Correction

DEVELOPMENTAL BIOLOGYCorrection for “Solute carrier family 3 member 2 (Slc3a2) con-trols yolk syncytial layer (YSL) formation by regulating micro-tubule networks in the zebrafish embryo,” by Aya Takesono,Julian Moger, Sumera Faroq, Emma Cartwright, Igor B. Dawid,Stephen W. Wilson, and Tetsuhiro Kudoh, which appearedin issue 9, February 28, 2012, of Proc Natl Acad Sci USA(109:3371–3376; first published February 13, 2012; 10.1073/pnas.1200642109).The authors note that the author name Sumera Faroq should

instead appear as Sumera Farooq. The corrected author lineappears below. The online version has been corrected.

Aya Takesono, Julian Moger, Sumera Farooq,Emma Cartwright, Igor B. Dawid, Stephen W. Wilson,and Tetsuhiro Kudoh

www.pnas.org/cgi/doi/10.1073/pnas.1203335109

5548 | PNAS | April 3, 2012 | vol. 109 | no. 14 www.pnas.org

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0 D

ownl

oade

d by

gue

st o

n N

ovem

ber

1, 2

020

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0 D

ownl

oade

d by

gue

st o

n N

ovem

ber

1, 2

020

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0 D

ownl

oade

d by

gue

st o

n N

ovem

ber

1, 2

020

Dow

nloa

ded

by g

uest

on

Nov

embe

r 1,

202

0 D

ownl

oade

d by

gue

st o

n N

ovem

ber

1, 2

020

Solute carrier family 3 member 2 (Slc3a2) controlsyolk syncytial layer (YSL) formation by regulatingmicrotubule networks in the zebrafish embryoAya Takesonoa,1, Julian Mogerb, Sumera Farooqc, Emma Cartwrighta, Igor B. Dawidd,1, Stephen W. Wilsonc,and Tetsuhiro Kudoha,1

aBiosciences, College of Life and Environmental Sciences, University of Exeter, Exeter, EX4 4QD, United Kingdom; bPhysics and Astronomy, College ofEngineering, Mathematics and Physical Sciences, University of Exeter, Exeter, EX4 4QL, United Kingdom; cDepartment of Cell and Developmental Biology,University College London, London WC1E 6BT, United Kingdom; and dNational Institute of Child Health and Human Development, National Institutes ofHealth, Bethesda, MD 20892

Contributed by Igor B. Dawid, January 14, 2012 (sent for review September 12, 2011)

The yolk syncytial layer (YSL) in the zebrafish embryo is a multinu-cleated syncytium essential for embryo development, but themolecular mechanisms underlying YSL formation remain largelyunknown. Here we show that zebrafish solute carrier family 3member 2 (Slc3a2) is expressed specifically in the YSL and that slc3a2knockdown causes severe YSL defects including clustering of theyolk syncytial nuclei and enhanced cell fusion, accompanied by dis-ruption of microtubule networks. Expression of a constitutively ac-tive RhoA mimics the YSL phenotypes caused by slc3a2 knockdown,whereas attenuation of RhoA or ROCK activity rescues the slc3a2-knockdown phenotypes. Furthermore, slc3a2 knockdown signifi-cantly reduces tyrosine phosphorylation of c-Src, and overexpressionof a constitutively active Src restores the slc3a2-knockdown pheno-types. Our data demonstrate a signaling pathway regulating YSLformation in which Slc3a2 inhibits the RhoA/ROCK pathway viaphosphorylation of c-Src to modulate YSL microtubule dynamics.This work illuminates processes at a very early stage of zebrafishembryogenesis and more generally informs the mechanism of celldynamics during syncytium formation.

CD98 | epiboly | morphogenesis

The yolk syncytial layer (YSL) in the zebrafish embryo is anextraembryonic structure composed of a multinucleated syn-

cytium located at the margin between the yolk and the blastoderm(1). Formation of the YSL is one of the earliest differentiationevents in the embryo and plays a vital role in embryo development:The YSL is crucial in regulating dorso-anterior axis formation (2–6), epiboly movements involved in embryo patterning and mor-phogenesis (7–12), and cardiac progenitor cell movements (13).The YSL forms between the 512-cell and 1k-cell stage just

after the midblastula transition, the onset of zygotic transcrip-tion, when the marginal blastomeres undergo acute membranecollapse, thereby fusing to each other and to the adjoining yolkcell to form a thin syncytial ring around the blastodisc edge (14,15). The YSL contains microtubules emanating from microtu-bule asters adjacent to the individual yolk syncytial nuclei (YSN)and extending vertically toward the vegetal pole (7–9). Disrup-tion of the YSL microtubules by depolymerizing reagents such asnocodazole, UV irradiation, or cold temperature causes an al-tered and enlarged YSL and aggregation of the YSN, leading toa severe delay of epiboly (7–9). This effect suggests that themicrotubule networks in the YSL act as towing guides to directthe dynamic movements of the YSL and blastoderm toward thevegetal pole during epiboly (7).Loss of function of some genes highly expressed in the YSL,

such as β4.1 and β4.2 (voltage-gated Ca2+ channel β subunits)and cyp11a1 (a steroidogenic enzyme), cause aberrant YSN lo-calization (12) or altered YSL structure with unstable microtu-bule arrays (10), leading to a severe delay of epiboly. However,these genes do not have any known molecular link to down-stream pathways that modulate microtubules, and thus the

signaling pathways by which microtubule dynamics in the YSLare regulated remain undefined (10, 14, 12).A zebrafish slc3a2 gene (here named “slc3a2-a”) has been

