Core PCP mutations affect short-time mechanical properties but not tissue morphogenesis in the Drosophila pupal wing

  1. Romina Piscitello-Gómez
  2. Franz S Gruber
  3. Abhijeet Krishna
  4. Charlie Duclut
  5. Carl D Modes
  6. Marko Popović
  7. Frank Jülicher
  8. Natalie A Dye  Is a corresponding author
  9. Suzanne Eaton
  1. Max Planck Institute of Molecular Cell Biology and Genetics, Germany
  2. DFG Excellence Cluster Physics of Life, Technische Universität Dresden, Germany
  3. National Phenotypic Screening Centre, University of Dundee, United Kingdom
  4. Center for Systems Biology Dresden, Germany
  5. Laboratoire Physico-Chimie Curie, CNRS UMR 168, Institut Curie, Université PSL, Sorbonne Université, France
  6. Max Planck Institute for Physics of Complex Systems, Germany
  7. Université Paris Cité, Laboratoire Matière et Systèmes Complexes, France
  8. Mildred Scheel Nachwuchszentrum P2, Medical Faculty, Technische Universität Dresden, Germany
  9. Biotechnologisches Zentrum, Technische Universität Dresden, Germany

Abstract

How morphogenetic movements are robustly coordinated in space and time is a fundamental open question in biology. We study this question using the wing of Drosophila melanogaster, an epithelial tissue that undergoes large-scale tissue flows during pupal stages. Previously, we showed that pupal wing morphogenesis involves both cellular behaviors that allow relaxation of mechanical tissue stress, as well as cellular behaviors that appear to be actively patterned (Etournay et al., 2015). Here, we show that these active cellular behaviors are not guided by the core planar cell polarity (PCP) pathway, a conserved signaling system that guides tissue development in many other contexts. We find no significant phenotype on the cellular dynamics underlying pupal morphogenesis in mutants of core PCP. Furthermore, using laser ablation experiments, coupled with a rheological model to describe the dynamics of the response to laser ablation, we conclude that while core PCP mutations affect the fast timescale response to laser ablation they do not significantly affect overall tissue mechanics. In conclusion, our work shows that cellular dynamics and tissue shape changes during Drosophila pupal wing morphogenesis do not require core PCP as an orientational guiding cue.

Editor's evaluation

This valuable study shows that the core PCP pathway, which commonly orients morphogenetic processes in development, does not establish global cues for cellular movements in the Drosophila pupal wing. Using a combination of laser-ablation experiments and mathematical modelling, it provides compelling evidence that the core PCP pathway only affects the fast timescale cellular response, but does not appear to drive overall tissue dynamics. While the signalling pathway guiding this morphogenetic process remains to be elucidated, these relevant findings challenge the role of the core PCP pathway in morphogenesis.

https://doi.org/10.7554/eLife.85581.sa0

Introduction

The spatial–temporal pattern of mechanical deformation during tissue morphogenesis is often guided by patterns of chemical signaling. Precisely how chemical signaling couples with the mechanics of morphogenesis, however, remains an active area of research. One conserved chemical signaling pathway that is known to be patterned across tissues is the core planar cell polarity (PCP) pathway, composed of a dynamic set of interacting membrane proteins that polarizes intracellularly within the plane of a tissue. Tissue-scale alignment of this pathway is known to orient cellular structures, such as hairs and cilia, and influence dynamic cellular behaviors during morphogenesis, such as cellular movements and cell divisions, through interactions with the cytoskeleton (reviewed in Devenport, 2014; Butler and Wallingford, 2017; Deans, 2021).

Here, we examine a potential role for the core PCP pathway in the dynamics and mechanics of morphogenesis using the Drosophila pupal wing. The Drosophila wing is a flat epithelium that can be imaged at high spatial–temporal resolution in vivo during large-scale tissue flows that elongate the wing blade (Aigouy et al., 2010; Etournay et al., 2015; Guirao et al., 2015). During the pupal stage, the proximal hinge region of the wing contracts and pulls on the blade region, generating mechanical stress that is counteracted by marginal connections mediated by the extracellular matrix protein Dumpy (Etournay et al., 2015; Ray et al., 2015). As a consequence, the tissue elongates along the proximal–distal (PD) axis and narrows along the anterior–posterior (AP) axis to resemble the adult wing. Both cell elongation changes and cell rearrangements are important for tissue deformation. To some extent, mechanical stress induces these cell behaviors. However, the reduction of mechanical stress in a dumpy mutant does not completely eliminate cell rearrangements, suggesting that there could be other patterning cues that drive oriented cell rearrangements (Etournay et al., 2015). We therefore wondered whether chemical PCP systems could orient cell behaviors, such as cell rearrangements, during pupal blade elongation flows.

In the Drosophila wing, there are two PCP systems termed Fat and core PCP (Matis and Axelrod, 2013; Adler, 2012; Devenport, 2014; Butler and Wallingford, 2017). The Fat PCP system consists of two cadherins Fat and Dachsous, a cytoplasmic kinase Four-jointed, and an atypical myosin Dachs. The core PCP system is composed of two transmembrane proteins Frizzled (Fz) and Flamingo or Starry night (Fmi, Stan), the transmembrane protein Strabismus or Van Gogh (Stbm, Vang), and the cytosolic components Dishevelled (Dsh), Prickle (Pk), and Diego (Dgo).

Our group has shown that tissue-scale patterns of PCP emerge during larval stages and then are dynamically reoriented during pupal tissue flows (Sagner et al., 2012; Aigouy et al., 2010; Merkel et al., 2014). At the onset of blade elongation flows, both systems are margin oriented, however as morphogenesis proceeds, core PCP reorients to point along the PD axis, whereas Fat PCP remains margin oriented until very late, when it reorients toward veins (Figure 1—figure supplement 1A, B; Merkel et al., 2014). Whether these PCP systems and their reorientation influence tissue dynamics and mechanics during blade elongation flows is unknown.

The core PCP pathway has been shown to influence numerous processes in Drosophila tissue development. These include hexagonal cell packing in the late pupal wing (Classen et al., 2005; Sugimura et al., 2016), as well as patterning of ommatidial clusters in the developing eye (Zheng et al., 1995; Jenny, 2010), orientation of cell division in sensory organ precursors (Gho and Schweisguth, 1998), formation of joints in the legs (Capilla et al., 2012), and regulation of tracheal tube length (Chung et al., 2009). In many cases, the mechanism connecting the core PCP pathway to cell dynamics and tissue mechanics is unclear. Recent studies suggest, however, that core PCP may act in concert with Nemo kinase to regulate cell rearrangements in the eye (Mirkovic et al., 2011; Founounou et al., 2021) and with the Drosophila NuMA ortholog Mud to orient cell division orientation in the sensory organ precursors (Ségalen et al., 2010).

Here, we examine cellular dynamics in tissues mutant for core PCP and we find that they are largely unperturbed, indicating that core PCP does not have an essential role in organizing global patterns of cell rearrangements in the pupal wing. We also performed an extensive analysis of the mechanics using laser ablation, developing a rheological model to interpret the results. We find that mutants in core PCP differ from wild type in the initial recoil velocity upon laser ablation. We find, however, that this difference is produced from the very fast timescale response, which does not appear to affect morphogenesis and overall tissue stresses, consistent with the lack of phenotype in cellular dynamics.

Results

Core PCP does not guide cellular dynamics during pupal blade elongation flows

To investigate the role of core PCP in orienting cell behaviors during pupal blade elongation flows, we analyzed cell dynamics in wild type (wt) and three different core PCP mutant tissues: prickle (pk30, abbreviated as pk), strabismus (stbm6, abbreviated as stbm), and flamingo (fmifrz3, aka stanfrz3, abbreviated as fmi). In pk, the core and Fat PCP systems remain aligned together toward the margin and the magnitude of Stbm polarity is reduced (Merkel et al., 2014). The mutants stbm and fmi are strong hypomorphs, where the core PCP network is strongly reduced (Figure 1—figure supplement 1B; Merkel et al., 2014). We analyzed shape changes of the wing blade during blade elongation flows and decomposed these changes into contributions from cell elongation changes and cell rearrangements, which include cell neighbor exchanges, cell divisions, cell extrusions, and correlation effects (Figure 1; Etournay et al., 2015; Merkel et al., 2017).

Figure 1 with 5 supplements see all
Core planar cell polarity (PCP) does not orient cellular behaviors and tissue reshaping during pupal blade elongation flows: (A) Cartoon of pupal wing dissection at 16 hAPF and imaging using a spinning disk microscope.

(B–B″) Images of a wt wing at −7, 0, and 9 hRPCE (for this movie these times correspond to 16, 23, and 32 hAPF). The green and orange regions correspond to the hinge and blade, respectively. Anterior is up; proximal to the left. Scale bar, 100 μm. (C) Schematic of the cellular contributions underlying anisotropic tissue deformation. The tissue shear rate component vxx, which quantifies the rate of anisotropic tissue deformation along the proximal–distal wing axis, is decomposed into deformations arising from the rate of change of cell shapes DQxx/Dt and the deformations arising from the cellular rearrangements Rxx (Etournay et al., 2015; Merkel et al., 2017). Total shear is the sum of cell elongation changes (green) and cell rearrangements (magenta). (D–D′″) Accumulated proximal–distal (Cum PD) tissue shear during blade elongation flows in the blade region averaged for (D) wt (n = 4), (D′) pk (n = 3), (D″) stbm (n = 3), and (D′″) fmi (n = 2) movies. Solid line indicates the mean, and the shaded regions enclose ± standard error of the mean (SEM). Differences in total accumulated shear are not statistically significant (Figure 1—figure supplement 2C). Time is relative to peak cell elongation (hRPCE).

Figure 1—source data 1

Numerical data for Figure 1D–D′′′, accumulated proximal–distal tissue shear during blade elongation flows in the blade region for wt and core PCP mutants.

https://cdn.elifesciences.org/articles/85581/elife-85581-fig1-data1-v3.zip

In wt, the wing blade elongates along the PD axis (blue line in Figure 1D). Cells first elongate along the PD axis and then relax to more isotropic shapes (green line in Figure 1D). Cell rearrangements, however, go the opposite direction, initially contributing to AP deformation, before turning around to contribute to PD deformation (magenta line in Figure 1D). We introduce here a relative timescale, where we measure time in hours relative to the peak in cell elongation (hRPCE). This new scale allows us to handle variation in the timing of the onset of the blade elongation flows, which we have observed recently (see Appendix 1).

In core PCP mutants, we find that the dynamics of tissue shear, cell elongation changes, and cell rearrangements, when averaged across the entire blade, occur normally (Figure 1D–D′′′). We observe that by the end of the process, only slightly lower total shear appears to occur in the core PCP mutants, caused by slightly less cell rearrangements, but these subtle changes are not statistically significant (Figure 1—figure supplement 2C). The cellular dynamics contributing to isotropic tissue deformation are also broadly the same between wt and core PCP mutant tissues (Figure 1—figure supplement 2D, E). We also looked for differences in the behavior of regions of the wing blade subdivided along the PD axis (Figure 1—figure supplement 3), as previous work has shown that distal regions of the wing blade shear more at early times, whereas proximal regions start deforming later (Merkel et al., 2017). Again, we do not find strong differences between core PCP mutants and wt when we subdivide the wing into regions along the PD axis (Figure 1—figure supplement 3, Figure 1—figure supplement 4).

From this analysis, we conclude that core PCP is not required to determine the global patterns of cell dynamics during blade elongation flows. Interestingly, core PCP mutants do have a subtle but significant phenotype in the adult wing shape: pk and stbm (but not fmi) mutant wings are slightly rounder and wider than wt (Figure 1—figure supplement 5I). In principle, these small differences could arise after the blade elongation flows studied here. However, it is also possible that the we could not reliably detect these subtle differences in pupal wings due to the small number of wings per genotype that we were able to analyze (n = 2−4). To illustrate this point, we used the pool of adult wings (n=53 for wt, n=47 for pk, n=74 for stbm, and n=56 for fmi), where the phenotype is significant, to understand the probability that a sample of smaller size m would provide a significant signal, see Figure 1—figure supplement 5J. For m=3, corresponding to the number of pupal wings we analyzed, we find that only about 20% of samples show a significant phenotype. In other words, if the same magnitude of difference occurred during the blade elongation flows as in the adult, we would have only about a 20% chance to observe it. Therefore, core PCP could subtly influence the cell dynamics occurring at this stage. To investigate this possibility, we next looked for a possible difference in mechanical stresses in core PCP mutants.

