1. Developmental Biology and Stem Cells
Download icon

aPKC-mediated displacement and actomyosin-mediated retention polarize Miranda in Drosophila neuroblasts

  1. Matthew Robert Hannaford
  2. Anne Ramat
  3. Nicolas Loyer
  4. Jens Januschke  Is a corresponding author
  1. University of Dundee, United Kingdom
Research Article
  • Cited 0
  • Views 1,037
  • Annotations
Cite as: eLife 2018;7:e29939 doi: 10.7554/eLife.29939

Abstract

Cell fate assignment in the nervous system of vertebrates and invertebrates often hinges on the unequal distribution of molecules during progenitor cell division. We address asymmetric fate determinant localization in the developing Drosophila nervous system, specifically the control of the polarized distribution of the cell fate adapter protein Miranda. We reveal a step-wise polarization of Miranda in larval neuroblasts and find that Miranda’s dynamics and cortical association are differently regulated between interphase and mitosis. In interphase, Miranda binds to the plasma membrane. Then, before nuclear envelope breakdown, Miranda is phosphorylated by aPKC and displaced into the cytoplasm. This clearance is necessary for the subsequent establishment of asymmetric Miranda localization. After nuclear envelope breakdown, actomyosin activity is required to maintain Miranda asymmetry. Therefore, phosphorylation by aPKC and differential binding to the actomyosin network are required at distinct phases of the cell cycle to polarize fate determinant localization in neuroblasts.

https://doi.org/10.7554/eLife.29939.001

Introduction

The development of the central nervous system depends on asymmetric cell divisions for the balanced production of progenitor and differentiating cells. During vertebrate and invertebrate neurogenesis, cell fates can be established through the asymmetric inheritance of cortical domains or fate determinants during asymmetric division of progenitor cells (Alexandre et al., 2010; Doe, 2008; Knoblich, 2008; Marthiens and ffrench-Constant, 2009).

A vital step in asymmetric cell division is the establishment of a polarity axis. Asymmetrically dividing Drosophila neuroblasts (NBs) establish an axis of polarity at the onset of mitosis and, as in many other polarized cells, this depends on the activity of the Par protein complex (Goldstein and Macara, 2007). As NBs enter prophase, the Par complex, comprising Par3/Bazooka (Baz), aPKC and Par-6, assembles at the apical NB pole, and this drives the localization of fate determinants to the opposite (basal) NB pole, thus establishing the apico–basal polarity axis (Betschinger et al., 2003; Homem and Knoblich, 2012; Petronczki and Knoblich, 2001; Prehoda, 2009; Rolls et al., 2003; Wodarz et al., 2000; 1999).

Upon NB division, the basally-localized fate determinants segregate to the daughter cell, which then commits to differentiation. Two adapter proteins localize the fate determinants to the basal NB cortex in mitosis: Partner of Numb (Pon), localizes the Notch signaling regulator Numb (Lu et al., 1998; Uemura et al., 1989), and Miranda (Mira), (Ikeshima-Kataoka et al., 1997; Shen et al., 1997) localizes fate determinants including the homeobox transcription factor Prospero (Pros) and the translational repressor Brat (Betschinger et al., 2006; Ikeshima-Kataoka et al., 1997; Lee et al., 2006). In the absence of Mira, fate determination is impaired and tumor-like growth can occur in larval NB lineages (Caussinus and Gonzalez, 2005; Ikeshima-Kataoka et al., 1997).

Intriguingly, Mira is uniformly cortical in interphase larval brain NBs (Sousa-Nunes et al., 2009). In embryonic NBs, Mira and its cargo Pros co-localize, at the interphase cortex (Spana and Doe, 1995). The cortical localization of Pros seems to depend on Mira, since in interphase mira mutant NBs, Pros is found in the NB nucleus (Matsuzaki et al., 1998). Given that the levels of nuclear Pros in NBs are important for the regulation of entry and exit from quiescence (Lai and Doe, 2014), Mira localization and its regulation are likely to be relevant for regulating nuclear Pros levels in interphase NBs, but how this might be achieved is unknown.

How the asymmetric localization of Miranda in mitotic NBs is achieved is also a long-standing question, however, the mechanism is still not fully understood. In embryonic NBs, Mira localization requires actin (Shen et al., 1998) and myosin activity, since mutation in the myosin regulatory light chain spaghetti squash (sqh, Barros et al., 2003) or the Myosin VI jaguar (Petritsch et al., 2003) lead to Mira localization defects. Moreover, Mira does not achieve a polarized distribution in embryos into which the Rho kinase (ROCK) inhibitor Y-27632 was injected. ROCK can regulate Myosin activity by phosphorylating the light chain of Myosin (Amano et al., 1996). Furthermore, the effect of pharmacological ROCK inhibition on Mira in embryonic NBs could be rescued by the expression of a phosphomimetic version of Myosin’s regulatory light chain, Spaghetti Squash (called SqhEE, Winter et al., 2001), which led to the idea that Myosin II might play a critical role in Mira localization and that aPKC affects Mira localization indirectly through regulating Myosin II (Barros et al., 2003).

However, Y-27632 can inhibit aPKC directly, and Mira is a substrate of aPKC (Atwood and Prehoda, 2009; Wirtz-Peitz et al., 2008). In fact many aPKC substrates, including Numb and Mira, contain a basic and hydrophobic (BH) motif that can be phosphorylated by aPKC. When phosphorylated, the substrates can no longer directly bind phospholipids of the plasma membrane (PM), (Bailey and Prehoda, 2015; Dong et al., 2015; Smith et al., 2007). Thus, asymmetric Mira localization in mitotic NBs can, in principle, be explained by keeping the activity of aPKC restricted to the apical pole. According to this model, Mira retention at the cortex is primarily mediated through direct interaction with the PM mediated by its BH motif. Therefore, although the contribution of actin was revealed in pharmacological and genetic experiments, it remains unclear how actin contributes to fate determinant localization.

To understand the regulation of Mira localization throughout the NB cell cycle, we set out to determine differences and similarities in the parameters of Mira binding in interphase and mitosis, and analysed the transition between the two different localization patterns. Mira has been shown to be able to localize to microtubules (Mollinari et al., 2002) and directly bind actin (Sousa-Nunes et al., 2009) and phospholipids of the PM (Bailey and Prehoda, 2015). Therefore, we re-examined the role of the cytoskeleton and PM interaction for Mira localization. We reveal that Mira uses two modes to interact with the cortex: in interphase, to retain Mira uniformly at the cortex direct interaction of Mira’s BH motif with phospholipids of the PM are necessary and likely sufficient. This interaction is inhibited by aPKC-dependent phosphorylation of the BH motif at prophase. After nuclear envelope breakdown Mira requires BH motif and actomyosin-dependent processes for asymmetric retention at the cortex. Therefore, we propose that Mira binds to the PM in interphase and to the actomyosin cortex in mitosis, both of which appear BH motif dependent.

Results

Uniform Miranda is cleared from the cortex during prophase and reappears asymmetrically localized after nuclear envelope breakdown

We confirmed that Mira localizes uniformly to the cortex in interphase larval NBs and also find that its cargo Pros localizes to the interphase cortex in a Mira-dependent manner (Figure 1A, Figure 1—figure supplement 1). These results suggest that Mira regulates the localization of its cargoes throughout the cell cycle. Therefore, we sought to address the regulation of cortical Mira in interphase and mitosis, and the transition between these localizations patterns of Mira.

Figure 1 with 3 supplements see all
Miranda is cleared from the cortex before localizing in a basal crescent in mitosis.

(A) Larval brain NBs fixed and stained as labeled at the indicated cell cycle stage. (B) Selected frames from Video 1. NB in primary cell culture expressing Baz::GFP (green) and Mira::mCherry (red) in the transition from interphase to mitosis. Arrowheads point at Mira being cleared (−8) and at basal Mira crescent (+4). (B’) Quantification of cortical Mira::mCherry signal plotting the fluorescence intensities from the apical to the basal pole computationally straightening (Kocsis et al., 1991) the cortices of five NBs against the distance in percent. Fluorescence was background subtracted and normalized to background subtracted cytoplasmic signal (1, dotted line). Cortical signal (yellow dotted line) and signal after NEB (green dotted line). Error bars, standard deviation. (C) Schematic of Mira localization. BAC{mira::mcherry-MS2} was the source of Mira::mCherry. Scale bar 10 µm. Time stamp: minutes.

https://doi.org/10.7554/eLife.29939.002

To monitor the establishment of asymmetric Mira localization, we used a BAC construct in which Mira was tagged with mCherry at its C-terminus (see also Figure 1—figure supplement 2). This tagged Mira recapitulated uniform cortical localization in interphase (Figure 1B, −33 to NEB) and polarized localization to the basal pole in mitosis (Figure 1B, +4), showing a 2.5-fold increase in intensity at the basal cortex (n = 5, Figure 1B’). This transition occurred in two distinct steps: First, during prophase, Mira was rapidly excluded from the apical pole, where Baz (Video 1) and aPKC (Video 2) began to accumulate (Figure 1B -8 to NEB, Figure 1B’ −15). Subsequently, Mira was progressively cleared from most of the rest of the cortex in an apical-to-basal direction (Figure 1B,B’ -4, −1, respectively to NEB); Second, following NEB, Mira reappeared at the cortex in a basal crescent (Figure 1B,B’ +4, +5 respectively). We recapitulated these steps using overexpression of GFP::Mira and by antibody staining of endogenous, non-tagged Mira (Figure 1—figure supplement 3).

Video 1
Interphase cortical Miranda is removed at the onset of mitosis.

Spinning disc confocal image of a neuroblast expressing Baz::GFP (red) and Mira::mCherry (green). For this and all subsequent videos maximum projection after a 3D Gaussian blur (FIJI, radius 8/.8/1) of 7 consecutive equatorial planes taken at 0.4 µm spacing is shown. Z-stacks taken every minute. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.006
Video 2
Interphase cortical Miranda is removed at the onset of mitosis.

Spinning disc confocal image of a neuroblast expressing aPKC::GFP (green) and Mira::mCherry (red). Z-stacks taken every minute. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.007

In conclusion, Miranda transitions from a uniformly cortical localization with low intensity levels in interphase, to a basal localization with high intensity levels in metaphase (Figure 1B,B’, C). These cell-cycle dependent differences in cortical Mira intensities prompted the idea that Mira might use different modes of binding to the cortex in interphase versus mitosis. Therefore, we assayed for potential differences in cortical binding of Mira in interphase versus mitosis to address whether Mira is retained at the cortex primarily by BH motif interaction with the PM, or whether other modes of cortical retention contribute.

Actomyosin is required for both establishment and maintenance of Miranda crescents

We started by re-examining the role of the actin cytoskeleton by disrupting it with Latrunculin A (LatA). F-actin has been shown to be involved in Mira localization in mitotic embryonic NBs (Shen et al., 1998). Therefore, we tested if an intact actin network was also required for Miranda localization in larval NBs. Despite efficient disruption of F-actin (Figure 2—figure supplement 1) causing cytokinesis failure (Figure 2A, 3:02, related to Video 3), LatA treatment did not affect the uniform interphase cortical localization of Mira (Figure 2A, 2:06) and Mira was driven into the cytoplasm during prophase (Figure 2A, 2:21). However, Mira failed to relocalize to a basal crescent following NEB (Figure 2A, 2:33). Thus, an actin cytoskeleton is not required for the interphase localization of Miranda, nor its removal from the interphase cortex. However, it is required for Mira basal localization following NEB.

Figure 2 with 1 supplement see all
Differential response of Mira localization in interphase and mitosis to disruption of the actin cytoskeleton.