reported based on genome sequence (16), and another slc3a2 gene,slc3a2-b, was found in our in situ-based gene-discovery screen (17).Zebrafish solute carrier family 3 member 2 (Slc3a2) proteins haveabout 60% sequence identity to the mammalian Slc3a2/4F2hc/CD98 heavy chain (CD98hc), an 85-kDa glycosylated type IImembrane protein, with conservation in intracellular, trans-membrane, and extracellular domains. Mammalian Slc3a2/4F2hc/CD98hc binds to integrin β1 and β3 subunits through its in-tracellular domain and hence is involved in various integrin-me-diated functions independent of the CD98 light chain (18–22).Mammalian Slc3a2/4F2hc/CD98hc regulates integrin-mediatedcell spreading, cell migration, and adhesion in mouse embryonicstem cells (19) and human placenta trophoblasts (21). Further-more, anti–CD98hc/FRP-1 antibodies regulate virus-mediated cellfusion (23, 24) and formation of multinucleated giant monocytes(25), and knockdown of Slc3a2/4F2hc/CD98hc inhibits placentalsyncytial trophoblast formation (26).Here we show that zebrafish Slc3a2 is indispensable for reg-

ulating microtubule dynamics in the YSL and that RhoA,ROCK, and Src act downstream of Slc3a2 in YSL formation.

ResultsZebrafish slc3a2 Is Expressed Specifically in the YSL. slc3a2-a andslc3a2-b transcripts were detected from the midblastula stageonward exclusively in the YSL. Expression of both genes in-creased through epiboly stages, when the mRNA was localizedpreferentially in the vicinity of the YSN (Fig. 1A), and continuedthrough somitogenesis stages. The YSL-specific expression ofslc3a2-b was confirmed by sectioning of in situ-stained embryos(Fig. S1). Specific expression of slc3a2 in the YSL suggests a rolein the function and/or formation of this structure.

Slc3a2 Is Required for Normal Embryonic Morphogenesis. To exam-ine the functions of Slc3a2 (hereafter, “slc3a2” refers to slc3a2-a and slc3a2-b) in embryogenesis, loss of function was achieved byinjecting morpholino (MO) antisense oligonucleotides againstslc3a2-a and slc3a2-b. Single knockdown of slc3a2-a or slc3a2-b induced relatively mild morphological defects (a shorter tail,smaller head) in embryos at 24 h postfertilization (hpf) (Fig. S2A,b and c). In contrast, double knockdown of both slc3a2-a andslc3a2-b (slc3a2-MOs) showed marked developmental defects at

Author contributions: A.T. and T.K. designed research; A.T., J.M., S.F., E.C., and T.K. per-formed research; J.M., I.B.D., S.W.W., and T.K. contributed new reagents/analytic tools;A.T. analyzed data; and A.T., J.M., I.B.D., S.W.W., and T.K. wrote the paper.

The authors declare no conflict of interest.1To whom correspondence may be addressed. E-mail: A.Takesono@exeter.ac.uk, idawid@mail.nih.gov, or t.kudoh@exeter.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1200642109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1200642109 PNAS | February 28, 2012 | vol. 109 | no. 9 | 3371–3376

DEV

ELOPM

ENTA

LBIOLO

GY

24 hpf (Fig. 1B and Fig. S2A, d) including lethality (38.1 ± 6.0%),a highly reduced head and truncated body axis (25.3 ± 4.8%), anda milder phenotype of a small head and short tail (21.6 ± 9.0%)(Fig. 1C). To confirm the specificity of the phenotypes caused bythe slc3a2-MOs, we used two different MOs for each slc3a2-a andslc3a2-b gene (Materials and Methods) and observed similarmorphological phenotypes (Fig. S2A, e). Rescue experiments arepresented below in the context of YSL phenotypes. These dataindicate slc3a2 is indispensable for normal embryogenesis inzebrafish and that there is at least partial functional redundancybetween the slc3a2-a and slc3a2-b paralogues.

Slc3a2 Regulates Epiboly Movements. To examine how loss offunction of Slc3a2 affects embryonic patterning, we analyzed theexpression of different marker genes by whole-mount in situ hy-bridization. sox3 (a neural ectoderm marker), no-tail (ntl, a me-sodermmarker), and sox17 (an endodermmarker) were expressedin their normal spatial relationships after slc3a2 knockdown, butthe embryos showed a marked delay of epiboly (Fig. 1D). At 24hpf, despite the severe morphological defects in slc3a2morphants(slc3a2-MOs), the expression domains of tissue-specific markers,such asmyoD in somitic muscle (Fig. S2C, a and b) and pax2 at theisthmus, optic stalk, otic vesicle, spinal cord, and pronephric duct(Fig. S2C, c and d), were comparable to those in control embryos.These data suggest that Slc3a2 is essential for gastrulationmovements and for embryonic morphogenesis rather than fortissue specification.

Slc3a2 Knockdown Enhances Plasma Membrane Fusion in the YSL andEnveloping Layer. Human SLC3a2/4F2hc/CD98hc regulates cellfusion in a variety of cell types (23–26). We therefore speculatedthat abrogation of Slc3a2 in the zebrafish embryo might alter