A rheological model for the response to laser ablation

We investigated cell and tissue mechanics in core PCP mutants using laser ablation in a small region of the wing blade. We used a region located between the second and third sensory organs in the intervein region between the L3 and L4 longitudinal veins, which is a region that is easy to identify throughout blade elongation flows (Figure 2A). We cut three to four cells in a line along the AP axis and measured the displacement of the tissue (Figure 2A, Video 1). We calculate the initial recoil velocity v by measuring the average displacement of ablated cell membranes at the first observed timepoint after the ablation, δt=0.65s (see Materials and methods, Linear laser ablations to calculate the initial recoil velocity). Previously, we reported that initial recoil velocity measured along the PD axis in wt peaks around −8 hRPCE (20 hAPF in Iyer et al., 2019), and therefore we first focus on this timepoint. We find that core PCP mutants have significantly lower initial recoil velocity (Figure 2B, Figure 2—figure supplement 1A), suggesting that there is a mechanical defect in these mutants.

Figure 2 with 1 supplement see all
Rheological model for the response to laser ablation: (A) Schematic of a wt wing at −8 hRPCE.

Linear laser ablation experiments were performed in the blade region enclosed by the red square. Dots on the wing cartoon indicate sensory organs. The red line corresponds to the ablation, and the kymograph was drawn perpendicularly to the cut (yellow). Scale bar, 5 μm. (B) Initial recoil velocity upon ablation (simplified as recoil velocity in the y-axis title) along the proximal–distal (PD) axis at −8 hRPCE for wt (gray) and pk (blue) tissues (n ≥ 9). Significance is estimated using the Mann–Whitney U-test. ***p-val ≤0.001. (C) Example of the measured displacement after laser ablation (black dots) and corresponding exponential fit of the mechanical model (red curve). The blue and green regions highlight the displacement in the fast and slow timescale, respectively. (D) Description of the mechanical model that was devised to analyze the tissue response upon laser ablation. After the cut, the spring with elastic constant k is ablated (red scissor), and the tissue response is given by the combination of the two Kelvin–Voigt models arranged in series. These two correspond to the fast response given by kf and ηf and the slow response given by ks and ηs. The mechanical stress σ is constant. The membrane displacement Δx(t) is calculated as a sum of the displacement (Xf) associated with the fast timescale (τf) and the displacement (Xs) associated with the slow timescale (τs). (E–E′′′) Values obtained for each of the four fitting parameters when fit to the data. (E) Displacement associated with the fast and (E′) slow timescale for wt (gray) and pk (blue). (E′′) Fast and (E′′′) slow timescale for wt (gray) and pk (blue) (n ≥ 5). Significance is estimated using the Student’s t-test. **p-val ≤0.01; ns, p-val >0.05. (F) Example of a circular laser ablation used for analysis with elliptical shape after circular ablation (ESCA). The left image shows the final shape of the ablation around 2 min after cut, and the right image shows the corresponding segmented image, where the inner and outer pieces were fit with ellipses. After the fitting, the model outputs the anisotropic and isotropic stress (equations shown on the right side). Scale bar, 20 μm. A = anterior, P* = posterior, D = distal, P = proximal. (F′) Anisotropic stress σ~/2K for wt (gray) and pk (blue) tissues at −8 hRPCE (n ≥ 4). Significance is estimated using the Mann–Whitney U-test. ns, p-val >0.05. (F″) Isotropic stress σ0/K¯ for wt (gray) and pk (blue) tissues at −8 hRPCE (n ≥ 4). Significance is estimated using the Mann–Whitney U-test. ns, p-val >0.05. Time is relative to peak cell elongation (hRPCE). In all plots, each empty circle indicates one cut, and the box plots summarize the data: thick black line indicates the median; the boxes enclose the first and third quartiles; lines extend to the minimum and maximum without outliers, and filled circles mark outliers.

Figure 2—source data 1

Numerical data for Figure 2B, initial recoil velocity upon ablation along the proximal–distal (PD) axis for wt and pk tissues.

https://cdn.elifesciences.org/articles/85581/elife-85581-fig2-data1-v3.zip
Figure 2—source data 2

Numerical data for Figure 2E, values for fitted parameters of the rheological model.

https://cdn.elifesciences.org/articles/85581/elife-85581-fig2-data2-v3.zip
Figure 2—source data 3

Numerical data for Figure 2F–F″, values for anisotropic and isotropic stress deteremined with elliptical shape after circular ablation (ESCA).

https://cdn.elifesciences.org/articles/85581/elife-85581-fig2-data3-v3.zip
Video 1
Shown here is an example of a linear laser ablation, cutting three to four cells, in wt (left) or pk pupal wings.

The movie goes dark during the ablation itself. Thereafter, the tissue displaces. Anterior is up; proximal is left.

As initial recoil velocity is often used as a proxy for mechanical stress (e.g. Mayer et al., 2010; Etournay et al., 2015; Iyer et al., 2019; Farhadifar et al., 2007), this result seems to suggest that the PCP mutant wings generate less mechanical stress during blade elongation flows, even though the cellular dynamics are at best only subtly perturbed. To explore this phenotype in more detail, we considered that the response to laser ablation is not exactly a direct measure of mechanical stress, as it is also affected by cellular material properties. We thus further analyzed the full kinetics of the linear laser ablations, focusing on the pk mutant, and developed a rheological model to interpret the results. When plotting displacement of the nearest bond to ablation over time, we realized that a single exponential relaxation cannot account for the observed behavior (Figure 2—figure supplement 1B, right). We obtained a good fit of the data by introducing a second relaxation timescale (Figure 2C). The slow timescale (∼20 s) accounts for most of the timecourse of displacement changes, but the fast timescale (<1 s) is required to account for first 5–10 datapoints, see Figure 2—figure supplement 1B, left. We therefore developed a model consisting of two Kelvin–Voigt (KV) elements in series (Figure 2D) to represent the tissue after ablation. The two KV elements have different elastic constants (kf and ks) and viscosities (ηf and ηs). Before ablation, the system is subjected to a constant stress (σ) and contains a spring with elastic constant k, which represents the cell patch that will be ablated. Upon ablation, the third spring is removed which leads to change in strain of our rheological model. We represent this strain by a displacement Δx as a function of time given by

(1) Δx(t)=Xf(1et/τf)+Xs(1et/τs),

where Xf=σκ/kf is the displacement associated with the fast timescale, τf=ηf/kf, and Xs=σκ/ks is the displacement associated with the slow timescale, τs=ηs/ks. Here, κ=k/(k+k¯) is the fraction of the overall system elasticity lost due to ablation (see Materials and methods, Kymograph analysis and fit to model) and k¯=kskf/(ks+kf) is the elasticity of the two KV elements connected in series. With this model, we presume the properties of the ablated cell itself, including its membrane, adhesion proteins, and acto-myosin cortex likely dominate the fast timescale response. The slow timescale response is a collective effect emerging from the ablated cell together with its surrounding cellular network.

We analyzed the experimentally measured displacement over time for each ablation and then fit the data to our model with four parameters (Xf, Xs, τf, and τs) (Equation 1, Figure 2E–E′′′). Surprisingly, we find that the only parameter that changes between pk and wt is Xf, the displacement associated with the fast timescale (Figure 2E–E′′′). To interpret this result, we consider that these four fitted parameters constrain the five mechanical model parameters (Figure 2D) but do not provide a unique solution. Since only one measured parameter changes, we asked what is the simplest set of model parameter changes that could have such an effect. To this end, we first note that the measured values of Xf and Xs (1.8-2.6μm vs 6-8μm, respectively) indicate kfks and therefore the overall elasticity of our rheological model is largely determined by the elasticity of the slow relaxation k¯ks. If we also consider that the contribution to the elasticity of the cellular patch from the ablated cells, represented by k in the model, is small, then we can approximate κk/ks and therefore Xfσk/(kskf) (see Materials and methods, Kymograph analysis and fit to model). Is the observed phenotype in the fast timescale displacement Xf due to a change in tissue stress σ or a change in the elastic constants?

To address this question, we sought to probe mechanical stress in the wt and pk mutant, independent of the ablation recoil velocity. To do so, we used a method called ESCA (elliptical shape after circular ablation) (Dye et al., 2021), which uses circular laser ablation and quantifies the resulting elliptical tissue outline once the mechanical equilibrium is established (Figure 2F and Materials and methods, Elliptical shape after circular ablation). Analysis of the elliptical tissue outline provides information about two-dimensional stresses present in the tissue before the ablation. In particular, we measure the magnitude of the anisotropic shear stress tensor, normalized by the shear elastic modulus σ~/(2K) and the isotropic stress normalized by the area elastic modulus σ0/K¯. The stress σ in the simple rheological model presented above would correspond to tissue stress normal to the linear laser ablation axis and therefore it is a linear combination of both σ~/(2K) and σ0/K¯. ESCA also provides an estimate of the ratio of shear and area elastic constants 2K/K¯.

Using ESCA, we find no significant difference between wt and pk mutants in anisotropic and isotropic stress magnitudes, nor in the ratio of elastic constants (Figure 2F–F″ and Figure 2—figure supplement 1C). Since the ratio σ/ks defined in the rheological model is related to the normalized tissue stresses and elastic moduli, which do not change as shown by ESCA, we conclude that that σ/ks is not different between wt and pk. Therefore, we account for the observed changes of fast timescale displacement Xf in the pk mutant with a change of the single elastic constant kf. In this scenario, ηf changes together with the kf, such that τf=ηf/kf is conserved. This suggests that fast elasticity and viscosity are not independent but stem from a microscopic mechanism that controls the relaxation timescale. An example of such mechanism is turnover of the acto-myosin network, although this mechanism would be too slow to account for the fast relaxation timescale we observe. The conclusion that only the short-time response to the ablation, and not the tissue stress, is affected in the pk mutant is consistent with the lack of a clear phenotype in the large-scale tissue flows (Figure 1).

Dynamics of stress and cell elongation throughout blade elongation flows in wild type and core PCP mutants

To examine the effect of PCP mutation throughout blade elongation flows, we aimed to simplify the time intensive segmentation of the full ablation dynamics. To this end, we measured only the initial recoil velocity at different developmental timepoints. In terms of our model, the initial recoil velocity measured during first δt=0.65s can be expressed as v=(Xf(1eδt/τf)+Xs(1eδt/τs))/δt. Since the value of δt is comparable to the fast timescale τf, about 63% of the Xf value relaxes over that time, while at the same time only about 5% of the Xs value is relaxed. Using the measured values of Xf and Xs, we estimate that the fast timescale dynamics contributes about 80% of the v value. Therefore, the initial recoil velocity is a good proxy for the fast displacement Xf.

We find that the initial recoil velocity along the PD axis peaks at −8 hRPCE before declining again by 4 hRPCE (Figure 3A), consistent with previous work (Iyer et al., 2019). The behavior of the initial recoil velocity in the pk mutant is qualitatively similar throughout blade elongation flows, however, with significantly lower magnitude than wt (Figure 3A). We also observed this behavior in stbm and fmi mutant tissues (Figure 3—figure supplement 1A). This result indicates that Xf is lower in core PCP mutants than in wt throughout blade elongation flows.

Figure 3 with 2 supplements see all
Dynamics of stress and cell elongation throughout blade elongation flows in wt and pk mutant.

(A) Initial recoil velocity upon ablation (simplified as recoil velocity in the y-axis title) along the proximal–distal (PD) axis throughout blade elongation flows for wt (gray) and pk (blue) tissues (n ≥ 3). Significance is estimated using the Mann–Whitney U-test. ****p-val ≤0.0001; ***p-val ≤0.001; **p-val ≤0.01; ns, p-val >0.05. (B) Elliptical shape after circular ablation (ESCA) results for anisotropic stress σ~/2K for wt (gray) and pk (blue) tissues throughout blade elongation flows (n ≥ 3). Significance is estimated using the Mann–Whitney U-test. *p-val <0.05; ns, p-val >0.05. (B′) ESCA results for isotropic stress σ/K¯ for wt (gray) and pk (blue) throughout blade elongation flows (n ≥ 3). Significance is estimated using the Mann–Whitney U-test. *p-val <0.05; ns, p-val >0.05. (C) Color-coded PD component of cell elongation Q in the blade region between the second and third sensory organs found in the intervein region between L2 and L3. The images correspond to wt (top row) and pk (bottom row) wings throughout blade elongation flows. Scale bar, 5 μm. (D) Quantification of the PD component of cell elongation Q in this region throughout blade elongation flows for wt (gray) and pk (blue) (n ≥ 3). Significance is estimated using the Mann–Whitney U-test. ns, p-val >0.05. Time is relative to peak cell elongation (hRPCE). In all plots, each empty circle indicates one experiment, and the box plots summarize the data: thick black line indicates the median; the boxes enclose the first and third quartiles; lines extend to the minimum and maximum without outliers, and filled circles mark outliers.

