Ajuba family proteins are implicated in the assembly of cell junctions and have been reported to antagonize Hippo signaling in response to cytoskeletal tension. To assess the role of these proteins in actomyosin contractility, we examined the localization and function of Wtip, a member of the Ajuba family, in Xenopus early embryos. Targeted in vivo depletion of Wtip inhibited apical constriction in neuroepithelial cells and elicited neural tube defects. Fluorescent protein-tagged Wtip showed predominant punctate localization along the cell junctions in the epidermis and a linear junctional pattern in the neuroectoderm. In cells undergoing Shroom3-induced apical constriction, the punctate distribution was reorganized into a linear pattern. Conversely, the linear junctional pattern of Wtip in neuroectoderm changed to a more punctate distribution in cells with reduced myosin II activity. The C-terminal fragment of Wtip physically associated with Shroom3 and interfered with Shroom3 activity and neural fold formation. We therefore propose that Wtip is a tension-sensitive cytoskeletal adaptor that regulates apical constriction during vertebrate neurulation.
Ajuba LIM family members are ubiquitous LIM domain-containing proteins implicated in the regulation of cell junctions and signaling (Kadrmas and Beckerle, 2004; Schimizzi and Longmore, 2015). The N- and C-terminal domains of these proteins have been proposed to physically interact, although the evidence for this interaction has been largely indirect (Sun and Irvine, 2013). The single Drosophila gene homologue Jub regulates organ size in fly embryo development (Rauskolb et al., 2014). In vertebrates, in vivo functions of this family remain poorly characterized because of potential redundancy of the three closely related genes encoding Limd1, Ajuba and Wilms tumor-1-interacting protein (Wtip) (Schimizzi and Longmore, 2015).
Ajuba family proteins are present in multiple locations in the cell, possibly reflecting diverse molecular functions. Drosophila Jub is present at apical cell junctions in wing disc epithelial cells and this distribution depends on mechanical tension (Rauskolb et al., 2014; Sabino et al., 2011). Similarly, mammalian Ajuba localizes to adherens junctions and was shown to bind α-catenin, thereby linking adherens junctions and the actin cytoskeleton (Marie et al., 2003). At the junctions, Ajuba modulates Rac1 activity and inactivates Lats kinase to inhibit Hippo signaling (Das Thakur et al., 2010; Nola et al., 2011). Ajuba and Wtip were also found in the nucleus (Kanungo et al., 2000; Kim et al., 2010; Langer et al., 2008; Srichai et al., 2004), and Ajuba family proteins are known to interact with Snail2 transcriptional repressors to activate neural crest specific transcription (Langer et al., 2008). In addition, Ajuba family proteins were found at the centrosomes of mammalian cells and fly neuroblasts (Hirota et al., 2003; Sabino et al., 2011). This localization has been associated with the control of cell division, centrosome positioning and mitotic spindle orientation (Sabino et al., 2011; Schimizzi and Longmore, 2015). Moreover, Wtip has been implicated in planar polarity and ciliogenesis in Xenopus and zebrafish embryos (Bubenshchikova et al., 2012; Chu et al., 2016). Taken together, these observations highlight the roles of Ajuba family proteins as adaptors mediating signaling between different cell compartments.
To gain additional knowledge about this protein family, we examined the localization and functions of Wtip in Xenopus embryonic ectoderm. Gain- and loss-of-function studies of Wtip revealed its role as a mechanosensitive cytoskeletal adaptor required for apical constriction during vertebrate neural tube closure. The N- and C-terminal domains of Wtip exhibited different localization in ectoderm cells but colocalized when expressed together. Moreover, we observed that Wtip associates with the actin-binding protein Shroom3, a key player in apical constriction (Haigo et al., 2003; Hildebrand and Soriano, 1999), providing a potential mechanism for Wtip to modulate actomyosin contractility.