(A) Stills from Video 3. LatA was added to a cycling NB in primary cell culture expressing Baz::GFP (green) and Mira::mCherry (red). Arrowheads point at cortical Mira after culturing ~1 hr with LatA (2:06). At 1 min to NEB, Mira::mCherry is cleared from the cortex (2:21). Mira forms no crescent in the next mitosis (2:33), but after cytokinesis fails (note bi-nucleated cell at 3:02), Mira is recruited to the cortex (arrowheads). Bottom panels: Colcemid-arrested NBs expressing Baz::GFP and Mira::mCherry. LatA was added at 5 µM prior to imaging at 15 s. intervals. Mira crescents (arrows) are lost upon LatA treatment. (B) Cycling NB in primary cell culture expressing Mira::mCherry, that remains cortical upon ML-7 addition (15 µM; interphase: 0’ and 138’, arrowheads), is cleared 1 min prior to NEB (174’), does not form a crescent after NEB (278’, arrow), but accumulates on the spindle (seen in cross section). After ML-7 washout, a basal Mira::mCherry crescent recovers (arrowhead, 328’). (C) Related to Video 5. Colcemid-arrested NB in primary cell culture expressing Baz::GFP (green) and Mira::mCherry (red). After addition of 20 µM ML-7 Mira (arrowhead, 0’) becomes cytoplasmic (arrow,+9’), but upon ML-7 washout a Mira crescent recovers. (D) The effect of 20 µM ML-7 can be quenched by overexpressing a phospho-mimetic form Sqh (SqhEE). Colcemid arrested NBs (ctrl: Mira::mCherry: SqhEE: Mira::mCherry co-expressing SqhEE by worniuGal4). Ctrl and SqhEE were co-cultured and ML-7 added (related to Video 6). Quantification of the time required to cause Mira::mCherry to become cytoplasmic shown on the left. Two-tailed t test for independent means revealed significance. BAC{mira::mcherry-MS2} was the source of Mira::mCherry. Scale bar: 10 µm.

https://doi.org/10.7554/eLife.29939.008
Video 3
Interphase cortical Miranda is actin independent.

Spinning disc confocal image of a NB expressing Baz::GFP (red) and Mira::mCherry (green) showing a control division before 1 µM LatA was added. Z-stacks taken every minute. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.010

We next tested whether F-actin was required to maintain Mira crescents. To this end, we arrested NBs in metaphase by depolymerising microtubules with colcemid (at which point Mira crescents are established) and then treated them with LatA. In this situation, LatA treatment caused Mira to relocalize to the cytoplasm (Figure 2A), indicative of a role for F-actin in retaining Mira basally after NEB.

However, LatA treatment after NEB also led to the redistribution of Baz/Par3 and aPKC to the entire NB periphery. Therefore, the observed effect on Mira could be indirect, caused by changes in aPKC localization when F-actin is compromised. Assuming that aPKC activity is restricted to the cortex (Atwood et al., 2007; Rodriguez et al., 2017) we sought to distinguish between direct and indirect effects on Mira by determining if Mira loss preceded (indicative of direct effect of loss of actin) or followed (indicative of an indirect effect caused by changes in aPKC localization) changes in Par complex distribution in colcemid arrested NBs upon LatA treatment. We found that Mira loss preceded changes in cortical aPKC/Baz localization in response to LatA. This occurred about 2.8 ± 1 min (n = 13) before aPKC (Video 4) or Baz (Video 5) became detectable at the basal cortex (Figure 2—figure supplement 1). Therefore, changes in cortical aPKC localization upon LatA treatment are unlikely to drive Mira off the cortex. We conclude that an intact actin network facilitates both establishment and maintenance of asymmetric Mira localization in mitotic larval NBs, though it is not clear how actin carries out this function.

Video 4
Colcemid-arrested NB expressing aPKC::GFP and Mira::mCherry that was treated with 5 µM LatA at the beginning of the recording at 16 s intervals.

The cortex was straightened out and split at the apical pole such that aPKC::GFP appears right and left and Mira in the center. Fluorescence profiles shown below. Note that Mira falls off homogenously from the cortex and becomes cytoplasmic at 5:36 (red arrowhead), while the detectable borders of cortical aPKC (green arrowheads) have not yet changed. Only from 07:12 onward aPKC rise above cytoplasmic levels where Mira was localized. Time stamp: mm:ss.

https://doi.org/10.7554/eLife.29939.011
Video 5
Colcemid-arrested NB expressing Baz::GFP and Mira::mCherry that were treated with 5 µM LatA at the beginning of the recording at 16 s intervals.

The cortex was straightened out and split at the apical pole such that Baz::GFP appears right and left and Mira in the center. Fluorescence profiles shown below. Note that Mira falls off homogenously from the cortex and becomes cytoplasmic at 9:00 (asterisks), while the detectable borders of cortical Baz (arrowheads) have not yet changed. Only from 12:30 onward Baz rise above cytoplasmic levels where Mira was localized. Time stamp: mm:ss.

https://doi.org/10.7554/eLife.29939.012

One possibility is that actin-associated Myosins (Barros et al., 2003; Erben et al., 2008; Petritsch et al., 2003) stabilize Mira at the basal cortex in mitosis. Therefore, we tested which step of Mira localization involved myosin. Myosin motor activity is enhanced by the phosphorylation of myosin regulatory light chain, encoded by the sqh gene in Drosophila (Jordan and Karess, 1997). We disrupted this phosphorylation using ML-7, a specific inhibitor of myosin light chain kinase (MLCK, Bain et al., 2003). As with LatA, ML-7 treatment of cycling NBs affected neither the uniform interphase cortical localization of Mira (Figure 2B, 0’) nor its clearance during prophase (Figure 2B, 138-278’) but resulted in a failure to establish a basal crescent after NEB, which was restored upon drug washout (Figure 2B, 328’). In colcemid arrested NBs, ML-7 treatment also resulted in Mira redistributing from the basal crescent to the cytoplasm, which also was restored upon ML-7 washout. However, unlike LatA treatment, ML-7 treatment did not cause the Par complex to spread over to the entire cortex (Figure 2C, Video 6). Finally, we demonstrated the specificity of the effect of ML-7 by counteracting its effect with the phosphomimetic version of myosin regulatory light chain SqhEE, as in (Das and Storey, 2014). Overexpressing SqhEE significantly delayed loss of cortical Mira after ML-7 addition in colcemid-arrested NBs (Figure 2D, Video 7).

Video 6
Myosin inhibition reversibly affects basal Mira anchoring in a polarized NB.

A colcemid arrested NB expressing Baz::GFP (green) and Mira::mCherry (red) in primary cell culture was treated with 20 µM ML-7 which was washed out when indicated. Left panel Baz::GFP, middle panel Mira::mCherry, right panel merge. Z-stacks taken every minute. Time stamp: mm:ss.

https://doi.org/10.7554/eLife.29939.013
Video 7
The effect of ML-7 on cortical Mira localization in mitosis can be delayed by overexpressing SqhEE.

Mira::mCherry NB (ctrl) and Mira::mCherry NB co-expressing SqhEE (rescue) were co cultured in neighboring clots in the same dish and the effect of ML-7 on cortical Mira recorded. Z-stacks taken every two minutes. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.014

In summary, these results support the notion that Mira interacts differently with the cortex in interphase and in mitosis since F-actin and myosin activity contribute to establish and maintain asymmetric Mira crescents at the basal cortex following NEB, but they are not essential for uniform cortical localization of Mira in interphase nor for Mira clearance during prophase.

Miranda binds directly to the plasma membrane in interphase NBs

The observation that Mira continues to localize to the cortex in interphase upon F-actin disruption suggested that it is directly retained at the PM. In Drosophila S2 cells Mira binds directly to phospholipids of the PM via its BH motif, and phosphorylation of this motif by aPKC abolishes this binding (Bailey and Prehoda, 2015). Therefore, we tested the role of the BH motif in Mira localization in interphase and mitosis.

Mira’s membrane interaction is sensitive to the phosphorylation by aPKC of a Serine which resides in the BH motif (S96), (Bailey and Prehoda, 2015). To understand the influence membrane binding has on the dynamic localization in NBs, we used a CrispR based approach to generate four mCherry-tagged Miranda alleles (Figure 3A): (1) control (S96 unchanged, ctrl, able to rescue embryonic lethality); (2) a phosphomutant (S96A, homozygous embryonic lethal); (3) a phosphomimetic (S96D, homozygous embryonic lethal and shown in vitro to reduce phospholipid binding and Mira recruitment to the PM when overexpressed in S2 cells); and (4) a complete deletion of the BH motif (ΔBH, homozygous embryonic lethal).

Figure 3 with 1 supplement see all
Miranda binds to the plasma membrane in interphase NBs via its BH motif.

(A) Schematic indicating the different Mira alleles used. Mira::mCherry localizes cortically uniform in interphase (arrowheads t1), is cleared from the cortex shortly before NEB and forms a crescent (arrowheads t3) thereafter that is inherited by daughter cells (related to Video 8). The phosphomutant S96A is uniformly cortical in interphase, accumulates apically shortly before NEB (arrow, t2), and is uniformly cortical after NEB (arrowheads t3) and in telophase (t4, related to Video 9). The phosphomimetic S96D localizes to cortical microtubules in interphase (arrow t1), is cleared from the cortex before NEB and asymmetric after NEB (arrowheads t3) and segregates to daughter cells (related to Video 10). Deletion of the BH motif leads to cortical microtubule localization in interphase (arrow t1), cytoplasmic localization before and after NEB and reappearance on microtubules around cytokinesis (related to Video 11). (B) Neuroblasts expressing the indicated Mira alleles were treated with 1 µM LatA or 50 µM colcemid for 60 min. Cortical localization of S96A is insensitive to LatA treatment. Below: While the control remains cortical, S96D and ΔBH become cytoplasmic upon colcemid treatment. (C) Frequency of indicated localization of the different Mira mutants. (D) Schematic of the localization of the different Mira alleles. Scale bar: 10 µm.

https://doi.org/10.7554/eLife.29939.015

The control localized uniformly to the cortex in interphase, was cleared from the cortex in prophase and reappeared as a basal crescent after NEB (Figure 3A, Video 8). In contrast, the phosphomutant S96A localized uniformly to the cortex in interphase but was not cleared at the onset of prophase and remained detectable on the entire cortex throughout mitosis, when it was also transiently enriched at the apical pole (possibly because it forms abnormally stable interactions with apically localized aPKC). After NEB, S96A mutant protein remained localized uniformly at the cortex, even in the presence of LatA (Figure 3A,B, Video 9). In contrast, the phosphomimetic S96D did not localize robustly to the cortex in interphase and instead accumulated predominantly on cortical microtubules, as evidenced by its relocalization to the cytoplasm upon colcemid treatment (Figure 3A,B). Nevertheless, S96D always localized asymmetrically, with a basal bias at the cortex following NEB, which appeared to occur at reduced levels compared to controls (Figure 3A, Video 10).

Video 8
Control miramCherry allele generated by CrispR/Cas9.

Mira localizes to the interphase cortex, from where it is cleared before NEB. Then Mira relocalizes to a larger crescent. Therefore this allele and Mira::mCherry (BAC rescue) are undistinguishable in terms of Mira dynamics. This control further shows that the MS2 binding site in the BAC rescue construct does not interfere with Mira cortical dynamics. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.017
Video 9
Phosphomutant S96A allele of Mira tagged with mCherry at the C-terminus.

Mira localizes uniformly to the interphase cortex. Shortly before NEB, S96A is apically enriched, before being uniformly cortical after NEB and during division. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.018
Video 10
Phosphomimetic S96D allele of Mira tagged with mCherry at the C-terminus.

S96D localizes to microtubules in interphase, but is asymmetric in mitosis. Note the signal resembling subcortical microtubules in interphase converging at the apical pole. After NEB a basal crescent is detectable. At 115:30 a z-stack spanning the entire NB was collected and the maximum projection is frozen. After this, 50 µM colcemid was added to reveal if MiraS96D::mCherry binds to the cortex. Next frozen frame: similar stack after 30 min in colcemid. Next frozen frame: 50 min in colcemid – no cortical signal is detectable. Last frozen frame 65 min in colcemid. Time stamp: mm:ss. Scale 15 µm.

https://doi.org/10.7554/eLife.29939.019

Deletion of the BH motif abolished both uniform cortical localization in interphase and asymmetric cortical localization after NEB, when it was entirely cytoplasmic (Figure 3A, Video 11, see Figure 3C for quantification and Figure 3D for summary of the localization of the Mira). Finally, ectopic activation of aPKC by overexpression of constitutively active aPKCΔN (Betschinger et al., 2003) also prevented Mira cortical localization in interphase and most of mitosis. Of note, Mira localization was rescued in telophase in aPKCΔN overexpressing NBs (Figure 3—figure supplement 1A,A’), suggesting that even in the presence of deregulated aPKC activity, Mira cortical association is not completely lost.

Video 11
Mira requires its BH motif for interphase cortical localization (see main text) and basal localization in mitosis.

The BH motif in Mira has been deleted by gene editing and this Mira mutant tagged with mCherry at the C-terminus (miraΔBHmCherry). MiraΔBH::mCherry when homozygous is found on the interphase microtubule network and in the cytoplasm during mitosis. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.020

These findings support the idea that, in interphase, the BH motif is necessary and likely to be sufficient to mediate interactions with phospholipids of the PM leading to uniform cortical localization of Mira. These findings also show that phospho-regulation of the BH motif affects Mira localization in both phases of the cell cycle. Therefore, these observations support the model that Mira uses only one mode, BH motif mediated PM interactions, for cortical association throughout the cell cycle. However, this model does not readily explain differences in the response of Mira to LatA and ML-7 in interphase versus mitosis (Figure 2).

