Lzts2 Regulates Embryonic Cell Movements and Dorsoventral Patterning through Interaction with and Export of Nuclear β-Catenin in Zebrafish*

Background: The activities of leucine zipper tumor suppressor 2 (Lzts2) in vertebrate embryogenesis remain unclear. Results: Lzts2 regulates gastrula CE movements, dorsoventral patterning, and midline convergence and specification of organ precursors in zebrafish. Conclusion: Lzts2 functions through interaction with and the export of nuclear β-catenins. Significance: This work has uncovered the developmental roles of Lzts2 and regulatory mechanisms in zebrafish. Leucine zipper tumor suppressor 2 (Lzts2) functions in the development and progression of various tumors, but its activities in vertebrate embryogenesis remain unclear. Here, we demonstrate that lzts2 transcripts are of maternal origin in zebrafish embryos. Activation of BMP signaling up-regulates zygotic expression of lzts2, whereas canonical Wnt signaling acts upstream of BMP signaling to inhibit lzts2 expression. Abrogation of lzts2 expression by its specific morpholino-enhanced gastrula convergence and extension (CE) movements, dorsalized early embryos, and inhibited specification of midline progenitors for pancreas, liver, and heart. In contrast, ectopic expression of lzts2 led to the delay of CE movements and midline convergence of organ progenitors and resulted in a certain ratio of ventralized embryos. Mechanistically, Lzts2 regulates the migration of embryonic cells and dorsoventral patterning through its limitation of Wnt/β-catenin activity, because it physically interacts with β-catenin-1 and -2 and transports them out of the nucleus. In addition, both β-catenin-1 and -2 exhibit redundant functions in activation of Stat3 signaling and in induction of Wnt5/11 expression through inhibition of BMP signaling and stimulation of Cyclops and Squint expression. Thus, Lzts2 regulates gastrula CE movements, dorsoventral patterning, and midline convergence and specification of organ precursors through interaction with and the export of nuclear β-catenins in zebrafish.

Proper cell movements during embryonic development of vertebrates are essential for establishment of an early threedimensional body and formation of later primitive organs (1,2). Principal gastrulation movements include epiboly, internalization of mesoendoderm, and convergence and extension (CE), 2 through which the dorsoventral and rostrocaudal embryonic axes are finally defined (3). Epibolic movements spread and thin the embryonic tissues; internalization of prospective mesodermal and endodermal cells at the blastopore lips leads to formation of three germ layers; convergence movements narrow germ layers mediolaterally, whereas extension movements lengthen them anteroposteriorly (1). Previous studies have shown that the intercalation of cells within mesoderm and ectoderm layers of developing Xenopus embryos drives simultaneous and interdependent mediolateral narrowing and anteroposterior elongation of embryonic tissues (4,5). This process is known as convergent extension and is accompanied by epiboly, mesoderm internalization, and anterior migration of prechordal mesoderm (6). In contrast to frog, impairment of CE movements does not interfere with mesoendoderm internalization and epiboly in zebrafish (7) and Fundulus (8). Moreover, mesoendoderm cell populations along the dorsoventral axis behave differently during fish gastrulation as follows: ventral cells engage in epibolic but not CE movements; lateral mesoendodermal cell populations converge and extend with increasing speed as they move dorsally; cells within the dorsal domain show strong ability of extension and moderate convergence (1,9). Furthermore, epiboly and mesoderm internalization are largely completed by the end of gastrulation, although CE movements continue to shape the embryonic body during segmentation (1). It has been shown that precursor cells of many organs initially appear as bilateral cell populations and subsequently migrate toward the embryonic midline, where they assemble to form the primitive organs such as heart, duodenum, liver, and pancreas (10 -12).
Vertebrate CE movements are known to be controlled by a large number of factors and signaling pathways, such as Ephrin (13), Slit (14), Ca 2ϩ signaling that is activated by noncanonical Wnt (nc-Wnt) signaling (6,15); Stat3 signaling that acts downstream of canonical Wnt signaling (16,17); JNK that regulates convergent extension movements downstream of Wnt5 in Xenopus (18); Rho kinase 2 that functions in a signaling pathway downstream of Wnt5 and Wnt11 (19,20), and BMP signaling that inhibits activity of nc-Wnt signaling (21). Additionally, the ventral-to-dorsal BMP gradient can guide dorsal convergence of cells in lateral regions of the zebrafish gastrula independently of nc-Wnt and Fgf signaling (22), and Nodal-related genes squint and cyclops are necessary for involution movements of marginal cells to mesoendoderm in the zebrafish embryos (23). Moreover, the Dishevelled-RhoA branch of nc-Wnt signaling is critical for gastrula CE movements and midline convergence of heart and foregut precursors during zebrafish development (2), and Wnt5b/11 serve as the main ligands for activation of the nc-Wnt pathway (24 -28). Furthermore, Stat3 and Wnt11 signaling pathways appear to act in parallel in the control of convergence and extension movements (17). These findings clearly indicate that the nc-Wnt signaling is the key mediator of CE movements in both zebrafish and Xenopus.