Figure 3—source data 1

Numerical data for Figure 3A, initial recoil velocity upon ablation along the proximal–distal (PD) axis throughout blade elongation flows for wt and pk mutant tissues.

https://cdn.elifesciences.org/articles/85581/elife-85581-fig3-data1-v3.zip
Figure 3—source data 2

Numerical data for Figure 3B–B′, elliptical shape after circular ablation (ESCA) results for anisotropic and isotropic stress in wt and pk mutant tissues throughout blade elongation flows.

https://cdn.elifesciences.org/articles/85581/elife-85581-fig3-data2-v3.zip
Figure 3—source data 3

Numerical data for Figure 3D, proximal–distal (PD) component of cell elongation Q throughout blade elongation flows for wt and pk mutant tissues.

https://cdn.elifesciences.org/articles/85581/elife-85581-fig3-data3-v3.zip

We also performed ESCA at different timepoints in pk mutants and observe that anisotropic stress (σ~/2K) rises early during blade elongation flows before eventually declining (Figure 3B), whereas isotropic stress (σ0/K¯) remains fairly constant (Figure 3B′). Strikingly, ESCA does not report any difference in measured stresses between pk and wt, nor in the ratio of elastic constants (2K/K¯, Figure 3—figure supplement 1B) throughout blade elongation flows. To further compare the information contained in the initial recoil velocity with the anisotropic stress measured by ESCA, we performed linear ablations also in the perpendicular orientation. With such data, we could quantify the difference in initial recoil velocity between the two orientations δv=vPDvAP, which is expected to be proportional to the shear stress along the PD wing axis. We then quantified how δv evolves throughout the blade elongation flows (Figure 3—figure supplement 1C, D). Whereas ESCA clearly shows that stresses in the tissue remain the same in wt and pk throughout blade elongation flows, the difference in initial recoil velocity δv is significantly lower in pk compared to wt. This result indicates that our conclusions based on the −8 hRPCE timepoint are true throughout blade elongation flows, namely that the differences in Xf between wt and pk stem from the fast elastic constant kf and not from the differences in mechanical stresses in the tissue.

To further probe the possible role of core PCP in epithelial mechanics, we also measured the dynamics of the PD component of cell elongation (Q) in wt and core PCP mutants (Figure 3C, Figure 3—figure supplement 2E). Interestingly, in both wt and pk, anisotropic stress peaks around −6 hRPCE (Figure 3B), whereas Q peaks significantly later, between −4 and 0 hRPCE. We have previously related the tissue stress and cell elongation through a constitutive relation σ~=2KQ+ζ, where ζ represented an active anisotropic stress component (Etournay et al., 2015). The difference in timing of the peaks in stress and cell elongation indicate that the active stresses change over time. However, we observe no differences between wt and core PCP in the peak of cell elongation (Figure 3D, Figure 3—figure supplement 2E), showing that core PCP also does not affect active anisotropic stresses underlying the dynamics of cell elongation during blade elongation flows.

Discussion

Here, we used the Drosophila pupal wing as a model for studying the interplay between planar polarized chemical signaling components, specifically the core PCP pathway, and the mechanical forces underlying tissue morphogenesis.

An extensive analysis of core PCP mutants shows no significant phenotype in pupal wing morphogenesis during the blade elongation flows. We find no significant differences in overall tissue shape change, nor in the pattern or dynamics of underlying cellular contributions. Even if a larger sample size of pupal wings would reveal a statistically significant phenotype, as indicated by our analysis of adult wings, the differences to the wild type would be subtle. Furthermore, we found no significant differences in tissue mechanical stress or in cell elongation over time. Generally these results are consistent in mutants that greatly reduce core PCP polarity (stbm and fmi) or prevent its decoupling from Fat (pk).

Interestingly, we do observe a phenotype in the initial recoil velocity upon laser ablation between core PCP mutants and wt, but this is not reflected in tissue stresses or large-scale morphogenetic flows that shape the wing. A detailed analysis of wt and pk suggests that the phenotype arises from a difference in the elastic constant kf underlying the fast timescale response (τf=0.65s) to the ablation. In our simple model, the fast and stiff spring kf has a small contribution to the effective tissue elasticity k¯, which is dominated by the slow and soft spring ks (See Results, A rheological model for the response to laser ablation). The observation that core PCP only affects kf is therefore consistent with the lack of phenotype at larger scales. What is the the biophysical nature of the fast response to laser ablation? We hypothesize that processes that react on timescaless <1 s to a laser ablation could be related to cortical mechanics of cell bonds or possibly changes in cell hydraulics, and it is unclear how core PCP would affect these processes. Whether this core PCP phenotype in kf leads to very subtle changes in tissue development not detected in our analyses here, or is only visible in response to a laser ablation, also remains unknown and would require a much larger sample size to address. For the adult wing, we have a sufficient sample size to reveal a weak but significant shape phenotype in core PCP mutant wings. This result suggests that a weak phenotype arises during pupal development that we could not reliably detect in our analysis of cell dynamics. We also cannot rule out the possibility that there is a compensating mechanism that prevents the phenotype from appearing at larger scales.

Initial recoil velocity after a laser ablation is often used a proxy for tissue mechanical stresses. However, our results highlight a limitation of this approach for looking at how stress changes in different genotypes, as here we show how initial recoil velocity is influenced by differences in mechanics on small scales that are not necessarily related to differences in overall tissue stress.

While we have shown that core PCP is not required to organize the dynamic patterns of cellular events underlying blade elongation flows, it might still affect later stages of wing development. Furthermore, there may still be other patterning systems acting redundantly or independently with core PCP. For example, the Fat PCP system and Toll-like receptors have been shown to influence the orientation of cellular rearrangements and cell divisions in other contexts (Bosveld et al., 2012; Mao et al., 2006; Paré et al., 2014; Lavalou et al., 2021, reviewed in Umetsu, 2022). Whether and how other polarity systems influence pupal wing morphogenesis remains unknown. Alternatively, anisotropic mechanical stress induced by hinge contraction could itself provide a polarity cue through mechano-sensitive activity of the cytoskeleton. Our recent work in the larval wing disc shows that the cell polarity that drives the patterning of cell shape and mechanical stress contains a mechano-sensitive component (Dye et al., 2021). Here, we show a detailed analysis of tissues stress dynamics and cell elongation in the pupal wing revealing that the active cellular stresses that are relevant for pupal wing morphogenesis (Etournay et al., 2015) change in time (Figure 3). Whether the same mechano-sensitive mechanism established in the larval wing can also account for the dynamics of active stresses during the pupal blade elongation flows will be an important question to answer in the future.

Materials and methods

Key resources table
Reagent type (species) or resourceDesignationSource or referenceIdentifiersAdditional information
gene (Drosophila melanogaster)w-NAFLYB:FBal0018186
gene (Drosophila melanogaster)shg (shotgun; E-cadherin)NAFLYB:FBgn0003391
gene (Drosophila melanogaster)pk30NAFLYB:FBal0101223
gene (Drosophila melanogaster)stbm6NAFLYB:FBal0062423
gene (Drosophila melanogaster)fmifrz3NAFLYB:FBal0143193
strain, strain background (Drosophila melanogaster, male)wtOtherPMID:19429710Huang et al., 2009. Genotype: wt: w-; EcadGFP;
strain, strain background (Drosophila melanogaster, male)pkBloomington Drosophila Stock CenterRRID:BDSC_44229Gubb et al., 1999. Genotype: w-; EcadGFP, pk30;
strain, strain background (Drosophila melanogaster, male)stbmBloomington Drosophila Stock CenterRRID:BDSC_6918Wolff and Rubin, 1998. Genotype: w-; EcadGFP, stbm6;
strain, strain background (Drosophila melanogaster, male)fmiBloomington Drosophila Stock CenterRRID:BDSC_6967Wolff and Rubin, 1998. Genotype: w-; EcadGFP, fmifrz3;
chemical compound, drugEuparalCarl Roth7356.1
chemical compound, drugHolocarbon oil 700Sigma-AldrichH8898
chemical compound, drugIsopropanol (2-propanol)Sigma-Aldrich1.0104
software, algorithmFijiOtherv. 2.0.0-rc-68/1.52eSchindelin et al., 2012
software, algorithmIlastikOtherv. 1.2.2Berg et al., 2019
software, algorithmMATLABOtherv. 9.2.0.1226206 (R2017a)MATLAB, 2017
software, algorithmPreMosaOtherBlasse et al., 2017
software, algorithmROtherv. 3.4.1R Development Core Team, 2020
software, algorithmRstudioOtherv. 3.6.1RStudio Team, 2020
software, algorithmTissueMinerOtherv. TM_1.0.2Etournay et al., 2016
otherCoverslipPaul Marienfeld GmbH107052
otherMicroscope slidesPaul Marienfeld GmbH1000200
otherDumont #55 ForcepsFine Science Tools11295–51
otherVannas Spring ScissorsFine Science Tools15000–08

Fly husbandry

Request a detailed protocol

Flies were maintained at 25°C and fed with standard fly food containing cornmeal, yeast extract, soy flour, malt, agar, methyl 4-hydroxybenzoate, sugar beet syrup, and propionic acid. Flies were kept at 25°C in a 12-hr light/dark cycle. Vials were flipped every 2–3 days to maintain a continuous production of pupae and adult flies. All experiments were performed with male flies, since they are slightly smaller and therefore the wings require less tiling on the microscope to be imaged than females.

Long-term timelapse imaging of pupal wing morphogenesis

Acquisition

Request a detailed protocol

White male pupae were collected, slightly washed with a wet brush, and transferred to a vial containing standard food. At 16 hAPF, the pupal case was carefully dissected so that the wing would be accessible. The pupae was then mounted onto a 0.017-mm coverslip on a self-built metal dish with a drop of Holocarbon oil 700 (Classen et al., 2008). Pupal wing morphogenesis was imaged every 5 min for approximately 24 hr, as in Etournay et al., 2015. Wings that did not develop after 4–5 hr of imaging were discarded and not analyzed.

Two different microscopes were used for acquisition of long-term timelapses. All wt, pk, and stbm movies were acquired using a Zeiss spinning disk microscope driven by ZEN 2.6 (blue edition). This microscope consists of a motorized XYZ stage, an inverted stand, a Yokogawa CSU-X1 scan head, and a temperature-controlled chamber set to 25°C. The sample was illuminated with a 488-nm laser, and the emission was collected using a 470/40 bandpass filter, through a Zeiss 63 × 1.3 W/Gly LCI Plan-Neofluar objective and a Zeiss AxioCam Monochrome CCD camera with 2 × 2 binning. The whole wing was imaged in 24 tiles with an 8 % overlap. Each tile consisted of 50–60 stacks with a Z-spacing of 1 μm. The laser power was set to 0.1 mW.

The two fmi movies were acquired with an Olympus IX 83 inverted stand driven by the Andor iQ 3.6 software. The microscope is equipped with a motorized xyz stage, a Yokogawa CSU-W1 scan head, and an Olympus 60 × 1.3 Sil Plan SApo objective. The setup was located inside a temperature-controlled chamber set to 25°C. The sample was illuminated with a 488-nm laser, and the emission was collected using a 525/50 bandpass filter. The whole wing was imaged by tiling with eight tiles with a 10 % overlap. Each tile consisted of 50–60 stacks with a distance of 1 μm between them. The laser power was set to 0.75 mW.

Table 1 summarizes the date when the long-term timelapses were acquired and the age of the pupae during the imaging.

