The T Box Transcription Factor No Tail in Ciliated Cells Controls Zebrafish Left-Right Asymmetry

Transcription

1 Current Biology, Vol. 14, , April 20, 2004, 2004 Elsevier Ltd. All rights reserved. DOI /j.cub The T Box Transcription Factor No Tail in Ciliated Cells Controls Zebrafish Left-Right Asymmetry Jeffrey D. Amack and H. Joseph Yost* Huntsman Cancer Institute University of Utah Salt Lake City, Utah Summary The heart, brain, and gut develop essential left-right (LR) asymmetries. Specialized groups of ciliated cells have been implicated in LR patterning in mouse, chick, frog, and zebrafish embryos [1]. In zebrafish, these ciliated cells are found in Kupffer s vesicle (KV) and are progeny of dorsal forerunner cells (DFCs) [1 3]. However, there is no direct evidence in any vertebrate that the genes involved in LR development are specifi- cally required in ciliated cells. By using a novel method in zebrafish, we knocked down the function of no tail (ntl, homologous to mouse brachyury) in DFCs without affecting its expression in other cells in the embryo. We find that the Ntl transcription factor functions cell autonomously in DFCs to regulate KV morphogenesis and LR determination. This is the first evidence that loss-of-gene function exclusively in ciliated cells per- turbs vertebrate LR patterning. Our results demon- strate that the ciliated KV, a transient embryonic organ of previously unknown function, is involved in the earli- est known step in zebrafish LR development, suggesting that a ciliary-based mechanism establishes the LR axis in all vertebrate embryos. Results and Discussion Embryonic monocilia protruding from ventral node cells in mouse embryos have been proposed to generate fluid flow that breaks bilateral symmetry during development [4, 5]. Several genes implicated in LR patterning have been proposed to function in ciliated node cells, includ- ing inversus viscerum (left/right dynein) [6], kif3b [4], kif3a [7, 8], and polaris [9]. However, defining the spe- cific role(s) of these genes in ciliated cells by traditional genetic analysis has not been possible for two reasons: first, these genes are expressed in other embryonic cells at the time of LR development, and second, mutations in these genes result in severe pleiotropic phenotypes affecting multiple tissues. Similarly, zebrafish LR mutants [10 12] all have pleiotropic phenotypes. Fate map- ping studies have shown that the precursors of ciliated cells in zebrafish are a group of cells known as dorsal forerunner cells (DFCs). DFCs migrate ahead of involuting dorsal blastoderm during gastrulation and then organize in the tailbud to form a ciliated epithelium in the spherical Kupffer s vesicle (KV) during somitogen- esis [1 3]. Here, we have directly tested gene function in zebrafish ciliated cells by silencing gene expression exclusively in DFCs. By using this cell-specific loss-of- *Correspondence: joseph.yost@hci.utah.edu function approach, we show that Ntl in ciliated cells has roles in LR development that are distinct from the roles of Ntl in other embryonic cells. Molecules injected into the yolk cell of zebrafish embryos prior to the 32-cell stage (2 hr postfertilization, hpf) enter every embryonic cell via cytoplasmic bridges that connect the yolk with the embryo [13]. When injected at these stages, antisense morpholino oligonucleotides (MO) will knock down function of a specifically targeted gene in all cells for several days, thereby facilitating global genetic loss-of-function studies [14]. Most cytoplasmic bridges are closed by 2 hpf, with the excep- tion of the progenitors of DFCs, which retain cyto- plasmic bridges with the yolk cell up to 4 hpf [2]. We have taken advantage of these yolk-dfc bridges to deliver fluorescein-tagged MOs to DFCs via injection into the yolk cell at mid-blastula stages (between 3 4 hpf). The efficiency of delivering MO to DFCs by mid-blastula stage yolk injection varied, but the fluorescein tag allowed the identification and selection of embryos in which MO had accumulated in DFCs (Figures 1B, 1C, and 1D). In 60% of embryos (n 317) injected at