (not shown), we performed tetra- development, we performed an embryoid body (EB) formation ploid rescue of KO and wild-type (WT) ESCs to examine the assay that could overcome the difficulty in observing embryos embryological phenotypes in postimplantation development. during the implantation stage (Figure 2; Figures S2, and S3) Because the tetraploid cells can only develop extraembryonic (Chen et al., 2000; Coucouvanis and Martin, 1995; Li et al., trophectoderm cells, the inner cell mass (ICM) lineage was 2001, 2004; Robertson, 1987). The ESCs were allowed to aggre- completely derived from the ESCs. Accordingly, this procedure gate in the absence of leukemia inhibitory factor (LIF). WT ESCs is used to generate embryos completely derived from ESCs, first developed ‘‘simple EBs’’ and then ‘‘cystic EBs’’ as and provides phenotype evaluation during embryonic develop- described (Figure S2C) (Doetschman et al., 1985; Robertson, ment (Nagy and Rossant, 1993). 1987). They first formed bilaminar germ layers by day 5 of According to histological examination of implanted uteri with culture, which are termed simple EBs. They appeared spherical, tetraploid-rescued blastocysts, WT embryos developed egg with epiblast in the core covered with a thin layer of primitive cylinders carrying primitive endoderm and epithelialized primi- endoderm. In EBs by day 6, the epiblast lineage further devel- tive ectoderm at 5.5 days post coitum (dpc) (Figure 1A). On the oped a proamnionic cavity lined by a basophilic columnar contrary, KO embryos did not develop to be delivered, and epithelium (Figure 2A, arrowheads), which are termed cystic were apparently arrested at the blastocyst stage to be lethal EBs. Accordingly, EBs have been utilized as a good ex utero (Figure 1B). Their primitive endoderm and epiblast could not model of early postimplantation development. be distinguished and appeared as cell clumps resembling the The KO EBs were significantly underdeveloped compared ICM of the blastocyst (Figure 1B). They did not develop an with WT EBs by day 8 (Figures S2A and S2B). Their parental epiblast epithelium (0/10), but wild-type (WT) embryos did (3/3; heterozygous ES clone was simultaneously subjected to the p < 0.001 by c2 test). The uterine reaction appeared to be also analyses to exclude the clonal artifact. Sixty-five percent of KO incomplete. Because the epiblast lineage is the primordium of EBs were <500 mm in diameter, while >50% of WT EBs were the three definitive germ layers of the fetus, KIF16B is considered >700 mm (n = 100, p < 0.001, c2 test). Histological examination to be essential for a fundamental process in embryonic of cultured EBs revealed that the morphology of cystic EBs development. was critically affected by KIF16B deficiency (Figures 2A and Development of the primitive endoderm and a basement 2B; Figures S2C and S2D). Although KO EBs first developed membrane derived from it was severely affected in KO embryos apparently normal simple EBs with primordial cysts by day 5 at 4.5 dpc. First, the expression of laminin surrounding the (Figure 2A, arrows), extensive cell death was observed by implanted region of KO embryos was greatly reduced (WT, 8/8; day 6 (Figure 2A, lower right panel). The percentage of EBs KO, 0/5; p < 0.001, c2 test) (Figures 1C and 1D, purple signals). carrying ectodermal epithelium over the whole EBs by day 10 Furthermore, immunohistochemistry against primitive endoderm derived from either of two independent lines of KO ESCs was markers Pem (Chazaud et al., 2006) and GATA6 (Li et al., 2004) significantly lower than that from WT cells (Figure 2B, n = 100, similarly revealed signals in cells facing the blastocyst cavities p < 0.01 by c2 test). Thus, it appeared that KO EBs initially formed or the blastocoele of WT ICM (n = 3 and 14, respectively), primordial ectoderm to some extent, but failed to undergo

Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. 61 Developmental Cell A Kinesin Transporting FGFR for Early Development

A B Figure 1. Defective Postimplantation Devel- WT KO opment of Embryos Directly Derived From Kif16b–/– ESCs (A and B) Histology of implanted Kif16b+/+ (WT; A) and Kif16b–/– (KO; B) embryos at 5.5 days post coitum (dpc) stages generated by tetraploid rescue. PEc, primitive ectoderm; ICM, inner cell mass; EM, endometrium. Note that KO primitive PEc ectoderm and primitive endoderm cannot be ICM distinguished. Bars represent 25 mm. (C–H) Immunohistochemistry of sections of pregnant uteri at 4.5 dpc stage with tetraploid-rescued

5.5 dpc HE embryos of indicated genotypes against laminin (C and D), Pem (E and F), and GATA6 (G and H). Arrowheads, inner cell mass. Note a significant decrease of signal intensities due to KIF16B deficiency. Bars represent 10 mm. ME ME (I) Immunoblotting of tetraploid-rescued blastocyst lysates with the indicated antibodies against differentiation markers, KIFs, and alpha tubulin (a-Tub). Note a significant decrease in the levels of GATA6, Oct4, Sox2 and KIF16B in KO lysates WT WT KO C D but not in those of Cdx2, KIF5B, and a-tubulin, suggesting defective ICM differentiation. (J) Schematic representation of essential cellular components in early postimplantation development. A basement membrane secreted from primitive endoderm lineage supports the survival and differentiation of primitive ectodermal cells. (K) Immunofluorescence microscopy of a tetra- 4.5 dpc Laminin ploid-rescued blastocyst against beta-galactosi- dase protein knocked into the Kif16b locus (Kif16b3loxP/+, clone D14-25), representing an expression of Kif16b gene in the inner cell mass (green). Bar represents 100 mm. E WT F KO G WT H KO

nin 1 and ﬁbronectin (Figures 2C–2E). A thin layer of basement membrane 4.5 dpc Pem