Table 1
Date of acquisition of all long-term timelapses.
GenotypeDate of acquisitionStart [hAPF]End [hAPF]
wtMarch 30, 20161639.83
April 2, 20161636.58
April 3, 20161636.50
April 13, 20161632.58
June 20, 20181641.17
pkApril 9, 20161639.00
June 28, 20161636.00
June 29, 20161639.92
stbmNovember 25, 20151640.67
November 28, 20151635.33
December 11, 20141637.58
fmiOctober 20, 20181639.17
July 20, 20191638.17

Processing, segmentation, tracking, and database generation

Request a detailed protocol

Raw stacks were projected, corrected for illumination artifacts, and stitched using PreMosa (Blasse et al., 2017). The stitched images of individual timepoints were cropped to fit the wing size, registered using the Fiji plugin ‘Descriptor-based series registration (2D/3D + t)’, and converted to 8 bit with Fiji (Schindelin et al., 2012). The segmentation was performed with the Fiji plugin TissueAnalyzer (Schindelin et al., 2012; Aigouy et al., 2010; Aigouy et al., 2016). Segmentation errors were identified and manually corrected by looking at the cell divisions and deaths masks.

Subsequent processing and quantifications were performed using TissueMiner (Etournay et al., 2016). Before generating the relational database, we rotated the movies so that the angle formed by a manually drawn line connecting the sensory organs would be 0. We manually defined the regions of interest, such as the blade, hinge, and the anterior and posterior regions, using the last frame of the movie. Next, we generated the relational database containing information about the cellular dynamics during blade elongation flows using TissueMiner (Etournay et al., 2016).

We queried and worked with the data using the Dockerized version of RStudio (Nickoloff, 2016), which loads all packages and functions required to work with TissueMiner. Movies were aligned by the peak of cell elongation by fitting a quadratic function around the cell elongation values 40 frames before and after the absolute maximum of cell elongation in the blade region for each movie. The maximum of this curve was identified and set as the timepoint 0 hRPCE.

Adult wing preparation and analysis of wing shape

Adult male flies were fixed in isopropanol for at least 12 hr. One wing per fly was dissected in isopropanol, transferred to a microscope slide and covered with 50% euparal in isopropanol. Wings were mounted with 50–70 μl 75 % euparal/isopropanol.

wt, pk, and stbm wings were imaged using a Zeiss widefield Axioscan Z1 microscope equipped with a Zeiss 10 × 0.45 air objective. fmi wings were imaged using a Zeiss widefield Axiovert 200 M microscope equipped with a Zeiss 5 × 0.15 Plan-Neofluar air objective.

Wing blade parameters were quantified using a custom-written Fiji macro (provided as Source code 1) (Schindelin et al., 2012). The shape or major-to-minor ratio was calculated using a custom RStudio script (R Development Core Team, 2020; RStudio Team, 2020).

Subsampling and statistical analysis

Request a detailed protocol

Random sampling was done using a custom written RStudio pipeline (R Development Core Team, 2020; RStudio Team, 2020). A group of a given sample size was randomly selected with replacement for each group (wt, pk, stbm, and fmi), and a Kruskal–Wallis test was ran to compare them. This analysis was repeated 10,000 times. The sample sizes analyzed were 3, 4, 5, 6, 7, 8, 9, 10, 20, and 40.

Quantification of the PD component of cell elongation Q

Request a detailed protocol

Prior to all laser ablation experiments, we acquired a stack of 50 μm thick that was projected using PreMosa (Blasse et al., 2017). We cropped a region that enclosed the region that was ablated, segmented cells using TissueAnalyzer (Aigouy et al., 2010; Aigouy et al., 2016), and generated a relational database with TissueMiner (Etournay et al., 2016).

The definition of cell elongation was first presented in Aigouy et al., 2010 and it describes the angle and magnitude of the tensor. The cell elongation tensor is given by

(2) (ϵxxϵxyϵxyϵxx),

where

(3) ϵxx=1Accos(2ϕ)dA

and

(4) ϵxy=1Acsin(2ϕ)dA.

Cell elongation is normalized by the cell area (Ac) of each cell. The magnitude of cell elongation is:

(5) ϵ=(ϵxx2+ϵxy2)12

Here, we plot ϵxx as Q, which we describe as the PD component of cell elongation.

Laser ablation experiments

Pupae were dissected and mounted as described for the long-term timelapses. Ablations were always performed in the same region of the wing blade, found in the intervein region between the longitudinal veins L3 and L4 and between the second and third sensory organs. This region was chosen because these landmarks are easily visible in all timepoints. Laser ablations were performed using a Zeiss spinning disk microscope equipped with a CSU-X1 Yokogawa scan head, an EMCCD Andor camera, a Zeiss 63 × 1.2 water immersion Korr UV-VIS-IR objective, and a custom-built laser ablation system using a 355-nm, 1000-Hz pulsed ultraviolet (UV) laser (Grill et al., 2001; Mayer et al., 2010). The imaging and cutting parameters for line and circular laser ablations are shown in Table 2. All laser ablation experiments were performed between January 2018 and July 2020, after the delay in pupal wing morphogenesis was identified.

Table 2
Parameters used to perform laser ablations.
Linear ablationsCircular ablations
Exposure time [s]0.050.05
488-nm laser intensity [%]5050
Time interval [s]0.092.55
Pulses per shot2525
Shots per µm22
Shooting time [s]0.67147.28
Thickness of stack ablated [µm]120

Linear laser ablations to calculate the initial recoil velocity

Request a detailed protocol

We performed both types of linear ablations in only one plane of the tissue, in order to minimize the time required for ablation and therefore be able to acquire the initial recoil velocity upon ablation (no imaging is possible during ablation). The length of the linear laser ablations was 10 μm, ablating three to four cells. We drew kymographs perpendicularly to the cut to follow the two edges of one ablated cell using Fiji (Schindelin et al., 2012). The initial recoil velocity was calculated as the average displacement of two membranes of the same cell that occurred during the black frames of the ablation itself. This calculation was made using a self-written MATLAB script (MATLAB, 2017). Scripts used to make kymographs and analyze the laser ablations are provided in Source code 2; Source code 3; Source code 4. The image acquired prior to the laser ablation was used to compute Q in that region, as described in Quantification of the PD component of cell elongation Q, and the time corresponding to the maximum of cell elongation was defined as 0 hRPCE.

Elliptical shape after circular ablation

Request a detailed protocol

Circular laser ablations used for ESCA were 20 μm in radius (approximately 10 cells). This radius was selected such that it would fit into the same blade region throughout the blade elongation flows. Due to the bigger size of these cuts and the curvature of the tissue, we cut the tissue along a stack of 20 μm thick. Approximately 2 min after the ablation, we acquired a stack of 50 μm. This image was projected using PreMosa (Blasse et al., 2017) and preprocessed by applying Gausian blur (σ = 1) and background subtraction filters (rolling ball radius = 30) in Fiji (Schindelin et al., 2012). The next steps were performed as in Dye et al., 2021: the image of the final shape of the cut was segmented using Ilastik (Berg et al., 2019) by defining three regions: membrane, cell, and dark regions. The segmented image was thresholded to obtain a binary image of the final shape of the cut. We fitted two ellipses to this image: one to the inner piece and another one to the outer outline of the cut. Based on the shape of these ellipses, the method outputs the anisotropic σ~2K and isotropic stress σK¯ as a function of their respective elastic constants, and the ratio of elastic constants 2KK¯. A small number of experiments were fitted poorly (defined as an error per point greater than 0.3) and were therefore excluded from analysis. Prior to the circular ablation, a stack of 50 μm was acquired and used to calculate cell elongation before ablation (Quantification of the PD component of cell elongation Q). The time corresponding to the maximum of cell elongation was set to be 0 hRPCE.

Kymograph analysis and fit to model

Request a detailed protocol

The ablations used to calculate the mechanical stress along the PD axis for wt and pk were further analyzed with the rheological model. To do so, we processed the kymographs by applying a Gaussian blur (σ = 1) (Schindelin et al., 2012), and then we segmented these kymographs with Ilastik (Berg et al., 2019). Using a self-written Fiji macro (Schindelin et al., 2012), we extracted the intensity profile for each timepoint. Next, we wrote an R script (R Development Core Team, 2020; RStudio Team, 2020) to identify the membrane displacement over time and obtained a unique curve per kymograph, which could be fitted with our model. We modeled a local patch of tissue as a combination of a spring with spring constant k, representing the ablated cells, and two KV elements with spring constants kf and ks and viscosity coefficients ηf and ηs, representing the unablated cells, as shown in Figure 2C, D. Because the local tissue strain in the experimental measurement is expressed by the displacement of the bond nearest to the ablation, in the rheological model we represent tissue strain by displacements of the two KV elements. In principle, the strain can be recovered by normalizing the displacements by the width of ablated cells. Displacements of the two KV elements are defined as a change in the distance between the end points of the KV elements xi(t), relative to their initial values xi(0), where i{f,s} for fast (f) and slow (s) element.

Mechanical stress in the tissue is represented by the σ acting on our model, and we assume that σ is not changed by the ablation. Before the ablation, the model is in mechanical equilibrium and we can write

(6) σ=(k+k¯)x(0),

where x(0) is the initial distance between the two end points of the model, and k¯=kfks/(kf+ks) is the elastic constant of the two KV elements connected in series. Upon ablation, the spring k is removed and stresses in the model are imbalanced. The distance between the end points of the model x(t) then evolves toward the new equilibrium position. The distance x(t) can be decomposed as x(t)=xf(t)+xs(t), where xf(t) and xs(t) are the time-dependent distances between end points of the two KV elements, representing their strains. The dynamics of x(t) is then obtained by writing the force balance equation for the two KV elements

(7) σ=kfxf(t)+ηfdxf(t)dt,
(8) σ=ksxs(t)+ηsdxs(t)dt,

We solve for xf(t) and xs(t) to obtain

(9) xf(t)=σkf(1et/τf)+xf(0)et/τf,
(10) xs(t)=σks(1et/τs)+xs(0)et/τs,

where

(11) xf,s(0)=σ(1κ)kf,s,

and κ=k/(k+k¯) is the fraction of the overall model elasticity k+k¯ destroyed by the ablation. The displacement relative to the initial configuration Δx(t)=x(t)x(0) is therefore

(12) Δx(t)=Xf(1et/τf)+Xs(1et/τs),

where we introduced the long time displacements associated with the two KV elements

(13) Xf,s=σκkf,s.

For simplicity, in the main text we refer to the long time displacements Xf and Xs of the two KV elements simply as displacements.

Statistical analysis

Request a detailed protocol

Statistical analysis was done using R (R Development Core Team, 2020; RStudio Team, 2020). We first tested normality of the data using the Shapiro–Wilk test. When data were normal, we used Student’s t-test to test statistical significance between two groups and analysis of variance test for multiple groups. When data were not normally distributed, significance was tested using the Mann–Whitney U-test for two groups and Kruskal–Wallis test for multiple groups. Statistical test results are shown on the figure captions.

Appendix 1

During the course of this work, we identified a delay in the onset of blade elongation flows compared to previous work (Appendix 1—figure 1A; Etournay et al., 2015; Piscitello-Gómez et al., 2023). In the past, cells reached their maximum of cell elongation at 22.9 ± 0.4 hAPF, while now they reach it at 28 hAPF. Although we do not know the cause of this delay, we have ruled out differences in temperature (either on the microscope or during development), nutrition (plant vs yeast-based foods), genetic background, presence of the parasite Wolbachia, or circadian gating (Piscitello-Gómez et al., 2023). To deal with this variation and combine data acquired over the years, we present cell dynamics data aligned in time by the peaks of cell elongation, and we refer to this timepoint as 0 hRPCE (relative to peak cell elongation) (Appendix 1—figure 1A). We investigated the cell dynamics underlying blade elongation flows in the delayed flies and observed that the shear rates were comparable with the older flies (Appendix 1—figure 1B). Thus, it is reasonable to shift the curves by aligning them to a new reference time.

Appendix 1—figure 1
Delay and time alignment of old and newer flies: (A) Left: Snapshots of the blade region of long-term timelapses of wt pupal wing morphogenesis acquired in different years.