mid- blastula stages (256 cell to 1k cell stages), control MO (contmo) [14] diffused throughout the yolk cell, localized in yolk nuclei, and incorporated into DFCs, but not other embryonic cells. These results indicated that we could efficiently create chimeric embryos in which expression of a particular protein is blocked by MO in DFCs but remains wild-type in rest of the embryo. This methodol- ogy provides the means to rapidly characterize the spe- cific functions of genes in the ciliated cells of KV. The ntl gene was previously thought to regulate LR axis formation [10, 11, 15, 16] by controlling development of the embryonic midline (e.g., notochord and floorplate), which was proposed to serve as a barrier between the embryo s left and right sides [10]. However, in addition to being expressed in the developing notochord, ntl is also expressed in DFCs [3, 17]. We have used our mid-blastula stage-injection technique to test whether ntl plays a unique role in DFCs during development. A MO that knocks down Ntl function (ntlmo) [14] was injected into the yolk cell at mid-blastula stages to abrogate Ntl expression specifically in DFCs. Anti-Ntl immunohistochemistry [18] indicated that ntlmo was active in DFCs. Fluorescein-tagged contmo injected at mid-blastula stages colocalized with Ntl protein accumulation in DFCs (Figures 1E G), whereas ntlmo abolished Ntl protein accumulation in DFCs without affecting Ntl expression in adjacent dorsal margin cells (Figures 1H 1J). These chimeric embryos (referred to as DFC ntlmo embryos) have allowed us to analyze the specific func- tion(s) of Ntl in DFCs. ntl mutants [19] and ntl morphants (i.e., embryos in which ntlmo delivered to all cells eliminates all Ntl expression and produces a phenocopy of ntl mutants, Figures 2C and 2L) have significant defects in meso- derm, notochord, and tail development. In contrast, DFC ntlmo embryos developed normal morphology (Figures 2E and 2N). In addition, analysis of the midline

2 Current Biology 686 Figure 1. Gene Knockdown Exclusively in DFCs (A) Diagram of yolk cell injection in the mid-blastula stage zebrafish embryo. (B D) When examined 4 5 hr postinjection, fluorescein-tagged MO injected into mid-blastula stage embryos were detected in DFCs, but not other embryonic cells. Confocal fluorescent microscopy shows MO (green) localized in DFCs (arrow in [C]) migrating just below the dorsal margin (arrowhead in [B]) during epiboly. (E J) Confocal fluorescent images of embryos at 70% epiboly (dorsal view). Ntl antibody [18] and alexa568-conjugated secondary antibody label Ntl protein (red) in margin cells and DFCs that lie below the dorsal margin (marked by dashed line). (E G) contmo (green) injected at mid-blastula stages entered DFCs (E) and colocalized with Ntl protein in these cells only (G). (H-J) ntlmo (1.7 ng) injected at mid-blastula stages incorporated in DFCs (H) and knocked down expression of Ntl protein in DFCs, but not adjacent margin cells (I and J). Figure 2. Midline, Anterioposterior, and Tail Development Are Normal in DFC ntlmo Embryos (A F) Live embryos at 28 hpf. (A), (C), and (E) are transmitted light images and (B), (D), and (F) detect fluorescein-tagged ntlmo. MO were present in all cells of embryos injected at the one-cell stage (ntl morphants) (D), and tail development was aberrant (C). ntlmo injected into mid-blastula stage embryos were found primarily in the contiguous yolk cell (yc) and yolk tube (yt) (E and F), indicating cells other than DFCs and yolk did not incorporate MO. These DFC ntlmo embryos developed a normal morphology (E) similar to control embryos (A). (G I) RNA in situ hybridization (ISH) analysis of ntl expression, performed as previously described [15], which marks the notochord in the midline (G). The notochord failed to develop in ntl morphants (H). In contrast, DFC ntlmo embryos developed a normal notochord that expressed ntl (I). (J O) Live embryos at 54 hpf. ntlmo remained in all cells of morphant embryos (M), resulting in a phenocopy of ntl mutants (L). DFC ntlmo embryos, in which MO remained confined to yolk (O), appeared normal (N).