Wtip is required for apical constriction in the Xenopus neural plate
To interfere with the function of Wtip in developing Xenopus embryos, we used previously described antisense morpholino oligonucleotides (MOs) that efficiently block Wtip mRNA translation in vivo (Chu et al., 2016). Depletion of Wtip inhibited neural fold formation and resulted in neural tube defects (Fig. 1A-D). The defects were predominantly in the anterior portion of the neural tube, consistent with the notion that the neural tube ‘zippers’ in Xenopus embryos from the posterior towards the anterior aspect. The severity and frequency of these phenotypes were suppressed by wtip mRNA, confirming specificity. These defects were unlikely to be due to abnormal cell division, since the nuclear morphology appeared to be normal in Wtip morphants (Fig. S1A,B). Similar neural tube defects were observed in embryos injected with a second Wtip MO with a different sequence, but not in those injected with a control MO (Fig. 1D and data not shown). At later stages, neural tube deficiencies became less pronounced; however, mild axial and eye defects have been observed (Fig. S1C-E′). These observations suggest that Wtip functions in vertebrate neural tube closure.
Apical constriction in the Xenopus neural plate is visible by the formation of the dorsolateral hinges (Colas and Schoenwolf, 2001; Suzuki et al., 2012), which is suppressed in Wtip-depleted embryos (Fig. 1B,C, arrowheads). To examine whether this hinge formation defect is due to the failure of apical constriction, cryosections of embryos were stained with Phalloidin to visualize F-actin. F-actin was enriched at the apical surfaces of constricting neuroepithelial cells, as previously reported (Rolo et al., 2009), but this enrichment was not detected in cells containing Wtip MO and lineage tracer (Fig. 1E,E′). In addition, cells depleted of Wtip had wider apical surfaces compared with those in cells in the uninjected half of the neural plate (Fig. 1E-G). Together, these findings indicate that Wtip is required for apical constriction during vertebrate neurulation.
Subcellular localization of Wtip and its fragments in embryonic ectoderm
To investigate the subcellular localization of Wtip in Xenopus ectoderm, fusions of Wtip with fluorescent proteins were constructed and expressed in early embryos. In superficial ectoderm cells during gastrulation, HA-RFP-Wtip was localized at the cellular junctions and in cytoplasmic puncta (Fig. 2A). At stage 12, the junctional puncta became more predominant in the epidermis (Fig. 2B). Frequently, the junctional puncta were visible in both neighboring cells as doublets positioned across the cell junctions (Fig. 2B′,C-C″). At stage 14, the cytoplasmic puncta of Wtip were more abundant (Fig. 2D-D″). In contrast, in neural ectoderm, Wtip showed continuous junctional staining and was enriched at the tricellular junctions, which was similar to the distribution of F-actin (Fig. 2E-E″). Tricellulin, a marker for tricellular junctions, was surrounded by Wtip (Fig. 2F-F″, arrows), implying the involvement of Wtip in tension sensing at the tricellular junction (Higashi and Miller, 2017). At later stages, Wtip was retained in cytoplasmic puncta and localized to the basal bodies of skin multiciliated cells (Fig. S2A). This is consistent with the reported centrosomal localization of Ajuba family proteins in several cell types (Chu et al., 2016; Hirota et al., 2003; Sabino et al., 2011). GFP-tagged Wtip exhibited a similar subcellular distribution (data not shown).
We next wanted to define specific protein sequences needed to localize Wtip to the puncta and the centrosome/basal body. Several truncated constructs were generated, including the N-terminal fragment (WtipN, amino acids 1-480) containing repeated serine-rich motifs and the C-terminal fragment (WtipC, amino acids 481-690) with three conserved LIM domains (Fig. 3A). As observed for full-length Wtip, WtipN also formed highly organized peri-junctional puncta visualized in both live embryos and fixed samples (Fig. 3B,B′ and data not shown). In contrast, WtipC was largely present at the cortex and in the nucleus (Fig. 3C,F), in agreement with the reported nuclear localization signal (Kanungo et al., 2000; Srichai et al., 2004). Of note, the small region between amino acids 1-250 was sufficient to target GFP to the centrosome and basal body, but not cortical puncta (Fig. S2B,C). These experiments reveal distinct targeting signals in Wtip.
