Junctional β-Catenin Stabilization Links Wnt Signaling and Force Generation

Reviewer #1 (Public review):

In this study, Otgonbaatar and colleagues investigate the stability of Armadillo (Arm) during Drosophila development using a creative tandem fluorescent protein timer approach via endogenous tagging of Arm. The tagging strategy allows for newly synthesised and longer-term stabilised Arm pools to be distinguished from one another. Specifically, the authors address the functional relevance of and mechanism behind the stabilisation of junctional Arm during dorsal closure.

The authors show that Arm is stabilised at the leading edge during dorsal closure. Using a sophisticated optogenetics approach, which allows for acute perturbations, they show that stabilised Arm is functionally required for dorsal closure. Increasing Wg (by overexpression) did not affect dorsal closure or Arm stability, in contrast to Axin overexpression, which reduces Wg/Arm signalling. In line with canonical signalling control of Arm levels being critical, stabilisation of Arm by N-terminal mutations disrupted dorsal closure. However, the same deletion is also expected to affect interaction with alpha-catenin. Co-localisation with E-cadherin and actin suggests a junctional role of leading-edge localised Arm. Optogenetic targeting of alpha-catenin points towards a key role of adherence junctions in dorsal closure. Allele replacement with mutant variants of Arm to affect adherence junction complex assembly further indicates an important contribution of coupling between Arm and alpha-catenin. Using overexpression approaches, the authors suggest that Dsh and Jnk contribute to dorsal closure.

This microscopy- and optogenetics-based study is generally well-conducted and provides strong evidence for stabilised Arm during dorsal closure, as well as its functional importance. This is an important discovery relevant to morphogenesis and potentially mechanotransduction. From a technical perspective, the validated beta-catenin timer provides a valuable tool for the field. The timer has revealed that Arm stabilisation does not coincide with Wg stripes, suggesting a Wg-independent stabilisation mechanism that may instead depend on adherence junction assembly, especially the interaction of Arm with alpha-catenin. However, as N-terminal deletion within Arm and Axin overexpression also disrupted dorsal closure, substantial ambiguity remains. Can suppression of the beta-catenin degradation machinery be ruled out as a regulatory mechanism? An expansion of ArmTimer mutant variants could contribute to testing the authors' conclusion further. Structural insights into junctional interactions involving Arm (e.g., 10.1074/jbc.M114.554709) could, for example, be used for further functional exploration by mutagenesis. The direct mechanistic impact of JNK and its potential link to Dsh in dorsal closure remains less compelling.

In summary, this is a highly relevant and important study, potentially pointing to a novel stabilisation mechanism of beta-catenin in development. Further corroboration of the mechanism, to test whether it is indeed distinct from canonical signalling, would be needed to support the conclusions.

Reviewer #2 (Public review):

Summary:

Otgonbaatar et al. sought to investigate β-catenin/Arm protein lifetime and stabilization dynamics in vivo during embryonic development. To address this question, the authors developed an endogenous tandem fluorescent protein timer (tFP) system that enables the visualization of newly synthesized versus long-lived Arm protein in vivo. Using this approach, the authors sought to determine where stabilized Arm accumulates during development and how it contributes to dorsal closure.

Strengths:

A major strength of the study is the development and application of the endogenous Arm timer system, which provides a powerful approach for monitoring protein stabilization dynamics in living tissues. Using this system, the authors unexpectedly found that the strongest Arm stabilization occurs not in Wnt signaling regions, but at the leading edge cells during dorsal closure. The study combines quantitative live imaging, optogenetic perturbation, genetic analysis, and structure-function approaches to demonstrate that stabilized junctional Arm interacts with α-catenin and contributes to tissue mechanics required at the leading edge for dorsal closure. Particularly compelling is the combination of multiple perturbations, including optogenetic disruption of Arm or α-catenin, Axin overexpression, and Arm mutants, which produce consistent dorsal closure defects.

Some conclusions are generally supported by the presented data. The work provides strong evidence that Arm plays an important role in dorsal closure. The identification of a requirement for the Dishevelled DEP domain and JNK signaling supports a non-canonical regulatory mechanism controlling dorsal closure.

Weaknesses:

(1) Conclusions are made regarding force transmission;(however, no experimental evidence is provided to support these conclusions.