plasma membrane fusion during YSL formation. To examine thispossibility, plasma membrane dynamics during YSL formationwas analyzed using two-photon microscopy to image membrane-tagged GFP-labeled cells (27). Acquisition of both two-photonfluorescence (TPF) (for GFP) and coherent anti-Stokes Ramanscattering (CARS) (for detecting lipids) enabled visualization ofplasma membranes of the blastomeres and the yolk cellthroughout cell-division cycles. Consistent with previous reports(1), metasynchronous membrane fusion in the marginal blasto-meres was observed during transition between the 512- and 1k-cell stages in control embryos (Fig. 2A and Movie S1). For thesubsequent YSL division cycles (from the 1k-cell to the oblongstage), the positions of the YSN in control embryos did notchange; thus the YSN remained stationary underneath the blas-toderm margin (Fig. 2A and Movie S1). In contrast, in the slc3a2-MOs, cell–cell fusion continued until the oblong stage, resultingin enlargement of the YSL (Fig. 2 A and B). Some of the YSN inslc3a2-MOs were clustered (Fig. 2B) and moved freely around,unlike in control embryos (Movie S2). Interestingly, some of theenveloping layer (EVL) cells that share a common cellular originwith the YSL (15) were able to fuse to become syncytial cells bothin control and in slc3a2-MOs (Fig. 2B and Movies S2 and S3).However, the quantity of fused EVL cells and, consequently, thenumber of nuclei within the syncytial region were increased sig-nificantly in slc3a2-MOs (Fig. 2B). Collectively, these data in-dicate that the stage-specific cell-to-cell membrane-fusion eventsare disorganized in the absence of Slc3a2, suggesting that Slc3a2regulates YSL formation through the spatiotemporal regulationof membrane collapse during syncytium formation.

Slc3a2 Regulates the Distribution of the YSN. To examine furtherthe roles of Slc3a2 in YSL organization, the YSN were labeledfluorescently by injecting Sytox green into the embryos. Consis-tent with previous studies (9, 12, 28), the YSN in control embryoswere distributed evenly underneath the blastoderm (Fig. 3A, a),whereas the YSN in slc3a2-MOs often coalesced and were dis-tributed unevenly (Fig. 3A, b; see also Fig. 2). In most cases, theexternal YSL in slc3a2-MOs was expanded from the blastodermmargin, spreading toward the vegetal pole, as compared withcontrol embryos (Fig. 3A, b). The proportion of embryos showingnormal YSL organization therefore was greatly reduced inslc3a2-MOs (Fig. 3B and Table S1). We confirmed that the ab-normalities in YSL organization were caused specifically byslc3a2 knockdown, because overexpression of slc3a2-b not tar-geted by the MOs substantially rescued these phenotypes (Figs. 3A, e and B and Table S1). These data indicate that Slc3a2 reg-ulates the structure of the YSL and distribution of the YSN.

Intracellular Domain of Slc3a2 Is Required to Regulate YSNLocalization. To examine the contribution of functional domainsof Slc3a2 to the establishment of YSL structure and YSN locali-zation, we generated a putative dominant-negative form of Slc3a2-b (HN-Slc3a2-b) in which the intracellular domain was replaced bythe cytosolic domain (24 amino acids) of human parainfluenzavirus type-2 hemagglutinin-neuraminidase (29). Overexpression ofHN-Slc3a2-b caused morphological changes in 24-hpf embryoscomparable to those seen in slc3a2-MOs (Fig. S2A, f). Consistently,many of theYSN in theHN-slc3a2-b–overexpressing embryos wereclustered, and the YSL expanded toward the vegetal pole, againsimilar to the phenotype of slc3a2-MOs (Fig. 3 A, c and B). Thesedata support the conclusion that Slc3a2 acts through its in-tracellular domain to regulate YSL structure andYSN localization.

Slc3a2 Is Required to Maintain Microtubule Arrays in the YSL. Thickarrays of microtubules extend from the YSL toward the yolk alongthe animal–vegetal axis; disrupting these arrays by microtubuledepolymerizing reagents or UV irradiation causes premature YSLformation and altered YSL organization (Fig. S3) (7–9). Wetherefore examined the microtubules in the YSL by α-tubulin an-tibody staining. In control embryos, dense arrays ofmicrotubules inthe YSL were seen underneath the blastoderm edge, where the

Fig. 1. Zebrafish slc3a2 is expressed specifically in the YSL and is essentialfor embryo development. (A) Expression of zebrafish slc3a2-a and slc3a3-b atdifferent embryonic stages. Intensive slc3a2-a and slc3a2-b expression wasobserved around the YSN (arrowheads). (B) slc3a2 knockdown (MOs) causesa small head and short body axis at 24 hpf. Shown are examples of mild andsevere morphological phenotypes. (C) Morphology at 24 hpf was catego-rized as normal (white), mild (light gray), or severe (dark gray) phenotypespresenting a small head and short body axis or as dead (black). Data shownare from three experiments; the total number of embryos is shown beloweach bar. (D) slc3a2 knockdown induces severe epiboly delay. Whole-mountin situ analyses for sox3, no-tail (ntl), and sox17 at the late gastrula stage areshown. The distance between the animal pole and the blastoderm margin isindicated by a black bracket.

3372 | www.pnas.org/cgi/doi/10.1073/pnas.1200642109 Takesono et al.

YSN were dispersed uniformly (Fig. 3C, a–d). In contrast, YSLmicrotubule arrays in embryos injected withHN-slc3a2-b or slc3a2-MOs were disrupted, with regions devoid of microtubules (Fig. 3C,e, g–i, k, and l), and the YSNwere arranged irregularly (Fig. 3C, f, g,j, and k). We further confirmed that overexpression of slc3a2-b rescued microtubule networks in the YSL and YSN localizationin slc3a2-MOs (Fig. 3 C, m–p). These data indicate that Slc3a2 isrequired to retain microtubule networks in the YSL.