4.5 dpc GATA6 (marked by laminin and fibronectin) just beneath a layer of primitive endoderm at the surface of EBs (marked by GATA6) IJWT KO K was observed only in WT EBs, but not in Primitive ectoderm (epiblast) GATA6 supported by basement membrane KO EBs, which could be rescued by Oct4 ICM KIF16B cDNA (Figure 2E, the right panel). Sox2 ICM These data were consistent in all the EBs Cdx2 examined (n = 9/9 [WT] and 7/7 [KO] in KIF16B Primitive endoderm Figure 2C; 9/9 [WT] and 3/3 [KO] in KIF5B Figure 2D; 10/10 [WT], 10/10 [KO], and Basement membrane α-Tub assembled by primitive endoderm 11/11 [rescued] in Figure 2E; p < 0.01, by c2 test). Thus, a failure of basement membrane assembly due to a problem in correct differentiation of primitive correct epithelialization of the epiblast lineage. Simultaneously, endoderm could explain the cause of the extensive cell death apoptosis in KO epiblast lineage was significantly more exten- of KO epiblasts. sive than that in the WT cells according to a terminal deoxy- nucleotidyl transferase dUTP nick end labeling (TUNEL) assay Kif16b–/– Cells Show Defective Surface Presentation (Figure S3, n = 100, p < 0.01 by c2 test). These findings sug- of FGFR2 gested that KIF16B is essential for a survival and differentiation The reduction in the amount of basement membrane and GATA6 pathway for cells in the epiblast lineage. expression both in KO blastocysts and KO EBs (Figure 1 and 2) We labeled primitive endoderm and the basement membrane suggested an upstream signaling defect in differentiation of respectively using immunohistochemistry against GATA6, lami- primitive endoderm caused by the KIF16B deficiency. On this

62 Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. Developmental Cell A Kinesin Transporting FGFR for Early Development

ABC WT KO WT KO

Development of ectodermal epithelium

*** ***

Day 5 Day 100 GATA6 80

0 DIC Day 6 Day Cav WT KO1 KO2 Ectoderm - Number of embryoid bodies [%] Number of embryoid M VE Ectoderm +

DEWT KO WT KO Rescued Laminin 1 Fibronectin DIC DIC

Figure 2. Developmental Failure in Kif16b–/– Embryoid Bodies (A and B) Developmental analyses of embryoid bodies. (A) Histological examination of WT and KO EBs on days 5–6. Note only WT EBs developed epithelialized primitive ectoderm (arrowheads) on the surface of proamnionic cavity (Cav) from primordial cysts (arrows). Bar represents 100 mm. (B) Quantification of ectodermal epithelialization on day 10. KO1, clone C18. KO2, clone G46. ***p < 0.001 by c2 test, n = 100. (C–E) Immunohistochemistry of EBs of the indicated genotypes on day 5 against GATA6 (C) for primitive endoderm; and laminin 1 (D) and fibronectin (E) for basement membrane; with corresponding differential interference contrast (DIC) images (lower columns). These phenotypes were consistent among all EBs examined (n = 3–11). Note that expression of Kif16b cDNA resumed fibronectin signals (E, Rescued). Bars represent 100 mm.

occasion, we noticed that the developmental phenotype of We established clonal mouse embryonic fibroblasts (MEFs) Kif16b–/– embryos, including the morphology of implanted with Kif16b–/– (KO) and control Kif16b3lox/II (control; CT) geno- embryos and explanted blastocysts, was very similar to the types using chimeric embryos recovered at 13.5 dpc. The Fgfr2 KO embryo. Although partial KO of the immunoglobulin- expression of KIF16B protein in CT cells and its deficiency in like domain III of mouse Fgfr2 gene resulted in midgestation KO cells were verified with immunoblotting using a newly gener- lethality (Xu et al., 1998), complete disruption of the Fgfr2 ated specific antibody (Figure 3A). Two separate lines of KO gene in mice resulted in a recessive embryonic lethal mutation MEFs independently revealed the following results for FGFR in early postimplantation development (Arman et al., 1998). signaling and transport. Preimplantation development of Fgfr2À/À embryos was normal First, immunoblotting of FGF-stimulated cell lysates revealed until the blastocyst stage. However, they died within few hours a significant decrease in the levels of phospho-Akt [Thr308] after implantation, with the blastocoele collapsed. Mutant blas- (pAkt) and phospho-Erk1/2 [Thr202/Thr204] (pErk) throughout tocysts could hatch, adhere, and form a layer of trophoblast the examination period (Figure 3B), suggesting that the PI3K giant cells in vitro. After prolonged culture, the growth of the inner and MAPK pathways (Eswarakumar et al., 2005) downstream cell mass halted, no primitive endoderm formed and finally the of FGF signaling were both impaired in KO MEFs. Second, the egg cylinder disintegrated. Therefore, we investigated whether surface FGFR2 levels of KO MEFs were lower than those of FGFR signaling is really affected by KIF16B deficiency using WT MEFs in an immuno-isolation assay using a FGFR2 extracel- cell lines intrinsically expressing FGFR2 derived from the genet- lular domain-specific antibody, although the total levels of ically engineered embryos. FGFR2 were not altered (Figure 3C). Surface biotinylation of

Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. 63 Developmental Cell A Kinesin Transporting FGFR for Early Development

AB C D MEF Immunoblot FGF Signaling Assay Immunoisolation of Surface biotinylation CT KO Surface FGFR2 Mr CT KO [kDa] 200 0' 5' 10' 30' 60' 0' 5' 10' 30' 60' Whole Cell CT KO cells surface KIF16B p-Akt WT KO WT KO 116 Akt 97 FGFR2 Input FGFR2 NRG FGFR2 Input NRG 66 p-Erk IB: FGFR1 FGFR2 Dye front Erk N-Cadherin KIF5B

E F CT KO1 ECFP FGFR2 MEF/FGFR2

ECFP-KIF16B FGFR2

G *** *** *** 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 ECFP- (Perinuclear/Whole cell) ECFP CT KO1 KO2 KIF16B