Failure to clear interphase Miranda in apkc mutants results in persistence of uniform plasma membrane bound Miranda in mitosis

NBs mutant for apkc fail to localize Mira asymmetrically in mitosis (Rolls et al., 2003). Since we found that clearance of uniform cortical Mira at the onset of mitosis fails when S96 cannot be phosphorylated, we predicted that in apkc mutant NBs at the onset of mitosis, Mira should not be cleared from the PM. To test this, we expressed fluorescently tagged Mira in apkck06403 mutant brains; a loss of function condition for aPKC (Wodarz et al., 2000). As in controls, Mira localized uniformly to the cortex of interphase apkck06403 NBs, but did not clear from the cortex during prophase and remained uniformly localized throughout mitosis (Video 12, Figure 4A).

Figure 4 with 1 supplement see all
Lateral diffusion and cytoplasmic exchange of cortical Miranda are different in control and aPKC impaired mitotic NBs.

(A) Stills from Video 12 of an apkck06403 mutant NB (MARCM clone labeled with nlsGFP, green) expressing Mira::mCherry (grey). Mira is cortical in interphase, as the NB enters mitosis, and after NEB (arrowheads, t1 – t4). (B) Conditions analyzed by FRAP. (C) Fluorescence redistribution curves of cortical Mira::mCherry at the indicated conditions. (C’) Estimates of t1/2 [sec.] for cortical Mira::mCherry under the indicated conditions derived from curve fitting (Rapsomaniki et al., 2012). (D) Photo-conversion experiment monitoring loss of myr-EOS converted signal over time. (D’) Estimates of t1/2 [sec.] for cortical Mira::mCherry under the indicated conditions from curve fitting. Overexpression was driven by worniu-Gal4. p values: two-tailed t test for independent means. Scale bar: 10 µm.

https://doi.org/10.7554/eLife.29939.021
Video 12
Miranda remains at the cortex throughout the cell cycle in apkck04603 mutant NBs.

Mutant NB, labeled with nlsGFP (green) expressing Mira::mCherry (white). Z-stacks taken every minute. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.023

From these observations, we made two predictions: First, the abnormal cortical localization of Mira in metaphase in the absence of aPKC should occur independently of F-actin, similar to normal cortical Mira in control interphase (Figure 2A). Second, if Mira binding to the cortex is differently controlled in interphase and mitosis, the turnover of Mira at the cortex when bound to the PM in interphase and when interacting with F-actin after NEB should be different. Turnover can be measured by fluorescence recovery after photo-bleaching (FRAP). Mira recovery should be different in interphase and mitosis and controls, but similar between control interphase NBs and mitotic apkc mutant NBs, as we suspect Mira to bind the PM in these mutants throughout the cell cycle.

Indeed, Mira localization to the cortex in mitosis was insensitive to LatA treatment in NBs depleted for aPKC (Figure 4—figure supplement 1). Consistent with the second prediction, FRAP measurements revealed that Mira recovery was significantly different in interphase and mitosis (Figure 4B–C’). However, while Mira recovery in mitosis was faster when aPKC was knocked down by RNAi (or aPKC inhibition by Lgl3A overexpression, Betschinger et al., 2003), this did not result in Mira recovery in mitosis becoming as fast as in interphase (Figure 4C,C’). It is known that changes in the actin network caused by progression through the cell cycle (Ramanathan et al., 2015) can influence dynamics of membrane-associated proteins in general (Heinemann et al., 2013). This is certainly the case in NBs, as a photo-convertible membrane-associated reporter that attaches to the entire NB PM via a myristoylation signal (myr-Eos) showed slower dynamics in mitosis compared to interphase (~four fold, Figure 4D,D’). Thus, these general cell cycle-driven changes in the actin cytoskeleton could account for the difference between Mira recovery in interphase and in mitosis in aPKC-impaired NBs.

To test the effect of such changes in the actin cytoskeleton on mobility of PM interacting proteins, we treated mitotic NBs with LatA, which resulted, in myr-Eos dynamics falling within a range similar to that observed for interphase cells (Figure 4D,D’). Importantly, myr-Eos dynamics did not change in response to Lgl3A overexpression, arguing against Lgl3A overexpression causing changes in the actin cytoskeleton to explain the resulting accelerated redistribution of Mira in mitosis. Finally, in mitotic NBs overexpressing Lgl3A, LatA treatment resulted in Mira recovery becoming as fast as in interphase (Figure 4C,C’).

In conclusion, these results show that, in unperturbed NBs, Mira turnover at the PM in interphase and at the basal cortex in mitosis are different, supporting the notion that Mira has different binding modes in interphase versus mitosis. Furthermore, in apkc mutant NBs, instead of being cleared, Mira may persist throughout mitosis with the same actin-insensitive uniform localization, and similar turnover, as in interphase.

High doses of Y-27632 inhibit aPKC and partially disrupt maintenance of Mira asymmetry after NEB

We assessed the relative contributions of actomyosin and aPKC to Mira localization throughout the cell cycle, and show that at the onset of mitosis aPKC displaces Mira from the PM. After NEB, however, Mira localization becomes actomyosin-dependent. We next attempted to address whether aPKC regulates Mira localization after NEB.

To dissect a role for aPKC during the cell cycle, temporal control over its activity is required. This can be achieved with temperature sensitive (ts) alleles or small molecule inhibition. We found that the available ts allele of apkc (Guilgur et al., 2012) is already hypomorphic at permissive temperatures resulting in Mira localization defects (not shown). Therefore, we made use of the non-specific effects of the ROCK inhibitor Y-27632, which inhibits aPKC with an IC50 of ~10 µM (Atwood and Prehoda, 2009).

We determined the concentration at which Y-27632 treatment phenocopied the apkc phenotype, and resulted in LatA insensitive uniform cortical Mira after NEB. This was the case when at least 200 µM Y-27632 was added to cycling NBs (Video 13, n = 25, Video 14, n = 12). This resulted in only partial loss of Mira asymmetry: Mira signal became detectable faintly at the apical pole 52 ± 11 min (n = 15) following this treatment, but Mira remained basally enriched (Figure 5A). This basal enrichment was only lost upon the addition of LatA (Figure 5A, Video 15, n = 15). Therefore, as reported previously (Barros et al., 2003), high doses of Y-27632 lead to Mira localization defects that are likely to reflect aPKC inhibition (Atwood and Prehoda, 2009). Intriguingly, there seems to be a difference in sensitivity to Y-27632 when added before or after NEB: Addition of high doses of Y-27632 added to cycling cells results in uniform cortical Mira whereas (Video 13), when the same dose is added to metaphase-arrested NBs, Mira appears with a delay (~52 min) and only faintly at the apical pole, but retains a LatA-sensitive basal bias (Figure 5A). These results suggest that in addition to the role of aPKC in clearing uniform PM bound Mira at the onset of mitosis, aPKC may contribute to Mira asymmetry after NEB, possibly by clearing it from the apical cortex. However, given the likelihood of several targets for Y-27632, a precise role for aPKC after NEB cannot be determined in this way.

Figure 5 with 1 supplement see all
Mira crescent size is affected by a Y-27632-sensitive mechanism that operates before NEB.

(A) Stills from Video 15. Colcemid arrested NBs were treated with 200 µM Y-27632. After >50 min Mira becomes faintly detectable apically, but retains a basal bias. LatA addition (5 µM) abolishes that asymmetric bias and Mira is uniformly distributed on the membrane. (B) Culturing colcemid-arrested NBs in 50 µM Y-27632 did not alter Mira crescent size (yellow arrowheads, quantified in E). (C) NBs polarizing in the presence of 25 µM Y-27632 show enlarged Mira crescents. Control division (−62’ to −35’) with normally sized Mira crescent and daughter cell size (−60’; yellow arrowheads, bracket, respectively). Dividing in the presence of Y-27632 (−3, NEB +1) leads to an enlarged Mira crescent (NEB +1, white arrowheads) and enlarged daughter cell size (+24’, brackets, 2). (D) NBs were allowed to polarize in the absence (upper row) or presence of 25 µM Y-27632 (middle and lower row) followed by colcemid arrest. upper row: Control NB with normal Mira crescent (yellow arrowheads) was depolarized by 1 µM LatA. Mira was displaced into the cytoplasm. middle row: adding 1 µM LatA leads to displacement of the enlarged Mira crescent (yellow arrowheads) in the cytoplasm. Lower row: adding 20 µM ML-7 drives Mira into the cytoplasm (+8’). Upon ML-7 washout, Mira recovered to an enlarged crescent (+14’, white arrowheads). (E) Quantification of Mira crescent size in the aforementioned experiments (unpaired t test). (F) Plot of the ratio of daughter cell to NB nuclei as a measure of the effect of Y-27632 on daughter cell size. NBs expressing NLSGFP were imaged by DIC to follow daughter cell birth order during three consecutive divisions [(1) pre-treatment; (2) division in the presence of 25 µM Y-27632; (3) division after drug washout]. A high-resolution z-stack of nlsGFP was recorded, and the nuclear volumes rendered and calculated using IMARIS to plot their ratio. p values: Dunn’s test. Time stamp: min. Labels as indicated. BAC{mira::mcherry-MS2} was the source of Mira::mCherry. Scale bar: 10 µm.

https://doi.org/10.7554/eLife.29939.024
Video 13
Miranda remains at the entire cell cortex throughout the cell cycle in NBs treated with 200 µM Y-27632.

A Baz::GFP (Green) and Miranda::mCherry (Red) NB was imaged through one cell cycle in the presence of 200 µM Y-27632. Miranda remained cortical throughout while Bazooka still localised apically in mitosis (n = 12). Z stacks taken every 2 min. Time stamp: hh:mm. Scale: 10 µM.

https://doi.org/10.7554/eLife.29939.026
Video 14
200 µM Y-27632 induced uniform cortical Mira in mitosis localizes independently of an intact actin network.

A Baz::GFP and Mira::mCherry expressing NB was cultured in the presence of 200 µM Y-27632 and then arrested with colcemid. 5 µM LatA was added after the first frame of the movie. LatA induces loss of Baz asymmetry, yet Mira remains cortical. Z-stacks shown. Z stacks taken every 2 min. Time stamp: hh:mm.

https://doi.org/10.7554/eLife.29939.027
Video 15
Colcemid arrested NB expressing Baz::GFP and Mira::mCherry, treated with 200 µM Y-27632.

Mira starts to become visible ~36 min after Y-27632 addition in this example, but remains asymmetrically distributed, until LatA is added. Z-stacks shown. Z stacks taken every 2 min. Time stamp: mm:ss.

https://doi.org/10.7554/eLife.29939.028

Low doses of Y-27632 can affect Mira crescent size independently of aPKC inhibition

In the course of determining the lowest concentration of Y-27632 to phenocopy aPKC loss of function, we observed that addition of 50 µM of Y-27632 to metaphase-arrested NBs, did not produce any significant changes in Mira or aPKC crescent size (n = 22, Figure 5B, and Figure 5—figure supplement 1). However, addition of just 25 µM Y-27632 to cycling NBs, produced aPKC crescent size comparable to controls (Figure 5—figure supplement 1) but significantly enlarged Mira crescents (Figure 5C).

Furthermore, the enlarged Mira crescents resulting from Y-27632 addition to cycling NBs were sensitive to LatA and ML-7 treatment, as were normal Mira crescents in controls. This suggests that also under this condition, actomyosin is important to retain Mira at the cortex mitosis, even when the size of the crescents is enlarged (Figure 5D; see Figure 5E for Mira crescent size quantification under the different conditions). Finally, NBs that divided in the presence of 25 µM Y-27632 produced larger daughter cells than controls (n = 12, Figure 5C,F). Therefore, enlarged Mira crescents induced by Y-27632 are correlated with an increase of NB daughter cell size.

In conclusion, while Y-27632 at higher concentration, can indeed mimic the effect of apkc mutation on Mira, these results suggest that when NBs polarize in the presence of low concentrations of Y-27632, Mira crescent size is affected, which is likely to occur independently of aPKC inhibition. These results suggest that Mira cortical retention has different mechanisms of regulation in interphase and in mitosis. They also hint at an additional, Y-27632 sensitive layer of regulation controlling basal Mira crescent size.