3 No Tail in Ciliated Cells Controls LR Asymmetry 687 markers ntl (Figures 2G 2I), col2a and shh (not shown) revealed that the midline defects observed in ntl morphants were not seen in DFC ntlmo embryos. After 3 days of development, when the no tail phenotype was obvious in ntl morphants, DFC ntlmo embryos appeared indistinguishable from wild-type controls (Figures 2J, 2L, and 2N). Although previous studies have suggested roles for DFCs in embryonic axis, midline, and/or tail development [2, 3, 20], our results indicate that these processes are independent of ntl function in DFCs. To determine whether LR patterning is dependent on Ntl expression in DFCs, we assessed molecular asymmetries by analyzing lefty1 (lft1) and lefty2 (lft2) gene expression in DFC ntlmo embryos. lft1 and lft2, which are normally expressed on the left side of the brain and heart field in lateral plate mesoderm, respectively (Figures 3A, 3D, and 3E), are expressed bilaterally in ntl mutants [15] and ntl morphants (Figures 3B, 3D, and 3E). In contrast, DFC ntlmo embryos showed a randomized array of lft1 and lft2 expression including a high frequency of reversed laterality (Figures 3C 3E), which was distinct from the bilateral phenotype observed in ntl morphants and ntl mutants. Consistent with the randomized molecular LR phenotype, heart laterality was aberrant in 35% of DFC ntlmo embryos (n 163). As an important control indicating that the ntlmo must be delivered to DFCs in order to elicit LR defects, embryos injected with ntlmo into the yolk at late blastula stages (4 5 hpf), when cytoplasmic bridges between the yolk and DFCs are closed, had normal heart and brain LR development (Figure S1). In additional control experiments to test whether MO injected at mid-blastula stages were active only in DFCs and not in other cells in close proximity, we targeted another gene in the LR pathway, lft1. During gastrulation, ntl is expressed in all mesoderm precursor cells, including dorsal margin cells, notochord precursor cells, and Figure 3. ntl in DFCs Controls Heart and Brain LR Patterning DFCs, whereas lft1 is expressed in dorsal margin cells (A C) In situ hybridization analysis of lft1 and lft2 expression [15] in and midline cells, but not DFCs (Figures 4A 4C). Consis somite stage embryos (dorsal view, the dotted line indicates tent with recently reported results [21], MO against lft1 the midline). (A) In control embryos, lft1 is expressed in the left side (lft1mo) delivered to all cells resulted in aberrant bilateral of the brain, and lft2 is expressed in the left heart field. (B) ntl expression of heart and brain LR markers (Figures 4J morphants showed bilateral lft1 and lft2 expression due to midline defects. (C) LR markers had a distinct randomized expression patand 4K). Since lft1 is not expressed in DFCs, LR defects tern in DFC ntlmo embryos, including right-sided expression. in DFC lft1mo embryos would indicate that the MO was (D and E) Weighted averages from at least four experiments asaffecting cells other than DFCs. In contrast with LR ef- sessing lft2 (D) and lft1 (E) expression (L, left-sided; R, right-sided; fects in DFC ntlmo embryos, DFC lft1mo embryos displayed B, bilateral; or A, absent) in uninjected control (n 814), ntl morphant normal heart and brain LR development (Figures 4J and (n 263), and DFC ntlmo (n 633) embryos. The portion of DFC ntlmo 4K). This indicates that lft1mo injected into the yolk at embryos that showed bilateral and right-sided expression of lft1 and lft2 were significantly different from ntl morphants (p mid-blastula stages cannot block gene function in cells and p respectively, using logistic regression). Error bars adjacent to DFCs, further supporting the conclusion that represent one weighted standard deviation. LR defects observed in DFC ntlmo embryos result from the loss of Ntl specifically in DFCs and not in other embrymesoderm cells are responsible for each of these pheonic cells. Taken together, these