Relative fluorescence intensity KO1

Figure 3. KIF16B Is Essential for FGFR2 Localization and Signaling (A) Expression of KIF16B in MEF cell lines. Immunoblotting of cell lysates with Kif16b3loxP/II (control, CT) and Kif16b–/– (KO) genotypes against KIF16B and KIF5B is presented. Note the KIF16B signal specifically labeled in CT lysates. (B) FGF signaling assay of MEFs on the time course. Activation profiles of PI3K and MAPK pathways were respectively detected using immunoblotting against phosphorylated and total Akt and Erk. Note that KIF16B deficiency affected both signaling pathways downstream to FGFRs. (C) Immuno-isolation of MEF cell surface FGFR2 by an FGFR2 antibody or normal rabbit IgG (NRG), detected by immunoblotting (IB). Note a decrease in surface FGFR2 level of KO cells. (D) Assessment of surface proteins of MEFs by biotinylation assay. Note that the surface levels of FGFR1 and FGFR2 but not that of N-cadherin are decreased in KO cells. (E–G) Distribution of FGFR2 in the MEF cytoplasm. (E) Immunofluorescence microscopy of MEFs against FGFR2 with the cell margin traced using a camera lucida. Inset, low magnification view. Bar represents 20 mm. (F) Rescue experiment of the KO1 MEFs with ECFP or ECFP-KIF16B expression vectors, labeled for FGFR2 by immunofluorescence microscopy. Arrowheads, transfected cells. Note a reversal of perinuclear clustering (lower right panel). Bar represents 20 mm. (G) Quantification of the data from (E) and (F) by relative fluorescence intensity of the perinuclear region. CT, clone D8; KO1, clone D8-2; KO2, clone D5-15. ***p < 0.001 by c2 test, n = 10. cultured MEFs further supported this result (Figure 3D), in which vesicular structures throughout the cytoplasm, as with CT surface FGFR1 and FGFR2 levels, but not surface N-cadherin MEFs (Figures 3F and 3G), suggesting that KIF16B is essential level, were specifically downregulated by KIF16B deficiency. and sufficient for the correct distribution of FGFR2, which is Third, immunocytochemistry of MEFs revealed a significant critical for FGFR signal transduction. change in FGFR2 distribution (Figures 3E and 3G). In CT MEFs, FGFR2 was distributed in punctate and vesicular structures KIF16B Is Associated with FGFR2 Vesicles throughout the cytoplasm; by contrast, in KO MEFs, it was accu- To pursue the relationship between FGFR2 transport and mulated in perinuclear dots, suggesting that the FGFR2 trans- KIF16B, we examined the association of FGFR2-containing vesi- port toward the cell periphery was impaired. This centrifugal cles with KIF16B by biochemistry and live cell imaging. direction of transport was consistent with the microtubule- First, we fractionated low-density membrane organelles using plus-end-directed motility of the kinesin-3 family of proteins Nycodenz density gradient centrifugation. KIF16B, FGFR2 and (Hirokawa et al., 2009). Finally, as a rescue experiment, ECFP- EGFR on vesicles were found in the same peak fraction (Fig- KIF16B was transfected into KO MEFs. In KIF16B-rescued ure 4A, fraction 13, arrow), other than membrane peak fractions KO MEFs, FGFR2 restored the distribution in punctate and of trans-Golgi vesicles or lysosomes. These data suggested that

64 Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. Developmental Cell A Kinesin Transporting FGFR for Early Development

ABFloatation Vesicle IP Figure 4. Association and Comigration of KIF16B with FGFR2-Containing Exocytotic IP: Bottom Top Vesicles (A) Nycodenz density gradient centrifugation. The Input Input NRIgG KIF16B FGFR2 NMIgG KIF3A post-nuclear supernatant (Input) of HeLa cells KIF16B IB: was separated using Nycodenz density gradient FGFR2 KIF16B EGFR centrifugation into 20 fractions. Note that KIF16B cofractionated with the FGFR2- and EGFR-con- TGN38 FGFR2 LAMP-1 taining vesicles. The trans-Golgi network (TGN) KIF3A 1234567891011121314151617181920 marker TGN38 and the lysosome marker LAMP- 1 did not cofractionate with KIF16B. Arrow, a co- C D ﬂoating fraction. FGFR2-ECFP EYFP-KIF16B DIC [sec] Merge [sec] Merge (B) Vesicle immunoprecipitation (IP) from MEFs. Postnuclear supernatants (input) were precipi- 0 tated with normal rabbit IgG (NRIgG), normal Nuc G 0 mouse IgG (NMIgG), or the indicated antibodies and probed with KIF16B, FGFR2, and KIF3A anti- 5 27 bodies by immunoblotting (IB). Note that KIF16B and FGFR2 coprecipitated with each other, but KIF3A did not coprecipitate with FGFR2. 10 54 (C and D) Comigration of KIF16B and FGFR2 on anterograde transported vesicles (arrowheads) 81 from the perinuclear region (C) toward a hub-like 15 peripheral structure (D, blue arrow). Time-lapse microscopy of a COS-7 cell cotransfected with 107 FGFR2-EYFP (green) and ECFP-KIF16B (red) is indicated, corresponding to Movies S1–S3. 20 134 Arrowhead, a doubly labeled organelle departing from the Golgi area. Nuc, nucleus. G, Golgi area. Bars represent 2 mm. 40 161 (E) Exocytosis of FGFR2-EYFP from hub-like structures observed by time-lapse TIRF micros- 188 copy of a serum-starved MEF cell corresponding 60 to Movie S4. Arrows, abrupt diffusion events from spot-like signals. Arrowhead, a remaining 215 spot as an internal control. Bar represents 2 mm.

E 0 10 20 30 40 [sec] arrows in Figure 4D; Movie S3). Time- lapse TIRF microscopy of FGFR2-EYFP- transfected CT MEFs further suggested that these subplasmalemmal compo- FGFR2-EYFP nents with an endosome-like appearance tended to undergo exocytosis triggered by serum deprivation (Figure 4E; Movies S4 and S5). Thus KIF16B may serve for KIF16B was associated with vesicles that might contain FGFR post-Golgi transport toward a hub, and probably an actin-depen- and/or EGFR. Second, we performed vesicle immunoprecipita- dent mechanism will regulate the exocytosis of FGFR2 from the tion from control MEFs. FGFR2 was specifically coprecipitated hub to the plasma membrane. with a KIF16B antibody, but not with negative controls. These observations suggest that KIF16B is associated with Moreover, KIF16B was coprecipitated with the FGFR2 antibody FGFR2-carrying vesicles and transports them from the Golgi (Figure 4B). These experiments suggested that FGFR2-contain- toward a hub-like apparatus below the plasma membrane. ing vesicles are specifically recognized by KIF16B. To examine comigration, enhanced-cyan-fluorescent-protein KIF16B Binds Directly to Rab14-GTP and Transports It (ECFP)-tagged FGFR2 (FGFR2-ECFP) and enhanced-yellow- A direct mechanistic link between KIF16B and vesicles was fluorescent-protein (EYFP)-tagged KIF16B (EYFP-KIF16B) were investigated using a yeast two-hybrid assay with the tail domain cotransfected into cells and observed using confocal micros- of KIF16B (amino acids 1068–1323) as a bait. Two independent copy. Colocalizing punctate signals were distributed throughout clones of Rab14 were found to bind directly to this domain the cytoplasm, with a density in the perinuclear Golgi region (Fig- (Figure 5A). This protein was considered to be a good candidate ure 4C). Time lapse analyses revealed that these colocalizing to regulate the binding between the motor and vesicles, because dots left the Golgi region (Movie S1), moved mostly toward the Rab14 has been reported to be a specific regulator of the Golgi- plasma membrane (Movie S2) and again accumulated in to-endosome pathway and essential for apical targeting of raft- a specific ‘‘hub’’-like component inside the plasmalemma (blue associated proteins (Junutula et al., 2004; Kitt et al., 2008).

Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. 65 Developmental Cell A Kinesin Transporting FGFR for Early Development

A B Lysate IP C Lysate IP Figure 5. KIF16B Binds Directly to the GTP IP: Form of Rab14 Yeast Two Hybrid NRIgG KIF16B

Input (A) Yeast two-hybrid assay. Rab14 was identiﬁed IP: GTP-γ-S - + - + - KISc FHA Coiled-coil PX to bind directly to the tail domain of KIF16B (amino GDP-β-S - - + - + KIF16B IB: acids 1068–1323). 1 366477529608 1032119012921323 IB: KIF16B Input KIF16B NRIgG FLAG NMIgG (B) Lysate IP of HeLa cells transfected with the FLAG- KIF16B Rab14 Bait indicated FLAG-Rab14 constructs detected by Rab14QL Rab4 FLAG IB. Note that KIF16B was preferentially associated Rab5 Rab14 FLAG- KIF16B with GTP-bound Rab14 (Rab14QL), but not with Rab8 GDP-bound Rab14 (Rab14SN). Rab14SN FLAG Rab11 (C) Lysate IP of HeLa cells precipitated with an a KIF16B antibody or normal rabbit IgG under D E conditions in the presence of GTP-g-S or GDP- EGFP- KIF16B Merge b-S, respectively. They are probed with indicated Rab14QL *** Rab antibodies. Note KIF16B is coprecipitated 80 with only Rab14 in the presence of GTP-g-S. (D and E) Fluorescence microscopy (D) of control 70 MEFs transfected with EGFP-Rab14 mutants 60 (green) and immunostained for KIF16B (red) and 50 its quantiﬁcation (E). Arrowheads in (D), colocalized particles. ***p < 0.001 by Welch’s t test, EGFP- KIF16B Merge 40 n = 20. Error bars indicate the standard error of Rab14SN 30 the mean (SEM). Scale bar represents 20 mm. 20 (F and G) Fluorescence images (F) of cells of

Colocalizing particles [%] 10 the indicated genotypes transfected with EGFP- Rab14 mutants and its quantiﬁcation (G). CT, 0 3loxP/II –/– SN QL Kif16b MEFs. KO, Kif16b MEFs. ***p < 0.001 by c2 test; n = 10 each. Scale bar represents 20 mm.

FGCT KO *** *** the cytoplasm, whereas Rab14SN was 0.9 not colocalized with KIF16B but was clus- 0.8 tered in the perinuclear region (Figures 5D 0.7 and 5E). These two consistent ﬁndings

EGFP-Rab14QL 0.6 suggested that KIF16B preferentially binds 0.5 to the GTP-bound form of Rab14 and 0.4 transports it; thus, KIF16B could be consid- 0.3 ered as a newly discovered ‘‘effector’’ of

(Perinuclear/Whole cell) 0.2 Rab14 GTPase.

Relative fluorescence intensity 0.1 To test the relevance of KIF16B and 0 the nucleotide cycle of Rab14 for the

EGFP-Rab14SN CT KO CT KO EGFP- EGFP- localization of Rab14, we transfected Rab14QL Rab14SN MEFs of each genotype with enhanced- green-fluorescent-protein (EGFP)-tagged Rab14QL or EGFP-Rab14SN via lipofec- To test the nucleotide dependency of KIF16B–Rab14 binding, tion. Rab14QL vesicles in KIF16B-KO MEFs and Rab14SN Rab14QL and Rab14SN mutants, respectively mimicking the vesicles in either genotype of the MEFs were significantly GTP- and GDP-bound forms of Rab14 (Junutula et al., 2004), more centrally distributed than Rab14QL vesicles in CT MEFs were tagged with FLAG and transfected into cells. Lysate (Figures 5F and 5G, p < 0.001 by c2 test). This may be because immunoprecipitation (IP) revealed that a KIF16B antibody specifi- dissociation from the plus-end-directed motor results in domi- cally coprecipitated Rab14QL, but not Rab14SN (Figure 5B), sug- nance of the minus-end-directed motor associated with the gesting that KIF16B had a binding preference for the GTP-bound organelle (Hendricks et al., 2010; Tanaka et al., 1998). Thus, form of Rab14. To test the specificity of this binding, we performed KIF16B appears to be an essential anterograde motor for coimmunoprecipitation of KIF16B with several Rab family Rab14-GTP but not for Rab14-GDP. members in the presence of GTP-g-S or GDP-b-S (Figure 5C). Lysate IP revealed that KIF16B was specifically coprecipitated Rab14 Localizes to FGFR2-Carrying Vesicles with Rab14 in the presence of GTP-g-S, but not in the presence and Regulates Vesicle-Motor Association of GDP-b-S. Furthermore, Rab4, 5, 8, and 11 were not coprecipi- Because the above two lines of evidence suggested indepen- tated with KIF16B, supporting the specificity of this binding. Immu- dently that KIF16B is essential for transporting FGFR2 vesicles nofluorescence microscopy revealed that Rab14QL was signifi- and that KIF16B preferentially binds to Rab14-GTP, we sought cantly colocalized with KIF16B on vesicles distributed throughout to examine the relationship between these two cargos of KIF16B.

66 Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. Developmental Cell A Kinesin Transporting FGFR for Early Development

LQ41baR-PFGE 2RFGF egreM

NS41baR-PFGE 2RFGF egreM

BC D E Colocalization Distribution Vesicle IP MAPK signaling CT EGFP- Rab14-SN *** FLAG- FLAG- 100 0.9 Rab14-SN Rab14-QL 0' 5' 10' 30' 60' 0' 5' 10' 30' 60' 0.8 Erk 80 0.7 IP: IP: 0.6 60 p-Erk 0.5 40 0.4 EGFP-

Input KIF16B FLAG FGFR2 CT 0.3 IB: Input KIF16B FLAG FGFR2 Rab14-QL 20 0.2 KIF16B 0' 5' 10' 30' 60' 0' 5' 10' 30' 60' % of colocalization 0.1

(Perinuclear/Whole cell) FLAG 0 0 Erk

QL SN Relative fluorescence intensity QL SN FGFR2 p-Erk

Figure 6. Rab14 Is a Molecular Switch for FGFR2 Transport by KIF16B (A–C) Fluorescence microscopy of CT MEFs transfected with EGFP-Rab14 mutants (A, green) and immunostained for FGFR2 (A, red). Merged images are presented on the right in A. (B) Colocalization statistics between Rab14 and FGFR2 signals. Note 80% of colocalization irrespective of the nucleotide state. n = 10. Error bars indicate the SEM. (C) Quantiﬁcation of FGFR2 distribution. ***p < 0.001 by c2 test; n = 10 each. QL, EGFP-Rab14QL-transfected MEFs. SN, EGFP- Rab14SN-transfected MEFs. (D) Vesicle IP of HeLa cells transfected with FLAG-tagged Rab14 mutants. Note that the binding capacity between Rab14 and KIF16B, but not Rab14 and FGFR2, is dependent on the nucleotide state of Rab14. (E) Time course of FGF signaling. Immunoblotting of transduced MEF lysates with the indicated constructs after the indicated time of stimulation with basic FGF, labeled against Erk and phospho-Erk (p-Erk). Note that EGFP-Rab14-SN dominant negatively inhibited the transduction of FGF signaling.