Inhibition of RhoA/ROCK Restores YSN Localization and YSLMicrotubule Arrays After slc3a2 Knockdown. Microtubule dynam-ics both regulate and are regulated by intracellular signalingpathways involving RhoA and the downstream kinase ROCK(30, 31). Using mammalian cell lines, we previously demon-strated an interrelationship between microtubules and RhoA/ROCK in which depolymerized microtubules activate RhoA,which in turn promotes depolymerization of microtubulesthrough ROCK (31). This negative regulation of microtubules byRho/ROCK signaling also has been shown to be critical for di-rectional migration of macrophages in live zebrafish embryos(32) as well as for contact inhibition of migrating chicken em-bryonic heart fibroblasts (33). We therefore hypothesized thatdisruption of microtubules in the YSL of slc3a2-MOs could becaused by increased RhoA/ROCK activity. Indeed, over-expression of a constitutively active form of RhoA (G14V) inzebrafish embryos induced severe YSL deformation and YSNaggregation, similar to the phenotypes in slc3a2-MOs (Fig. 3 A,d and B and Table S1). These observations led us to examinewhether inhibiting RhoA or ROCK activity could restore YSLstructure and YSN localization after slc3a2-MOs knockdown.Embryos microinjected with a mixture of MOs against slc3a2 andRhoA (MOs+RhoA MO) retained intact YSL structure in whichthe YSN were distributed uniformly underneath the blastodermmargin (Fig. 3 A, f and B and Table S1). Similarly, treatingslc3a2-MOs–injected embryos with a ROCK inhibitor, Y-27632,rescued the morphology of the YSL with even distribution of theYSN (Fig. 3 A, g and B and Table S1). We further examinedwhether the attenuation of RhoA or ROCK activity can restore

microtubule networks and found that microtubule arrays wererestored in embryos injected with MOs+RhoA-MO (Fig. 3C, q–t) as well as in the slc3a2-MOs–injected embryos treated with Y-27632 (Fig. 3C, u–x). These data indicate that Slc3a2 regulatesmicrotubule networks in the YSL and distribution of the YSN byinhibiting the RhoA/ROCK pathway.

Slc3a2 Inhibits RhoA/ROCK Pathway via c-Src Kinase. MammalianSLC3a2/4F2hc/CD98hc contributes to integrin outside–in signal-ing to regulate various integrin-mediated cell functions, includingcell adhesion and cell–cell fusion (19, 21–24). A downstream ef-fector of integrin outside–in signaling is c-Src, which is phos-phorylated at tyrosine 416 upon integrin–ligand binding (34).Because activation of c-Src transiently inhibits RhoA activity viap190RhoGAP (35), we hypothesized that zebrafish Slc3a2 mightinhibit RhoA via c-Src activation. To examine this possibility, weanalyzed the phosphorylation level of c-Src tyrosine 416 in slc3a2-MOs by Western blotting using c-Src phosphotyrosine–specificantibody. We found that slc3a2 knockdown substantially reducedtyrosine phosphorylation of c-Src (pY416) (Fig. 4A). Similarly, co-knockdown of Slc3a2 and RhoA also reduced the level of phos-phorylated c-Src, even though this treatment restored normalYSN distribution and YSLmicrotubule networks (Fig. 4A). Thesedata indicate that Slc3a2 is required for c-Src activation and thatRhoA acts downstream of these two molecules in the signalingcascade regulating YSL organization. Finally, we examinedwhether overexpression of a constitutively active c-Src (v-Src)could rescue YSL phenotypes in slc3a2-MOs–injected embryos.The level of v-Src mRNA, 1.25 pg per embryo, did not affectembryogenesis through 24 hpf.We found that overexpression of v-Src restored YSL deformation (YSN clustering and YSL expan-sion) (Figs. 3B and 4B) and YSL microtubules (Fig. 4C) in slc3a2-MOs–treated embryos. Together, these data indicate Slc3a2 reg-ulates c-Src to inhibit the RhoA/ROCK signaling pathway,thereby contributing to the even distribution of the YSN and tothe preservation of YSL microtubule networks.

Fig. 2. Loss of function of Slc3a2 enhances cell–cell fusion and YSN clustering. (A) Sequences of TPF time-lapse images showing GFP membrane (Upper) andoutlines of marginal cells and the YSL (Lower). Fusing cells are indicated by colored outlines, and fusion locations are indicated by colored arrows. The widthof the YSL at the outer edge is shown by a white bracket. Cell–cell fusion occurs mostly around the 512-cell stage (red arrow in control, red and green arrowsin MOs) but persists through the following stages in slc3a2-MOs (MOs) (pink and cyan arrows in MOs). (B) Merged images of CARS (for lipid; blue) and TPF (forGFP membrane; green) of control and MOs. Uniformly distributed YSN in the control are indicated by white arrowheads, and clustered YSN in MOs areindicated by red asterisks. Multinucleated EVL cells are indicated by red brackets.

Takesono et al. PNAS | February 28, 2012 | vol. 109 | no. 9 | 3373

DEV

ELOPM

ENTA

LBIOLO

GY

DiscussionIn this study, we demonstrate that Slc3a2 plays a critical role inYSLformation and YSL organization by regulating microtubule net-works in the YSL. We find that knockdown of slc3a2 causes thedisruption of microtubule networks in the YSL accompanied byclustering of the YSN and ectopic syncytium formation, leading tosevere morphological defects in the zebrafish embryo. By rescueexperiments, we found that all the slc3a2-knockdown phenotypesare rescued by attenuating RhoA orROCK activity, suggesting thatthese phenotypes are caused by increased RhoA/ROCK activity.Furthermore, we showed that slc3a2 knockdown reduces tyrosinephosphorylation of c-Src, which links to integrin-mediated RhoAinhibition (35), suggesting that Slc3a2 normally activates c-Src toinhibit RhoA during YSL formation. Consistently, overexpressionof constitutively active c-Src also rescues the YSL phenotypescaused by slc3a2 knockdown. Our data led us to construct a modelin which Slc3a2 inhibits a RhoA/ROCK signaling pathway via c-Srcactivation, thereby regulating YSL microtubules and YSL organi-zation (Fig. 5).Depolymerization of microtubules induces both premature