(2) The conclusion was made that Wingless does not affect dorsal closure. However, this was based solely on Wingless overexpression in the amnioserosa, and the level of Wingless expression was not quantified. One possibility is that this level was not sufficient to see an effect. Alternatively, Wingless may have a role in migrating epithelium rather than the amnioserosa. Indeed, it is known that wingless mutants display a defect in dorsal closure.

(3) The effect of JNK knockdown on Arm localization maybe is indirect, and due to a secondary consequence on disruption of epithelial morphology rather than a direct effect of JNK on Arm.

(4) Some conclusions rely on overexpression-based perturbations (e.g., Axin or Arm mutants), which may not fully recapitulate endogenous physiological regulation.

(5) The Arm timer was not able to detect Wingless-dependent Arm stabilization in stripes. This finding demonstrates that the timer is not sensitive enough to thoroughly analyze Arm dynamics.

Overall, this work provides important conceptual advances in understanding junctional β-catenin/Arm function during dorsal closure. The endogenous fluorescent timer approach will likely be broadly useful to the community for studying protein stability dynamics in vivo, and the findings expand current views of β-catenin by highlighting its mechanical and junctional functions during tissue morphogenesis.

Author response:

We are glad the reviewers found the tandem fluorescent timer approach valuable and the leading-edge Arm stabilization finding significant.

We agree with the Assessment that our evidence for three specific claims Wingless-independence, JNK-mediated regulation of Arm stability, and a direct mechanical/force-transmission role for stabilized Arm is currently incomplete, and we will revise the text throughout to reflect this more precisely rather than overstating the current data. In addition, we commit to two new experiments, both using existing reagents and fly stocks, that speak directly to the two most experimentally tractable points raised by the reviewers:

(1) Re-staining our existing JNK-RNAi and JNK-overexpression embryos for E-cadherin (reagent already validated in Figure 4), to test whether JNK acts directly on junctional architecture or only indirectly, via broader epithelial disruption.

(2) Imaging ArmTimer in a wingless loss-of-function background, to directly test Wingless-dependence of leading-edge Arm stabilization as the reciprocal of our existing overexpression data.

We address each public review point below and outline the accompanying text revisions.

On Wingless independence (Reviewer #1; Reviewer #2, Weaknesses #2 and #5; Recommendation #1):

We agree that our current evidence unquantified Wg overexpression restricted to the amnioserosa (C381-Gal4) and uniform overexpression, alongside the absence of detectable Wg-stripe-associated Arm-Timer signal supports a more limited conclusion than "Wingless-independent" as currently stated. We will revise our language throughout the Abstract, Results, and Discussion to state that canonical Wg overexpression does not detectably enhance leading-edge Arm stabilization or perturb dorsal closure under our conditions, rather than asserting pathway independence. As noted above, we commit to imaging ArmTimer in a wg mutant background to test this directly, complementing our overexpression data with the reciprocal loss-of-function manipulation.

On the related point that the Timer's failure to detect a Wg-stripe-associated stabilization signal could reflect a sensitivity limitation rather than a true absence of stabilization (Reviewer #2, Weakness #5): we agree and will state this explicitly rather than treating absence of signal as evidence of absence. This does not undermine the positive leading-edge finding, which is not defined relative to the stripe comparison: all embryos, channels, and time points were imaged and rendered using identical laser power and brightness/sensitivity settings, and the leading-edge RFP signal clearly exceeds background under those same acquisition conditions. We also note that detection limits of this kind are a recognized challenge for endogenously tagged reporters of canonical Wnt/β-catenin signaling generally, including in mammalian systems, and cite two studies already in our bibliography that report the same class of limitation: de Man et al. (2021, eLife 10:e66440) and Ambrosi et al. (2022, eLife 11:e64498). We have added this clarification, with these citations, to the Results (paragraph describing Figure 3).