First, we transfected CT MEFs with EGFP-tagged Rab14 coprecipitated with both FLAG-tagged Rab14QL and Rab14SN mutant proteins and examined the distribution of Rab14 and proteins (Figure 6D), suggesting that the nucleotide state of FGFR2 using confocal microscopy (Figures 6A–C). FGFR2 and Rab14 does not essentially modulate its binding to FGFR2-con- Rab14 were clearly colocalized in both cases (arrowheads in taining vesicles. However, the Rab14/FGFR2-containing vesi- Figure 6A and statistics in Figure 6B). However, the colocalized cles coprecipitated with KIF16B only when Rab14 was in its signals tended to be dispersed throughout the cytoplasm and GTP-bound form; thus, Rab14 appears to critically modulate detected at the cell periphery in the case of Rab14QL, but this motor-cargo association. were clustered at the cell center in the case of Rab14SN (Figures Finally, we examined whether the nucleotide state of Rab14 6A and 6C). Thus, the nucleotide states of Rab14 was veriﬁed to could modulate the levels of FGF signaling as a consequence modulate the distribution of cargo vesicles containing Rab14 of altered FGFR transport. Cells were transduced via lentivirus itself and FGFR2. with EGFP-tagged Rab14 mutant proteins and a time course of Second, vesicle IP was conducted using cell lysates trans- pErk levels was measured by immunoblotting after FGF stimula- fected with FLAG-tagged Rab14 mutant proteins. FGFR2 tion (Figure 6E). As a result, the stimulated pErk levels of

Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. 67 Developmental Cell A Kinesin Transporting FGFR for Early Development

A B C EGFP EGFP-Rab14SN Figure 7. Dominant-Negative Rab14 Mimics KIF16B Deficiency in Embryoid Body Devel- opment (A) Characterization of stably transfected ESCs. IB: Anti-GFP Immunoblotting (IB) of the lysates of nontrans- fected (WT), EGFP- (Clone A-20), and EGFP- EGFP Rab14SN-expressing (Clone B-8) ESC lines using a GFP antibody. Anti-fibronectin WT A-20 (EGFP) B-8 (EGFP-Rab14SN) (B–F) Examination of the effect of dominant-nega- [kDa] tive Rab14 overexpression on EB differentiation by 50 general appearance on day 8 (B and E), immunos- taining against fibronectin on day 5 (C), and 37 histology on day 8 (D and F). Note that development of EBs was significantly affected by express-

25 DIC/GFP

EGFP-Rab14SN ing EGFP-Rab14SN compared with control. Bars represent 1 mm (B) and 100 mm (C and D). ***p < 0.001 by c2 test, n = 100. (G) Schematic diagram of KIF16B function. KIF16B-mediated D E F transport of FGFR2-containing vesicles is dynam- Size of EBs Development of ically regulated by the nucleotide state of Rab14 ectodermal epithelium 100 *** and modulates FGF signaling. 100 *** 80 80 60 60 40 in vivo relevance of the nucleotide states 40 of Rab14 in developmental regulation 20 20 (Figures 7B–7F). First, although both ESCs could form 0 EGFP EGFP- 0

Number of embryoid bodies [%] Number of embryoid Rab14SN EGFP EGFP- normal simple EBs, clone B-8 expressing < 500 μm Rab14SN 500 - 700 μm Ectoderm - EGFP-Rab14SN was significantly > 700 μm Ectoderm + retarded in developing cystic EBs compared with the control clone A-20 G FGFR vesicle on day 8 (Figures 7B and 7E; n = 100, p < 0.001 by c2 test). Second, clone B-8 FGFR vesicle Rab14-GDP significantly impaired the assembly of the basement membrane of egg cylinders KIF16B Rab14-GTP (0/10), compared with clone A-20 (Fig- EGFP-Rab14SN EGFP 2 KIF16B ure 7C, 10/10; p < 0.001 by c test). Finally, A-20 EBs developed basophilic (-) Microtubules (+) ectodermal epithelium on day 8, but B-8 EBs failed to do so (Figures 7D and 7F; n = 100, p < 0.01 by c2 test). Because Rab14SN-transduced cells were significantly lower than those these phenotypes were in parallel to those of KIF16B-deficient in nontransduced or Rab14QL-transduced cells, revealing embryos (Figure 1), KIF16B-deficient EBs (Figure 2), and that Rab14-GDP overexpression can dominant-negatively previously described FGFR2-dominant-negative EBs (Li et al., downregulate FGF signaling. These in vitro data suggested 2004; Li et al., 2001); and that Rab14 has been suggested that conditional association of Rab14 with KIF16B motor acts to be an essential dynamic adaptor between KIF16B and as a molecular switch regulating FGF signal transduction. FGFR2-carrying vesicles from the above experiments in vitro, Rab14-mediated regulation of biosynthetic KIF16B-mediated Rab14 Critically Regulates FGF Signaling FGFR2 transport will be essential in embryonic development. for Early Embryonic Development These data collectively suggest that KIF16B/Rab14-mediated To investigate the significance of transport regulation by Rab14 Golgi-to-endosome transport of FGFR2-carrying vesicles criti- in vivo, we examined whether the dominant negative effect of cally regulates both the surface presentation of FGFR2 and Rab14-GDP mutant in FGFR signal transduction further affected FGF signal transduction, which is essential for embryogenesis embryonic development in an EB formation assay. We estab- in early postimplantation development (Figure 7G). lished ESC clones stably expressing EGFP-Rab14SN (Clone B-8) and EGFP only (Clone A-20). We verified that both of these DISCUSSION clones exhibited EGFP signals strongly, using fluorescence microscopy (not shown). Using immunoblotting of the cell A Newly Discovered Biosynthetic Pathway lysates, we confirmed expected mobility of the respective for the FGF Receptor recombinant proteins in SDS-PAGE (Figure 7A). Then, these In this study, we have presented evidence that the KIF16B/Rab14 clones were subjected to EB formation assays to test the complex is responsible for biosynthetic Golgi-to-endosome