Scale bar, 10 μm. Right: Cell elongation norm during blade elongation flows for old flies (orange palette, 2016 flies) and new flies (green curve, 2018 fly). Top plot: Cell elongation magnitude for each movie not aligned in time. The peak of cell elongation is delayed from around 23 to 28 hAPF. Bottom plot: Cell elongation magnitude after alignment in time to the peak of cell elongation. Time is now expressed in hours relative to peak cell elongation (hRPCE). (B) Cell dynamics underlying anisotropic tissue deformation for 2016 wt (n = 4, top left), 2018 wt flies (n = 1, top right), and 2 fmi flies imaged in 2018 (bottom left) and 2019 (bottom right). The vertical dashed line marks the timepoint where cell rearrangements flip from AP- to PD-oriented per movie. The two dotted lines mark the start and the end of the analyzed wt long-term timelapses acquired in 2016. The time is relative to the peak of cell elongation (hRPCE).

Data availability

Source data and code are provided for each figure.

References

  1. Book
    1. Adler PN
    (2012)
    The frizzled/Stan pathway and planar cell polarity in the Drosophila wing
    In: Adler PN, editors. Planar Cell Polarity During Development. Elsevier. pp. 1–31.
  2. Book
    1. Aigouy B
    2. Umetsu D
    3. Eaton S
    (2016)
    Segmentation and quantitative analysis of epithelial tissues
    In: Aigouy B, Umetsu D, editors. Drosophila. Springer. pp. 227–239.
  3. Book
    1. Classen AK
    2. Aigouy B
    3. Giangrande A
    4. Eaton S
    (2008)
    Imaging Drosophila Pupal wing Morphogenesis
    In: Classen AK, Aigouy B, editors. Drosophila. Springer. pp. 265–275.
  4. Book
    1. Jenny A
    (2010)
    Planar cell polarity signaling in the Drosophila eye
    In: Jenny A, editors. Current Topics in Developmental Biology. Elsevier. pp. 189–227.
  5. Book
    1. Nickoloff J
    (2016)
    Docker in Action
    Manning Publications Co.
  6. Software
    1. R Development Core Team
    (2020) R: A language andEnvironment for statistical computing
    R Foundation for Statistical Computing, Vienna, Austria.

Decision letter

  1. Marcos Nahmad
    Reviewing Editor; Center for Research and Advanced Studies (Cinvestav), Mexico
  2. Detlef Weigel
    Senior Editor; Max Planck Institute for Biology Tübingen, Germany
  3. Marcos Nahmad
    Reviewer; Center for Research and Advanced Studies (Cinvestav), Mexico
  4. M Lisa Manning
    Reviewer; Syracuse University, United States

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Core PCP mutations affect short time mechanical properties but not tissue morphogenesis in the Drosophila pupal wing" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, including Marcos Nahmad as Reviewing Editor and Reviewer #1, and the evaluation has been overseen by Michael Eisen as the Senior Editor. The following individual involved in the review of your submission has agreed to reveal their identity: M. Lisa Manning (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) Please revise the text to be more clear about how you reconcile your observations that PCP mutants have an effect on the local tissue dynamics (e.g., recoil velocity upon laser ablation) but this has no global consequences on wing morphogenesis. In particular, could the local effects explain the subtle, but significant effects on adult wing shapes in these mutants?

2) Please justify your choice of pk30 mutants for most of the analysis, instead of using stbm6 or fmi mutants that completely disturb the core-PCP pathway.

3) Please explain in your rheological model why you believe that the recoil velocity has two phases.

4) Please reconsider the way you plot your data so that readers do not need to extract subtle differences on curves plotted in different graphs with different scales.

4) Please revise the discussion on the ESCA analysis, as reviewers have a hard time following it.

Reviewer #1 (Recommendations for the authors):

Overall the manuscript is well written and the results support the authors' conclusions; therefore I am supportive of publication. Here, I list a few comments that will improve the manuscript's readability, but will likely not affect the conclusions of the study:

– Considering that pk30 mutants maintain the polarity towards the margin i.e., both core-PCP and Fat-PCP remain aligned (Figure S1.1), while stbm6 and fmi mutants completely disturb the core-PCP pathway, why did the authors focus their analysis in the pk30 mutants (line 118)? If the purpose of the study is to examine the perturbations of the core-PCP, why not choose stbm6 or fmi mutants? The authors should justify this further.

– While it is clear that this study only focuses on the core-PCP pathway, perturbing only this pathway will not eliminate polarity and therefore it is not so surprising that oriented cell rearrangements persist. The authors should discuss which experiments (mutants) they suggest to perturb polarity altogether.

– The authors chose a specific region of the wing (between veins L3 and L4) to perform their perturbations. The justification is that this is a region that is easy to identify. But, do the results depend on this choice? Isn't the strength of the polarity signal larger closer to the wing margin?

– Use consistent English spelling throughout the manuscript (e.g., behaviour or behavior lines 169 vs. 171).

Reviewer #2 (Recommendations for the authors):

The authors have done a good job of trying to bring together what appear to be datasets generated over a fairly long period by different authors.

I do have a few issues regarding the presentation and interpretation of the data (specific points noted below). The main concern is that the authors keep flipping between saying core PCP has no effect on wing morphogenesis, to saying it has subtle effects to saying it has effects but they are not statistically significant. This is hard to reconcile with the convincing effect on tissue mechanics reveals by the laser ablation work, and possibly also the changes in adult wing shape and known roles of core PCP in both the wing and other tissues in the fly. Mostly I think it's just a question of careful choice of words and most likely leaving the question of effects of core PCP on morphogenesis open until all the data is considered together in the Discussion e.g. try to stick to observations in the Results and try to only do interpretation in the Discussion.

I also have queries about the rheological model and why the authors believe the recoil has 2 phases based on the data they show, and also about the way developmental timings have been recalibrated.

Specific points:

The references on lines 21 and 22 are an eclectic selection. The sentence seems to be about how polarized PCP proteins orient structures in epithelia, but the references include a mix of reviews and papers, some from before the polarization had been reported. For instance, Eaton 1997 was a seminal review that suggested planar polarization of stereocilia might be under the control of planar polarity pathways, but does not report such a result (I think the first papers were Montcouquiol et al. 2003 and Curtin et al. 2003?). It would probably be better for readers if the authors either cited reviews or if they want to cite primary papers, than just the first reports of polarized localization determining the orientation of structures (e.g. I think Usui et al. 1999 is the first in flies? Then Axelrod and Strutt?).

The references on lines 27 and 28 are mostly referring to Fat-Dachsous while the rest of the paragraph is about "core". There are papers about core PCP in flies affecting cell rearrangements/divisions. Segalen et al. 2010 is a good example, but not the first to report core PCP controlling SOP division orientation (Gho and Schweisguth 1998?). Sugimura and Ishihara 2013 report changes in cell rearrangements in the wing of a core PCP mutant. Others (moving away from the wing/notum) might include many papers about ommatidial rotation in the eye, joint formation in the legs, cell rearrangements in the tracheal system (and probably others). A better review of core PCP roles in tissue morphogenesis in flies would put the work in a better context.

In general, I would recommend thinking carefully about the references in the Introduction so that they are of the most help to readers from outside the field. It is hard to see an organizing principle behind the mixture of reviews from different decades, primary papers (but often not first reports), cloning, or genetic or biochemical characterization studies. As above, I would recommend recent reviews (where available) and primary papers that are the first reports, for each point the authors wish to make.

Line 44 – Fz is not a cadherin.

Figure S1A -I don't fully agree that 30 hAPF is the "End of morphogenesis" as suggested by the figure label. Even if you ignore the emergence of hairs and wing ridges, the epithelium itself will go on expanding and folding in subsequent hours. Similar statements are made on line 73 and on Figure 1.

Figure S1A legend – do "tissue flows reorganize Fat PCP"? My guess is Fat PCP continues to follow the Ds and Fj gradients and its insensitivity to the tissue flows is why it ends up perpendicular to core PCP. Of course, tissue flow might also have an effect but do the authors have evidence for this (I'm not sure Merkel et al. show this?). The statement in the legend is also not the same as what lines 51-52 say, where the message is that Fat PCP remains margin oriented.

Lines 66 and 67 – in normal nomenclature pk30 = pk[30] (i.e. 30 in superscript) and stbm6 = stbm[6], consistent with fmi[frz3]. I think it would be simplest to refer to stbm[6] as stbm and pk[30] as pk after the first definition, in the same way, fmi[frz3] is referred to as fmi, but this is of course up to the authors. As an aside, fmi[frz3] was described as a strong hypomorph in Chae et al. 1999, so it seems unlikely that in frz3/frz3 the "PCP network is absent", as stronger phenotypes are seen for frz3/Df, implying some residual activity.

Figure S1.2C – typo in the first word of the label.

Line 84 "these subtle changes are not statistically significant". This all looks reasonable, but possibly the authors need to qualify this by saying something like "considering the small numbers of wings analyzed". I appreciate the work involved to get these datasets was enormous, so higher n numbers are not feasible. However, with larger n numbers there might be a difference. The average shear values for all three mutant conditions are lower than wt. Ideally, a power calculation would be done to place a numerical value on the likelihood of the shear being the same in each condition, but again I appreciate this is difficult and possibly not appropriate post hoc.

Line 85 "also do not show differences" – see the previous point. Without stats, you can't say they don't differ, but here it is implied that the data in Figure S1.2C and D have been tested for differences and none was found. This is true for S1.2C with the caveat of small n numbers and no power calculation, but for S1.2D appears to be a judgment made by eye. However, by eye total tissue area at -2h looks lower for pk[3] and stbm[6] than for wt, and at e.g. 6h all 3 mutants show lower cell area changes than wild-type. I think the most the authors can say (unless I've missed something) is that the patterns of tissue behavior look broadly the same for the mutants as for wt, suggesting no gross differences. Note lines 89-90 say "Again, we find only subtle differences" which is not the same as no differences.

Lines 94-95 and Figure S1.4G – again I think it is deceptive to put weight on statistics with low n numbers and no power calculation. I'm not confident that you can "conclude that core PCP does not guide the global patterns of cell dynamics during pupal morphogenesis". The best that you can conclude is that changes are at best subtle (and in the experiments presented might be due to "noise" but could also be due to roles for core PCP in cell behavior). The authors rather undermine themselves by immediately going on to say that there is a reproducible difference in adult wing shape, although admittedly we don't know that this is due to events between 16-32h as the authors note. I think more nuanced conclusions would be better.

Lines 115-116 "cellular dynamics is [are?] basically unperturbed" is another example of wording that implies the authors think cellular dynamics are perturbed to some extent, but they want to minimize the significance of this observation. Similarly, lines 198-199 "no significant phenotype" belies the difficulty in making such an assertion when the genotypes do look different in some aspects and the tests have (apparently) minimal statistical power.

Lines 119-120 "When plotting displacement of the nearest bond to ablation over time, we observe both a fast (<1 s) and a slow (<20 s) regime (Figure 2C)." I think the authors are saying that their data fits a 2 phase exponential with a fast and slow phase, but I'm a bit nervous about this as the first phase seems to be based on a single data point at 1 sec. I'm not an expert on curve fitting, but only having 1 point to demonstrate the presence of a "fast" phase seems a bit unsafe. I couldn't find anything in the manuscript that describes why the authors decided a 2 phase fit was the correct choice. The subsequent model appears to assume 2 phases, but I'm interested to know if you can get just as good a fit with 1 phase and if you build a model with 1 phase does that still show pk30 having a significantly lower displacement over time? I'm concerned that a 1 phase model may have been unnecessarily ruled out.

Appendix 1 – I understand the issue confronting the authors. Earlier studies from the lab (from Aigouy et al. 2010 onwards) consistently saw peak cell elongation at ~23h, then in Iyer et al. peak elongation appears to be at ~28h (I'm unsure if Iyer et al. comments on this). I have a couple of comments:

Line 519 the statement "we identified a delay on the onset of pupal wing morphogenesis" is unclear to me. Do the authors mean that morphogenesis proceeds at a normal rate but just start late (similar to a train being delayed but running at the normal speed), or is everything just happening slower? The curves in panels A and B really look like the latter (e.g. in panel B the same sequence of events in 2016 takes 12h while in 2018 it takes 20h), but the authors seem to be saying it's the former.

The most obvious explanation would be lower incubation temperature (or possibly nutrition?) which is known to slow the rate of development, at least if it's not background genotype (which I'm assuming it isn't, as I assume the authors are saying all genotypes are now "delayed"). If it is a slower rate of development, then the rate of development needs recalibrating, not the reference time.