In zebrafish, the asymmetric accumulation of maternal ␤-catenin in nuclei of dorsal margin blastomeres leads to the organizer formation and dorsoventral patterning by activation of the expression of a number of zygotic genes, including bozozok/dharma (boz), chordin, dickkopf1 (dkk1), squint, and FGF signals (29). These ␤-catenin targets then act to inhibit the action of ventralizing factors such as BMP signals and regulate CE movements through multiple signal molecules, including Squint and Stat3 (1). Although multiple factors, such as Frizzled 2 (30), Stbm (31,32), Daam1 (33), and glypican knypek (34), have been shown to regulate the activity of nc-Wnt signaling, the precise molecular network and how nc-Wnt signaling polarizes cells and promotes CE movements remain largely unknown. Recently, we have shown that zebrafish ␤-catenin-1 and -2 have redundant roles in the control of dorsal formation (35), but it remains unclear whether they have distinct roles in regulation of CE movements during embryogenesis.
Leucine zipper tumor suppressor 2 (LZTS2)/LAPSER1 gene is located at a subregion of human chromosome 10q24.3 and deleted in various cancers, including prostate tumors (36,37). It is a negative regulator of katanin-mediated microtubule severing and releasing from the centrosome and is required for central spindle formation and completion of cytokinesis (38). Lzts2 is able to repress the transactivation of ␤-catenin by affecting its subcellular localization (39). Moreover, reciprocal cross-talk between ␤-catenin/Tcf pathway and NF-B signaling in human adipose tissue is mediated through the regulation of Lzts2 expression (40). However, little is known about its developmental roles in vertebrates.
In this study, we investigated the regulation of gastrula CE movements, dorsoventral patterning, and midline convergence and specification of organ precursors by Lzts2 during zebrafish embryogenesis. We examined molecular mechanisms underlying the inhibitory effect of Lzts2 on cell migration and dissected the activities of two zebrafish ␤-catenins in regulation of nc-Wnt signaling. These findings will extend our understanding on the roles of Lzts2 in embryonic morphogenesis and tumorigenesis.

EXPERIMENTAL PROCEDURES
Fish Strains and Maintenance-AB inbred strain and Tg (insu: GFP) transgenic line (41) of zebrafish (Danio rerio) were reared in a recirculating water system and maintained at standard conditions. Naturally fertilized zebrafish embryos were staged by morphological features or by hours post-fertilization (42).
Plasmids Construction-Total RNA was isolated from embryos at 48 hpf with TRIzol reagent from Invitrogen following the manufacturer's instructions. First-strand cDNA was synthesized using the RevertAid TM first strand cDNA synthesis kit from Fermentas. Two PCR primers, 5Ј-TCAGAAGCTTG-CCACCATGGCTCTAGTTCAGGCACTGC-3Ј and 5Ј-CAG-TGAATTCTTAGATTTCAGTGGCAGTTATCTCTTC-3Ј, that contain specific restriction sites for subcloning zebrafish lzts2 cDNA into vectors for in vitro transcription of capped mRNA and overexpression were designed according to the predicted full-length mRNA sequence for zebrafish lzts2 gene in the GenBank TM data base (XM_694459.3). A mutant of Lzts2 (lzts2-M) that contains mutations in its nuclear export signal was generated by site-directed mutagenesis with two PCR primers, 5Ј-GCTAGTGCGGAGG-CGGAAAACAGGGACATG-3Ј and 5Ј-CGCCTCCGCAC-TAGCTTCCCTCATCACAC-3Ј. Three Lzts2 mutants that contain mutations in the potential ␤-catenin-binding sites were created with PCR primer pairs: Cell-tracing Experiments-The Kaeda protein emits bright green fluorescence that can be irreversibly converted to red upon irradiation with a beam of ultraviolet or violet light (350 -400 nm) (43). The full-length kaeda cDNA was inserted into the pCS2ϩ vector. Capped mRNAs were synthesized using the mMESSAGE mMACHINE Kit (Ambion, Austin, TX). Morpholino (MO) oligonucleotides for zebrafish Lzts2 (lzts2-MO, 5Ј-CATGGTGAACACCCGACCAGGACAC-3Ј) and standard control (STD-MO) were obtained from Gene Tools, LLC. (Corvallis, OR). 200 pg of capped kaeda mRNA was co-injected with 10 ng of lzts2-MO, 300 p of lzts2 capped mRNA, 300 pg of lzts2-M capped mRNA, or 10 ng of lzts2-MO plus 300 pg of lzts2 capped mRNA into one cell stage embryos. Injected embryos were maintained in a dark room until shield stage. To convert green fluorescence to red, a beam of ultraviolet light that was generated by a UV filter set (350 -400 nm) on a Zeiss fluorescence microscope was directed for 30 s at the dorsal or lateral blastoderm margin of embryos. The location of red fluorescent cells was then photographed at indicated time points using the SteReo Lumar V12 microscope from Carl Zeiss.
Whole-mount in Situ Hybridization-Spatiotemporal expression of lzts2 was determined by the whole-mount in situ hybridization (WISH). Digoxigenin-labeled antisense and sense RNA probes for lzts2, CE movements, and dorsoventral patterning were transcribed with T7 or T3 RNA polymerases. In situ hybridization on crystal sections was performed as described previously (44) with minor modifications. Images were taken using the SteReo Lumar V12 and LSM710 microscope from Carl Zeiss.
Cell Culture and Transfection-ZF4 cells were grown in DMEM/F-12 from Invitrogen supplemented with 10% fetal calf serum (FCS), 100 units/ml penicillin, and 100 g/ml streptomycin and incubated at 28°C in 5% CO 2 and 95% air. Cells grown to 50 -60% confluence on 35-mm Petri dishes were transfected using the FuGENE HD transfection reagent from Roche Applied Science according to the manufacturer's instructions. Optimal ratio of DNA (g) to Transfection Reagent (l) was determined to be ϳ1:3. After 24 h of transfection, cells were used for further treatments.