68 Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. Developmental Cell A Kinesin Transporting FGFR for Early Development

trafficking of the FGF receptor both in vitro and in vivo. This will be activity of AS160, a GTPase-activating protein (GAP) for Rab14 the first report that the motor of Golgi-to-endosome trafficking (Ishikura et al., 2007; Miinea et al., 2005). This suggests that has been identified, and the first to reveal the molecular mecha- Akt positively regulates Rab14-mediated biosynthetic FGF nism of the FGF receptor’s biosynthetic pathway. Because receptor traffic. Because Akt is activated by FGFR signaling, KIF16B has a PX domain in its C-terminal domain, its association this could partially explain the molecular mechanism of autocrine with endosomes has been suggested (Hoepfner et al., 2005). or paracrine loops in cancers where FGF signals are responsible However, the precise characteristics have now been revealed (Turner and Grose, 2010). Although the early developmental by our immuno-isolation of the cargo organelles and by identi- phenotypes could be solely explained by deficiency in FGFR2 fying Rab14 as its direct binding partner using a yeast two-hybrid traffic, it is also likely that this molecular machinery is responsible assay. Previous reports suggesting that Rab14 is a regulator of for modulating other receptor tyrosine kinase signaling than Golgi-to-endosome biosynthetic transport (Junutula et al., FGFR2 in later stages. Conditional inactivation of the Kif16b 2004; Kitt et al., 2008) is consistent with our observation that gene should reveal other signaling processes involving KIF16B is involved in post-Golgi transport of FGFR2 toward KIF16B/Rab14 in adult tissues and cancers. a biosynthetic endosome-like peripheral structure. Because essential housekeeping components of the endosomal system Molecular-Motor-Mediated Membrane Traffic Promotes might be transported simultaneously, the disruption of KIF16B’s a Developmental Signaling function can affect the kinetics of the whole endosome system, Early lineage segregation between primitive ectoderm (epiblast) as reported by Hoepfner et al. (2005). The precise nature of this and primitive endoderm represents the first indication of transporting organelle associated with both KIF16B and FGFR2 dorsoventral specification in mouse embryogenesis. The devel- was directly visualized by our dual-color time-lapse microscopy opmental fates of these two lineages are mutually exclusive and (Figure 4). We could observe those organelles that are derived are reported to be segregated by the 3.5 dpc peri-implantation from the Golgi region, and transported toward a hub beneath blastocyst stage (Chazaud et al., 2006). FGF4/FGFR2/MAPK the plasma membrane. This is one of the first visualizations signaling has been suggested to be the earliest inductive of the process by which the newly synthesized FGF receptor is signal in the post-implantation embryo (Ang and Constam, actually transported toward the plasma membrane by a respon- 2004; Feldman et al., 1995). Loss of the primitive endoderm sible kinesin. lineage has been observed similarly in FGFR2 KO embryos The hub-like structure is likely to comprise recycling endo- (Arman et al., 1998), FGFR2-dominant-negative EBs (Li et al., somes (RE) rather than early endosomes (Ang et al., 2004) and 2001, 2004), FGF4 KO embryos (Feldman et al., 1995; Wilder to be responsible for polarized sorting for the apical membrane et al., 1997), Grb2 KO embryos (Chazaud et al., 2006) and blas- (Kitt et al., 2008), because our TIRF analyses suggested that tocysts treated with FGF/MAPK-inhibitors (Yamanaka et al., exocytosis occurs through this structure (Figure 4E; Movies S4 2010). Intriguingly, KIF16B KO embryos, KIF16B KO EBs and and S5). RE activity has been suggested to be essential for the Rab14 dominant-negative EBs similarly revealed this phenotype surface presentation of apical proteins (Ang et al., 2004). Thus and reduced FGFR/MAPK signaling, suggesting that KIF16B/ the involvement of RE in this biosynthetic pathway will be needed Rab14-mediated vesicular traffic is a newly discovered essential for the apical sorting of FGFR2, for which the mechanism has component for FGFR2 signaling in germ layer formation. long been elusive (Yan et al., 2005). Because KIF16B motor Because activin-A signaling rather than FGF signaling appears appeared not to reach the plasma membrane in our observa- to be essential for mouse epiblast stem cell development (Greber tions, reduction of the surface level of FGFR2 by the inhibition et al., 2010), the phenotype in survival and differentiation in of KIF16B-mediated traffic will be a close consequence of the KIF16B/Rab14/FGFR2/MAPK-ablated embryos was considered reduction in Golgi-to-endosome traffic, which might in turn to be secondary to the impaired differentiation of the primitive reduce the endosome-to-cell-surface traffic of FGFR2. What endoderm lineage. It has been supposed that FGF induces the regulates the endosome-to-cell-surface traffic should be differentiation of primitive endoderm, to express GATA6, and a subject of future study. Our preliminary observation that serum to assemble the basement membrane. Here laminin-1 signaling starvation tended to increase the exocytosis and that FGF changes the fate of epiblast cells into survival and differentiation stimulation increased the tendency of endocytosis of surface (Li et al., 2001, 2004). Our present results, showing that KIF16B- receptors (not shown) will give a hint for such a regulation mech- or Rab14-ablated models similarly lacked this basement anism. If this is dependent on myosin-V-based short-range membrane, and that supplementation of KIF16B restored the motility (Kogel et al., 2010; Wang et al., 2010) or myosin-II-based phenotype in KO EBs, strongly support this hypothesis. It also actin reorganization (Aoki et al., 2010; Doreian et al., 2008; suggests that KIF16B/Rab14-mediated traffic is an essential Mochida et al., 1994; Neco et al., 2008) as suggested in other component of this FGFR signaling pathway. systems, the advantage of transporting to the RE nearby the The specification of epiblast and primitive endoderm lineages plasma membrane by microtubule motors will be further is a critical event in peri-implantation embryos and precedes clarified. gastrulation and formation of the anteroposterior and left–right The GTP/GDP cycle of Rab14 GTPase is a candidate molec- axes. The epiblast lineage is the primordium of the embryo ular switch for receptor tyrosine kinase signaling, because proper and will differentiate into the definitive ectoderm/meso- overexpression of a dominant negative form of Rab14 could derm/endoderm layers during gastrulation. Thus, the KIF16B/ downregulate FGF signal transduction, mimicking the pheno- Rab14 complex serves as a fundamental process for developing type of KIF16B null cells and embryos (Figures 6 and 7). It has all germ layers of the fetus, by regulating the biosynthetic been reported that Akt protein kinase negatively regulates the membrane traffic of FGFR2 directly. This provides one of the first

Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. 69 Developmental Cell A Kinesin Transporting FGFR for Early Development

lines of genetic evidence that microtubule-based membrane intermediates during transport from the Golgi to the plasma membrane of trafficking directly promotes signal transduction in early MDCK cells. J. Cell Biol. 167, 531–543. embryogenesis. Aoki, R., Kitaguchi, T., Oya, M., Yanagihara, Y., Sato, M., Miyawaki, A., and Tsuboi, T. (2010). Duration of fusion pore opening and the amount of hormone released are regulated by myosin II during kiss-and-run exocytosis. EXPERIMENTAL PROCEDURES Biochem. J. 429, 497–504. ESC Procedure Arman, E., Haffner-Krausz, R., Chen, Y., Heath, J.K., and Lonai, P. (1998). Kif16b3loxP/II and Kif16bÀ/À ESC lines were established by Cre/loxP-mediated Targeted disruption of fibroblast growth factor (FGF) receptor 2 suggests repeated homologous recombinations with a 3loxP targeting vector into the a role for FGF signaling in pregastrulation mammalian development. cell line CMTI-1 (Specialty Media). EGFP and EGFP-Rab14N expression Proc. Natl. Acad. Sci. USA 95, 5082–5087. vectors were also stably transfected to this cell line. Kif16b3loxP/II and Kif16bÀ/À Chazaud, C., Yamanaka, Y., Pawson, T., and Rossant, J. (2006). Early lineage MEFs were established from 13.5 dpc chimeric embryos generated by uterine segregation between epiblast and primitive endoderm in mouse blastocysts transfer of blastocysts injected with Kif16b3loxP/II ESCs. Kif16b+/+ and through the Grb2-MAPK pathway. Dev. Cell 10, 615–624. À/À 3loxP/II Kif16b embryos were generated by tetraploid rescue. Kif16b , Chen, Y., Li, X., Eswarakumar, V.P., Seger, R., and Lonai, P. (2000). Fibroblast +/+ À/À Kif16b , and Kif16b embryoid bodies were developed by aggregating growth factor (FGF) signaling through PI 3-kinase and Akt/PKB is required for the ESCs in vitro. Paraffin-embedded pregnant uteri were sectioned and embryoid body differentiation. Oncogene 19, 3750–3756. subjected to immunohistochemistry using an HRP-DAB system. Coucouvanis, E., and Martin, G.R. (1995). Signals for death and survival: a two- step mechanism for cavitation in the vertebrate embryo. Cell 83, 279–287. Yeast Two-Hybrid Assays Di Fiore, P.P., and De Camilli, P. (2001). Endocytosis and signaling: an insep- A mouse Kif16b cDNA fragment corresponding to amino acids 1068–1323 arable partnership. Cell 106, 1–4. was subjected as a bait to yeast two-hybrid assays using the Matchmaker Two-Hybrid System 3 (Clontech), to screen mouse kidney cDNA libraries Doetschman, T.C., Eistetter, H., Katz, M., Schmidt, W., and Kemler, R. (1985). (Clontech) of about 107 clones with a medium stringency selection. The in vitro development of blastocyst-derived embryonic stem cell lines: formation of visceral yolk sac, blood islands and myocardium. J. Embryol. Cell Surface Protein Assays Exp. Morphol. 87, 27–45. For immuno-isolation of surface proteins, MEFs were incubated with an Doreian, B.W., Fulop, T.G., and Smith, C.B. (2008). Myosin II activation and FGFR2 antibody or normal rabbit IgG (Invitrogen) in HBSS on ice for 30 min. actin reorganization regulate the mode of quantal exocytosis in mouse adrenal Their lysates were subjected to immunoprecipitation using protein A Sephar- chromaffin cells. J. Neurosci. 28, 4470–4478. ose 6MB beads (GE Healthcare). Surface biotinylation assay was performed Eswarakumar, V.P., Lax, I., and Schlessinger, J. (2005). Cellular signaling by using Pierce Cell Surface Protein Isolation Kit (Thermo scientific) following fibroblast growth factor receptors. Cytokine Growth Factor Rev. 16, 139–149. the manufacturer’s protocol. Feldman, B., Poueymirou, W., Papaioannou, V.E., DeChiara, T.M., and Goldfarb, M. (1995). Requirement of FGF-4 for postimplantation mouse devel- Microscopy opment. Science 267, 246–249. Immunocytochemistry and time lapse microscopy were performed using confocal laser scanning microscopes (Carl Zeiss, model LSM510 and George, E.L., Georges-Labouesse, E.N., Patel-King, R.S., Rayburn, H., and LSM710) or a TIRF microscope (Carl Zeiss) equipped with an EMCCD camera Hynes, R.O. (1993). Defects in mesoderm, neural tube and vascular develop- (Andor). The obtained images were quantified using ImageJ software (Abram- ment in mouse embryos lacking fibronectin. Development 119, 1079–1091. off et al., 2004). Greber, B., Wu, G., Bernemann, C., Joo, J.Y., Han, D.W., Ko, K., Tapia, N., Sabour, D., Sterneckert, J., Tesar, P., et al. (2010). Conserved and divergent SUPPLEMENTAL INFORMATION roles of FGF signaling in mouse epiblast stem cells and human embryonic stem cells. Cell Stem Cell 6, 215–226. Supplemental Information includes Supplemental Experimental Procedures, Hendricks, A.G., Perlson, E., Ross, J.L., Schroeder, H.W., 3rd, Tokito, M., and three figures, and five movies and can be found with this article online at Holzbaur, E.L. (2010). Motor coordination via a tug-of-war mechanism drives doi:10.1016/j.devcel.2010.11.008. bidirectional vesicle transport. Curr. Biol. 20, 697–702. Hirokawa, N., Noda, Y., and Okada, Y. (1998). Kinesin and dynein superfamily ACKNOWLEDGMENTS proteins in organelle transport and cell division. Curr. Opin. Cell Biol. 10, 60–73. Hirokawa, N., Noda, Y., Tanaka, Y., and Niwa, S. (2009). Kinesin super- The authors thank H. Fukuda, H. Sato, T. Aizawa, T. Akamatsu, N. Onouchi and family motor proteins and intracellular transport. Nat. Rev. Mol. Cell Biol. 10, others from the Hirokawa lab for technical assistance. We are also grateful 682–696. to Dr. R. Hynes for the fibronectin antibody. This study was supported by Hoepfner, S., Severin, F., Cabezas, A., Habermann, B., Runge, A., Gillooly, D., a Grant-in-Aid for Specially Promoted Research to N.H. and a Global COE Stenmark, H., and Zerial, M. (2005). Modulation of receptor recycling and program to the University of Tokyo from the Ministry of Education, Culture, degradation by the endosomal kinesin KIF16B. Cell 121, 437–450. Sports, Science and Technology of Japan. Ishikura, S., Bilan, P.J., and Klip, A. (2007). Rabs 8A and 14 are targets of the Received: May 19, 2010 insulin-regulated Rab-GAP AS160 regulating GLUT4 traffic in muscle cells. Revised: October 4, 2010 Biochem. Biophys. Res. Commun. 353, 1074–1079. Accepted: November 10, 2010 Junutula, J.R., De Maziere, A.M., Peden, A.A., Ervin, K.E., Advani, R.J., van Published: January 18, 2011 Dijk, S.M., Klumperman, J., and Scheller, R.H. (2004). Rab14 is involved in membrane trafficking between the Golgi complex and endosomes. REFERENCES Mol. Biol. Cell 15, 2218–2229. Kitt, K.N., Hernandez-Deviez, D., Ballantyne, S.D., Spiliotis, E.T., Casanova, Abramoff, M.D., Magelhaes, P.J., and Ram, S.J. (2004). Image processing with J.E., and Wilson, J.M. (2008). Rab14 regulates apical targeting in polarized ImageJ. Biophotonics Int. 11, 36–42. epithelial cells. Traffic 9, 1218–1231. Ang, S.L., and Constam, D.B. (2004). A gene network establishing polarity in Kogel, T., Rudolf, R., Hodneland, E., Hellwig, A., Kuznetsov, S.A., Seiler, F., the early mouse embryo. Semin. Cell Dev. Biol. 15, 555–561. Sollner, T.H., Barroso, J., and Gerdes, H.H. (2010). Distinct roles of myosin Ang, A.L., Taguchi, T., Francis, S., Folsch, H., Murrells, L.J., Pypaert, M., Va in membrane remodeling and exocytosis of secretory granules. Traffic Warren, G., and Mellman, I. (2004). Recycling endosomes can serve as 11, 637–650.