results demonstrate that Ntl expression specifically in DFCs is critical for notypes. sox17 expression in DFCs during gastrulation proper LR development. Furthermore, our molecular [22] is unaltered in DFC ntlmo embryos (data not shown), analyses reveal that the function of Ntl in DFCs is distinct suggesting that DFCs persist and migrate normally dur- from its role in establishing the midline barrier in neighabove (see Figure 2) indicate that ntl function in DFCs ing gastrulation. Furthermore, the results described boring mesodermal cells. This is the first example in which a gene knocked down exclusively in ciliated cells is not necessary for midline or tail development. In con- disrupts vertebrate LR development. trast, we found that DFCs failed to organize into KV in How does Ntl in DFCs control LR patterning? ntl mun 192) of DFC ntlmo embryos (Figure 5B), whereas KV DFC ntlmo embryos. KV was absent in the majority (70%, tants display pleiotropic alterations during gastrulation, KV morphogenesis, midline development, and tail formation, but it has remained unclear which ntl expressing embryos (n 213 and n 80, respectively). It is failed to form in only 3% 6% of uninjected or DFC contmo likely

4 Current Biology 688 Figure 4. MO in the Yolk Cell and DFCs Do Not Block Gene Expression in Adjacent Cells (A) The spherical KV (arrow) forms in the tailbud during early somite stages in uninjected and DFC contmo embryos. (B) KV failed to develop in 70% of DFC ntlmo embryos (n 192, arrow indicates the tailbud region). (C and D) Confocal images of cilia (red) labeled by fluorescent anti- acetylated tubulin immunohistochemistry. Cilia were organized within KV in wild-type embryos (C) but were disorganized in the tailbud of KV DFC ntlmo embryos (D). (E) Heart laterality was determined in live DFC ntlmo embryos at 52 hpf. DFC ntlmo embryos that formed a normal KV (n 37) had predominantly normal heart orientation, whereas laterality was randomized in DFC ntlmo embryos that failed to form KV (n 80). (A C) RNA in situ analysis of ntl (A), lft1 (B), and both ntl and lft1 (C) in embryos at 85% epiboly. (A) and (B) are dorsal views and (C) is a lateral view. ntl is expressed in midline precursor cells, margin cells, and DFCs, whereas lft1 is expressed in midline and margin cells, but not DFCs. (D I) Live embryos at 24 hpf. (D), (F), and (H) are transmitted light images and (E), (G), and (I) are fluorescent images. lft1 morphants, in which 1.7 ng fluorescein-tagged contmo and 1.4 ng untagged lft1mo was delivered to all cells (G), had a reduced head and short- ened body axis (F), while embryos yolk-injected at mid-blastula stages (DFC lft1mo embryos) developed normally (H). (J and K) Weighted averages from at least three experiments assessing lft2 (J) and lft1 (K) expression in uninjected (n 277), lft1 morphant (n 111), and DFC lft1mo (n 152) embryos, and DFC contmo embryos injected with control MO alone (n 198). L, left-sided; R, right-sided; B, bilateral; and A, absent. Error bars correspond to one weighted standard deviation. Figure 5. Ntl Regulates Morphogenesis of Kupffer s Vesicle, Which Is Required for Normal LR Development tochemistry [1] detected ciliated DFCs in the tailbud region where KV should have formed. However, these cells were disorganized and did not resemble the spheri- cal pattern of ciliated cells observed in KV control embryos (Figures 5C and 5D). Together, these analyses indicate that Ntl is required cell autonomously in DFCs that the 70% penetrance in DFC ntlmo embryos is indicative of the portion of embryos in which sufficient ntlmo role in DFC formation or migration during gastrulation, for the morphogenesis of KV, that Ntl does not play a entered DFCs to knock down Ntl function below the and that the roles for Ntl in gastrulation and midline threshold of normal activity. In DFC ntlmo embryos that development occur through its functions in non-dfc failed to develop KV, anti-acetylated tubulin immunohis- tissues.