Immunofluorescence Staining-Immunofluorescence staining of transfected ZF4 cells was performed following our previous protocol (35). The primary and secondary antibodies are mouse anti-HA and goat anti-mouse FITC, respectively.
Co-immunoprecipitation and Western Blotting-Embryonic cells were isolated following a previous protocol (52). Co-immunoprecipitation and Western blotting of embryonic cells were performed following our previous protocol (35).
Statistical Analysis-Student's t test was performed using GraphPad Prism version 3.0 for Windows from GraphPad Software, San Diego. Statistical significance was defined at p Ͻ 0.05.

RESULTS
Spatiotemporal Expression of Zebrafish Lzts2 and Its Transcriptional Regulation during Embryogenesis-To identify the lzts2 gene in zebrafish genome, amino acid sequence of human Lzts2 was used to blast the NCBI protein data base. One protein (XP_699551.31) was found to share a high degree of similarity with human Lzts2 (E-values Ͻ4e-132). According to the guideline for naming zebrafish genes, this gene was named as lzts2. Amino acid sequence of zebrafish Lzts2 was aligned with those from human and mouse (supplemental Fig. S1). Zebrafish Lzts2 consists of 709 amino acids and shares 47.8% amino acid sequence identity with human Lzts2, 45.2% with mouse Lzts2, and 58.5% identity with Xenopus Lzts2. Importantly, all Lzts2s contain a leucine zipper region with high similarity to the DNAbinding domain of cAMP-responsive activating transcription factor 5 (53). This region, known as the FEZ1 domain, is closely associated with inhibition of cancer cell growth through regulation of mitosis, and its alterations result in abnormal cell growth (36,54). In addition, a consensus sequence of leucinerich nuclear export signal (NES) was identified in the C terminus of zebrafish Lzts2 (supplemental Fig. S1; amino acids 673-684).
We further examined how transcriptional expression of lzts2 was controlled by major signaling pathways involved in early embryonic development. As shown in Fig. 1B, expression of lzts2 at shield stage (6 hpf) was markedly induced on the dorsal side by ectopic expression of a Wnt inhibitor frzb and on the ventral side by overexpression of bmp4 but inhibited by expression of a constitutively active ␤-catenin-1 (␤-cat) or a truncated BMP receptor I (bmpRI). Moreover, expres-sion of lzts2 was inhibited in embryos injected with equal amount mRNAs for frzb and bmpRI and mainly induced on the ventral side of embryos injected with equal amount mRNAs for ␤-cat and bmp4. However, inhibition of FGF signaling by injection of zebrafish sef mRNA or activation of Nodal signaling by injection of mRNA for zebrafish Nodalrelated factor, Cyclops, showed no significant effects on lzts2 transcript levels (data not shown). Similar results were obtained by detection of lzts2 expression in pooled embryos with real time PCR (Fig. 1C). These data indicate that activation of BMP signaling positively regulates lzts2 expression and Wnt signaling acts upstream of BMP signaling to negatively control transcriptional expression of lzts2 in developing zebrafish embryos.
Alterations of Lzts2 Expression Lead to Developmental Defects in Zebrafish-To elucidate developmental functions of Lzts2 in zebrafish, we checked its effects on embryonic phenotypes after microinjection of lzts2-MO or synthetic lzts2 mRNA into one-cell stage embryos. To determine the efficiency of lzts2-MO, a construct named lzts2/EGFP was generated to produce an in-frame transcript of EGFP that contains the lzts2-MO target at its N terminus. We found that EGFP was expressed in Ͼ90% of embryos after co-injection of 10 ng of STD-MO with 200 pg of lzts2/EGFP plasmid, but it weakly appeared in Ͻ5% of embryos when STD-MO was substituted by equal amounts of lzts2-MO (supplemental Fig. S2, A and B). These data indicate that lzts2-MO can efficiently inhibit endogenous lzts2 expression.
Injection of 10 ng of lzts2-MO resulted in abnormal embryos with thickened organizer at 6 hpf and elongated anterior-posterior (AP) axis at 12 hpf ( Fig. 2A). Nevertheless, injection of 300 pg of capped lzts2 mRNA led to abnormal embryos with thickened ventrals at 6 hpf and broadened AP axis at 12 hpf ( Fig.  2A). In addition, injection of 300 pg of capped lzts2 mRNA rescued most of the abnormal embryos caused by injection of 10 ng of lzts2-MO ( Fig. 2A). Total number of embryos examined and ratios of embryos with shown phenotypes are summarized in Fig. 2B. Moreover, knockdown of Lzts2 by MO resulted in abnormal embryos with small eyes and head, shortened and twisted tails, and/or shortened AP axis at 48 hpf (Fig. 2C). These knockdown phenotypes were not affected by coinjection of 1-3 ng of p53 MO (data not shown), indicating that lzts2-MO-mediated phenotypes are not an artifact of MO toxicity. In contrast, overexpression of lzts2 led to embryos with a fused eye, shortened distance between the eyes, loss of head, and/or expanded somites and tails at 48 hpf (Fig. 2C).