marginal blastomere fusion and YSN clustering (7–9), suggestingthat microtubule dynamics contribute to two events essential forYSL organization: cell–cell fusion and YSN localization (Fig. 5).Realignment of microtubules into parallel arrays along the axis ofthe fusing cells has been shown to be crucial for syncytial myotubeformation (36–38). These data support the view that microtubulescontrol the spatiotemporal aspects of plasma membrane fusion inYSL progenitors. How, then, can Slc3a2 modulate plasmamembrane dynamics via microtubules? It is known that micro-tubule dynamics affect the regional distribution of adhesionmolecules (38–40) and turnover of adherence junctions (41),thereby regulating adhesive strength and the integrity of cell–cellcontacts. One possibility is that the local distribution of Slc3a2regionally protects the microtubule networks, which in turn feedback on the localization of Slc3a2 or affect cellular distribution ofother adhesion molecules (e.g., E-cadherin), thereby regulatingthe plasma membrane dynamics. Indeed, we found that slc3a2knockdown appears to reduce the integrity of cell–cell contactsbetween the YSL and the marginal blastomeres; hence the mar-ginal cells actively crawl or tumble within the looser extracellularspace (Movie S2). In contrast, the marginal blastomeres in controlembryos are tightly attached to each other and to the YSL, leavingless extracellular space at the marginal area (Movies S1 and S3).We observed that not only cell–cell fusion in the YSL pro-

genitors but also YSN localization correlates strongly with theintegrity of microtubule networks in the YSL. The combined TPFand CARS time-lapse imaging demonstrates that the YSN movequickly in the enlarged YSL of slc3a2-MOs but are stationary incontrol embryos. The persistent and regularly distributed YSN incontrol embryos may ensure the direction of epiboly movements,because vertical microtubule arrays extending toward the vegetalpole emanate from microtubule asters close to the aligned

Fig. 3. Loss of function of Slc3a2 severely alters YSL organization, which isrestored by inhibition of RhoA or ROCK. (A) The YSN are visualized by Sytoxgreen. Shown are lateral views of dome-stage embryos from control (a);slc3a2-MOs (MOs) (b); HN-slc3a2-b (HNslc3a2b) (c); RhoA-G14V (G14V) (d);slc3a2-MOs+slc3a2-b (MOs+slc3a2b) (e); slc3a2-MOs+RhoA-MO (MOs+RhoAMO) (f); slc3a2-MOs + treatment with 50 μM Y-27632 (MOs+Y27632)(g); and control treated with 50 μM Y-27632 (Y27632) (h). Clustered YSN areindicated by white arrowheads. Note rescue of YSL phenotypes by coinjectingslc3a2-b-mRNA and RhoA-MO or by treatment with 50 μM Y-27632. (B) Lossof function of Slc3a2 severely alters YSL organization, which is restored byinhibition of RhoA or ROCK or by overexpression of v-Src. Phenotype in thedome-stage embryo is categorized as normal (white) showing uniform dis-tribution of the YSN without YSL deformation/expansion, abnormal (gray)showing altered YSN localization including uneven distribution of the YSN

and YSN clustering with YSL deformation/expansion, or dead (black). Dataare from more than three different experiments. Total number of embryos isshown below each bar. (C) Inhibition of RhoA or ROCK restores microtubulenetworks in the YSL in slc3a2-MOs. (a–d) control; (e–h) HNslc3a2b; (i–l) MOs;(m–p) MOs+slc3a2b; (q–t) MOs + RhoA-MO; (u–x) MOs + Y27632. Microtu-bule networks were visualized by α-tubulin antibody staining (a, e, i, m, q,and u), and the YSN were visualized by Sytox green (b, f, j, n, r, and v). Shownare merged images of α-tubulin (red) and Sytox green (green) (c, g, k, o, s,and w) and reconstituted section views of the merged images rotated by 90°(d, h, l, p, t, and x). Absence or disruption of microtubule arrays in the YSL isindicated by red arrowheads in e and i, and the clustered YSN are indicatedby green asterisks in f and j. Control shows overlap of the YSN and dense YSLmicrotubule networks (white square in d), which are missing in bothHNslc3a2b and MOs (h and l). YSL microtubule networks, YSN distribution,and overlapping of the YSN and YSL microtubules are all restored in MOs+slc3a2b, MOs+RhoA-MO, and MOs+Y27632 (m–p, q–t, and u–x).

3374 | www.pnas.org/cgi/doi/10.1073/pnas.1200642109 Takesono et al.

individual YSN (8, 9). The YSN might be associated with thedense YSL microtubule mesh and therefore persist underneaththe blastoderm margin. Indeed, α-tubulin staining shows that theYSN in control embryos locate with the dense microtubule net-works in the YSL, whereas the YSN in the slc3a2-MOs–injectedembryos are separated from the YSL microtubules.Our data suggest that Slc3a2 normally inhibits RhoA in the