Discussion

In this study, we addressed the localization of the adapter protein Mira throughout the cell cycle of Drosophila larval NBs to shed light on how polarized fate determinant localization is achieved. It was previously demonstrated that actomyosin is essential for Mira polarization in mitosis (Barros et al., 2003; Petritsch et al., 2003; Shen et al., 1998). More recent work showed that Mira can directly bind to the PM via its BH motif. As a result, spatially controlled phosphorylation of this BH motif by aPKC, leading to displacement of Mira from the cortex where aPKC is active, can in principle explain Mira asymmetry without the need to evoke a role for actomyosin (Atwood and Prehoda, 2009; Bailey and Prehoda, 2015).

To address this apparent inconsistency, we reassessed in vivo the relative contribution of aPKC and actomyosin throughout the cell cycle analysing Mira localization using endogenously expressed reporters in living NBs. This has allowed us to resolve this problem as we find that asymmetric Mira localization is established stepwise and involves both aPKC-dependent phosphorylation and actomyosin-dependent anchoring, which are required at different time points in mitosis.

We propose that Mira has two different modes by which it can be retained at the cortex (Figure 6). In interphase, Mira localizes uniformly to the cortex via direct interactions with the PM for which its BH motif is necessary and likely to be sufficient and which occurs independently of an intact F-actin cortex (Figure 2A). After NEB, Mira still relies on the BH motif to localize in a basal crescent, but at this stage of the cell cycle it might be required to mediate actomyosin-dependent basal retention of Mira (Figure 3A, Figure 2A–D). The transition between these localizations depends on phosphorylation by aPKC (Figure 3A, Figure 4A).

Model.

Mira associates with the cortex using two different modes, which characteristics are detailed in the bottom row. During interphase, Mira directly binds to the phospholipids of the PM via its BH motif (black double arrow). During prophase, aPKC-dependent phosphorylation of this motif abolishes this interaction, resulting in the progressive clearance of Mira from the cortex, in an apical-to-basal manner driven Mira into the cytoplasm. This clearance in prophase is necessary for Mira to associate with the basal cortex after NEB, via Actomyosin-dependent retention. Both the precise phosphoregulation and molecular characteristics of this mode remain to be determined. The BH motif, also required at this step, may directly or indirectly mediate interactions between Mira and actomyosin (green double arrow). PM interactions via its BH motif (black double arrow) may still contribute, but are not sufficient to mediate Mira basal retention after NEB.

https://doi.org/10.7554/eLife.29939.029

We observe that deletion of the BH motif as well as overexpression of aPKCΔN disrupt cortical localization of Mira in interphase and mitosis (Figure 3A and Figure 3—figure supplement 1) and that the phosphomimetic S96D mutation reduces Mira localization in interphase as well as in mitosis (Figure 3A). These findings by themselves argue for the model that throughout the cell cycle Mira cortical association depends solely on BH motif-mediated interaction with the PM, that is negatively regulated by locally controlled aPKC phosphorylation (Atwood and Prehoda, 2009; Bailey and Prehoda, 2015).

What could be the role of F-Actin for Mira localization in this model? F-Actin clearly contributes to aPKC regulation of Miranda localization by restricting the localization of the Par complex to the apical pole as LatA addition changes the distribution of aPKC and Baz (Figure 2A). However, at least with the resolution with which we can observe live NBs, basal Miranda relocates into the cytoplasm in LatA treated metaphase arrested NBs before changes in the localization of the Par complex are induced (Figure 2—figure supplement 1). Furthermore, inhibiting Myosin activity in metaphase arrested NBs with ML-7 also caused the relocation of Mira into the cytoplasm. However, the localization of the Par-complex was unchanged (Figure 2B,C). This argues that actomyosin plays an additional anchoring function that contributes to retain Mira basally after NEB.

The intensity (Figure 1B,B’) and turnover (Figure 4C,C’) of Mira differ in interphase and mitosis. If BH motif mediated retention at the PM sufficed to mediate Mira cortical binding throughout the cell cycle, these observations might be explained by changes in Mira properties such as dimerization. Mira can form homo-dimers (Yousef et al., 2008) and targeting of proteins with phospholipid-binding domains to membranes often requires their dimerization (Lemmon, 2008). However, a point mutation preventing dimerization of Mira’s cargo binding domain (required to bind Pros) does not prevent Mira’s asymmetric localization in mitosis, while Pros localization is lost (Jia et al., 2015). We found that Pros localizes to the PM in a Mira-dependent manner in interphase (Figure 1—figure supplement 1), suggesting that Mira is already a dimer at this stage. Thus, changes in dimerization are unlikely to explain the differences in Mira turnover detectable by FRAP.

Differences in turnover may be explained by different modes of cortical association of Mira in mitosis and interphase. We propose that in mitosis, additional stabilizing interactions might retain Mira basally, which are not operating in interphase. Those stabilizing interactions require the actomyosin network, but rather than being pushed towards the basal pole by Myosin (Barros et al., 2003), actomyosin may provide a Mira anchoring function (Figure 6). Mira might directly bind to F-actin as shown in vitro (Sousa-Nunes et al., 2009) or be anchored basally by myosin activity. Alternatively, additional processes could be involved such as Mira’s interaction with mira mRNA (Ramat et al., 2017), potentially providing an anchoring scaffold to maintain Mira basally. The phosphomimetic S96D mutation strongly disrupts uniform localization to the PM in interphase, but localizes at the basal cortex in mitosis, albeit at reduced levels (Figure 3A). This is consistent with the existence of Mira stabilizing interactions in mitosis, that are not present in interphase, which reduce the effect of a negative charge provided by the Aspartate in the phosphomimetic mutant on cortical Mira localization.

A surprising finding is that the BH motif is essential for Mira localization in interphase and in mitosis. In mitosis, BH motif mediated PM binding is no longer sufficient to localize Miranda to the basal pole. This is indicated by the requirement of the actomyosin cytoskeleton after NEB (Figure 2C,D). However, the BH motif is still necessary for Mira localization after NEB. It is possible that the BH-phospholipid interactions still play a role in mitosis. Deletion of the BH motif could also cause more indirect effects. For example, Mira∆BH is not found on mitotic microtubules, where Mira is typically observed in conditions where it is unable to localize correctly (Albertson and Doe, 2003; Barros et al., 2003; Rolls et al., 2003; Slack et al., 2007). We propose that the BH motif may mediate Mira’s interaction with actomyosin, which remains to be tested.

Interfering with aPKC-dependent Mira displacement from the cortex during prophase, either by apkc mutation or by directly preventing phosphorylation of the BH motif (S96A) results in the persistence of uniform, PM bound Mira in metaphase; indicated by its abnormally fast dynamics and its actomyosin independence (Figure 4C,C’ and Figure 4—figure supplement 1). This aPKC-dependent step in prophase might be a prerequisite for basal crescent formation in metaphase. One possibility is that phosphorylation of the BH motif might potentiate Mira’s ability to engage with actomyosin for basal retention after NEB. Mira phosphorylation might need to be properly balanced and locally controlled to allow for Mira asymmetric localization. This could explain why the phosphomimetic S96D mutant, displays reduced basal localization in metaphase and that upon overexpression of aPKCΔN Mira does not form basal crescents in metaphase (Figure 3—figure supplement 1). Another unexpected observation is that despite never localizing to the PM in interphase or to the cortex in metaphase, Mira localization is rescued at telophase in aPKCΔN overexpressing NBs, which does not occur when the BH motif is deleted (Video 11). This suggests that Mira retains at least in telophase the ability to engage with the cortex when aPKCΔN is expressed and that the BH motif might be important for the telophase rescue phenomenon (Peng et al., 2000).

In addition to its role in displacing Mira into the cytoplasm during prophase, aPKC might contribute to Mira localization after NEB. High doses of Y-27632 when added to colcemid arrested NBs lead to apical and lateral accumulation of Mira at the cortex (Figure 5A), suggesting that also after NEB aPKC contributes to keep Mira off the apical membrane. However, at 200 µM Y-27632 is likely to inhibit multiple processes. Therefore, the precise contribution of aPKC at different time points during the cell cycle remains to be determined.

Compartmentalized aPKC activity provides an explanation for Mira crescent size in a model in which, throughout the cell cycle, Mira solely relies on BH mediated PM interactions to localize to regions of the cortex where aPKC is inactive. This would hold true in a model where Mira uses one mode to interact with PM in interphase and another to bind to actomyosin in mitosis, if phosphorylation of the BH motif was required for basal Mira retention by actomyosin after NEB. An interesting possibility is that spatial information for Mira crescent size is provided by the actomyosin network itself. Low doses of Y-27632 yield enlarged basal Mira crescents that are correlated with an increase in daughter cell size (Figure 5C,E,F), the control of which involves actomyosin regulation (Roubinet and Cabernard, 2014). Therefore, spatial information for determinant localization in NBs could be coupled to the machinery that regulates daughter cell size. Indeed, ROCK has recently been shown to accumulate at the apical pole in prophase generating an asymmetry in the actomyosin network (Tsankova et al., 2017).

The IC50 of Y-27632 for ROCK is about an order of magnitude lower than the IC50 determined for aPKC (Atwood and Prehoda, 2009; Uehata et al., 1997). Therefore, enlarged Mira crescents we observe for 25 µM Y-27632 on Mira in cycling NBs might stem from effects on ROCK. These may trigger changes in actomyosin configuration and/or cortical tensions (Matthews et al., 2006; Tsankova et al., 2017). It is thus possible that local tension anisotropies in the actomyosin network provide spatial information for Mira localization. ROCK and MLCK both affect myosin activity (Amano et al., 1996; Saitoh et al., 1987; Ueda et al., 2002; Watanabe et al., 2007) yet the effects of MLCK (ML-7) and ROCK (Y-27632) inhibition on Mira differ (Figure 2 versus Figure 5). In MCDK II cells, Y-27632 and ML-7 treatment had different effects on myosin regulatory light chain phosphorylation (Watanabe et al., 2007). This could result in different effects on myosin activity that may explain the different effects on Mira also in NBs.

What could be the advantage of relying on PM binding in interphase and actomyosin retention in mitosis? Nuclear levels of Mira’s cargo Pros in NBs affect quiescence and differentiation (Choksi et al., 2006; Lai and Doe, 2014). It is possible that PM-bound Mira sequesters Pros at the interphase NB PM. Two modes of cortical retention might allow regulation of nuclear Pros levels in interphase NBs and ensure segregation of elevated determinant levels to daughter cells in mitosis to achieve correct thresholds of cell fate information in the differentiating daughter cells.

Materials and methods

Fly stocks and genetics

Flies were reared on standard cornmeal food at 25°C. Lines used were:

(1) Baz::GFP trap (Buszczak et al., 2007); (2) w1118 (Bloomington); (3) MARCM: hsFlp tubGal4 UASnlsGFP; FRT42B tubGal80/Cyo and FRT82B gal80 (Lee and Luo, 1999); (4) worniu-Gal4 (Albertson et al., 2004); (5) UAS-Lgl3A (Betschinger et al., 2003); (6) w1118, y,w, hsp70-flp; tubP-FRT >cd2>FRT-Gal4, UAS-GFP (Gift from M. Gho); (7) Mz1061 (Ito et al., 1995); (8) UAS-GFP::Mira (Mollinari et al., 2002). (9) FRT82B miraL44 (Matsuzaki et al., 1998). (10) Df(3R)oraI9 (Shen et al., 1997). (11) UAS-aPKCRNAi: P{y[+t7.7] v[+t1.8]=TRiP.HMS01320}attP2 (BL#34332); (12) Numb::GFP (Couturier et al., 2013); (13) FRT42B apkck06403 (Wodarz et al., 2000); (14) UAS-aPKCΔN; (15) P{UASp-sqh.E20E21}3 (BL#64411); (16) P{10xUAS-IVS-myr::tdEos]attP2 (BL #32226); y[1] w[*]; P{y[+t*] w[+mC]=UAS-Lifeact-Ruby}VIE-19A (BL# 35545); (17) aPKC::GFP (Besson et al., 2015); Source 1 of Mira::mCherry: BAC{mira::mcherry-MS2} (Ramat et al., 2017). aPKCRNAi clones were generated by heat shocking larvae of the genotype y,w, hsp70-flp; tub-FRT >cd2>FRT-Gal4, UAS-GFP; P{y[+t7.7] v[+t1.8]=TRiP.HMS01320}attP2. Heat shocks were performed 24hph and 48hph for 1 hr at 37°C. MARCM clones were generated by heat shocking L1 larvae for 2 hr at 37°C.