Phenotypes of some Lzts2 morphants are similar to dorsalized embryos caused by overexpression of Boz (55), LiCl treatment (56), or knockdown of caveolin-1 (35). However, Lzts2overexpressing embryos during gastrulation and some of embryos at later stages resemble those seen in slb/wnt11 and ppt/wnt5 mutants with defects in CE movements (24,57). Importantly, most of the abnormal embryos caused by injection of 10 ng of lzts2-MO were rescued by injection of 300 pg of capped lzts2 mRNA. Ratios of embryos with shown phenotypes after injection of lzts2-MO and/or lzts2 mRNA are summarized in supplemental Table S1. These findings indicate that Lzts2 may be involved in the control of embryonic cell movements and dorsoventral patterning through regulation of nc-Wnt and/or Wnt/␤-catenin signaling.
Lzts2 Negatively Regulates CE Movements during Gastrulation-To further determine whether Lzts2 is required for embryonic cell movements, we performed cell tracing assays to measure cell movements at indicated time points by ultraviolet light stimulation of different cell populations in the lateral margin or dorsal organizer of developing embryos at 5.7 hpf. In wild type embryos, red fluorescence-labeled cells migrated to the dorsal and, at the same time, spread to the anterior and then underwent mediolateral intercalation in the axial tissue at 24 hpf (Fig. 3A, WT of lateral cells). Next, the lzts2-MO was utilized to knock down the expression of endogenous lzts2 gene, and a lzts2 mutant (lzts2-M) was generated by substitution of three residues (Leu-680, Leu-682, and Leu-684) in its putative NES with three alanine residues to block the interaction of Lzts2 with an uncharacterized signal molecule in the CRM1/ exportin-␣ pathway. Injection of lzts2-MO or lzts2-M mRNA hastened the migration of lateral cells to the dorsal without interruption of their intercalation with axial cells; however, migration of labeled lateral cells was clearly delayed in embryos overexpressing Lzts2 from 7 to 9.5 hpf, and these cells failed to intercalate in the axial tissues at 24 hpf (Fig. 3A, Lateral cells). In addition, the quick lateral cell movements by lzts2-MO were rescued to normal by co-injection of capped lzts2 mRNA. Dorsal convergence was quantified by the position of labeled cells at 9.5 hpf relative to their initial position at 5.7 hpf (Fig. 3B).
To examine the extension of axial mesoderm cells, we directed the ultraviolet light at cell population in the dorsal organizer at 5.7 hpf. In wild type embryos, the labeled cells extended along the AP axis and intercalated in nonlabeled cells to form an elongated cell array at 24 hpf (Fig. 3A, WT of dorsal cells). Injection of lzts2-MO or lzts2-M mRNA promoted the migration of dorsal organizer cells to the anterior without interruption of their intercalation with axial cells; however, in lzts2 mRNA-injected embryos, the initial size of labeled cells at 5.7 hpf was the same as that in the wild type embryos, but subse- quent elongation of these labeled cells along the AP axis was clearly inhibited (Fig. 3A, Dorsal cells). Moreover, the quick dorsal cell movements by 10 ng of lzts2-MO were rescued to normal by co-injection of 300 pg of capped lzts2 mRNA. Anterior extension was quantified by the position of labeled cells at 9.5 hpf relative to their initial position at 5.7 hpf (Fig. 3C). These data indicate that Lzts2 plays crucial roles in the limitation of lateral cell CE movements and dorsal cell extension during gastrulation.
We next investigated the effects of lzts2-MO or lzts2 mRNA on the expression of marker genes for the specification of endoderm (foxd3 and sox17), mesoendoderm (hgg1), mesoderm (shh, ntl, myoD, and papc), and ectoderm (dlx3). As shown in Fig. 4A, expression of these genes was slightly affected in most of embryos injected with lzts2-MO or capped lzts2 mRNA; however, distributions of their transcripts were clearly narrowed along the dorsoventral axis and extended along the AP axis in embryos injected with lzts2-MO. In contrast, transcriptional expression of these genes were markedly expanded along the dorsoventral axis and shortened along the AP axis in embryos overexpressing Lzts2. Moreover, expression patterns of sox17, myoD, and ntl in embryos overexpressing Lzts2 assemble those seen in sqt Ϫ/Ϫ embryos (58). Total embryos and ratios of embryos with the expression patterns as shown are summarized in Fig. 4B. These findings indicate that Lzts2 regulates the migration of cells in three germ layers without significant effects on their fate specification in early developing embryos of zebrafish. Taken together, the analysis of embryonic morphology, cell migration, and molecular markers has provided strong evidence that Lzts2 is a negative regulator of CE movements during gastrulation of zebrafish embryos.

Lzts2 Regulates Midline Convergence and Specification of
Precursors for Heart, Liver, and Pancreas-We have shown that lzts2 transcripts are mainly distributed in the anterior and midline structures of later stage embryos (Fig. 1A), suggesting that Lzts2 may function in the formation of midline organs. To address this possibility, we performed WISH assays to examine the expression patterns of marker genes for liver, heart, and pancreas in embryos injected with lzts2-MO or capped lzts2 mRNAs. As shown in Fig. 5A, knockdown of Lzts2 inhibited the expression of heart markers cmlc and bmp4 at 24 hpf, liver marker cp at 48 hpf, and markers (insulin, sst, glu, and pdx1) for pancreas specification and differentiation at 48 hpf; however, overexpression of Lzts2 led to the expression of heart markers cmlc and bmp4, and liver marker cp at two separate sites near the midline. In addition, pancreas markers insulin, sst, glu, and pdx1 exhibited dispersed expression patterns in embryos expressing Lzts2. Total number of embryos examined and ratios of embryos with shown expression patterns of marker genes are summarized in Fig. 5B. These data indicate that Lzts2 plays crucial roles in the control of midline convergence and/or specification of precursors for heart, liver, and pancreas in zebrafish.