Given the prior literature (much of it from the Eaton lab) uses a linear time scheme with 0h at the start of pupal life, I'm unsure why the authors didn't just stick with this scheme and rescale the timings so peak elongation is at 23h? This would make the data much more comparable both internally in the manuscript and to past work.

Reviewer #3 (Recommendations for the authors):

We believe the manuscript is likely ultimately suitable for publication in eLife, but there are a few major comments we would like the authors to address before publication.

1. We find the second paragraph of the discussion confusing. In it, the authors highlight that the difference in the initial recoil velocity between wt and core PCP mutants does not lead to a phenotype in tissue morphogenesis. Then they try to explain why this difference does not generate a global phenotype, but the explanation, "the proportionality factor can depend on the genotype and can change in time", is quite unclear.

In the paragraph that starts with line 153, it seems the authors are saying their data suggests that both $k_f$ and $\eta_f$ are changing in the pk30 mutant. Is that the "proportionality factor" they mean above? If so, why not just come out and say more explicitly that the in the mutants the fast viscosity apparently changes along with $k_f$ so as to make the time scale constant? (As a side note isn't that an interesting coincidence? Would it be interesting/useful to speculate that there might be some sort of feedback loop or mechanical coupling that drives the compensating change in the viscosity?)

And we believe the "change in time" clause above refers to the paragraph that starts at line 179, which is also confusing (see next comment). But isn't the change in viscosity alone already enough to explain the lack of large-scale phenotype? Do you need this change in time also? Would such a change in time be alone sufficient to explain the lack of a global phenotype?

2. The paragraph starting at line 179 is apparently asking the reader to compare the data for the pk30 mutants in Figure 3B (blue data) to the data for the pk30 mutant in S3.1C (blue line). But it's very difficult to do that, as the axes for the two plots are different, and there are not even units given for the y-axis in Figure 3B. We think we're supposed to see that the blue curves in the two plots have different shapes in time. But that's really difficult to see toggling between them. Would it be possible to show both sets of data on the same plot (perhaps one set rescaled) so that we could directly compare the shapes? Moreover, it would be useful to state something more about the data than "its different". It looks like maybe the ESCA data peaks earlier in time than the initial recoil velocity. What does that mean?

3. What is the possible reason for different times for the peaks in anisotropic stress and cell elongation (discussion in the last paragraph of section 2.3)? It seems the response time for the stresses is almost 2 to 4 h.

4. The discussion on ESCA analysis is incomplete. It is not clear how σ/Ks is related to anisotropic ((σ¯/2K)) and isotropic stress (σ/K¯). The variables K,K¯ and σ¯ are not defined. It is mentioned in the caption of Figure 2 that the equations for anisotropic and isotropic stress are on the right side. However, there is no equation in Figure 2F.

https://doi.org/10.7554/eLife.85581.sa1

Author response

Essential revisions:

1) Please revise the text to be more clear about how you reconcile your observations that PCP mutants have an effect on the local tissue dynamics (e.g., recoil velocity upon laser ablation) but this has no global consequences on wing morphogenesis. In particular, could the local effects explain the subtle, but significant effects on adult wing shapes in these mutants?

We now mention this possibility in the Results section, lines 102-110, and then again in the discussion, 242-248. In short, our result that PCP mutations only affect k_f is consistent with the lack of strong phenotype at the tissue scale. Whether changes in k_f can lead to more subtle defects in tissue morphogenesis is possible, but we have no mechanism as yet to propose, and this would require a much larger sample size to address.

2) Please justify your choice of pk30 mutants for most of the analysis, instead of using stbm6 or fmi mutants that completely disturb the core-PCP pathway.

Most of the analysis was actually done with all three mutants: analysis of cell dynamics (Figure 1, Figure 1-figure supplement 2D, Figure 1-figure supplement 3, Figure 1-figure supplement 4), tissue shape (Figure 1-figure supplement 2C,E, Figure 1-figure supplement 4H), adult wing shape (Figure 1-figure supplement 5I, J), initial recoil velocity over time (Figure 2B, Figure 2-figure supplement 1A, Figure 3A, Figure 3-figure supplement 1A), cell elongation over time (Figure 3D, Figure 3-figure supplement 2E). The only analysis that was done with the pk mutant alone was the full dynamics of the response to laser ablation and the comparison of initial recoil velocity over time with ESCA (Figure 2C-F’’, Figure 2-figure supplement 1BC, Figure 3-figure supplement 1B-D). This mutant was chosen because the line grows the most reliably in our hands, but since none of the other results differed between mutants, we have no reason to expect any differences in these more focused analyses or in our conclusions about the core PCP pathway in general.

3) Please explain in your rheological model why you believe that the recoil velocity has two phases.

We have added an explanation in the Results section (lines 129-34), as well as a new supplemental figure (Figure 2—figure supplement 1B) to address this point. We also provide a detailed response to this point in answer to Reviewer 2 point 13.

4) Please reconsider the way you plot your data so that readers do not need to extract subtle differences on curves plotted in different graphs with different scales.

This is useful feedback, and we have changed the most relevant figures (Figure 3—figure supplement 1C and Appendix Figure 1).

4) Please revise the discussion on the ESCA analysis, as reviewers have a hard time following it.

Thank you for pointing out the lack of clarity. We have considerably changed the text of the results, lines 161-75 and 199-212, to amend this shortcoming.

Reviewer #1 (Recommendations for the authors):

Overall the manuscript is well written and the results support the authors' conclusions; therefore I am supportive of publication.

We thank the reviewer for a careful reading of the manuscript and for helpful and critical feedback.

Here, I list a few comments that will improve the manuscript's readability, but will likely not affect the conclusions of the study:

– Considering that pk30 mutants maintain the polarity towards the margin i.e., both core-PCP and Fat-PCP remain aligned (Figure S1.1), while stbm6 and fmi mutants completely disturb the core-PCP pathway, why did the authors focus their analysis in the pk30 mutants (line 118)? If the purpose of the study is to examine the perturbations of the core-PCP, why not choose stbm6 or fmi mutants? The authors should justify this further.

To clarify, most of the analysis (shown in Figure 1 with Figure 1—figure supplements 2-5, Figure 2B with Figure 2—figure supplement 1A, Figure 3A with Figure 3—figure supplement 1A, and Figure 3D with Figure 3—figure supplement 2E) is actually performed on all 3 mutants (pk, stbm, fmi), and they behave similarly. We find the effect on overall cellular dynamics, tissue shear, as well as initial recoil velocity over developmental time are all the same. The only place in which we analyze the pk mutant alone is the full dynamic analysis of the response to laser ablation used for the model (Figure 2C-E’’’) and the comparison of ESCA with initial recoil velocity (Figure 2F-F’’, Figure 2—figure supplement 1B-C, Figure 3—figure supplement 1B-D). We altered the title of section 2.3 to reflect this fact (previously it only referenced wt and pk mutants, whereas now we say wt and core PCP mutants). We also slightly altered the paragraph structure so that now all the Pk-only data in 2.3 are in one paragraph (lines 199-212). Pk mutant flies were best suited for this experiment because they grow more reliably and therefore we could obtain data with better statistics.

– While it is clear that this study only focuses on the core-PCP pathway, perturbing only this pathway will not eliminate polarity and therefore it is not so surprising that oriented cell rearrangements persist. The authors should discuss which experiments (mutants) they suggest to perturb polarity altogether.

We agree with the reviewer that we are not completely eliminating polarity, as there may be many other sources of polarity. We can only conclude from our data that core PCP is not required to orient cell dynamics. We mention two likely additional sources of planar polarity already in the manuscript, including Fat PCP and Toll-like receptors (lines 255-58), which may act independently or redundantly with core PCP or each other. Perturbing all sources of polarity would be very difficult and beyond the scope of this manuscript, as it would involve complicated genetics to combine multiple different types of mutations, which may not even be viable. In addition, it is still possible that there are unknown planar polarity pathways. Furthermore, anisotropic mechanical stress induced by hinge contraction could itself provide a source of polarity through mechanosensitive activity of the cytoskeleton, a possibility that we now suggest in the Discussion on lines 262-7.

– The authors chose a specific region of the wing (between veins L3 and L4) to perform their perturbations. The justification is that this is a region that is easy to identify. But, do the results depend on this choice? Isn't the strength of the polarity signal larger closer to the wing margin?

We cannot rule out the possibility that the results depend on the choice of location, since we did not do different locations. However, in contrast to what the reviewer suggests, the strength of core PCP polarity actually does not seem to vary significantly in space and there is no evidence of stronger polarity signal near the wing margin, according to data in Merkel et al. 2014 (Figure 1D, G, J – quantification of polarity nematics across the wing over time in Stbm::eYFP).

– Use consistent English spelling throughout the manuscript (e.g., behaviour or behavior lines 169 vs. 171).

We thank the reviewer for this comment. We have revised and now use US spelling convention.

Reviewer #2 (Recommendations for the authors):

The authors have done a good job of trying to bring together what appear to be datasets generated over a fairly long period by different authors.

I do have a few issues regarding the presentation and interpretation of the data (specific points noted below). The main concern is that the authors keep flipping between saying core PCP has no effect on wing morphogenesis, to saying it has subtle effects to saying it has effects but they are not statistically significant. This is hard to reconcile with the convincing effect on tissue mechanics reveals by the laser ablation work, and possibly also the changes in adult wing shape and known roles of core PCP in both the wing and other tissues in the fly. Mostly I think it's just a question of careful choice of words and most likely leaving the question of effects of core PCP on morphogenesis open until all the data is considered together in the Discussion e.g. try to stick to observations in the Results and try to only do interpretation in the Discussion.

I also have queries about the rheological model and why the authors believe the recoil has 2 phases based on the data they show, and also about the way developmental timings have been recalibrated.

We thank the reviewer for careful reading and well-informed, thoughtful feedback of our work.

Specific points:

The references on lines 21 and 22 are an eclectic selection. The sentence seems to be about how polarized PCP proteins orient structures in epithelia, but the references include a mix of reviews and papers, some from before the polarization had been reported. For instance, Eaton 1997 was a seminal review that suggested planar polarization of stereocilia might be under the control of planar polarity pathways, but does not report such a result (I think the first papers were Montcouquiol et al. 2003 and Curtin et al. 2003?). It would probably be better for readers if the authors either cited reviews or if they want to cite primary papers, than just the first reports of polarized localization determining the orientation of structures (e.g. I think Usui et al. 1999 is the first in flies? Then Axelrod and Strutt?).

The sentence has been changed and now reads:

“Tissue-scale alignment of this pathway is known to orient cellular structures, such as hairs and cilia, and influence dynamic cellular behaviors during morphogenesis, such as cellular movements and cell divisions, through interactions with the cytoskeleton”.

We now only cite reviews.

The references on lines 27 and 28 are mostly referring to Fat-Dachsous while the rest of the paragraph is about "core". There are papers about core PCP in flies affecting cell rearrangements/divisions. Segalen et al. 2010 is a good example, but not the first to report core PCP controlling SOP division orientation (Gho and Schweisguth 1998?). Sugimura and Ishihara 2013 report changes in cell rearrangements in the wing of a core PCP mutant. Others (moving away from the wing/notum) might include many papers about ommatidial rotation in the eye, joint formation in the legs, cell rearrangements in the tracheal system (and probably others). A better review of core PCP roles in tissue morphogenesis in flies would put the work in a better context.

We thank the reviewer for this feedback. We have now changed the introduction, shortening that first paragraph to the sentence mentioned in point 1, and removing references to the Fat/Ds pathway. We have also added a more focused introduction to core PCP’s role in morphogenesis in Drosophila on lines 50-59, including many of the references you suggest.

In general, I would recommend thinking carefully about the references in the Introduction so that they are of the most help to readers from outside the field. It is hard to see an organizing principle behind the mixture of reviews from different decades, primary papers (but often not first reports), cloning, or genetic or biochemical characterization studies. As above, I would recommend recent reviews (where available) and primary papers that are the first reports, for each point the authors wish to make.

We have revisited the references and adhered to the advice of the reviewer, focusing on reviews or first reports.

Line 44 – Fz is not a cadherin.

Fixed.

Figure S1A -I don't fully agree that 30 hAPF is the "End of morphogenesis" as suggested by the figure label. Even if you ignore the emergence of hairs and wing ridges, the epithelium itself will go on expanding and folding in subsequent hours. Similar statements are made on line 73 and on Figure 1.