YSL. This result might be similar to the previous report by Kabir-Salmani et al. (21) that stable expression of human CD98hc(Slc3a2, 4F2hc) in a CD98hc-null hepatocyte line facilitatesintegrin-mediated cell adhesion, which is accompanied by en-hanced transient inhibition in RhoA activity. In contrast, Féralet al. (20) have reported that CD98hc-null mouse embryonicfibroblasts fail to induce the integrin-mediated delayed activationof RhoA that follows transient RhoA inhibition and contributesto stress fiber formation and fibronectin matrix assembly. Thediscrepancy in observed effects of mammalian SLC3a2/4F2hc/CD98hc on RhoA activity could be the result of cell-type speci-ficity, so that SLC3a2/4F2hc/CD98hc can be influenced by and inturn affect different signaling molecules. Although mammalianSLC3a2/4F2hc/CD98hc is expressed in a wide variety of tissues(42), our in situ analyses revealed that zebrafish slc3a2 isexpressed specifically in the YSL in the embryo. Considering therelatively low similarity of zebrafish Slc3a2 with mammalian

SLC3a2/4F2hc/CD98hc (60%), it is possible that some functionaldiversifications may have occurred in different species.Mammalian SLC3a2 is involved in the formation of multinu-

cleated giant monocytes, placental syncytial trophoblast forma-tion, and virus-mediated cell–cell fusion. Thus it is highly possiblethat Slc3a2 regulates microtubule dynamics via RhoA/ROCKpathway in cell–cell fusion and syncytium formation in these cells(23–26).

Materials and MethodsZebrafish Strains and Maintenance. Wild-type and Tg[β-actin:GFP] strains ofzebrafish (27) were maintained according to standard procedures (1).

RNA Probe Synthesis and in Situ Hybridization. The procedures for whole-mount in situ hybridization have been described previously (17, 43).

Constructs. Details of plasmid constructions are described in SI Materialsand Methods.

Synthesis and Microinjection of mRNA. mRNAs were synthesized using themMessangemMachine SP6 kit (Ambion). One nanoliter of 150 pg/nL slc3a2-b-mRNA or HN-slc3a2-b-mRNA, 15 pg/nL RhoA-b-G14V-mRNA, or 1.25 pg/nLGFP-FL-v-Src-mRNA was injected through the intact chorion into the blas-tomere at the one-cell stage.

MO Analysis and Injection. The MOs against slc3a2-a (BC053256) (slc3a2-a-MO1: 5′-CCTTCATCTCGTCTTCTTTGTTCAT-3′ and slc3a2-a-MO2: 5′- TTTGTT-TGATAGTAGTTTCAGCACT-3′) and slc3a2-b [BC044497 (17)] (slc3a2-b-MO1:5′-CCTTCATCTCGTCTTCTTTGTTCAT-3′ and slc3a2-b-MO2: 5′-TATCCACTTCA-GTGTCGTTGCTCAT-3′) were purchased from GeneTools. The MO againstRhoA-b (RhoA-b-MO: 5′-TTCTTGCGAATTGCTGCCATATTTG-3′) was purchasedfrom Open Biosystems. Each MOwas used at 1:4 dilutions from a 1 mM stock.In all cases, 2 nL of solution was injected into the yolk proximal to the blas-tomeres of the embryo at the one- to four-cell stage. A combination of slc3a2-a-MO1 and slc3a2-b-MO1 (slc3a2-MOs) was used for all data shown except forthe data shown in Fig. S2 A and B.

Sytox Green Injection, Drug Treatment, and Whole-Mount Microscopy. To labelthe nuclei within the YSL, a total of ∼6 nL of 1 mM Sytox green (Invitrogen)was injected into the yolk at high stage (∼2 nL of Sytox green was injectedinto three or four different regions of the yolk proximal to the blastomeres).In some experiments, the embryos were treated with 50 μM of ROCK in-hibitor Y-27632 (Calbiochem) in embryo medium from the 256-cell stage,before YSL formation. YSL organization was analyzed at dome stage byfluorescent dissection microscopy (Nikon SMZ1500).

TPF and CARS Microscopy. Embryos from Tg[β-actin:GFP] fish were injectedwith slc3a2c-MOs as described above. Embryos then were dechorionatedand mounted onto glass-bottomed 35-mm dishes (MatTek) with 0.7% low-melting agarose in embryo medium. Embryos held in this condition con-tinued cleavage throughout the imaging period. TPF imaging and CARSmicroscopy were performed simultaneously; details have been describedpreviously (44, 45) and are described in SI Materials and Methods. Time-lapseimages of membrane-GFP dynamics were acquired using Flouroview v5

Fig. 4. Slc3a2 inhibits the RhoA/ROCK pathway via c-Src activation in reg-ulating YSL organization. (A) slc3a2 knockdown reduces tyrosine phos-phorylation of c-Src. Western blot shows pY416 c-Src and total c-Src in lysatesfrom control, slc3a2-morphants (MOs), or slc3a2/RhoA morphants (MOs+RhoA-MO). Data shown are representative of three independent experi-ments. (B) Overexpression of v-Src restores YSN localization after slc3a2knockdown. The YSN are visualized by Sytox green. Clustered YSN in MOs(white arrows) are restored by coinjecting v-Src mRNA (v-Src+MOs). (C)Overexpression of v-Src restores microtubule networks in the YSL in slc3a2-knockdown embryos. YSL microtubule networks (αTUB; red), YSN distribu-tion (SytG; green), and overlap of YSN and YSL microtubules (white squaresin merged images) are all restored by v-Src+MOs.

Fig. 5. Model for the Slc3a2-mediated signaling pathway regulating YSNlocalization and cell–cell fusion in the YSL. (A) Slc3a2 is involved in integrinoutside–in signaling, which activates c-Src, which in turn inhibits RhoA. Mi-crotubule arrays in the YSL then are retained, contributing to regular YSNlocalization and normal cell–cell fusion. (B) slc3a2 knockdown abolishes c-Srcactivation; hence, microtubule networks in the YSL are disrupted by in-creased RhoA/ROCK activity in the YSL, thereby causing YSN clustering andenhanced ectopic cell–cell fusion.