Lzts2 Regulates Dorsoventral Patterning of Zebrafish Embryos-The lzts2 transcripts are ubiquitously distributed in the blastula embryo, raising the question of whether Lzts2 contributes to the dorsoventral patterning. To analyze the spatial requirement for Lzts2 function within the early embryo, WISH assays were performed to examine expression of dorsal genes boz, chordin, and gsc and ventral genes bmp2b, bmp4, eve1, and gata1 in embryos injected with lzts2-MO or lzts2 mRNA. As shown in Fig. 6A, increased expression of dorsal genes and decreased expression of ventral genes were clearly seen in 30 -50% of embryos injected with lzts2-MO, and opposite expression patterns of these marker genes were seen in 30 -55% of lzts2 mRNA-injected embryos (Fig. 6A). Total number of embryos examined and ratios of embryos with shown expression patterns of marker genes are summarized in Fig. 6B. These results demonstrate that Lzts2 is also involved in the limitation of dorsal specification in zebrafish.
Lzts2 Regulates Activities of Canonical Wnt and BMP Signaling in Zebrafish-To dissect the mechanisms underlying the regulation of CE movements and dorsoventral patterning by Lzts2, activities of key signaling pathways, including canonical Wnt, BMP, Nodal, and FGF, were examined after depletion or ectopic expression of lzts2 in early developing embryos. We first performed WISH assays to determine effects of lzts2-MO or lzts2 mRNA on expression of genes that are regulated by key signaling pathways. These genes include spry4 and sef for FGF, flh and xbp1 for Nodal, ved and vent for BMP, dkk1, boz, and otx2 for maternal Wnt, and wnt8a and tbx6 for zygotic Wnt signaling. As discussed in our previous study (35), expression of these genes can be carefully used as indicators for activation of FGF, Nodal, BMP, or Wnt signaling, although some of them are regulated by multiple signaling pathways. We found that expression of sef, spry4, xbp1, and flh was minimally affected by knockdown of Lzts2 or ectopic expression of Lzts2 in early developing embryos (data not shown). However, expression of dkk1, boz, wnt8a, otx2, and tbx6 was markedly induced in Ͼ80% of embryos injected with 10 ng of lzts2-MO and inhibited in FIGURE 5. Effects of Lzts2 on midline convergence of precursors for liver, pancreas, and heart in zebrafish. A, expression patterns of heart markers cmlc and bmp4 at 24 hpf, liver marker cp at 48 hpf, and markers for pancreas specification and differentiation insulin, sst, glu and pdx1 at 48 hpf, were detected with WISH in wild type (WT) or embryos injected with 10 ng of lzts2-MO or 300 pg of lzts2 mRNA. Dorsal views, anterior toward the top. B, total number of embryos examined (indicated at the bottom) and percentage of embryos with shown expression patterns in A. The expression patterns of these markers in remaining embryos are similar to those in wild type.

Functions of Lzts2 in Zebrafish
Ͼ60% of embryos injected with 300 pg of capped lzts2 mRNA (Fig. 7A). In addition, expression of ved and vent was inhibited by injection of 10 ng of lzts2-MO and enhanced by injection of 300 pg of capped lzts2 mRNA. Furthermore, the increased expression of Wnt targets and decreased expression of BMP targets in Ͼ85% of Lzts2-knockdown embryos were rescued to normal by co-injection of 8 ng of ␤-catenin-1-MO or 8 ng of ␤-catenin-2-MO (Fig. 7A).
To further substantiate these observations, Top-flash, Bre, and (CAGA) 12 -Luc reporters were used to determine effects of Lzts2 on activities of Wnt, BMP, and Nodal/TGF-␤ signaling pathways, respectively. In transfected ZF4 cells, overexpression of ␤-catenin (HA-␤-cat) significantly induced the Top-flash luciferase activity (p Ͻ 0.01), but this induction was inhibited by co-expression of FLAG-lzts2 in a dose-dependent manner ( Fig. 7B; p Ͻ 0.01 in all cases). Overexpression of Lef1-VP16 fusion, the dominant positive of Wnt signaling, dramatically stimulated Wnt reporter activity (Ͼ100-fold), and the stimulation was not affected by co-expression of FLAG-lzts2 (Fig. 7C), implying that lzts2 inhibits the canonical Wnt signaling through ␤-catenin but not Lef1. Additionally, overexpression of Bmp4 (HA-bmp4) significantly induced the Bre luciferase activity (p Ͻ 0.01) and overexpressed FLAG-lzts2 demonstrated similar inductive effects on the Bre luciferase activity in a dose-dependent manner ( Fig. 7D; p Ͻ 0.05 or p Ͻ 0.01). Western blotting analysis indicated that HA-␤-cat, HA-lef1-VP16, HA-bmp4 and FLAG-lzts2 were properly expressed in corresponding groups of transfected cells (Fig. 7, B-D). However, activity of (CAGA) 12 -Luc reporter was not significantly affected by overexpressed Lzts2 in ZF4 cells (data not shown).
Taken together, our findings indicate that developmental roles of Lzts2 appear to be closely associated with its regulation of Wnt and BMP signaling in zebrafish, and Nodal signaling pathway may not be appreciably regulated by Lzts2.