70 Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. Developmental Cell A Kinesin Transporting FGFR for Early Development

Li, X., Chen, Y., Scheele, S., Arman, E., Haffner-Krausz, R., Ekblom, P., and Tanaka, Y., Kanai, Y., Okada, Y., Nonaka, S., Takeda, S., Harada, A., and Lonai, P. (2001). Fibroblast growth factor signaling and basement membrane Hirokawa, N. (1998). Targeted disruption of mouse conventional kinesin assembly are connected during epithelial morphogenesis of the embryoid heavy chain, kif5B, results in abnormal perinuclear clustering of mitochondria. body. J. Cell Biol. 153, 811–822. Cell 93, 1147–1158.

Li, L., Arman, E., Ekblom, P., Edgar, D., Murray, P., and Lonai, P. (2004). Turner, N., and Grose, R. (2010). Fibroblast growth factor signalling: from Distinct GATA6- and laminin-dependent mechanisms regulate endodermal development to cancer. Nat. Rev. Cancer 10, 116–129. and ectodermal embryonic stem cell fates. Development 131, 5277–5286. Miinea, C.P., Sano, H., Kane, S., Sano, E., Fukuda, M., Peranen, J., Lane, W.S., Vale, R.D. (2003). The molecular motor toolbox for intracellular transport. Cell and Lienhard, G.E. (2005). AS160, the Akt substrate regulating GLUT4 112, 467–480. translocation, has a functional Rab GTPase-activating protein domain. Wang, Q., Uhlirova, M., and Bohmann, D. (2010). Spatial restriction of FGF Biochem. J. 391, 87–93. signaling by a matrix metalloprotease controls branching morphogenesis. Mochida, S., Kobayashi, H., Matsuda, Y., Yuda, Y., Muramoto, K., and Dev. Cell 18, 157–164. Nonomura, Y. (1994). Myosin II is involved in transmitter release at synapses Wilder, P.J., Mountjoy, C., Macleod, M.C., and Rizzino, A. (1997). DNA formed between rat sympathetic neurons in culture. Neuron 13, 1131–1142. sequence and nucleosome placement on the murine ﬁbroblast growth Nagy, A., and Rossant, J. (1993). Production of completely ES cell-derived factor-4 gene. DNA Seq. 7, 117–121. fetus. In Gene Targeting: A Practical Approach. A.L. Joyner, ed. (Oxford, UK: IRL Press), pp. 147–180. Xu, X., Weinstein, M., Li, C., Naski, M., Cohen, R.I., Ornitz, D.M., Leder, P., and Deng, C. (1998). Fibroblast growth factor receptor 2 (FGFR2)-mediated recip- Nakata, T., Terada, S., and Hirokawa, N. (1998). Visualization of the dynamics rocal regulation loop between FGF8 and FGF10 is essential for limb induction. of synaptic vesicle and plasma membrane proteins in living axons. J. Cell Biol. Development 125, 753–765. 140, 659–674. Neco, P., Fernandez-Peruchena, C., Navas, S., Gutierrez, L.M., de Toledo, Yamanaka, Y., Lanner, F., and Rossant, J. (2010). FGF signal-dependent G.A., and Ales, E. (2008). Myosin II contributes to fusion pore expansion during segregation of primitive endoderm and epiblast in the mouse blastocyst. exocytosis. J. Biol. Chem. 283, 10949–10957. Development 137, 715–724. Robertson, E.J., ed. (1987). Teratocarcinomas and Embryonic Stem Cells: A Yan, X., Yokote, H., Jing, X., Yao, L., Sawada, T., Zhang, Y., Liang, S., and Practical Approach (Oxford, UK: IRL Press). Sakaguchi, K. (2005). Fibroblast growth factor 23 reduces expression of Seto, E.S., Bellen, H.J., and Lloyd, T.E. (2002). When cell biology meets type IIa Na+/Pi co-transporter by signaling through a receptor functionally development: endocytic regulation of signaling pathways. Genes Dev. 16, distinct from the known FGFRs in opossum kidney cells. Genes Cells 10, 1314–1336. 489–502.

Developmental Cell 20, 60–71, January 18, 2011 ª2011 Elsevier Inc. 71