Generation of Mira alleles: Source 2 of Mira::mCherry: miramCherry; miraΔBHmCherry (Ramat et al., 2017), miraS96AmCherry and miraS96DmCherry are derived from miraKO (Ramat et al., 2017). miramCherry was generated by inserting a modified wt genomic locus in which mCherry was fused to the C-terminus following a GSAGS linker into miraKO. For miraS96DmCherry: TCG (Serine96) was changed to GAC (aspartic acid). For miraS96AmCherry: TCG was replaced with GCG (alanine). CH322-11-P04 was the source for the mira sequences cloned using Gibson assembly into the RIV white vector (Baena-Lopez et al., 2013) that was injected using the attP site in miraKO as landing site. BAC{mira::mcherry-MS2} (Ramat et al., 2017) see Figure 1—figure supplement 2). While miramCherry behaves similarly to BAC{mira::mcherry-MS2} and rescues embryonic lethality, miraΔBHmCherry, miraS96AmCherry and miraS96DmCherryare homozygous lethal.

Live imaging: Live imaging was performed as described (Pampalona et al., 2015). Briefly, brains were dissected in collagenase buffer and incubated in collagenase for 20 min. Brains were transferred to a drop of fibrinogen (0.2mgml-1, Sigma f-3879) dissolved in Schneider’s medium (SLS-04-351Q) on a 25 mm Glass bottom dish (WPI). Brains were manually dissociated with needles before the fibrinogen was clotted by addition of thrombin (100Uml−1, Sigma T7513). Schneider’s medium supplemented with FCS, Fly serum and insulin was then added. A 3–4 µm slice at the center of the neuroblasts was then imaged every 30–90 s using a 100x OIL objective NA1.45 on a spinning disk confocal microscope. Data were processed (3D Gaussian blur 0.8/0.8/0.8 pixels) and analyzed using FIJI (Schindelin et al., 2012). Nuclear volume was measured using Imaris software. All other drugs were added to the media either prior to or during imaging: ML-7 (Sigma, I2764, dissolved in water), Y-27632 (Abcam, Ab120129, dissolved in water). Drugs were washed out by media replacement; in the polarity reconstitution assay colcemid concentrations were kept constant throughout the experiments. FRAP experiments were carried out on a Leica SP8 confocal using a 63x NA1.2 APO water immersion objective. To estimate t1/2 for the recovery curves, we used published curve fitting methods (Rapsomaniki et al., 2012).

Immunohistochemistry

Primary larval brain neuroblast cell culture

Brains were dissected in collagenase buffer and incubated for 20 min in collagenase, as for live imaging. Brains were then transferred into supplemented Schneider’s medium and manually dissociated by pipetting. Cells were pipetted onto a poly-lysine-coated 25 mm glass-bottomed dish and left to adhere for 40 min. Schneider’s was then replaced with 4% formaldehyde (Sigma) in PBS and cells were fixed for 10 min. Cells were permeabilized with 0.1% PBS–Triton for 10 min. Cells were then washed with PBS 3 × 10 min before antibody staining overnight at 4°C. All antibodies were dissolved in PBS–1%Tween. Whole mount brains: Brains were fixed in 4% formaldehyde (Sigma) for 20 min at room temperature. Primary antibodies: Rabbit anti-Miranda (1:200, gift from C. Gonzalez); Mouse anti-GFP (1:400, Abcam); Rabbit anti-Brat (1:200, a gift from J. Knoblich); Guinea pig anti-Dpn (1:500 a gift from J. Skeath); Mouse anti-Pros (1:40, DSHB). To stain F-actin we used Alexa Fluor 488 or 561 coupled Phalloidin (Molecular Probes, 5:200) for 20 min at room temperature. Secondary antibodies were from Life Technologies and raised in donkey: Anti-rabbit Alexa-594; Anti-mouse Alexa-488; Anti-rabbit Alexa-647; Anti-guinea pig Alexa-647. Microscopy was performed using a Leica-SP8 CLSM (60x Water objective, 1.2) and images were processed using FIJI.

In all cases the sample size n provided reflects all samples collected for one experimental condition. All experimental were repeated at least twice.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
    miranda localizes staufen and prospero asymmetrically in mitotic neuroblasts and epithelial cells in early Drosophila embryogenesis
    1. F Matsuzaki
    2. T Ohshiro
    3. H Ikeshima-Kataoka
    4. H Izumi
    (1998)
    Development 125:4089–4098.
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
    Selective inhibition of catalytic activity of smooth muscle myosin light chain kinase
    1. M Saitoh
    2. T Ishikawa
    3. S Matsushima
    4. M Naka
    5. H Hidaka
    (1987)
    The Journal of Biological Chemistry 262:7796–7801.
  54. 54
  55. 55
  56. 56
  57. 57
  58. 58
  59. 59
  60. 60
    The prospero transcription factor is asymmetrically localized to the cell cortex during neuroblast mitosis in Drosophila
    1. EP Spana
    2. CQ Doe
    (1995)
    Development 121:3187–3195.
  61. 61
  62. 62
  63. 63
  64. 64
  65. 65
  66. 66
  67. 67
  68. 68
  69. 69
  70. 70

Decision letter

  1. Yukiko M Yamashita
    Reviewing Editor; University of Michigan, Ann Arbor, HHMI, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Switching the cortical binding mode is required for asymmetric fate determinant localization in Drosophila neuroblasts" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom, Yukiko M Yamashita (Reviewer #1), is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Shigeki Yoshiura (Reviewer #2).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

Whereas all reviewers agreed that the idea of Miranda having two distinct modes of localization is interesting and of potential impact, they also felt that your data do not prove your hypothesis: whereas your data indicate that the existing model of Miranda localization may require a revision, your alternative hypothesis does not provide a better explanation than the current model, leaving unexplained observations as much as the current model does. The individual comments can be found at the bottom of this letter.

The reviewers felt that it is critically important for you to carefully assess the current model and all the data that support it. At least the revised model must be better than the current one in explaining all the observations/data in the field. For example, a key result supporting the current model is that the unphosphorylatable Miranda (Mira5A) is depolarized in mitosis. This would be inconsistent with your revised model. In reaching a new model, it is not sufficient to point out that the current model is imperfect, but it is important that there is a significant advancement in the understanding, instead of muddying the field.

Reviewer #1:

The major conclusion of this paper is that Miranda, a fate determinant that is segregated to the differentiating daughter during asymmetric neuroblast division in Drosophila, has two binding modes to the plasma membrane. First, during interphase, Mira localizes to the cortex (evenly). In G2/M, it localizes to the basal cortex in an actin dependent manner. As opposed to the previous model, in which aPKC was suggested to promote basal cortex localization by apical exclusion of Mira through phosphorylation, the present study shows that aPKC is required to clear Mira from cortex prior to its localization to the basal cortex. (i.e. Mira first disappear from the entire cortex, and reappear as a basal crescent in mitosis)

This work revises the current model of Miranda localization in NBs, and shows that Miranda cortical localization is regulated by two distinct mechanisms (even, cortical localization in interphase and basal crescent in mitosis). The authors' main message seems to be that these two mechanisms are independent of each other. The experimental data presented in this manuscript seem to leave rooms for alternative interpretations. Although this paper is right in that revision of the old model is required, this paper does not necessarily present 'better' revised model, as it still contains a few loose ends. Some of the inconsistencies I found in this paper are the following:

1) If Mira localization is regulated by two independent mechanisms, the significance of interphase cortical localization becomes unclear. The reason why Mira localization is important is because it functions as a fate determinant for NB daughter (GMC), thus only mitotic localization should matter.

2) However, the Figure 4D data (Mira-deltaBH cannot go to mitotic basal cortex, whereas Mira-SD goes to basal crescent, despite not being able to localize to the interphase cortex) seems to suggest an interdependence between interphase and mitotic localization. For example, Mira has to localize to the cortex in interphase so that it can be phosphorylated by aPKC, then it dissociates from the cortex but potentiated for basal cortex localization.

Overall, the data is of high quality, and clearly suggest the need of revision to the current model, but the study does not put forward an alternative, revised model that is more convincing than the old one. I do agree that the revision is needed to the current model, but all the data do not fit yet into a simpler model (i.e. something is still missing). The alternative model that they suggest to replace the existing model still leaves unexplained observations. As I am not exactly in this field, I cannot suggest what would be the way to go. I feel the work is important, but this work needs a bit more to be able to say that they have 'better' understanding of the process.

Reviewer #2:

This manuscript by Dr. Januschke and colleagues examined the mechanisms underlying the asymmetric localization of Miranda in the Drosophila larval neuroblasts. They proposed that there would be two different mechanisms that regulate the Miranda cortical localization; direct binding to the plasma membrane in interphase and actin-myosin dependent mechanism in mitotic phase. The concept that Miranda utilizes two different mechanisms to localize at the cell cortex in different cell cycle phases is new and interesting, however, there are some points to be clarified before published.

Major comments

1) In the most of the figures, authors utilized Lgl3A overexpression to abolish the function of aPKC, however, as shown in Figure 2G, the majority of the Lgl3A expressing neuroblasts achieved asymmetric localization of Miranda in mitosis while in the aPKC mutant clones Miranda localized at the entire cortex (Figure 4), suggesting that Lgl3A overexpression does not recapitulate aPKC loss of function. Although Lgl3A is thought to suppress aPKC function, it seems that, at least in these cases, Lgl3A expression is not enough to suppress aPKC function or it has some other functions than aPKC suppression, making it difficult to interpret Lgl3A overexpression phenotypes. Thus, it is crucial to check what is observed in Lgl3A overexpression is also the case in aPKC mutant or RNAi clones.

2) In Figure 4E, the authors showed that Mira[S96D] showed uniform cortical localization in colcemid and LatA treated neuroblast. Following the authors' model, Miranda localizes at the cortex by either direct binding to the plasma membrane or actin-dependent mechanism, however, in this situation Mira seems to localize at cell cortex in the absence of both machinery. Then how does Miranda localize at cell cortex in this situation? Authors should discuss about this point.

3) Authors utilized Y-27632 to temporally inhibit aPKC function but it is not clear whether Y-27632 treatment is really affecting the aPKC activity or not. As a control experiment, it is better to examine whether the addition of 50uM Y-27632 to the cycling neuroblasts results in the Lat-A-insensitive entire cortical localization of Miranda in mitotic phase (like aPKC loss of function situation).

Reviewer #3:

This revision, like the original, proposes that Miranda is polarized in Drosophila neuroblasts via two distinct mechanisms, one in interphase and one in mitosis – the former being by a previously described "BH" association mechanism, and the latter being a "myosin affinity zone". I had two main criticisms of this proposed mechanism. First, the authors did not separate these two functions, rather deletion of the BH caused loss of Miranda polarity in both interphase and mitosis. Second, the proposal of a "myosin affinity zone" is based on data using inhibitors leaving open the possibility that the effect is indirect. The author's work is potentially exciting because there are interesting questions related to Miranda's function in interphase, and the role of myosin in polarizing Miranda. Unfortunately, however, I do not believe that the revised manuscript comes close to adequately addressing the criticisms raised in the original review and therefore I do not recommend publication.

• A separation of function allele would be the most definitive test of the authors model – if two different mechanisms are used, then it should be possible to find Miranda alleles that only polarize in mitosis and not in interphase (unless the interphase mechanism is required for the mitotic one, which is actually what the title says – see below). Mira∆BH isn't polarized in interphase or mitosis indicating that the authors model is wrong or the BH motif is required for each binding mode. To distinguish between these possibilities, the authors examined MiraS96D, a phosphomimetic allele, and conclude that it is a bona fide separation of function allele, disrupting interphase localization but not mitotic. However, S96D is clearly a hypomorph (Figure 2Q,T Current Biology 19, 723-729, 2009) and my interpretation of the author's data (i.e. Video 9) is consistent with it behaving as a hypomorph in their assay – Mira96D basal crescent signal is reduced and spindle signal is increased not only in interphase but also in mitosis.