On the mechanistic link between Arm stability and destruction-complex activity (Reviewer #1):

We agree that because both ΔArm and Axin overexpression converge on the destruction complex, our data cannot yet fully separate "escape from degradation" from "impaired α-catenin/junctional coupling" as the operative mechanism. We will revise the Discussion to state this ambiguity explicitly and will treat the ArmTimer-AA result (partial α-catenin-binding disruption via phosphosite mutation, independent of destruction-complex regulation) as the strongest current evidence isolating the junctional-coupling mechanism. We also thank Reviewer 1 for pointing us to Pokutta, Choi, Ahlsen, Hansen & Weis (2014, J Biol Chem 289:13589-13601), which structurally and thermodynamically characterized the mammalian cadherin·β-catenin·α-catenin complex and showed that α-catenin binding to β-catenin is a distinct, allosterically regulated interface cadherin binding increases β-catenin's affinity for α-catenin roughly 10-fold, and α-catenin homodimerization independently competes with β-catenin binding. We have added this citation to the Discussion as structural support for treating cadherin engagement, α-catenin coupling, and destruction-complex regulation as mechanistically separable interfaces, and note that the crystallized β-catenin·α-catenin interface provides a structural template for future experiments for example, structure-guided point mutations at the homologous interface residues in Arm, or in vitro binding assays comparing wild-type and threonine-mutant (T111A/T121A) Arm affinity for α-catenin.

We do not, however, believe a destruction-complex-independent stabilizing allele of Arm is a tractable experiment to close this gap directly: any allele that stabilizes Arm without engaging the destruction complex is, by definition, a Wnt pathway gain-of-function allele, since destruction-complex-mediated degradation is the very regulatory step that canonical Wnt signaling controls. Nor would restricting the allele to a transcriptionally inactive form of Arm cleanly resolve the confound: Wnt/TCF target loci include dedicated repressive TCF-binding sites (Blauwkamp, Chang & Cadigan, 2008, EMBO J 27:1436-1446), so a transcriptionally "dead" stabilized Arm could still alter transcription by disrupting TCF-mediated repression. We therefore treat this as a genuine, currently unresolvable confound of the overexpression approach, and rely instead on the CRY2 optogenetic and ArmTimer-AA results as the strongest available evidence isolating a junctional-coupling contribution. We have added this reasoning, with both citations, to the Discussion.

On JNK acting on Arm directly vs. indirectly (Reviewer #1; Reviewer #2, Weakness #3 and Recommendation #2):

This is the most actionable point raised by both reviewers. As noted above, we commit to re-imaging and re-staining our existing JNK-RNAi and JNK-overexpression embryos for E-cadherin to determine whether junctional/polarity architecture is broadly disrupted under these conditions (indirect mechanism) or whether E-cadherin localization is comparatively preserved while Arm stabilization is specifically altered (direct mechanism). In the meantime, we note that a direct mechanism is biochemically plausible: in mammalian cells, JNK phosphorylates β-catenin directly and regulates adherens junction integrity, and JNK activity separately controls the binding of α-catenin to the junctional complex (Lee, Koria, Qu & Andreadis, 2009, FASEB J 23:3874-3883; Lee, Padmashali, Koria & Andreadis, 2011, FASEB J 25:613-623). We cite these as precedent that a direct route from JNK to junctional β-catenin/α-catenin regulation exists in another system, while being explicit that this does not establish the same mechanism in Drosophila dorsal closure that will be tested directly by the E-cadherin re-staining experiment. We have added these citations and this caveat to the Discussion.

On the Dsh-DEP-to-JNK mechanistic link (Reviewer #1, Public Review #2 and Recommendation #5):

We agree that our data show the Dsh-DEP requirement and the JNK requirement for dorsal closure as parallel, independent findings rather than a demonstrated linear pathway in our system. To provide context for why we consider a DEP-to-JNK connection a reasonable working hypothesis, we searched the literature in both Drosophila and vertebrates and will cite six additional studies establishing this link: Axelrod et al. (1998) and Axelrod (2001), establishing that DEP-dependent membrane recruitment and unipolar localization of Dishevelled are specifically required for planar polarity signaling, distinct from Wingless signaling; Paricio et al. (1999) and Fanto et al. (2000), showing Dishevelled acts through Misshapen and Rac1/RhoA to the same JNK module used in dorsal closure; and Moriguchi et al. (1999) and Yamanaka et al. (2002), showing biochemically in vertebrates that the DEP domain of Dvl-1 selectively activates JNK independent of β-catenin/TCF-LEF activity, and that this JNK requirement is conserved in Xenopus convergent extension, the vertebrate process most functionally analogous to dorsal closure. We will state explicitly that this precedent, while now cross-species, comes from planar-cell-polarity and convergent-extension assays rather than dorsal closure itself, so it supports the plausibility of a Dsh/Dvl-DEP-to-JNK connection without establishing that the identical pathway operates in our system.