Lzts2 Regulates Nuclear Export of ␤-Catenin in Zebrafish-A previous study using cells of monkey kidney and human tumors has shown that Lzts2 is able to repress activity of canonical Wnt signaling through its interaction with and nuclear export of ␤-catenin (39), suggesting that developmental roles of Lzts2 are mediated by its interaction with ␤-catenin in zebrafish. To test this hypothesis, physical interactions of Lzts2 with ␤-catenin-1 and -2 were first examined by co-immunoprecipitation assays and Western blotting. We found that overexpressed FLAG-lzts2 was reciprocally immunoprecipitated with HA-␤catenin-1 or -2 (HA-␤-cat1 or HA-␤-cat2) in embryonic cells (Fig. 8, A and B), indicating that both ␤-catenin-1 and -2 are binding partners of Lzts2 in zebrafish. Immunofluorescence staining assays were then performed after co-transfection of a plasmid expressing HA-␤-cat1 or HA-␤-cat2 with another plasmid expressing Lzts2 or Lzts2-M into ZF4 cells. Soon after transfection, ZF4 cells were cultured in medium containing 20 mM of LiCl, a well known GSK-3 inhibitor, to activate the canonical Wnt signaling. As shown in Fig. 8, C and D, treatment  DECEMBER 30, 2011 • VOLUME 286 • NUMBER 52

JOURNAL OF BIOLOGICAL CHEMISTRY 45123
with LiCl led to nuclear accumulation of HA-␤-cat1 or HA-␤-cat2, and the LiCl-induced nuclear accumulation of HA-␤catenins was blocked by Lzts2 but not Lzts2-M. Moreover, HA-␤-catenin signals were mainly located near the nuclear envelope of Lzts2-overexpressed cells (Fig. 8, C and D). Cells showing HA signals in nuclear or nuclear envelope of three visual fields are summarized in Fig. 8, E and F. These data indicate that interaction of Lzts2 with ␤-catenin and the Lzts2 NES signal are required for the export of nuclear ␤-catenin. Lzts2-M can be used as a dominant negative to block the export of nuclear ␤-catenin.
Moreover, injection of capped lzts2-M mRNA resulted in dorsalized phenotypes (supplemental Fig. S4A and supplemental Table S1), expanded dorsal gene expression, reduced ventral gene expression, elongated and narrowed expression of CE movement marker genes, reduced expression of cmlc in heart at 24 hpf, cp in liver, and insulin in pancreas at 48 hpf (supplemental Fig. S4B). These results are similar to those from Lzts2-MO-injected embryos. Injection of lzts2-M mRNA partially rescued the abnormal phenotypes and marker gene expression caused by lzts2 mRNA injection (supplemental Fig. S4 and supplemental Table S1), indicating the nuclear export of Lzts2 is closely associated with shown phenotypes and gene expression.
Furthermore, injection of capped Lzts2-M(650 -655) mRNA caused no obvious morphological defects in developing embryos, although slightly deceased expression of dorsal markers and increased expression of ventral markers were observed in 3-7% of injected embryos. In addition, marker gene expression for CE movements, midline convergence, and specification of midline organ precursors were little affected by ectopic expression of Lzts2-M(650 -655) (supplemental Fig. S4). Moreover, the abnormal phenotypes and marker gene expression caused by Lzts2 mRNA injection were not affected by co-injecting equal amount of Lzts2-M(650 -655) mRNA (supplemental Fig. S4 and supplemental Table S1). These results imply that interaction of Lzts2 with ␤-catenin is required for the inhibition of CE movements and dorsal development in zebrafish.
Lzts2 Regulates Embryonic Cell Movements through Multiple Signaling Pathways in Zebrafish-To further elucidate the mechanism(s) underlying the activities of Lzts2 in regulation of CE movements, we examined spatiotemporal expression of several key genes that are known to regulate the CE movements. These marker genes include squint, cyclops, wnt5, wnt11, and liv1. Expression of insulin was used to show the location and pattern of wnt11 expression in embryos at 24 hpf. As shown in Fig. 9A, expression of squint, cyclops, wnt5, wnt11, and a downstream gene of Stat3 signaling liv1 (60) was induced by injection of lzts2-MO or lzts2-M mRNA but was inhibited by injection of Lzts2 mRNA. Co-injection of 8 ng of ␤-catenin-1-MO or 8 ng of ␤-catenin-2-MO with 10 ng of lzts2-MO abolished the stimulatory effects of lzts2-MO on the expression of these targets. Importantly, co-injection of 5 pg of lef1-VP16 mRNA rescued the expression of squint, cyclops, wnt5, wnt11, and liv1 in most of lzts2 mRNA-injected embryos (Fig. 9A). Similar results were obtained by real time PCR assays of pooled embryos and summarized in Fig. 9B. Phenotypes and percentage of abnormal embryos were listed in supplemental Table S2 after co-injection of different doses of lef1-VP16 mRNA. These results indicate that Lzts2 regulates CE movements mainly through its direct interaction with ␤-catenin.