• I also criticized the original version of the manuscript because it proposed that mitotic Miranda polarization occurs through a "myosin anchor" in a "basal affinity zone" because the data supporting this relies on drug treatments. Nothing in the revised manuscript addresses this criticism. The author's rebuttal letter states, "We agree that the mode of action of myosins remains perhaps obscure since we do not know whether the effect is direct or indirect." That is precisely my criticism, that the mode of action of myosins remains obscure. The letter goes on to state, "However, given that the ML-7 effect on Mira basal localization in colcemid arrested NBs can be tempered by overexpressing a phospho-mimetic version of Sqh…, an involvement of myosin activity in anchoring Mira at the basal pole at NEB seems very likely." This statement is not very impactful if the involvement of myosin activity is very indirect, which the authors themselves acknowledge is entirely possible. The letter further states, "We hope that the reviewer would agree that the current apical exclusion model does not leave much room to explain the role of myosin activity in Mira localization…". I strongly disagree with this statement. First, pretty much any model would allow for an indirect role for the cytoskeleton. But more importantly, the authors do not seem to understand the "current apical exclusion model", at least as it is articulated in Bailey and Prehoda, Dev Cell, 2015, which they have referenced. In this model, BH motifs cooperate with accessory interactions to mediate cortical recruitment of Par substrates (e.g. see Discussion section, "Multivalent Interactions Mediate Par Substrate Cortical Localization"), and that aPKC phosphorylation of the BH motif disrupts this interaction. In this model an "accessory interaction" could certainly require interactions with myosin – the key point being that any such interaction must not be sufficient for cortical targeting (otherwise BH phosphorylation wouldn't have an effect on cortical localization). So, if the authors identified a direct interaction between Miranda and myosin that was required for cortical targeting, that would in an of itself not be inconsistent with the "current apical exclusion model", although it would certainly be interesting. All of this leads me to believe that the authors are attempting to frame the impact of their work in the context of refuting a current model of polarity which, in fact, they do not understand.

• The title of the paper is incorrect, according to the authors. The title states, "Switching the cortical binding mode is required for asymmetric fate determinant localization in Drosophila neuroblasts". However, the authors claim to have a Miranda allele that doesn't switch cortical binding modes and is still asymmetrically localized, directly refuting the statement made in the title. Stating one thing in the title and another in the text will confuse readers and make it very difficult for those in and outside the field to understand the impact of the paper. Furthermore, it suggests that the authors themselves do not understand the impact of their work.

• Along those lines, like the original manuscript, the revised version is difficult to follow, especially for those outside the field. For example, the current Abstract states, "Here we test the current model of how the Par-complex component aPKC directs the localization of Miranda". This places a large burden on readers, especially those outside the field, to have a familiarity with current polarity models. In my opinion, it is ultimately up to the authors to decide if they want to make their work accessible, but the journal may feel differently.

[Editors’ note: a second version of this study was rejected after peer review, but the authors submitted for reconsideration. The decision letter after peer review is shown below.]

Thank you for submitting your work entitled "aPKC-mediated displacement and actomyosin-mediated retention polarize Miranda in Drosophila neuroblasts" for consideration by eLife. Your article has been reviewed by four peer reviewers, one of whom, Yukiko M Yamashita (Reviewer #1), is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Shigeki Yoshiura (Reviewer #3).

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

As you remember, in the previous round of your submission, all reviewers agreed that your work has a good potential but poor presentation/writing, over-interpretation and some questionable interpretation precludes publication at that point. We hoped that you take our advice and revise the manuscript drastically. In this round, we have invited the new reviewer per your request, who provided positive comments, yet also felt that the manuscript requires considerable revision to be acceptable. Later in the discussion among reviewers, this reviewer commented " I agree that the poor presentation is major problem, and that they do not convincingly prove the existence of a basal affinity zone.", favoring "giving the authors the chance to resubmit a radically revised version."

We all remain agreed that your discovery that Miranda localization has two modes is interesting and important. However, poor presentation of manuscript (writing makes it unclear what is the most important discovery of this manuscript. Manuscript contains over-interpretations) makes it difficult to accept without 'radical revision'.

As you know, eLife does not allow revision unless it is straightforward. This does not exclude your future opportunity to submit the revised manuscript to eLife again. However, we sincerely hope that you take our advice to improve the manuscript.

Reviewer #1:

This is a revised version of previously revised manuscript by Januschke and colleagues. They report the molecular mechanism of asymmetric Miranda localization in Drosophila neuroblast.

Same as the previous versions of this manuscript, I found that each experiment is well done and well documented. However, the writing and presentation again makes it difficult for me to assess general impact of this manuscript (thus I have to rely on other reviewers with more expertise in the field). The main difficulty in reading this manuscript for me was 1) sometimes it is not clear whether all of their conclusions are consistent among themselves, 2) sometimes it is not clear whether their claims are indeed consistent with previous findings especially when they argue against existing models. (examples will be provided below). Here, I am not trying to be nit-picky about writing, but the lack of clarity throughout the text has made it difficult for me (and probably for other reviewers in the previous rounds) to grasp the (potential) impact of this paper. With that being said, I am not saying that writing only can be a reason to accept or reject a paper.

In addition, in the course of revisions, the authors often responded to reviewer comments only in their rebuttal/response letters, without incorporating it into the main text. This has also made it difficult to relate how the concern was resolved in a way presentable to the future readers. Response letters should be in the format of 1. Reviewer comments, 2. response by authors, 3. Explanation how the authors incorporated reviewer comments and responses within the main text ("based on these[…]we changed the main text as following.[…]"). In the previous versions, all the reviewers understood that the manuscript clearly demonstrates that the current model needs to be revised. However, the reviewers also commented that the authors' new model may not be consistent with the existing data. In response, the authors provided explanations only in the response letter, without changing much in the main text. The authors should keep in mind that the future readers may have the same questions as reviewers (very likely), and thus such questions from reviewers are better be addressed within the main text, not in the rebuttal letter.

Reviewer #2:

This manuscript contains a series of well-controlled and careful experiments that convincingly demonstrate that Miranda localises by different mechanisms in interphase and mitosis. The observations that Miranda is excluded from the cortex by aPKC and retained in a basal crescent by the acto-myosin cortex are not novel, but most previous work has presented these as competing models to explain how Miranda is asymmetrically localised. Thus, the main novelty in this manuscript is that both are true, but at different time points during the cell cycle. While I agree with the other referee's comments, I don't think that any of them challenge this basic conclusion, which in my opinion is sufficiently important to merit publication in eLife.

While I am very positive about these results and the quality of the data, I think that the presentation needs to be improved. My version was entirely lacking Figure 3 (which contains one of the key pieces of data showing that Miranda behaves differently in mitosis), and the Videos were not labelled in the same way for downloading as in the text, which made it difficult to relate the two. The text also needs to be improved to highlight the main conclusions and discuss how these relate to previous work. To give just one example, the manuscript presents strong evidence that myosin activity is required for the basal recruitment of Miranda after NEBD (ML-7 treatment, rescue of this effect with SqhEE, low doses of Y-27632), but does not discuss these data in light of previous work from Barros et al. arguing that myosin excludes Miranda from the cortex or the argument against Barros et al. that high concentrations of Y-27632 inhibit aPKC. Januschke et al. seem to have cleared up all of this confusion by using a different drug, proving that it is specific for myosin activation and by separating the effects of Y-27632 on Rho kinase and aPKC at different doses, but they fail to put this all together into a single argument in the discussion that makes their contribution clear.

Reviewer #3:

This manuscript by Januschke's group proposed that the asymmetric Miranda localization in the asymmetric cell division of the Drosophila neuroblast is mediated by two mechanisms: cortical exclusion through the phospho-regulation by aPKC and cortical retention by actomyosin network. The concept that Mira utilizes two different mechanisms in different mitotic phases is very interesting but their data does not seem to fully support their idea and seem to leave some room for alternative interpretations.

First of all, there was no Figure 3 in this manuscript.

I think we cannot accept such incomplete manuscript for a peer-reviewing process, but I at least leave some comments.

In the Lgl3A over-expression experiments, as authors mentioned in the letter, the degree of the aPKC suppression seems to be varied; 7/25 seem to completely suppress aPKC and 4/25 seem to fail to suppress aPKC function. I think it is natural to treat the rest 14/25 as the intermediate between these two; a partial suppression of aPKC function. In such situation, " Mira initially asymmetrically localized, but LatA treatment cause Mira to become uniformly localized on the PM". This result is mostly same as the result when authors introduced MiraS96D in which " MiraS96D::mCherry always achieves asymmetric localization in mitosis…MiraS96D::mCherry relocalizes to the entire cortex upon LatA treatment".

Thus, it is highly possible that MiraS96D is a hypomorphic allele with regards to the phospho-regulation by aPKC.

In this point of view, I think these results would not support the authors' idea that " the BH motif mediates two different binding modes", and rather it is possible to interpret that Mira in mitosis would require both PM binding and other supportive interactions mediated by actomyosin for basal localization.

In the most of the figures, authors used Baz-GFP as a marker for the Par complex localization. Baz localization and aPKC localization are not always the same as shown in Figure 6F. Thus, it would be important to examine the distribution of aPKC itself. Especially, it is important to examine the kinetics of aPKC in the presence of LatA (Figure 2C-E), because this is the only one result that distinguishing whether Mira was "cleared" by aPKC or "anchored" by actin.

It is difficult to understand the different outcomes of the ML-7 treatment and Y-27362 treatment. Myosin inhibition resulted in the displacement of Mira from the cortex and Myosin/actin inhibition resulted in the expansion of Mira crescent, why? Authors should mention their interpretation about these results.

Reviewer #4:

This manuscript proposes a mechanism for the polarization of the fate determinant Miranda in Drosophila neuroblasts. The current model for this process is that Miranda directly interacts with the plasma membrane and phosphorylation of the membrane-interacting peptide by aPKC displaces Miranda into the cytoplasm. The model proposed in the work under consideration is that Miranda polarization occurs via different mechanisms in interphase and mitosis, and the mitosic mechanism involves a "basal affinity zone" that somehow involves the cytoskeleton. In general, I feel that the authors have overinterpreted their data, which relies on drug treatments and in some cases, used flawed logic to draw their conclusions. Some of the main problems are summarized below:

• While the characteristics of interphase localization are interesting, they are not significant unless interphase localization is somehow important for Miranda function

• Drug treatments, such as LatA, lead to loss of Miranda cortical localization, which could result from drug-induced ectopic aPKC activity. The experiment the authors provide to discount this possibility ("anchoring vs. clearing") is not convincing for a number of reasons, including that Baz localization is used as a proxy for aPKC localization. The cortical localization of Miranda in LatA treated neuroblasts lacking aPKC function are consistent with a misinterpretation in this regard.

• The authors over interpret the S96D mutant, which only partially abrogates plasma membrane binding. They state in the text that the localization of mutant is the same as WT but the Videos they provided do not support this claim.

• Even if the previous problems are overlooked, the authors have not gone far enough to characterize the new interactions that they are proposing take place. The phenotype they observe when inactivating the cytoskeleton could arise from very indirect effects, yet they are proposing direct interactions with a "basal affinity zone". I do not believe that they have presented data supporting their claim that the zone is only basal, or that tell us anything about the interactions that would occur in the zone.

• The text is poorly-written and convoluted. It would be very difficult to follow for someone outside the field.

In summary, I do not believe that the authors have adequately disproven the possibility that the Miranda phenotypes they observe in drug-treated neuroblasts arise from indirect effects, especially through effects on the Par complex. Thus, they have not sufficiently demonstrated the existence of a "basal affinity zone". Furthermore, they have not sufficiently characterized the interactions that they propose are occurring in the "basal affinity zone", if it does indeed exist.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]

Thank you for submitting your article "aPKC-mediated displacement and actomyosin-mediated retention polarize Miranda in Drosophila neuroblasts" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom, Yukiko M Yamashita (Reviewer #1), is a member of our Board of Reviewing Editors, and the evaluation has been overseen by K VijayRaghavan as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Daniel St Johnston (Reviewer #2).

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

As you can see in the individual review comments, they agreed with the potential importance of your discovery, yet even the most positive reviewer shows a concern about the interpretation of the results. In addition, during reviewer discussion, all agreed that the presentation (writing, flow of the logic) requires considerable improvement such that clear and consistent message can be delivered to the readers. We must emphasize that the reviewers had major concerns regarding the fact that the presentation has not improved in the past rounds of submission-review cycles. Please pay serious attention in these aspects in preparing your revision.

Reviewer #1:

The authors describe interesting behavior of Miranda localization during Drosophila neuroblast divisions: Miranda directly binds to plasma membrane (and evenly around the neuroblast cortex) during interphase, it is cleared from the plasma membrane at transition to mitosis, and comes back as 'basal crescent' in mitosis, which is required for asymmetric outcome of the neuroblast division.

They show that:

-In interphase, Mira directly bins to plasma membrane (evenly) in an actin independent manner.

-F-actin/myosin-dependent localization of Mira operates only in mitosis.

-Defective clearance of interphase cortical Miranda by aPKC leads to a failure in asymmetric Miranda localization in mitosis.

This is a revised, new submission of previously rejected manuscript at eLife, and this revised version has some additional data that solidify their claims. Their data may provide interesting insights into Mira localization throughout NB cell cycle, clarifying certain aspects of existing models on Mira localization.