The junctional punctate signal of Wtip and WtipN often appeared as a doublet across the cell border. Images of neighboring cells expressing either HA-RFP-WtipN or GFP-WtipN revealed pairing of differently colored puncta along the cell junction (Fig. 3D-D″), confirming that their positions are coordinated across the cell membrane. Since this pattern is reminiscent of sarcomere-like actomyosin complexes observed in some epithelial cells (Choi et al., 2016; Ebrahim et al., 2013; Fanning et al., 2012), we examined colocalization of Wtip with several actin- and myosin-interacting proteins. We found that the distribution of Moesin actin-binding domain (Moe-GFP), a marker of F-actin (Edwards et al., 1997; Skoglund et al., 2008), occasionally alternated with the position of WtipN puncta (Fig. 3E-E″). However, Phalloidin or α-actinin staining did not exhibit a punctate pattern (Fig. 2C′ and data not shown), suggesting that a specific yet uncharacterized pool of F-actin may play a role in the regulation of Wtip localization.
Interaction of the N- and C-terminal Wtip domains
Zyxin is a LIM-domain protein that is closely related to the Ajuba family. Previous studies support a model in which Zyxin maintains a ‘closed’ conformation as a result of the intramolecular interaction between its N- and C-terminal domains (Hirota et al., 2000; Moody et al., 2009). We therefore examined whether a similar interaction exists in Wtip by coexpressing WtipC with WtipN or full-length Wtip. WtipC was recruited to the cortical puncta containing full-length Wtip or WtipN (Fig. 3F-J″), providing evidence for association of the two domains in vivo.
Wtip junctional localization is altered in a tension-dependent manner
Because apical constriction is known to generate actomyosin-dependent tension (Martin and Goldstein, 2014), we examined whether Wtip distribution is modulated by Shroom3, which induces ectopic apical constriction in Xenopus ectoderm (Haigo et al., 2003) by recruiting Rho-associated kinase to apical junctions (Das et al., 2014; Nishimura and Takeichi, 2008). In ectodermal cells overexpressing Shroom3, the punctate localization of HA-RFP-Wtip at the junctions was rearranged into a continuous pattern (Fig. 4A-D). This change was visible at the onset of apical constriction (Fig. 4A-B′) and became more evident over time (Fig. 4C,D). HA-RFP-Wtip was partially colocalized with Shroom3 at the junctions and cytoplasmic puncta (Fig. 4E-E″). This finding suggests that Wtip is recruited to apical junctional complexes in response to increased cytoskeletal tension.
Since Wtip is redistributed to junctional complexes in response to enhanced actomyosin contractions, we wanted to test its localization in cells with reduced myosin activity. This complementary experiment was carried out in the neuroectoderm, in which Wtip localization at cell junctions is continuous (Fig. 2E,F). Myosin function is inhibited by Mypt1T696A, a constitutively active form of the myosin light chain phosphatase subunit Mypt1 (Weiser et al., 2009). Upon coexpression of Mypt1T696A, Wtip distribution became more punctate, both in the cytoplasm and at the junctions, and the enrichment of Wtip at the tricellular junctions was diminished (Fig. 5A,B). Unlike full-length Wtip, neither WtipN nor WtipC revealed any sensitivity to Mypt1T696A (Fig. 5C-F). Our observations therefore suggest that Wtip responds to mechanical tension triggered by actomyosin contractions and that both the N- and C-terminal domains are required for this process.
The interaction of WtipC and Shroom3 modulates apical constriction
Since both Shroom3 and Wtip are junctional, and Shroom3 affects Wtip localization, we tested whether Wtip physically interacts with Shroom3. Immunoprecipitation analysis using transfected HEK293T cells revealed the binding of Shroom3 to Wtip (Fig. 6A). This binding was confirmed by a reciprocal pull-down experiment using mRNA-injected embryos, and we further demonstrated that Shroom3 binds specifically to WtipC rather than WtipN (Fig. 6B). When coexpressed with Shroom3 in the ectoderm, WtipC, but not WtipN, blocked Shroom3-mediated apical constriction almost completely (Fig. 6C-G). Moreover, overexpression of WtipC on one side of the embryo caused a unilateral neural tube defect (Fig. 6H,I), which was similar to the defect in Wtip-depleted embryos (Fig. 1A-D). Taken together, these observations suggest that Wtip regulates neural tube closure by modulating Shroom3-dependent apical constriction via its C-terminus.