Takesono et al. PNAS | February 28, 2012 | vol. 109 | no. 9 | 3375

DEV

ELOPM

ENTA

LBIOLO

GY

software (Olympus). Stacks of frame-averaged (four-frame) sections werecollected and reconstructed by using Image J (http://rsbweb.nih.gov/ij/).

Immunostaining and Confocal Imaging. Embryos were injected with MOs ormRNAs, and in some experiments Sytox green was injected subsequently asdescribed above. Staining with anti–α-tubulin antibody (DM1A; Sigma) wasperformed as described by Gard (46). Stained samples were analyzed bya Zeiss LSM510 inverted confocal microscope. Z-stacks of line-averaged(eight-line) sections were obtained by scanning an area of 102.4 × 102.4 μm(0.1 μm pixel−1) with 6-μm steps over a total vertical distance of 180–240 μμ,and were reconstituted using LSM510 meta.

Western Blot. Embryos at the shield stage were lysed in cold lysis buffer [50mM Tris (pH 7.5), 10 mM MgCl2, 0.5 M NaCl, and 2% (vol/vol) Igepal con-

taining 1× protease inhibitor mixture and 50 mM NaVaO3] at 20 embryosper 200 μL lysis buffer. Lysates were clarified by centrifugation. Super-natants were mixed with lithium dodecyl sulfate (LDS) sample buffer (Invi-trogen), heated, and analyzed by SDS/PAGE. pY416 Phospho-Src and totalSrc (Cell Signaling) blots were performed according to the manufacturer’sinstructions.

ACKNOWLEDGMENTS. We thank Y. Ito for providing HN-cDNA, Y. Fujita forproviding pCS2-GFP-FL-v-Src, M. Tada and S. J. Heasman for critical commentson the manuscript, and the fish facility in Biosciences, University of Exeter forfish maintenance. This work was supported by Biotechnology and BiologicalSciences Research Council Grant BB/F010222/1 (to T.K.) and in part by theintramural research program of the National Institute of Child Health andHuman Development, National Institutes of Health (to I.B.D.) and WellcomeTrust funding (to S.W.W.).

1. Kimmel CB, Ballard WW, Kimmel SR, Ullmann B, Schilling TF (1995) Stages of em-bryonic development of the zebrafish. Dev Dyn 203:253–310.

2. Mizuno T, Yamanaka M, Wakahara A, Kuroiwa A, Takeda H (1996) Mesoderm in-duction in zebrafish. Nature 383(6596):131–132.

3. Hong SK, Jang MK, Brown JL, McBride AA, Feldman B (2011) Embryonic mesodermand endoderm induction requires the actions of non-embryonic Nodal-related ligandsand Mxtx2. Development 138:787–795.

4. Feldman B, et al. (1998) Zebrafish organizer development and germ-layer formationrequire nodal-related signals. Nature 395:181–185.

5. Solnica-Krezel L, et al. (1996) Mutations affecting cell fates and cellular rearrange-ments during gastrulation in zebrafish. Development 123:67–80.

6. Yamanaka Y, et al. (1998) A novel homeobox gene, dharma, can induce the organizerin a non-cell-autonomous manner. Genes Dev 12:2345–2353.

7. Strähle U, Jesuthasan S (1993) Ultraviolet irradiation impairs epiboly in zebrafishembryos: Evidence for a microtubule-dependent mechanism of epiboly. Development119:909–919.

8. Jesuthasan S, Stähle U (1997) Dynamic microtubules and specification of the zebrafishembryonic axis. Curr Biol 7:31–42.

9. Solnica-Krezel L, Driever W (1994) Microtubule arrays of the zebrafish yolk cell: Or-ganization and function during epiboly. Development 120:2443–2455.

10. Hsu HJ, Liang MR, Chen CT, Chung BC (2006) Pregnenolone stabilizes microtubulesand promotes zebrafish embryonic cell movement. Nature 439:480–483.

11. Lachnit M, Kur E, Driever W (2008) Alterations of the cytoskeleton in all three em-bryonic lineages contribute to the epiboly defect of Pou5f1/Oct4 deficient MZspgzebrafish embryos. Dev Biol 315:1–17.

12. Ebert AM, et al. (2008) Ca2+ channel-independent requirement for MAGUK familyCACNB4 genes in initiation of zebrafish epiboly. Proc Natl Acad Sci USA 105:198–203.

13. Kawahara A, et al. (2009) The sphingolipid transporter spns2 functions in migrationof zebrafish myocardial precursors. Science 323:524–527.

14. Carvalho L, Heisenberg CP (2010) The yolk syncytial layer in early zebrafish de-velopment. Trends Cell Biol 20:586–592.

15. Kimmel CB, Warga RM, Schilling TF (1990) Origin and organization of the zebrafishfate map. Development 108:581–594.

16. Yoder JA, Litman GW (2000) The zebrafish fth1, slc3a2, men1, pc, fgf3 and cycd1genes define two regions of conserved synteny between linkage group 7 and humanchromosome 11q13. Gene 261:235–242.

17. Kudoh T, et al. (2001) A gene expression screen in zebrafish embryogenesis. GenomeRes 11:1979–1987.

18. Fenczik CA, et al. (2001) Distinct domains of CD98hc regulate integrins and aminoacid transport. J Biol Chem 276:8746–8752.

19. Feral CC, et al. (2005) CD98hc (SLC3A2) mediates integrin signaling. Proc Natl Acad SciUSA 102:355–360.

20. Féral CC, et al. (2007) CD98hc (SLC3A2) participates in fibronectin matrix assembly bymediating integrin signaling. J Cell Biol 178:701–711.