5 No Tail in Ciliated Cells Controls LR Asymmetry 689 To test whether the absence of KV correlated with LR Acknowledgments defects, heart laterality was scored in DFC ntlmo embryos We thank M. S. Cooper, M. Brueckner, and C. Tabin for discussions in which KV failed to form (70%) and in those that formed and M.L. Condic, B. Bisgrove, W. Branford, and P. Sacayon for a normal KV (30%). DFC ntlmo embryos that formed KV comments on the manuscript. Statistical analyses were performed showed predominantly normal rightward heart looping, by A. Szabo. This work was supported by grants from the National whereas those that failed to form KV showed randomthe Institutes of Health-National Heart, Lung and Blood Institute and ized heart looping (Figure 5E). These results demonstrate Huntsman Cancer Foundation to H.J.Y., and a National Research that KV is required for normal LR pattern formation. Service Award fellowship to J.D.A. We conclude that loss of Ntl in DFCs randomizes Received: January 30, 2004 LR patterning by disrupting KV morphogenesis and that Revised: February 19, 2004 as a transcription factor, Ntl controls a battery of genes Accepted: February 20, 2004 that are expressed cell autonomously in DFCs to drive Published: April 20, 2004 the formation of this ciliated embryonic organ. Here, by using a novel cell type-specific gene knock References down strategy, we provide the first genetic evidence for 1. Essner, J.J., Vogan, K.J., Wagner, M.K., Tabin, C.J., Yost, H.J., a developmental role for the enigmatic KV first described and Brueckner, M. (2002). Conserved function for embryonic in the 1860s: control of LR determination. The transcrip- nodal cilia. Nature 418, tion factor Ntl has a regulatory function in DFCs that is 2. Cooper, M.S., and D Amico, L.A. (1996). A cluster of noninvoluting distinct from its roles in mesoderm cells, suggesting that endocytic cells at the margin of the zebrafish blastoderm unique transcription circuitry controls morphogenesis of marks the site of embryonic shield formation. Dev. Biol. 180, KV. As the first elimination of gene function specifically in 3. Melby, A.E., Warga, R.M., and Kimmel, C.B. (1996). Specification specialized ciliated cells in any vertebrate, these results of cell fates at the dorsal margin of the zebrafish gastrula. Develdemonstrate that the precise organization of these cells opment 122, is essential for normal LR development. 4. Nonaka, S., Tanaka, Y., Okada, Y., Takeda, S., Harada, A., Kanai, Ciliated cells in the spherical KV are reminiscent of Y., Kido, M., and Hirokawa, N. (1998). Randomization of leftciliated cells in the mouse embryo that are organized in right asymmetry due to loss of nodal cilia generating leftward flow of extraembryonic fluid in mice lacking KIF3B motor proa transient semispherical pit at the node; the action tein. Cell 95, of cilia in a confined and highly structured space would 5. McGrath, J., Somlo, S., Makova, S., Tian, X., and Brueckner, facilitate the generation of fluid flow. There is evidence M. (2003). Two populations of node monocilia initiate left-right that motile cilia in the mouse node generate leftward asymmetry in the mouse. Cell 114, fluid flow ( nodal flow ) that is detected by a subset of 6. Supp, D.M., Witte, D.P., Potter, S.S., and Brueckner, M. (1997). nonmotile mechanosensory cilia at the periphery of the Mutation of an axonemal dynein affects left-right asymmetry in inversus viscerum mice. Nature 389, node [5, 23, 24]. These sensory cilia respond to nodal 7. Takeda, S., Yonekawa, Y., Tanaka, Y., Okada, Y., Nonaka, S., flow by initiating asymmetric Ca 2 signaling, which might and Hirokawa, N. (1999). Left-right asymmetry and kinesin sube the trigger of left-sided gene expression immediately perfamily protein KIF3A: new insights in determination of laterality adjacent to the node. We propose that KV cilia function and mesoderm induction by kif3a-/- mice analysis. J. Cell in an analogous manner, to create and sense directional Biol. 145, Marszalek, J.R., Ruiz-Lozano, P., Roberts, E., Chien, K.R., and fluid flow inside KV that signals cells on the left side