We next examined whether Wtip constructs can modulate the localization of Myc-Shroom3. At the onset of apical constriction, Shroom3 showed punctate distribution at the cell junctions and cytoplasm, and this pattern remained unchanged upon WtipN expression (Fig. 7A,B). However, in cells overexpressing WtipC, staining of Shroom3 was reduced at the junctions and became more diffuse in the cytoplasm (Fig. 7C). The protein level of Shroom3 was not affected by the expression of WtipN or WtipC (Fig. 7D). These experiments suggest that WtipC blocks Shroom3-induced apical constriction by displacing Shroom3 from the cell junctions.
We also tested whether binding of Wtip to Shroom3 is sensitive to tension. Mypt1T696A was co-expressed with Wtip and Shroom3 to reduce tension, and its effect was visualized by the reduction of apical constriction in all injected embryos (Fig. S3A). Immunoprecipitation analysis showed that similar amounts of Shroom3 were pulled down by Wtip regardless of the presence of Mypt1T696A (Fig. S3B), suggesting that binding of Wtip to Shroom3 is not regulated by a change in tension.
This study evaluated the functions of Xenopus Wtip, an Ajuba family protein, in embryonic ectoderm. Previous studies implicated Wtip in neural crest specification and ciliogenesis in Xenopus embryos (Chu et al., 2016; Langer et al., 2008) and, in vitro, in the adhesion of mouse podocytes (Kim et al., 2012). Our depletion and interference experiments suggest that an important developmental function of Wtip is to mediate apical constriction during neural tube closure. We propose that Wtip is a cytoskeletal adaptor that regulates actomyosin contractility at the apical junctions. This novel function of Wtip expands the list of the cellular and developmental processes that involve the Ajuba family.
Mechanical forces generated by actomyosin contractions can alter protein conformation, binding partners and signaling activity, as has been demonstrated for the binding of vinculin to α-catenin (Yonemura et al., 2010). Consistent with this notion, Wtip puncta at the junctions are converted into a linear (continuous) distribution in response to apical constriction triggered by Shroom3. Moreover, relaxation of tension by Mypt1, which inhibits myosin II activity, leads to a complementary result: Wtip distribution is changed from a linear to a punctate pattern. We speculate that these localization patterns reflect the formation of diverse Wtip-containing protein complexes that are rearranged in a tension-dependent manner, which is somewhat similar to the behavior of Jub in Drosophila wing discs (Rauskolb et al., 2014). Interestingly, Wtip is localized at the tricellular junctions in a similar manner to vinculin (Higashi and Miller, 2017), implying a connection between Wtip and the vinculin–α-catenin complex under high tension. Notably, Ajuba can be recruited to adherens junctions through an interaction with α-catenin (Marie et al., 2003; Rauskolb et al., 2014), suggesting that Wtip might also associate with α-catenin. Future studies are required to address this possibility.
Since neither WtipN nor WtipC respond to Mypt1 in the neural plate, as full-length Wtip does, both domains are likely to contribute to Wtip function at the junctions. While the interacting partners of WtipN at the apical junctions remain unknown, we identified an association of WtipC with Shroom3. Given that it is still unclear how Shroom3 becomes localized to apical junctions (Hildebrand, 2005; Lang et al., 2014), we propose that Wtip functions as an adaptor that recruits Shroom3 to apical junctions to allow spatially restricted actomyosin contractions. The contraction increases junctional tension and, in turn, positively regulates Wtip localization to the junctions (Fig. 7E). Consistent with this model, overexpression of WtipC reduced Shroom3 junctional localization, inhibited Shroom3-mediated apical constriction and caused neural tube closure defects. However, Wtip MO caused only a slight change in Shroom3 localization (data not shown), possibly due to insufficient depletion or the presence of other Ajuba family proteins in this tissue. Alternatively, because Wtip localization changes in response to Shroom3, Shroom3 may function upstream of Wtip. At present, potential redundancies and feedback regulation at the apical junctions leave open the question of the epistatic relationship between Wtip and Shroom3.