21. Kabir-Salmani M, et al. (2008) The membrane-spanning domain of CD98 heavy chainpromotes alpha(v)beta3 integrin signals in human extravillous trophoblasts. MolEndocrinol 22:707–715.

22. Prager GW, Féral CC, Kim C, Han J, Ginsberg MH (2007) CD98hc (SLC3A2) interactionwith the integrin beta subunit cytoplasmic domain mediates adhesive signaling. J BiolChem 282:24477–24484.

23. Ohta H, et al. (1994) Molecular and biological characterization of fusion regulatoryproteins (FRPs): Anti-FRP mAbs induced HIV-mediated cell fusion via an integrin sys-tem. EMBO J 13:2044–2055.

24. Okamoto K, et al. (1997) An anti-fusion regulatory protein-1 monoclonal antibodysuppresses human parainfluenza virus type 2-induced cell fusion. J Gen Virol 78:83–89.

25. Tajima M, et al. (1999) Suppression of FRP-1/CD98-mediated multinucleated giant celland osteoclast formation by an anti-FRP-1/CD98 mAb, HBJ 127, that inhibits c-srcexpression. Cell Immunol 193:162–169.

26. Kudo Y, Boyd CA, Millo J, Sargent IL, Redman CW (2003) Manipulation of CD98 ex-pression affects both trophoblast cell fusion and amino acid transport activity duringsyncytialization of human placental BeWo cells. J Physiol 550:3–9.

27. Cooper MS, et al. (2005) Visualizing morphogenesis in transgenic zebrafish embryosusing BODIPY TR methyl ester dye as a vital counterstain for GFP. Dev Dyn 232:359–368.

28. D’Amico LA, Cooper MS (2001) Morphogenetic domains in the yolk syncytial layer ofaxiating zebrafish embryos. Dev Dyn 222:611–624.

29. Okamoto K, et al. (1997) Paramyxovirus-induced syncytium cell formation is sup-pressed by a dominant negative fusion regulatory protein-1 (FRP-1)/CD98 mutatedconstruct: An important role of FRP-1 in virus-induced cell fusion. J Gen Virol 78:775–783.

30. Birkenfeld J, Nalbant P, Yoon SH, Bokoch GM (2008) Cellular functions of GEF-H1,a microtubule-regulated Rho-GEF: Is altered GEF-H1 activity a crucial determinant ofdisease pathogenesis? Trends Cell Biol 18:210–219.

31. Takesono A, Heasman SJ, Wojciak-Stothard B, Garg R, Ridley AJ (2010) Microtubulesregulate migratory polarity through Rho/ROCK signaling in T cells. PLoS ONE 5:e8774.

32. Redd MJ, Kelly G, Dunn G, Way M, Martin P (2006) Imaging macrophage chemotaxisin vivo: Studies of microtubule function in zebrafish wound inflammation. Cell MotilCytoskeleton 63:415–422.

33. Kadir S, Astin JW, Tahtamouni L, Martin P, Nobes CD (2011) Microtubule remodellingis required for the front-rear polarity switch during contact inhibition of locomotion.J Cell Sci 124:2642–2653.

34. Gong H, et al. (2010) G protein subunit Galpha13 binds to integrin alphaIIbbeta3 andmediates integrin “outside-in” signaling. Science 327:340–343.

35. Arthur WT, Petch LA, Burridge K (2000) Integrin engagement suppresses RhoA ac-tivity via a c-Src-dependent mechanism. Curr Biol 10:719–722.

36. Pizon V, Gerbal F, Diaz CC, Karsenti E (2005) Microtubule-dependent transport andorganization of sarcomeric myosin during skeletal muscle differentiation. EMBO J 24:3781–3792.

37. Straube A, Merdes A (2007) EB3 regulates microtubule dynamics at the cell cortex andis required for myoblast elongation and fusion. Curr Biol 17:1318–1325.

38. Zhang T, et al. (2009) Microtubule plus-end binding protein EB1 is necessary formuscle cell differentiation, elongation and fusion. J Cell Sci 122:1401–1409.

39. Stehbens SJ, et al. (2006) Dynamic microtubules regulate the local concentration of E-cadherin at cell-cell contacts. J Cell Sci 119:1801–1811.

40. Zaoui K, Honoré S, Isnardon D, Braguer D, Badache A (2008) Memo-RhoA-mDia1signaling controls microtubules, the actin network, and adhesion site formation inmigrating cells. J Cell Biol 183:401–408.

41. Mary S, et al. (2002) Biogenesis of N-cadherin-dependent cell-cell contacts in livingfibroblasts is a microtubule-dependent kinesin-driven mechanism. Mol Biol Cell 13:285–301.

42. Devés R, Boyd CA (2000) Surface antigen CD98(4F2): Not a single membrane protein,but a family of proteins with multiple functions. J Membr Biol 173:165–177.

43. Kudoh T, Concha ML, Houart C, Dawid IB, Wilson SW (2004) Combinatorial Fgf andBmp signalling patterns the gastrula ectoderm into prospective neural and epidermaldomains. Development 131:3581–3592.

44. Moger J, Johnston BD, Tyler CR (2008) Imaging metal oxide nanoparticles in biologicalstructures with CARS microscopy. Opt Express 16:3408–3419.

45. Mansfield J, et al. (2009) The elastin network: Its relationship with collagen and cellsin articular cartilage as visualized by multiphoton microscopy. J Anat 215:682–691.

46. Gard DL (1991) Organization, nucleation, and acetylation of microtubules in Xenopuslaevis oocytes: A study by confocal immunofluorescence microscopy. Dev Biol 143:346–362.

3376 | www.pnas.org/cgi/doi/10.1073/pnas.1200642109 Takesono et al.