We have now introduced the term 'end of blade elongation flows' to describe the end of our experimental time-window and have adjusted the text and figures accordingly.

Figure S1A legend – do "tissue flows reorganize Fat PCP"? My guess is Fat PCP continues to follow the Ds and Fj gradients and its insensitivity to the tissue flows is why it ends up perpendicular to core PCP. Of course, tissue flow might also have an effect but do the authors have evidence for this (I'm not sure Merkel et al. show this?). The statement in the legend is also not the same as what lines 51-52 say, where the message is that Fat PCP remains margin oriented.

We thank the reviewer for noticing this error, which we have now corrected.

Lines 66 and 67 – in normal nomenclature pk30 = pk[30] (i.e. 30 in superscript) and stbm6 = stbm[6], consistent with fmi[frz3]. I think it would be simplest to refer to stbm[6] as stbm and pk[30] as pk after the first definition, in the same way, fmi[frz3] is referred to as fmi, but this is of course up to the authors. As an aside, fmi[frz3] was described as a strong hypomorph in Chae et al. 1999, so it seems unlikely that in frz3/frz3 the "PCP network is absent", as stronger phenotypes are seen for frz3/Df, implying some residual activity.

This is a good suggestion. We followed the advice of the reviewer and altered the nomenclature in text and figures as suggested. We also now refer to fmi and stbm as strong hypomorphs that greatly/strongly reduce core PCP.

Figure S1.2C – typo in the first word of the label.

Fixed.

Line 84 "these subtle changes are not statistically significant". This all looks reasonable, but possibly the authors need to qualify this by saying something like "considering the small numbers of wings analyzed". I appreciate the work involved to get these datasets was enormous, so higher n numbers are not feasible. However, with larger n numbers there might be a difference. The average shear values for all three mutant conditions are lower than wt. Ideally, a power calculation would be done to place a numerical value on the likelihood of the shear being the same in each condition, but again I appreciate this is difficult and possibly not appropriate post hoc.

We have revised the text (lines 98-111) and included a new subsampling of the adult wing data to estimate the likelihood of detecting a significant result in pupal stages with small N (Figure S1.5J). If PCP does indeed have a subtle effect on pupal morphogenesis, by the same magnitude as is visible in the adult wing shape, we estimate that we would detect such a difference only ~20% of the time with N=3. Thus, as the reviewer suggests, we have limited statistical power with only N=3. Therefore, while we can rule out the possibility that PCP is absolutely required for the global pattern of cell dynamics and tissue shear, as these are largely unchanged, we cannot rule out the possibility that PCP has subtle effects at this stage.

Line 85 "also do not show differences" – see the previous point. Without stats, you can't say they don't differ, but here it is implied that the data in Figure S1.2C and D have been tested for differences and none was found. This is true for S1.2C with the caveat of small n numbers and no power calculation, but for S1.2D appears to be a judgment made by eye. However, by eye total tissue area at -2h looks lower for pk[3] and stbm[6] than for wt, and at e.g. 6h all 3 mutants show lower cell area changes than wild-type. I think the most the authors can say (unless I've missed something) is that the patterns of tissue behavior look broadly the same for the mutants as for wt, suggesting no gross differences. Note lines 89-90 say "Again, we find only subtle differences" which is not the same as no differences.

We now include a statistical test for the accumulated isotropic deformation by the end of the experiment (Figure 1—figure supplement 2E). There are no significant changes. We have changed the text as suggested to say the patterns are “broadly the same” (line 92). We have also shortened the discussion of the regional analysis along the PD axis (previous lines 89-90, now lines 93-97), removing the part about subtle differences.

Lines 94-95 and Figure S1.4G – again I think it is deceptive to put weight on statistics with low n numbers and no power calculation. I'm not confident that you can "conclude that core PCP does not guide the global patterns of cell dynamics during pupal morphogenesis". The best that you can conclude is that changes are at best subtle (and in the experiments presented might be due to "noise" but could also be due to roles for core PCP in cell behavior). The authors rather undermine themselves by immediately going on to say that there is a reproducible difference in adult wing shape, although admittedly we don't know that this is due to events between 16-32h as the authors note. I think more nuanced conclusions would be better.

We respond below together with point 12.

Lines 115-116 "cellular dynamics is [are?] basically unperturbed" is another example of wording that implies the authors think cellular dynamics are perturbed to some extent, but they want to minimize the significance of this observation. Similarly, lines 198-199 "no significant phenotype" belies the difficulty in making such an assertion when the genotypes do look different in some aspects and the tests have (apparently) minimal statistical power.

Given that we do not see considerable changes in the cellular dynamics in core PCP mutants, we are confident in concluding that core PCP is not the sole determinant of these patterns. We agree, however, that it remains possible that there are more subtle effects of PCP mutation on cell flows at this stage, which would only be confirmed with better statistics. We now state this point in the text (lines 102-11, 229-30, 242-48). However, this fact does not have an impact on the main results of our work, namely, that core PCP is not required for organizing large scale tissue flows.

Lines 119-120 "When plotting displacement of the nearest bond to ablation over time, we observe both a fast (<1 s) and a slow (<20 s) regime (Figure 2C)." I think the authors are saying that their data fits a 2 phase exponential with a fast and slow phase, but I'm a bit nervous about this as the first phase seems to be based on a single data point at 1 sec. I'm not an expert on curve fitting, but only having 1 point to demonstrate the presence of a "fast" phase seems a bit unsafe. I couldn't find anything in the manuscript that describes why the authors decided a 2 phase fit was the correct choice. The subsequent model appears to assume 2 phases, but I'm interested to know if you can get just as good a fit with 1 phase and if you build a model with 1 phase does that still show pk30 having a significantly lower displacement over time? I'm concerned that a 1 phase model may have been unnecessarily ruled out.

Thank you to the reviewer for pointing out this oversimplification in our previous version. We now include a new supplemental figure (Figure 2—figure supplement 1B), which better demonstrates the need for a double exponential, as well as an explanation in the Results section (lines 129-34). In short, the single exponential fit is not sufficient (as shown in Figure 2—figure supplement 1B, right). The fast timescale (<1~s) is required to account for first 5-10 datapoints, not just the first.

Appendix 1 – I understand the issue confronting the authors. Earlier studies from the lab (from Aigouy et al. 2010 onwards) consistently saw peak cell elongation at ~23h, then in Iyer et al. peak elongation appears to be at ~28h (I'm unsure if Iyer et al. comments on this). I have a couple of comments:

Line 519 the statement "we identified a delay on the onset of pupal wing morphogenesis" is unclear to me. Do the authors mean that morphogenesis proceeds at a normal rate but just start late (similar to a train being delayed but running at the normal speed), or is everything just happening slower? The curves in panels A and B really look like the latter (e.g. in panel B the same sequence of events in 2016 takes 12h while in 2018 it takes 20h), but the authors seem to be saying it's the former.

Yes, by saying that there is a delay in the onset of the process, we mean that the train is delayed but running at the same speed. Note that we have altered this figure to now include two more 2018 videos (fmi genotype), in response to comments from Reviewer 3. This means that we can now compare 3 (1 wt, 2 fmi) 2018-2019 videos to the average of our 2016 wt videos. To better facilitate comparison, we have now added landmark gridlines at particular locations: the start and end of the 2016 videos, as well as the crossing of the curves for cell rearrangements and cell elongation changes.

Note that the reviewer’s estimation of total duration of the process – 12h in 2016 (from -4 to 8hr) vs 20h in 2018 (from -12 to 8hr) – assumes that the process has only begun at the beginning of the experiment. This is true for the later videos, but not for the 2016 videos, where the rate of shear is already high at the start of the video. Thus, we cannot know from the 2016 videos how long the actual process takes, as we do not see the start point. It is better to compare the duration of processes that are visible in both sets of videos. For example, the duration of time spanning from the point where the curves for the cell elongation changes and cell rearrangements cross each other (dashed to second dotted line) is approximately 9hr in both 2016 and 2018 videos. Thus, we conclude that the delay in the onset of the process is a far greater difference than the difference in overall rate of the process.

We have performed many experiments to try to sort out the cause of the delay, which we recently published as a preprint. We now cite this work in the Appendix.

The most obvious explanation would be lower incubation temperature (or possibly nutrition?) which is known to slow the rate of development, at least if it's not background genotype (which I'm assuming it isn't, as I assume the authors are saying all genotypes are now "delayed"). If it is a slower rate of development, then the rate of development needs recalibrating, not the reference time.

We have actually done quite some experiments to test such hypotheses. We found that the significant shift in the peak of cell elongation between 2016 and 2018 cannot be attributed to a difference in temperature (either on the microscope or during development), nutrition (plant vs yeast-based foods), genetic background (now shown with the fmi videos in the updated Appendix figure), presence of Wolbachia, or circadian gating. With all these perturbations, we see more cell elongation at 28hr than at 24hr, in contrast to our 2016 data. Our best guess is that the change in developmental timing results from the change in the wavelength of light emitted by the incubators. In fact, two incubators at the same temperature/humidity with different light emission spectra produce slightly different timecourses of cell elongation, although this behavior has not yet been fully characterized. As these experiments are tangential to the topic of this paper, we did not include them. Some of the authors have now prepared a manuscript containing these data and have published it as a preprint, which we now cite in this revised manuscript. As noted above, the delay is more prominent than the change in rate, justifying our choice to merely shift the curves, rather than rescale.

Given the prior literature (much of it from the Eaton lab) uses a linear time scheme with 0h at the start of pupal life, I'm unsure why the authors didn't just stick with this scheme and rescale the timings so peak elongation is at 23h? This would make the data much more comparable both internally in the manuscript and to past work.

Indeed, this is what we have done in the past – all timelapse data were aligned on the peak of cell elongation to make them more comparable and account for slight differences in staging, but we called them all 16hr APF. In the past, however, the shift was only ~30min and thus referring to them all as “16hr APF” was still reasonable. Now, however, if we apply the same strategy, we would have to shift by ~6-8hrs. Continuing to call the peak of cell elongation “16hr APF” at that point, when in some videos that was actually 23hr APF, would likely cause considerable confusion and makes the “hAPF – after puparium formation” reference rather meaningless. Our data from Appendix 1 and the new preprint indicate that there is something happening after the onset of pupariation that delays the onset of pupal flows, but that once the flows start, they occur at more-or-less the same rate. Since the onset of pupal flows is not visible in all of the videos (in particular the 2016 data, when at the time we start imaging the shear rates are already high), we chose to orient the data based on the peak of cell elongation. We now introduce a relative time scale called “hRPCE” for hr relative to peak cell elongation.

Reviewer #3 (Recommendations for the authors):

We believe the manuscript is likely ultimately suitable for publication in eLife, but there are a few major comments we would like the authors to address before publication.

Thank you for the careful reading of the manuscript and the excellent suggestions for improvement.

1. We find the second paragraph of the discussion confusing. In it, the authors highlight that the difference in the initial recoil velocity between wt and core PCP mutants does not lead to a phenotype in tissue morphogenesis. Then they try to explain why this difference does not generate a global phenotype, but the explanation, "the proportionality factor can depend on the genotype and can change in time", is quite unclear.

We have now revised the discussion, as well as the results, in response to this and later comments. In the discussion, we now first address the phenotype on initial recoil velocity and why it may not be reflected in global tissue stress or cell dynamics. We have taken out the confusing statement about proportionality between stress and recoil velocity. Instead, we merely point out that recoil velocity is often used as a proxy for stress in other works, and that here we highlight a limitation of that approach in looking at differences between genotypes, as recoil velocity can also report differences in fast timescale mechanics that are actually not reflected in tissue stress.

In the paragraph that starts with line 153, it seems the authors are saying their data suggests that both $k_f$ and $\eta_f$ are changing in the pk30 mutant. Is that the "proportionality factor" they mean above? If so, why not just come out and say more explicitly that the in the mutants the fast viscosity apparently changes along with $k_f$ so as to make the time scale constant? (As a side note isn't that an interesting coincidence? Would it be interesting/useful to speculate that there might be some sort of feedback loop or mechanical coupling that drives the compensating change in the viscosity?)