However, the major weakness of this manuscript in writing remains. Aside from multiple clear grammatical mistakes, the lack of contexts when they describe their data really interferes readers' understanding (the most important question in the field that is addressed by each specific experiment is not explained). I listed up several specific places, where the way it's written confused me a lot. I recommend the authors to go through extensive editing focusing on 'reader friendly writing'. Given the history of this submission, we must insist that the next version will be the last. Please be certain that you have had others read and proof this so that the Board and reviewers will have an easier time deciding if you have made your essential points. Failing that, the paper will not be returned for another rounds of revisions.

Reviewer #2:

This manuscript contains a series of well-controlled and careful experiments that present evidence that Miranda localises by different mechanisms in interphase and mitosis. The observations that Miranda is excluded from the cortex by aPKC and retained in a basal crescent by the acto-myosin cortex are not novel, but most previous work has presented these as competing models to explain how Miranda is asymmetrically localised. Thus, the main novelty in this manuscript is that both may be true, but at different time points during the cell cycle. The experiments are carefully quantified and informative, but more care is needed in the interpretations of the results. The authors should discuss alternative explanations for their data, even if they are not their preferred model. Another obvious weakness is the use of Y-27632 as an aPKC inhibitor when it also affects myosin activity through ROCK, although this experiment is still valuable and provides a useful comparison with previous work. The manuscript is an improvement over the previous version, but needs to be revised to include a more comprehensive discussion of the meaning of the results that considers alternative interpretations and that is more accessible to a general audience.

Specific comments:

1) FRAP stands for Fluorescence Recovery after Photobleaching and not Fluorescence Redistribution.

2) Intriguingly, such enlarged crescents were LatA and ML-7 sensitive, both indicators that enlarged Mira crescents were not due to lack of phosphorylation of Mira by aPKC. This needs more explanation and perhaps some discussion as to whether the expansion depends on aPKC phosphorylation of another substrate.

Reviewer #3:

This revision from Hannaford et al. remains seriously flawed and I do not recommend it for publication. First, many of the author's conclusions are contradicted by their own data. For example, a key conclusion is that Miranda is cleared from the cortex at prophase. It's difficult or impossible to make this assessment from the data the authors show because of Miranda signal from attached GMCs. Video S2 (included for a different reason) shows a beautiful division without any visible GMCs. This video supports a conclusion opposite of the author's: Miranda remains on the cortex throughout the cell cycle including in prophase. Another example: if Miranda is cleared all at once from the cortex upon LatA treatment, as the authors conclude, then why does the kymograph show the Miranda crescent shrinking? This effect is obfuscated by compressing the transition over a small number of pixels but Video S4 shows it beautifully. Scrubbing video S4 back and forth around the time that Miranda disappears clearly shows the crescent growing and shrinking, not disappearing all at once. Another example: The MLCK inhibitor ML-7 causes cortical Miranda to be lost in metaphase, but the phenotype could be non-specific (SqhEE expression only causes delay of the phenotype), and even if it isn't this one result doesn't prove that Miranda loss is due to a direct effect. Thus, close inspection of the author's data does not support clearing of Miranda at prophase or loss of anchoring by cytoskeletal poisons (in fact, for two experiments the opposite conclusion is supported). The paper contains a fair amount of data, such as the difference in photobleaching dynamics of Miranda in interphase and mitosis, and the examination of Y-27632 treated neuroblasts, which appear to be superfluous without a more clear explanation.

Considering that the paper's foundational conclusions are flawed, it is not surprising that the strongest tests of the resulting model (that Miranda is cleared from the cortex by aPKC phosphorylation of its BH motif during prophase and anchored by actomyosin during metaphase) fail miserably. These tests come in three experiments. First, Miranda∆BH – if the BH is only required for interphase localization, then Miranda∆BH should be cytoplasmic and interphase and polarized during metaphase. However, the authors find that it's cytoplasmic in metaphase too. Expression of constitutively active aPKC (aPKC ∆N) is also a great test of the model. If aPKC activity is just required to clear interphase cortical localization, constitutively active aPKC should result in cytoplasmic Miranda in interphase and polarized Miranda in metaphase. The result again contradicts the model's prediction: Miranda is cytoplasmic in both. Remarkably the authors don't even discuss the metaphase result in the paper! Finally, Miranda S96D (reduces but does not ablate BH interactions with the membrane) should have reduced cortical interactions in interphase but polarized normally in metaphase. Here again the result, reduced cortical signal in both interphase and metaphase, is not consistent with the author's proposed model.

Remarkably, these results directly contradict the author's proposed model but are precisely as predicted from a simpler model in which aPKC regulated BH interactions with the membrane control Miranda cortical association throughout the cell cycle. Of course the more complex model could be saved by making it even more complex – perhaps the BH is also involved in the actomyosin interaction, perhaps aPKC also regulates the actomyosin anchor, perhaps a single point mutant that partially disrupts the BH lipid interaction also partially disrupts the actomyosin interaction – and the authors make a half-hearted attempt to do so (at least for the first experiment), but more data would need to be provided to support a more complex model (e.g. how aPKC might regulate the anchor).

I believe the problems outlined above represent deep and fundamental flaws and therefore preclude publication in its current form. I also note that the authors removed the model figure from the paper. In my opinion this is moving in the wrong direction and encourage the authors to revise the paper so it is more clear, not less.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for submitting your article "aPKC-mediated displacement and actomyosin-mediated retention polarize Miranda in Drosophila neuroblasts" for consideration by eLife. Your article has been reviewed by two peer reviewers, and the evaluation has been overseen by a Reviewing Editor and K VijayRaghavan as the Senior Editor. The reviewers have opted to remain anonymous.

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

The core discovery of this manuscript 'Miranda localization is regulated differently during interphase and mitosis' is interesting and worth publishing. Knowing that Miranda localization switches the mode (plasma membrane-bound in interphase, and anchored via actin cytoskeleton in mitosis) is an important stepping stone to fully understand how Mira localization is regulated to achieve asymmetric neuroblast division.

The reviewers had a lengthy discussion as to how to proceed with this manuscript. First, we agreed that the core discovery described above is worth publishing if accurately stated: that Miranda localization have two modes in interphase and mitosis. However, while several reviewers agreed that the data in the manuscript sufficiently supported their model, we were not able to reach a consensus on this point. In the end, the reviewers agreed to recommend publication provided that the authors clearly and accurately state their results in a manner that allows critical assessment of all the data, such that all the readers can judge on their own.

All reviewers noted concerns regarding the accuracy of descriptions. More specific comments will follow below, but the collective result of inaccurate statements/descriptions is that 1) the real contribution of the manuscript is blurred, 2) the manuscript reads as if the authors are proposing an alternative model which they claim to be better than the existing model, but the new model also leaves many unexplained observations, making readers wonder whether the new model is a real improvement or only confusing the field.

Therefore, we ask you to edit the text and submit the revised version. We decided to allow you one more round of revision because it will only require textual revision. With that said, all the reviewers expressed strong concerns that the current writing is not accurate enough, and if it is not fully taken care of in the next round, we must reject your manuscript, and no further revision will be allowed. In revising, we ask you to ensure that the revised main text is self-sufficient in conveying all the messages accurately and clearly, and not to utilize the rebuttal/response letter as a platform to explain things that are not entirely consistent with the main text.

Guidelines for revision:

Each Results section should clearly indicate how the data contributes (or doesn't) to the model. We also ask that in the discussion the authors recognize that there are observations they have made that are inconsistent with this model. They can do their best to try and explain them away, but at least readers will be able to more readily appreciate these inconsistencies and judge the author's explanations for themselves. Please ensure that the revised manuscript is self-sufficient in conveying these points, without the need of relying on response letter.

The reviewers consider that the most reasonable model to be put forward in this manuscript is "either Miranda BH-phospholipid or BH-actomyosin interactions are sufficient for cortical localization. In interphase BH-phospholipid interactions mediate uniform cortical association but early in prophase they start becoming inactivated by aPKC in an apical to basal fashion. By metaphase all BH-phospholipid interactions are inactivated and BH-actomyosin interactions take over (BH-phospholipid interaction might have supportive role in mitosis, too).

Overall, the reviewers agreed that there are many observations that do not exactly fit to the authors' model (e.g. overexpression of aPKC-deltaN). Some experiments were suggested to test authors' model, which resulted in observations that are inconsistent with the authors model. Each time, the authors provided 'possible explanations why the test did not support their model', yet remained that 'the model is still correct'. The authors should put more effort into explaining these inconsistent results (instead of 'explaining away' inconvenient results, trying to reach better explanation that makes sense as a whole). The reviewers do not expect the authors to figure out the mechanism of Miranda localization entirely. The reviewers appreciate that knowing that Miranda localization likely has two modes is sufficiently important progress, but writing is not conveying that this is the core message of the manuscript, and instead it claims more than the experimental data can support.

As an example of inaccurate statement, in the first section of the results, the authors start out by stating Miranda is 'cleared' and 'reappears' in mitosis, clearly leaving the impression that the previous model of expanding apical Par complex gradually displacing Miranda is wrong. However, as the video shows, Miranda 'clearance' occurs from the apical to basal. Although the authors explain that this clearing happens from apical to basal, but in subsequent sections, they do not come back to this fact and stick to the expression of 'clearance' and 'reappearance'. Here, the emphasis should be the fact that mitotic Miranda crescent is much stronger, prompting the investigation of the mechanism by which Miranda anchoring is promoted during mitosis (which they show to be actin dependent).

The weakest explanations in the current manuscript are the following:

1) If the authors' model is entirely correct, aPKC-deltaN expression should result in delocalization of Miranda in interphase, but normal, crescent localization in mitosis. But the actual observation is that aPKC-deltaN results in the delocalization of Miranda in interphase and mitosis. In the main text, the authors do not mention that Miranda fails to localize even in mitosis upon aPKC-deltaN expression, and conclude that their data is consistent with their model. In explaining this inconsistency, they state 'it is very difficult to predict the precise consequences of overexpressing a deregulated kinase'. A better explanation is required for the authors to be able to propose that their new model fits better than the existing model with all experimental results.

MiraS96D: the model predicts that MiraS96D should be cortically polarized in metaphase but it's partially cytoplasmic (S96D only partially destabilizes the BH-phospholipid interaction). The author's explanation is that cortical localization is affected more in interphase, which is not a thorough discussion. Unfortunately, readers will be wondering why Miranda5D (which completely abolishes BH-phospholipid interactions) wasn't tested because it would very easily resolve the explanations provided by the author for both of these inconsistencies (especially because the explanations are so poor).

2) According to the authors' model, Mira-deltaBH should be cytoplasmic in interphase, yet should localize to the basal crescent in mitosis. However, they find that Mira-deltaBH fails to localize to the crescent even in mitosis. In the main text, they simply conclude that 'this is consistent with Mira being localized to the cortex via interaction with plasma membrane'. Then, in response to the reviewers' comment that deltaBH shouldn't be cytoplasmic in mitosis, only in response letter, they point out that mitotic Miranda localization may require BH domain-mediated plasma membrane localization in addition to actomyosin. The authors should avoid any discrepancies between the main text and the response letter, and a cohesive story must be presented within the main text (and the observations that do not fit to the model must be clearly presented and acknowledged). Also, the reviewers do not agree that the data shows that Miranda disappears all at once upon LatA treatment. The revision should remove this argument and focus on the argument that Miranda disappears before aPKC arrives. Of course, we do not ask the authors to figure out everything about Mira localization, but they should explain 'why a particular observation betrays their prediction based on the model' in a cohesive manner, and more sincerely (i.e. do not bury important discussions within the response letter, which become inconsistent when placed in the main text).

We would like to provide an example of how reviewers' discussion proceeded. One reviewer noted: "I prefer an alternative interpretation, in which all of the localization after NEB is actomyosin dependent (ML-7 abolishes the crescent) and that this is mediated by the BH domain.": this clearly suggests that this reviewer appreciated the authors' discovery but still required to introduce his/her own interpretation to support authors' model. In response, another reviewer asked, "Then why is Miranda cortical in metaphase neuroblasts treated with LatA and lacking aPKC function (Figure 4—figure supplement 1 panel B)? If localization after NEB is actomyosin dependent, then surely it shouldn't localize after treatment with LatA." In response to this question, the first reviewer responded, "My interpretation of this experiment is that it indicates that aPKC-dependent clearing is an essential prerequisite for the formation of the basal crescent. In the absence of aPKC, Miranda remains in the interphase state where it binds to phospholipids. I could be wrong, as this result is entirely consistent with the simple (existing) model, but the latter doesn't explain the ML-7, Y26732 and FRAP data. What I find harder to understand is why the basal crescent doesn't form in the presence of constitutively-active aPKC, which is why the results aren't clear cut. It seems that Miranda needs to be phosphorylated by aPKC before NEB to form the basal crescent, but that too much aPKC activity or activity after NEB inhibits this. It is a shame that the manuscript didn't really get to grips with these questions."