We further investigated whether activation of nc-Wnt or Stat3 signaling by ectopic expression of wnt5, wnt11, rhoA, or stat3-C can rescue the Lzts2-mediated defects in midline convergence of organ precursors. As shown in supplemental  examined the effects of ␤-catenin-1 and -2 on expression of squint, cyclops, boz, chordin, bmp2b, bmp4, wnt5, wnt11, and liv1 by WISH. As shown in Fig. 10A, overexpression of ␤-catenin-1 or -2 induced expression of cyclops, boz, chordin, wnt5 (wnt5b), wnt11, and liv1 but repressed expression of bmp2b and bmp4. In comparison with those in wild type embryos, expression of squint at 4.3 hpf was markedly stimulated by overexpression of ␤-catenin-1 but not by ␤-catenin-2; however, expression of squint at 4.7 and 6 hpf was markedly stimulated by both ␤-catenin-1 and -2 (Fig. 10A). Similar results were obtained from real time PCR assays of pooled embryos injected with capped mRNAs for ␤-catenin-1 or -2 ( Fig. 10B; p Ͻ 0.05 or p Ͻ 0.01). These data indicate that ␤-catenin-1 and -2 in zebrafish have redundant functions in regulation of CE movements.
Based on findings from this and previous studies, we proposed a model for regulation of CE movements by Lzts2 (Fig.  10C). In this model, Lzts2 negatively regulates activities of canonical Wnt signaling through the export of nuclear ␤-catenin-1 and -2. Both ␤-catenin-1 and -2 are able to inhibit the activity of BMP signaling and stimulate the activity of Stat3 and expression of Cyclops and Squint. Squint and Cyclops stimulate expression of Wnt5 and Wnt11, and BMPs inhibits their expression. Therefore, Lzts2 negatively regulates gastrula CE movements through inhibition of nc-Wnt and Stat3 signaling in zebrafish. In addition, a regulatory loop appears to tightly control the expression of lzts2 by Wnt/␤-catenin and BMP signaling. However, we could not rule out the possibility that BMP signaling directly regulates CE movements because the BMP gradient has recently been shown to determine the dorsal convergence of lateral mesodermal cells through negative regula-tion of Ca 2ϩ /cadherin-dependent cell-cell adhesiveness during zebrafish gastrulation (22).

DISCUSSION
In this study, we have addressed developmental functions of Lzts2 in the control of CE movements and dorsoventral patterning in zebrafish. Knockdown of Lzts2 results in abnormal embryos with thickened organizer and elongated AP axis during gastrulation, whereas Lzts2 overexpression leads to abnormal embryos with thickened ventral and posterior regions and broadened AP axis. Phenotypes of Lzts2 overexpressing embryos resemble those seen in slb/wnt11 and ppt/wnt5 mutants with defects of CE movements at the onset of gastrulation (24,57). In addition, knockdown of Lzts2 leads to a certain ratio of dorsalized embryos similar to those overexpressing Boz (55) or ␤-catenin⌬N (61,62), or treated with LiCl (56). In contrast, Lzts2 overexpression inhibits the midline convergence of precursors for heart, liver, and pancreas and results in some of ventralized embryos at 24 -48 hpf. These ventralized phenotypes are similar to those seen in ichabod mutants and embryos overexpressing caveolin-1, in which dorsal accumulation of maternal ␤-catenin is disrupted (35,45,63). Roles of Lzts2 in CE movements and dorsoventral axis formation are supported by its effects on a number of marker genes for early embryonic cell migration, dorsoventral patterning, midline convergence, and specification of organ precursors. Mechanistically, Lzts2 functions through its interaction with ␤-catenin and subsequently inhibits activities of canonical and noncanonical Wnt signaling. Moreover, zebrafish ␤-catenin-1 and -2 have redundant roles in regulation of CE movements. Therefore, our findings provide the first genetic and biochemical evi-  DECEMBER 30, 2011 • VOLUME 286 • NUMBER 52 dence that Lzts2 functions in embryonic cell migration, dorsoventral patterning, and specification and midline convergence of organ precursors for heart, liver, and pancreas.

Functions of Lzts2 in Zebrafish
Multiple lines of evidence from this study indicate that activities of Lzts2 in early embryonic development are mainly associated with ␤-catenin. First, Lzts2 and ␤-catenin are both of maternal origin in zebrafish and physically interact with each other. Second, Lzts2 inhibits activity of Wnt/␤-catenin signaling and subsequent expression of its downstream genes during early development of zebrafish. Third, knockdown of Lzts2 leads to dorsalized embryos that resemble those embryos in which ␤-catenin is activated (35,61,62). Fourth, Lzts2 inhibits expression of dorsal marker genes boz, gsc, and chordin but stimulates expression of ventral marker genes bmp2b, bmp4, eve1, and gata1. The stimulatory effect of Lzts2 on BMP signaling appears to be mediated by its inhibition of ␤-catenin activity, because knockdown of Lzts2 represses expression of marker genes ved and vent. Fifth, mutations in the ␤-catenin binding domain of Lzts2 lead to no inhibitory effects on Top-flash activity and causes no obvious morphological defects during embryonic development. Moreover, Lzts2 appears to regulate the activity of canonical Wnt signaling in a Lef1-independent manner. Furthermore, our data indicate that a regulatory loop exists in early developing embryos to tightly regulate the transcriptional expression of lzts2 by Wnt/␤-catenin and BMP signaling. This regulatory loop may play crucial roles in the maintenance of asymmetric activation of Wnt/␤-catenin and BMP signaling along the dorsoventral axis.