No, this is not the proportionality factor we meant above. The proportionality factor was meant to describe how tissue stress may be proportional to recoil velocity. The reviewer is correct, however, that k_f changes with eta_f. We have now clarified this point, in particular we speculate that this proportionality suggests that both parameters originate from a single microscopic mechanism that constrains the relaxation time eta_f/k_f, such as cytoskeletal turnover time (lines 174-80).

And we believe the "change in time" clause above refers to the paragraph that starts at line 179, which is also confusing (see next comment). But isn't the change in viscosity alone already enough to explain the lack of large-scale phenotype? Do you need this change in time also? Would such a change in time be alone sufficient to explain the lack of a global phenotype?

We agree with the referee that this phrase was confusing and we have removed it.

2. The paragraph starting at line 179 is apparently asking the reader to compare the data for the pk30 mutants in Figure 3B (blue data) to the data for the pk30 mutant in S3.1C (blue line). But it's very difficult to do that, as the axes for the two plots are different, and there are not even units given for the y-axis in Figure 3B. We think we're supposed to see that the blue curves in the two plots have different shapes in time. But that's really difficult to see toggling between them. Would it be possible to show both sets of data on the same plot (perhaps one set rescaled) so that we could directly compare the shapes? Moreover, it would be useful to state something more about the data than "its different". It looks like maybe the ESCA data peaks earlier in time than the initial recoil velocity. What does that mean?

We thank the referee for the good suggestion, and we have now updated the supplemental figure (Figure 3—figure supplement 1D) to plot the relevant data in similar styles, side-by-side, to facilitate a direct comparison. We have now re-written the results to better present the ESCA method as well as the comparison between the ESCA and the initial recoil velocity results (lines 161-76, 199-212).

3. What is the possible reason for different times for the peaks in anisotropic stress and cell elongation (discussion in the last paragraph of section 2.3)? It seems the response time for the stresses is almost 2 to 4 h.

In previous work (Etournay et al., eLife 2015), we measured the relationship between initial recoil velocity and cell elongation, and we found that they follow a linear relationship with an offset. We have interpreted this offset as a signature of active stress in the tissue, but at the time we treated it as a constant in time. However, current results highlight the possibility that the active stress is dynamic. This opens interesting questions for future work, in particular the possibility that active stresses in the pupal wing are mechanosensitive, as we have recently proposed to account for cell elongation patterns in the larval wing disc (Dye et al., eLife 2021). We now explain this in results (lines 216-220) and in the discussion (lines 262-67).

4. The discussion on ESCA analysis is incomplete. It is not clear how σ/Ks is related to anisotropic ((σ¯/2K)) and isotropic stress (σ/K¯). The variables K, K¯ and σ¯ are not defined. It is mentioned in the caption of Figure 2 that the equations for anisotropic and isotropic stress are on the right side. However, there is no equation in Figure 2F.

We thank the reviewer for pointing out this confusion. We have rewritten this section of the Results, including a more extensive description of ESCA and why we use it (lines 161-76). We have also adjusted the variable names to prevent confusion between the parameters in ESCA and our rheological model and now introduce them in the text.

https://doi.org/10.7554/eLife.85581.sa2

Article and author information

Author details

  1. Romina Piscitello-Gómez

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. DFG Excellence Cluster Physics of Life, Technische Universität Dresden, Dresden, Germany
    Contribution
    Data curation, Formal analysis, Investigation, Software, Validation, Visualization, Writing – original draft, Writing – review and editing
    Contributed equally with
    Franz S Gruber
    Competing interests
    No competing interests declared
  2. Franz S Gruber

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. National Phenotypic Screening Centre, University of Dundee, Dundee, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Software, Validation, Writing – review and editing
    Contributed equally with
    Romina Piscitello-Gómez
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2008-8460
  3. Abhijeet Krishna

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. DFG Excellence Cluster Physics of Life, Technische Universität Dresden, Dresden, Germany
    3. Center for Systems Biology Dresden, Dresden, Germany
    Contribution
    Formal analysis, Methodology, Software, Writing – review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9291-500X
  4. Charlie Duclut

    1. Laboratoire Physico-Chimie Curie, CNRS UMR 168, Institut Curie, Université PSL, Sorbonne Université, Paris, France
    2. Max Planck Institute for Physics of Complex Systems, Dresden, Germany
    3. Université Paris Cité, Laboratoire Matière et Systèmes Complexes, Paris, France
    Contribution
    Formal analysis, Software
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-8595-6815
  5. Carl D Modes

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. DFG Excellence Cluster Physics of Life, Technische Universität Dresden, Dresden, Germany
    3. Center for Systems Biology Dresden, Dresden, Germany
    Contribution
    Validation, Investigation, Formal analysis, Software
    Competing interests
    No competing interests declared
  6. Marko Popović

    1. DFG Excellence Cluster Physics of Life, Technische Universität Dresden, Dresden, Germany
    2. Center for Systems Biology Dresden, Dresden, Germany
    3. Max Planck Institute for Physics of Complex Systems, Dresden, Germany
    Contribution
    Investigation, Formal analysis, Data curation, Software
    Competing interests
    No competing interests declared
  7. Frank Jülicher

    1. DFG Excellence Cluster Physics of Life, Technische Universität Dresden, Dresden, Germany
    2. Center for Systems Biology Dresden, Dresden, Germany
    3. Max Planck Institute for Physics of Complex Systems, Dresden, Germany
    Contribution
    Visualization, Validation, Investigation, Writing – original draft, Formal analysis, Writing – review and editing, Software
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-4731-9185
  8. Natalie A Dye

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. DFG Excellence Cluster Physics of Life, Technische Universität Dresden, Dresden, Germany
    3. Mildred Scheel Nachwuchszentrum P2, Medical Faculty, Technische Universität Dresden, Dresden, Germany
    Contribution
    Data curation, Supervision, Writing – original draft, Writing – review and editing
    For correspondence
    natalie_anne.dye@tu-dresden.de
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4859-6670
  9. Suzanne Eaton (deceased)

    1. Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany
    2. DFG Excellence Cluster Physics of Life, Technische Universität Dresden, Dresden, Germany
    3. Center for Systems Biology Dresden, Dresden, Germany
    4. Biotechnologisches Zentrum, Technische Universität Dresden, Dresden, Germany
    Contribution
    Visualization, Validation, Investigation, Writing – original draft, Formal analysis, Writing – review and editing
    Competing interests
    No competing interests declared

Funding

Max Planck Society

  • Franz S Gruber
  • Abhijeet Krishna
  • Charlie Duclut
  • Carl D Modes
  • Frank Jülicher
  • Natalie A Dye
  • Suzanne Eaton
  • Romina Piscitello-Gómez
  • Marko Popović

Deutsche Forschungsgemeinschaft (EXC-2068-390729961)

  • Natalie A Dye

Deutsche Forschungsgemeinschaft (SPP1782/EA4/10-1)

  • Suzanne Eaton
  • Natalie A Dye
  • Romina Piscitello-Gómez
  • Franz S Gruber

Deutsche Krebshilfe (MSNZ-P2 Dresden)

  • Natalie A Dye

Austrian Academy of Sciences (DOC Fellowship)

  • Franz S Gruber

Agence Nationale de la Recherche (ANR-11-LABX-0071)

  • Charlie Duclut

Agence Nationale de la Recherche (ANR-18-IDEX-0001)

  • Charlie Duclut

Deutsche Forschungsgemeinschaft (SPP1782/EA4/10-2)

  • Suzanne Eaton
  • Natalie A Dye
  • Romina Piscitello-Gómez

Open access funding provided by Max Planck Society. The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Stephan Grill for giving us access to the microscope used for laser ablation. We thank the Light Microscopy Facility, the Computer Department, and the Fly Keepers of the MPI-CBG for their support and expertise. We would like to thank Christian Dahmann and Jana Fuhrmann for comments on the manuscript prior to publication. This work was funded by Germany’s Excellence Strategy – EXC-2068-390729961– Cluster of Excellence Physics of Life of TU Dresden, as well as grants awarded to SE from the Deutsche Forschungsgemeinschaft (SPP1782, EA4/10-1, EA4/10-2) and core funding of the Max-Planck Society to SE and NAD. NAD additionally acknowledges funding from the Deutsche Krebshilfe (MSNZ P2 Dresden). AK and RPG were funded through the Elbe PhD program. FSG was supported by a DOC Fellowship of the Austrian Academy of Sciences. CD acknowledges the support of a postdoctoral fellowship from the LabEx 'Who Am I?' (ANR-11-LABX-0071) and the Université Paris Cité IdEx (ANR-18-IDEX-0001) funded by the French Government through its 'Investments for the Future'. We dedicate this work to our coauthor Prof. Dr. Suzanne Eaton, who tragically passed away before the finalization of the project.

Senior Editor

  1. Detlef Weigel, Max Planck Institute for Biology Tübingen, Germany

Reviewing Editor

  1. Marcos Nahmad, Center for Research and Advanced Studies (Cinvestav), Mexico

Reviewers

  1. Marcos Nahmad, Center for Research and Advanced Studies (Cinvestav), Mexico
  2. M Lisa Manning, Syracuse University, United States

Version history

  1. Preprint posted: December 10, 2022 (view preprint)
  2. Received: December 15, 2022
  3. Accepted: December 18, 2023
  4. Accepted Manuscript published: December 20, 2023 (version 1)
  5. Version of Record published: February 2, 2024 (version 2)
  6. Version of Record updated: February 5, 2024 (version 3)

Copyright

© 2023, Piscitello-Gómez, Gruber et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 428
    Page views
  • 122
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Romina Piscitello-Gómez
  2. Franz S Gruber
  3. Abhijeet Krishna
  4. Charlie Duclut
  5. Carl D Modes
  6. Marko Popović
  7. Frank Jülicher
  8. Natalie A Dye
  9. Suzanne Eaton
(2023)
Core PCP mutations affect short-time mechanical properties but not tissue morphogenesis in the Drosophila pupal wing
eLife 12:e85581.
https://doi.org/10.7554/eLife.85581

Share this article

https://doi.org/10.7554/eLife.85581

Further reading

    1. Developmental Biology
    2. Neuroscience
    Longjiang Xu, Zan Yuan ... Xiaozhong Peng
    Research Article

    Despite intense research on mice, the transcriptional regulation of neocortical neurogenesis remains limited in humans and non-human primates. Cortical development in rhesus macaque is known to recapitulate multiple facets of cortical development in humans, including the complex composition of neural stem cells and the thicker supragranular layer. To characterize temporal shifts in transcriptomic programming responsible for differentiation from stem cells to neurons, we sampled parietal lobes of rhesus macaque at E40, E50, E70, E80, and E90, spanning the full period of prenatal neurogenesis. Single-cell RNA sequencing produced a transcriptomic atlas of developing parietal lobe in rhesus macaque neocortex. Identification of distinct cell types and neural stem cells emerging in different developmental stages revealed a terminally bifurcating trajectory from stem cells to neurons. Notably, deep-layer neurons appear in the early stages of neurogenesis, while upper-layer neurons appear later. While these different lineages show overlap in their differentiation program, cell fates are determined post-mitotically. Trajectories analysis from ventricular radial glia (vRGs) to outer radial glia (oRGs) revealed dynamic gene expression profiles and identified differential activation of BMP, FGF, and WNT signaling pathways between vRGs and oRGs. These results provide a comprehensive overview of the temporal patterns of gene expression leading to different fates of radial glial progenitors during neocortex layer formation.

    1. Developmental Biology
    Marta Grzonka, Hisham Bazzi
    Research Article

    SAS‑6 (SASS6) is essential for centriole formation in human cells and other organisms but its function in mouse is unclear. Here, we report that Sass6‑mutant mouse embryos lack centrioles, activate the mitotic surveillance cell death pathway and arrest at mid‑gestation. In contrast, SAS‑6 is not required for centriole formation in mouse embryonic stem cells (mESCs), but is essential to maintain centriole architecture. Of note, centrioles appeared after just one day of culture of Sass6‑mutant blastocysts, from which mESCs are derived. Conversely, the number of cells with centrosomes is drastically decreased upon the exit from a mESC pluripotent state. At the mechanistic level, the activity of the master kinase in centriole formation, PLK4, associated with increased centriolar and centrosomal protein levels, endow mESCs with the robustness in using SAS‑6‑independent centriole-duplication pathways. Collectively, our data suggest a differential requirement for mouse SAS‑6 in centriole formation or integrity depending on PLK4 and centrosome composition.