Our experiments also revealed the interaction between the N- and C-terminal domains of Wtip. Upon coexpression, the puncta containing WtipN or Wtip also contained WtipC, consistent with an intramolecular or intermolecular interaction, in which the C-terminus of one Wtip molecule associates with the N-terminus of another, leading to head-to-tail aggregation. Notably, in Zyxin, this association has been proposed to lead to a ‘closed’ (i.e. inactive) conformation that prevents it from binding to phosphorylated VASP (Hirota et al., 2000; Moody et al., 2009). We found that Wtip and WtipC can pull down similar amounts of Shroom3, suggesting that binding of Shroom3 to WtipC is not blocked by WtipN. Future work will define the role of the WtipN-WtipC interaction in Wtip functions and evaluate whether it is regulated by mechanical forces and/or JNK-dependent phosphorylation, as proposed by recent studies (Rauskolb et al., 2014; Sun and Irvine, 2013).
MATERIALS AND METHODS
Plasmids, in vitro mRNA synthesis and morpholino oligonucleotides (MOs)
pCS2-Myc-Wtip (Xenopus) was a gift from Greg Longmore and Kris Kroll (Langer et al., 2008). The following plasmids have been previously described: pCS2-Myc-Shroom3 and pCS2-FLAG-Shroom3 (Chu et al., 2013; Plageman et al., 2010), pCS107-Moesin-GFP (Skoglund et al., 2008), pCS2-GFP-Tricellulin (Higashi et al., 2016), Mypt1T696A (Weiser et al., 2009), Histone-GFP (Petridou and Skourides, 2014). pCS2-GFP-Wtip and pCS105-HA-RFP-Wtip have been described (Chu et al., 2016). pCS2-GFP-WtipN and pCS105-HA-RFP-WtipN encodes amino acids 1-480 of Wtip. pXT7-GFP-WtipN251 encodes the first 251 amino acids of Wtip. pCS2-HA-RFP-WtipC and pCS2-GFP-WtipC encodes amino acids 481-690 of Wtip. Coding sequences of Wtip full-length, WtipN, and WtipC were subcloned into pCS2-FLAG. Details of plasmid construction are available upon request. Capped mRNAs were synthesized using mMessage mMachine kits (Ambion, Austin, TX), according to manufacturer's instructions. MOs were purchased from Gene Tools (Philomath, OR, USA). The following MOs were used: control MO, 5′-GCTTCAGCTAGTGACACATGCAT-3′ (Ossipova et al., 2015), WtipMO1, 5′-TGTCCTCATCGTACTTCTCCATGTC-3′ (Chu et al., 2016).
Xenopus embryo culture and microinjections
In vitro fertilization and culture of Xenopus laevis embryos were carried out as previously described (Dollar et al., 2005). The study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol 04-1295 was approved by the IACUC of the Icahn School of Medicine at Mount Sinai. Staging was according to Nieuwkoop and Faber (1994). For microinjections, four-cell embryos were transferred into 3% Ficoll in 0.5× MMR buffer (50 mM NaCl, 1 mM KCl, 1 mM CaCl2, 0.5 mM MgCl2, 2.5 mM HEPES pH 7.4) and 10 nl mRNA or MO solution was injected into one or more blastomeres. Amounts of injected mRNA per embryo have been optimized in preliminary dose-response experiments (data not shown).