Wnt/␤-catenin signaling is critical for embryonic development, organogenesis, and oncogenesis (64 -66). Lzts2 has been shown to inhibit activity of canonical Wnt signaling through its interaction with armadillo repeats of ␤-catenin in monkey and human cells (39). In zebrafish embryos, activation of Wnt/␤catenin signaling promotes cardiogenesis prior to gastrulation but inhibits heart specification at a later developmental stage (67). Additionally, Wnt/␤-catenin signaling represses liver and pancreas fate during earlier endoderm patterning stages but promotes hepatic growth in the liver bud stage of mouse (68,69). It is recently shown that zebrafish Wnt2bb from lateral plate mesoderm stimulates liver specification (70); however, Wnt ligands involved in liver development of mice remain to be characterized (71). Moreover, ␤-catenin activity is required for development of the exocrine pancreas but is not required for development of the endocrine compartment (72). In this study, we demonstrate that Lzts2 negatively regulates the midline convergence and is required for the specification of organ precursors for heart, liver, and pancreas through inhibition of zygotic ␤-catenin activity. Therefore, further investigation of ␤-catenin activity regulation by Lzts2 and other intracellular factors would resolve the discrepancy of Wnt/␤-catenin signaling in organogenesis.
Zebrafish genome contains two distinct genes that encode ␤-catenin-1 and -2, respectively. Previous studies have shown that activation of ␤-catenin regulates the dorsoventral patterning, formation of Spemann-Mangold organizer, and embryonic CE movements through activation of boz homeobox, nodalrelated squint, and/or stat3 genes (17, 73). Inactivation of boz leads to mesodermal and neural patterning defects but is not necessary for regulation of CE (73). Nodal signaling deficiency results in a dramatic extension defect and loss of mesoendoderm, whereas convergence is less affected (23,74). It is suggested that organizer formation is mainly associated with activity of ␤-catenin-2 (45), but findings from our previous study (35) and this study favor an idea that ␤-catenin-1 is also involved in early dorsal formation. Evidence supporting the notion includes the following: 1) overexpression of either ␤-catenin leads to severe dorsalization of zebrafish embryos; 2) ␤-catenin-1 and -2 induce expression of dorsal marker genes boz, gsc, and chordin; 3) ␤-catenin-1 and -2 inhibit expression of ventral marker genes eve1, gata1, bmp2b, and bmp4. In this study, we further demonstrate that both ␤-catenin-1 and -2 have played similar roles in regulation of multiple signaling molecules and pathways that are involved in CE movements. Thus, we conclude that zebrafish ␤-catenin-1 and -2 have redundant functions in the control of dorsal formation and CE movements. Furthermore, roles of Lzts2 in embryonic development appear to be mediated by ␤-catenin-1 and -2 through the same binding site, because their putative Lzts2-binding motifs contain the same amino acid sequence. However, it remains unknown how this interaction affects the specification of tissues during embryonic development.
Apart from extracellular factors that are able to regulate the activity of Wnt/␤-catenin signaling, multiple intracellular molecules are known to interact with active pools of ␤-catenin in cells. Examples of these molecules include E-cadherin (75)(76)(77) and caveolin-1 in the plasma membrane, components in the APC-axin-containing protein complex (78 -80) in the cytoplasm, and TCF-type transcription factors (81) and BCL9 (35,82) in the nucleus. The dissociation of ␤-catenin with E-cadherin was recently shown to feed into the Wnt pathway (75), suggesting the interaction switch among ␤-catenin pools plays crucial roles in the control of Wnt/␤-catenin signaling. Furthermore, it is suggested that an inactive pool of ␤-catenin exists in the cell through its interaction with inhibitory proteins, post-translational modifications, and/or conformational changes (83). In this study, we demonstrate that Lzts2 regulates activities of canonical and noncanonical Wnt signaling through its regulation of the nuclear export of active ␤-catenin in zebrafish.
It is known that multiple intracellular pathways control the transport of ␤-catenin between the cytoplasm and the nucleus (84). The nuclear import of ␤-catenin is shown to be mediated by binding its central armadillo repeats to the nuclear pore complex, because Lzts2 contains no recognizable nuclear localization signal (85,86). The nuclear export of many large proteins, including Lzts2, is mediated through a short leucine-rich motif, known as the NES sequence (39,87). Major factors that are associated with the export of nuclear ␤-catenin include APC and Lzts2, which appear to be regulated by the CRM-1/ exportin-␣ pathway (39,80). In this study, we demonstrate that overexpression of Lzts2 with mutations in its NES results in the accumulation of ␤-catenin in the nuclear pore complex embedded in the nuclear envelope. Thus, zebrafish Lzts2 serves as a binding partner of ␤-catenin in an undefined nuclear export complex to regulate subcellular localization of ␤-catenin and thereby activities of canonical and nc-Wnt signaling.
Because subcellular localization and the amount of active ␤-catenin are closely associated with the regulation of cell adhesion and activity of Wnt/␤-catenin signaling, its aberrant activation leads to uncontrolled cell division in various tumor tissues (88,89). Recent evidence demonstrates that ␤-catenin is essential for the development, regeneration, and cancer pathogenesis of liver (90 -92). In contrast to the ␤-catenin-dependent Wnt pathway, which is activated in many cancer cells and serves as a tumor promoter, roles of the ␤-catenin-independent pathway are still controversial. Our current findings indicate that Lzts2 is a critical regulator of both Wnt/␤-catenin and nc-Wnt signaling through the export of nuclear ␤-catenin. It will be of great interest to investigate effects of Lzts2 on activity of intracellular ␤-catenin in pathological contexts.