Goldstein, L.S. (1999). Situs inversus and embryonic ciliary morof the vesicle to activate asymmetric gene expression. phogenesis defects in mouse mutants lacking the KIF3A subunit Consistent with this model, we have identified several of kinesin-ii. Proc. Natl. Acad. Sci. USA 96, components of motile and sensory cilia that are ex- 9. Murcia, N.S., Richards, W.G., Yoder, B.K., Mucenski, M.L., Dunlap, pressed in KV cells (data not shown). The elimination of J.R., and Woychik, R.P. (2000). The Oak Ridge Polycystic Ntl function in DFCs/KV results in a disorganization of Kidney (orpk) disease gene is required for left-right axis determination. Development 127, these ciliated cells so that downstream signals are ran- 10. Danos, M.C., and Yost, H.J. (1996). Role of notochord in specifidomly activated on the two sides of the embryos. Since cation of cardiac left-right orientation in zebrafish and Xenopus. the midline is intact in DFC ntlmo embryos, these randomly Dev. Biol. 177, activated signals are then transmitted and maintained 11. Chen, J.N., van Eeden, F.J., Warren, K.S., Chin, A., Nusslein- on the side on which they were activated. The DFCpattern Volhard, C., Haffter, P., and Fishman, M.C. (1997). Left-right specific knockdown technique described here provides of cardiac BMP4 may drive asymmetry of the heart in zebrafish. Development 124, a powerful tool to selectively eliminate ciliary proteins 12. Yan, Y.T., Gritsman, K., Ding, J., Burdine, R.D., Corrales, J.D., in KV and directly test whether motile cilia and mechano- Price, S.M., Talbot, W.S., Schier, A.F., and Shen, M.M. (1999). sensory cilia function in KV to mediate LR development. Conserved requirement for EGF-CFC genes in vertebrate leftright The results presented here reveal an essential role for axis formation. Genes Dev. 13, ntl in the ciliated KV as the earliest known step in zebrablastomeres. 13. Kimmel, C.B., and Law, R.D. (1985). Cell lineage of zebrafish I. Cleavage pattern and cytoplasmic bridges be- fish LR determination. These findings support a model tween cells. Dev. Biol. 108, in which signaling from specialized and highly organized 14. Nasevicius, A., and Ekker, S.C. (2000). Effective targeted gene ciliated cells is a conserved mechanism used to estab- knockdown in zebrafish. Nat. Genet. 26, lish the LR axis in vertebrate embryos. 15. Bisgrove, B.W., Essner, J.J., and Yost, H.J. (2000). Multiple pathways in the midline regulate concordant brain, heart and gut left-right asymmetry. Development 127, Supplemental Data 16. Chin, A.J., Tsang, M., and Weinberg, E.S. (2000). Heart and gut A supplemental figure is available at com/cgi/content/full/14/8/685/dc1/. chiralities are controlled independently from initial heart position in the developing zebrafish. Dev. Biol. 227,

6 Current Biology Schulte-Merker, S., Hammerschmidt, M., Beuchle, D., Cho, K.W., De Robertis, E.M., and Nusslein-Volhard, C. (1994). Expression of zebrafish goosecoid and no tail gene products in wild-type and mutant no tail embryos. Development 120, Schulte-Merker, S., Ho, R.K., Herrmann, B.G., and Nusslein- Volhard, C. (1992). The protein product of the zebrafish homologue of the mouse T gene is expressed in nuclei of the germ ring and the notochord of the early embryo. Development 116, Halpern, M.E., Ho, R.K., Walker, C., and Kimmel, C.B. (1993). Induction of muscle pioneers and floor plate is distinguished by the zebrafish no tail mutation. Cell 75, Alexander, J., Rothenberg, M., Henry, G.L., and Stainier, D.Y. (1999). casanova plays an early and essential role in endoderm formation in zebrafish. Dev. Biol. 215, Feldman, B., Concha, M.L., Saude, L., Parsons, M.J., Adams, R.J., Wilson, S.W., and Stemple, D.L. (2002). Lefty antagonism of Squint is essential for normal gastrulation. Curr. Biol. 12, Alexander, J., and Stainier, D.Y. (1999). A molecular pathway leading to endoderm formation in zebrafish. Curr. Biol. 9, Tabin, C.J., and Vogan, K.J. (2003). A two-cilia model for vertebrate left-right axis specification. Genes Dev. 17, Yost, H.J. (2003). Left-right asymmetry: nodal cilia make and catch a wave. Curr. Biol. 13, R808 R809.