Sectioning, immunofluorescence staining and imaging
Ectodermal and neural plate explants were isolated from embryos fixed in MEMFA (0.1 M MOPS, pH 7.4, 2 mM EGTA, 1 mM MgSO4 and 3.7% formaldehyde) (Harland, 1991). For sectioning, embryos fixed in MEMFA (for Phalloidin staining) or Dent's fixative (80% methanol+20% DMSO, for β-catenin staining) were embedded and cryosectioned using a Leica cryostat CM3050 at 10 µm as described previously (Ossipova et al., 2014). Indirect immunofluorescence staining was performed as described (Ossipova et al., 2014), the blocking solution was PBS+0.1% Triton X-100+3% BSA+3% goat serum. Mouse anti-GFP (B-2, Santa Cruz), rabbit anti-β-catenin (c2206, Sigma) and mouse anti-Myc (9E10 hybridoma) antibodies were used at 1:200, and Alexa Fluor 488-conjugated goat anti-mouse antibody (Invitrogen) and Cy3-conjugated donkey anti-mouse antibody (Jackson ImmunoResearch Laboratories) were used at 1:400. Alexa Fluor 488- or 568-conjugated Phalloidin (Thermo Fisher) were used for F-actin staining. Explants were mounted for observation in the Vectashield mounting medium (Vector Laboratories). Images were captured using a Zeiss AxioImager microscope with the Apotome attachment. AxioVision software (Zeiss) was used for image processing and measurements. Quantification of apical constriction was carried out using images of β-catenin-stained sections from 12 embryos. Apical width was defined as the distance between apical junctions, and cell height was defined as the distance between the midpoints of the apical and basal domains (Chu et al., 2013). Up to five superficial neuroectodermal cells adjacent to the midline were measured. Results shown are representative images from 2-4 independent experiments with 8-12 embryos per group.
Cell culture and transfection
Human embryonic kidney 293T cells (ATCC) were maintained in DMEM (Corning) with 10% FBS (Gemini) and penicillin-streptomycin (Sigma). Cells growing at 70% confluence were transiently transfected using linear polyethylenimine (MW 25,000, Polysciences) as described (Ossipova et al., 2009). Each 35 mm dish of cells received 1.5 µg of pCS2 plasmids encoding FLAG-Shroom3 or Myc-Wtip as indicated. pCS2 vector DNA was added to plasmid DNA mixture to reach a total DNA amount of 3 µg.
Immunoprecipitation and western blot analysis
For immunoprecipitation, cells transfected for 24 h were lysed in immunoprecipitation buffer (10 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM EGTA, 1 mM MgCl2, 1% Triton X-100, 1 mM Na3VO4, 10 mM NaF, 25 mM β-glycerol phosphate), containing protease inhibitor cocktail (cOmplete Mini EDTA-free, Roche). For Xenopus gastrula embryos, 50 mM NaCl was used instead of 150 mM to remove the majority of yolk proteins. After centrifugation at 15,000 g, the supernatant was incubated with anti-FLAG agarose beads (Sigma) at room temperature for 2-3 h. The beads were washed, boiled in SDS-PAGE sample buffer, and subjected to SDS-PAGE and immunoblotting, essentially as described (Chu et al., 2013). The following primary antibodies were used: mouse anti-GFP (B-2, Santa Cruz), mouse anti-FLAG (M2, Sigma), mouse anti-Myc (9E10). Chemiluminescence was captured by the ChemiDoc MP imager (Bio-Rad).
We thank Ann Miller, Paul Skoglund, Ray Keller, Kristen Kroll and Greg Longmore for plasmids. We thank members of the Sokol laboratory for discussions.
Conceptualization: C.-W.C., B.X, S.Y.S.; Methodology: C.-W.C., B.X., O.O., A.I., S.Y.S.; Validation: A.I.; Formal analysis: B.X., O.O., A.I.; Investigation: C.-W.C., B.X., O.O., A.I., S.Y.S.; Writing - original draft: C.-W.C., S.Y.S.; Writing - review & editing: C.-W.C., B.X., O.O., S.Y.S.; Visualization: C.-W.C., O.O.; Supervision: S.Y.S.; Funding acquisition: S.Y.S.
This study was supported by the National Institutes of Health (GM122492 to S.Y.S.). Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.