On the Dsh DIX/DEP domain-separability argument (Reviewer #1, Recommendation #4):

We thank the reviewer for pointing us to Gammons, Renko, Johnson, Rutherford & Bienz (2016, Mol Cell 64:92-104), which showed that the Wnt signalosome itself is assembled by head-to-tail DEP domain swapping between Dishevelled molecules, and that this DEP-dependent oligomerization is directly required for canonical Wnt pathway activity not restricted to the non-canonical/planar-polarity branch as we had implied. We agree this evidence undercuts our previous interpretation of the DshΔDEP dorsal closure phenotype as evidence for a strong non-canonical/polarity-specific role for the DEP-dependent branch of Dsh. We have revised the Discussion accordingly: we now state that the DEP domain is required for the morphogenetic program culminating in dorsal closure, cite Gammons et al. directly, and note that this requirement does not by itself establish a non-canonical/polarity-specific role, since we cannot rule out a contribution from DEP-dependent canonical Wnt signalosome assembly.

On force transmission (Reviewer #2, Weakness #1):

We agree that we have not directly measured force or tension at the leading edge, and that our current data (colocalization with actin/E-cadherin, and functional requirement shown via CRY2 optogenetics and mutant analysis) are consistent with, but do not directly demonstrate, a role in force transmission. We do not have the in-house expertise to perform direct force/tension measurements (e.g., laser ablation, junctional tension assays), so we will not be adding such an experiment in this revision. Instead, we have revised the language throughout the manuscript including two Discussion section headings that previously stated a mechanical role for stabilized Arm as established fact to consistently present the mechanical/force-transmission role as a hypothesis raised by our data, not a demonstrated conclusion, and we retain a clear statement that direct force measurement (ideally in collaboration with groups with the relevant biophysical expertise) is future work rather than a claim we are making in this manuscript.

On the phosphomimetic threonine mutant (Reviewer #1, Recommendation #2):

ArmTimer-AA (T111A, T121A) was generated with the expectation that the tyrosine phosphosite mutants (ArmTimer-EE, ArmTimer-FF) would be the primary drivers of any dorsal closure phenotype, given their proposed role in E-cadherin binding; the pronounced zippering defect we observed in ArmTimer-AA was therefore an unanticipated finding rather than a predicted result. We have not generated the reciprocal phosphomimetic ArmTimer-EE(Thr) (T111E, T121E) allele. Generating and characterizing this allele is a substantial undertaking we estimate over a year including allele generation, validation, and phenotypic characterization and we will state this explicitly in the Discussion as planned future work rather than part of the current revision.

On confirmation of myristoylated-Dsh membrane targeting (Reviewer #1, Recommendation #3):

We cannot confirm that myristoylation localizes all Dsh protein to the membrane. However, this strategy has extensive prior genetic validation using the identical Src-derived myristoylation sequence: it was originally used to tether Armadillo and shown sufficient for constitutive Wnt pathway activation (Zecca, Basler & Struhl, 1996; Tolwinski & Wieschaus, 2001, 2004), and the same approach was subsequently applied to GSK3 and Dishevelled, in each case producing the expected pathway-activation phenotypes (Mannava & Tolwinski, 2015; Kaur et al., 2017). We have added these citations to the Results where the Myr-Dsh constructs are introduced.

On overexpression-based perturbations versus endogenous regulation (Reviewer #2, Weakness #4):

We would like to clarify that most of the Arm alleles used in this study including all of the point-mutant Timer alleles (ArmF1a, ArmTimer-FF, ArmTimer-EE, ArmTimer-AA) central to our mechanistic conclusions were generated as knock-ins at the endogenous ‘arm’ locus via MiMIC/RMCE, not overexpressed. The two exceptions are ΔArm and ArmS56A, expressed from UAS constructs because both are gain-of-function alleles anticipated to be lethal if expressed from the endogenous locus, based on prior experience with similarly stabilizing mutations. Axin overexpression was used because no Axin mutant or knock-in allele was generated for this study; we agree an endogenous Axin allele would be the ideal complement and will state this explicitly as a limitation, while noting that our CRY2 optogenetic perturbations of Arm and α-catenin which act acutely on the endogenous proteins provide an orthogonal line of evidence supporting the same conclusions. We have added this clarification to the Discussion.