In our opinion, a manuscript should not leave this much room of interpretation to the readers, and the authors must clearly present their data (or if that is not sufficient, more data would be clearly required. But the reviewers are not asking more experiments here).

Additionally, the model figure is currently extremely vague because it is hard to see what the authors are implying is taking place as far as Miranda-cortex interactions. Please revise the model figure to describe your model more clearly.

https://doi.org/10.7554/eLife.29939.032

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Whereas all reviewers agreed that the idea of Miranda having two distinct modes of localization is interesting and of potential impact, they also felt that your data do not prove your hypothesis: whereas your data indicate that the existing model of Miranda localization may require a revision, your alternative hypothesis does not provide a better explanation than the current model, leaving unexplained observations as much as the current model does. The individual comments can be found at the bottom of this letter.

The reviewers felt that it is critically important for you to carefully assess the current model and all the data that support it. At least the revised model must be better than the current one in explaining all the observations/data in the field. For example, a key result supporting the current model is that the unphosphorylatable Miranda (Mira5A) is depolarized in mitosis. This would be inconsistent with your revised model. In reaching a new model, it is not sufficient to point out that the current model is imperfect, but it is important that there is a significant advancement in the understanding, instead of muddying the field.

Reviewer #1:

[…]

1) If Mira localization is regulated by two independent mechanisms, the significance of interphase cortical localization becomes unclear. The reason why Mira localization is important is because it functions as a fate determinant for NB daughter (GMC), thus only mitotic localization should matter.

We have not yet been able to identify a function for interphase Mira, However, as stated before, we have a working hypothesis. How are fate determinants prevented from acting in the NB? Very little is known about this. In mitosis, Mira shuttles between cytoplasm and basal crescent, therefore it is expected that not all of Mira (and presumably its cargos) segregate upon mitosis to daughter cells. Nevertheless NBs are sensitive to low doses of the Mira cargo Pros (Lau and Doe 2014). Keeping Mira at the interphase cortex might help to sequester Pros from entering the NB nucleus.

2) However, the Figure 4D data (Mira-deltaBH cannot go to mitotic basal cortex, whereas Mira-SD goes to basal crescent, despite not being able to localize to the interphase cortex) seems to suggest an interdependence between interphase and mitotic localization. For example, Mira has to localize to the cortex in interphase so that it can be phosphorylated by aPKC, then it dissociates from the cortex but potentiated for basal cortex localization.

We are not sure we fully understand this comment. While it is unknown where exactly aPKC can phosphorylate Mira in NBs, cytoplasmic Lgl might indeed restrict aPKC activity to the cortex. Therefore we agree, that it is a possibility that the localization of Mira at the cortex may allow phosphorylation by aPKC. The potentiation of basal cortex localization, once phosphorylated is also a possibility. The current data situation in NBs does not allow to be sure whether phosphorylation of Mira by aPKC has only as consequence the inhibition of cortical localization. It remains a possibility that phosphorylation of Mira enables it to engage with the actomyosin cortex basally.

Reviewer #2:

This manuscript by Dr. Januschke and colleagues examined the mechanisms underlying the asymmetric localization of Miranda in the Drosophila larval neuroblasts. They proposed that there would be two different mechanisms that regulate the Miranda cortical localization; direct binding to the plasma membrane in interphase and actin-myosin dependent mechanism in mitotic phase. The concept that Miranda utilizes two different mechanisms to localize at the cell cortex in different cell cycle phases is new and interesting, however, there are some points to be clarified before published.

Major comments

1) In the most of the figures, authors utilized Lgl3A overexpression to abolish the function of aPKC, however, as shown in Figure 2G, the majority of the Lgl3A expressing neuroblasts achieved asymmetric localization of Miranda in mitosis while in the aPKC mutant clones Miranda localized at the entire cortex (Figure 4), suggesting that Lgl3A overexpression does not recapitulate aPKC loss of function.

We have no reasons to doubt that overexpression of Lgl3A inhibits aPKC activity. As we stated before it is possible that overexpression of Lgl3A has different consequences than aPKC inhibition by other means. In addition we noticed that worniu-Gal4 is not expressed homogenously in all NBs (not shown). Therefore, Lgl3A levels are likely to vary between individual NBs. We interpret the different effects on Mira by Lgl3A overexpression to mean that aPKC is inhibited to different degrees.

Although Lgl3A is thought to suppress aPKC function, it seems that, at least in these cases, Lgl3A expression is not enough to suppress aPKC function or it has some other functions than aPKC suppression, making it difficult to interpret Lgl3A overexpression phenotypes. Thus, it is crucial to check what is observed in Lgl3A overexpression is also the case in aPKC mutant or RNAi clones.

Failure to clear Mira from the interphase cortex has been observed in mitotic apkc NBs and actin network dependence of Mira localization was tested in aPKC RNAi NBs, which we confirm to have reduced aPKC levels (Figure 2—figure supplement 1). Therefore, the only experiment that solely relied on Lgl3A overexpression is the analysis of Mira cortical dynamics by FRAP (Figure 3). We have now performed FRAP of basal Mira in mitotic aPKC RNAi NBs and find it to be similar to when Lgl3A is overexpressed (new Figure 3B, B’).

We do see enlarged Mira crescents in a few NBs when aPKC RNAi is driven by worniu-Gal4 (not shown) in line with the Lgl3A overexpression results. Whatever the actual effect on aPKC, the importance of this experiment lays in the fact that disrupting the actin cortex, results in loss of basal enrichment of Mira. Importantly, we see a similar effect of LatA treatment on MiraS96D in which aPKC activity in principle should be normal unless MiraS96D has an inhibitory effect on aPKC, which seems unlikely.

2) In Figure 4E, the authors showed that Mira[S96D] showed uniform cortical localization in colcemid and LatA treated neuroblast. Following the authors' model, Miranda localizes at the cortex by either direct binding to the plasma membrane or actin-dependent mechanism, however, in this situation Mira seems to localize at cell cortex in the absence of both machinery. Then how does Miranda localize at cell cortex in this situation? Authors should discuss about this point.

We agree that this is a highly unexpected observation. In our view the phosphomimetic MiraS96D demonstrates the point that the way Mira localizes to the cortex in interphase (weak interaction with phospholipids drives plasma membrane localization) is different to that in mitosis (Mira is stabilized additionally by the actomyosin cortex basally). Even though MiraS96D localizes uniformly to the PM upon microtubule and actin network disruption, it does not do that when the actomyosin cortex is intact. MiraS96D is therefore able to read the information required for its asymmetry and stabilized basally.

The Aspartate replacing the Serine residue at position 96 only mimics phosphorylation and as is true for any phosphomimetic mutant, the ability to regulate the protein at the modified residue is gone. Therefore uniform plasma membrane binding upon LatA treatment could be an irrelevant consequence of the MiraS96D mutant. Alternatively, general aspects of Mira (ability to dimerize, other posttranslational modifications (Jia et al., 2015, Slack et al., 2007, Zhang et al., 2015)) might change upon NEB and contribute to Mira’s ability to localize to the PM.

3) Authors utilized Y-27632 to temporally inhibit aPKC function but it is not clear whether Y-27632 treatment is really affecting the aPKC activity or not. As a control experiment, it is better to examine whether the addition of 50uM Y-27632 to the cycling neuroblasts results in the Lat-A-insensitive entire cortical localization of Miranda in mitotic phase (like aPKC loss of function situation).

We have now performed the requested experiment. We added 50, 100 and 200µM Y-27632 to cycling NBs co-expressing Baz::GFP or aPKC::GFP with Mira::mCherry. 50 and 100µM resulted again in enlarged crescents. Only with 200µM we were able to detect uniform cortical Mira in mitotic NBs. When 200µM were added to cycling NBs, Baz::GFP started to look abnormal in interphase and mitosis and so did NB cell shape in interphase. However 8 out of 25 NBs in this condition divided, 7 of which had incomplete clearing of Mira at the onset of prophase and Mira was everywhere on the cortex upon NEB. When such NBs divided the daughter cells appeared similar in size (new MOV S15). We repeated this experiment co-expressing aPKC::GFP and Mira::mCherry, but arrested the NBs with colcemid in mitosis. We were able to detect in few cases apical aPKC crescents and uniform cortical Mira. Adding LatA to these cells, lead to loss of cortical aPKC asymmetry, but Mira remained at the cortex (new Figure 6F).

These experiments are in agreement with the idea that Mira phosphorylation by aPKC is not occurring in the presence of high doses of Y-27632.

Reviewer #3:

This revision, like the original, proposes that Miranda is polarized in Drosophila neuroblasts via two distinct mechanisms, one in interphase and one in mitosis – the former being by a previously described "BH" association mechanism, and the latter being a "myosin affinity zone". I had two main criticisms of this proposed mechanism. First, the authors did not separate these two functions, rather deletion of the BH caused loss of Miranda polarity in both interphase and mitosis. Second, the proposal of a "myosin affinity zone" is based on data using inhibitors leaving open the possibility that the effect is indirect. The author's work is potentially exciting because there are interesting questions related to Miranda's function in interphase, and the role of myosin in polarizing Miranda. Unfortunately, however, I do not believe that the revised manuscript comes close to adequately addressing the criticisms raised in the original review and therefore I do not recommend publication.

• A separation of function allele would be the most definitive test of the authors model – if two different mechanisms are used, then it should be possible to find Miranda alleles that only polarize in mitosis and not in interphase (unless the interphase mechanism is required for the mitotic one, which is actually what the title says – see below). Mira∆BH isn't polarized in interphase or mitosis indicating that the authors model is wrong or the BH motif is required for each binding mode. To distinguish between these possibilities, the authors examined MiraS96D, a phosphomimetic allele, and conclude that it is a bona fide separation of function allele, disrupting interphase localization but not mitotic. However, S96D is clearly a hypomorph (Figure 2Q,T Current Biology 19, 723-729, 2009) and my interpretation of the author's data (i.e. video 9) is consistent with it behaving as a hypomorph in their assay – Mira96D basal crescent signal is reduced and spindle signal is increased not only in interphase but also in mitosis.

Indeed elements in the Mira protein mediating the interphase mechanism are also required for the mitotic mechanisms, but the mechanism mediating binding in interphase is not sufficient in mitosis. That is what we propose.

We take the criticism that we apparently failed to be clear. But we respectfully would like to point out that we never argued that the BH motif is not required in mitosis nor did we claim that MiraS96D is not able to bind the PM, we also fully agree that miraS96Dis a hypomorph with regards to efficient PM binding.

However, the negative charge within the BH motif strongly reduces PM binding in interphase, but MiraS96D is always asymmetric in mitosis. This is an argument that the affinity of Mira for the cortex changes in mitosis and hence that the mechanisms of cortical retention in both cases are different. Somehow this has been perceived as if we were to use this experiment to rule out a role of the BH motif in mitosis. We have improved this point in the current version to avoid further misunderstanding.

Furthermore we would like to point out that additional features of Mira protein are altered when the entire BH motif is deleted. MiraS96D can bind microtubules and PM in interphase and in mitosis. MiraΔBH, however, is not enriched at the cortex in interphase or mitosis. Furthermore, unlike MiraS96D, MiraΔBH apparently does not bind spindle microtubules in metaphase. Therefore, the deletion of the entire BH motif affects other properties of Mira than just mediating binding to the PM.

• I also criticized the original version of the manuscript because it proposed that mitotic Miranda polarization occurs through a "myosin anchor" in a "basal affinity zone" because the data supporting this relies on drug treatments. Nothing in the revised manuscript addresses this criticism. The author's rebuttal letter states, "We agree that the mode of action of myosins remains perhaps obscure since we do not know whether the effect is direct or indirect." That is precisely my criticism, that the mode of action of myosins remains obscure. The letter goes on to state, "However, given that the ML-7 effect on Mira basal localization in colcemid arrested NBs can be tempered by overexpressing a phospho-mimetic version of Sqh[…]an involvement of myosin activity in anchoring Mira at the basal pole at NEB seems very likely." This statement is not very impactful if the involvement of myosin activity is very indirect, which the authors themselves acknowledge is entirely possible.

We agree that the data is based on drug treatments and therefore weaker than the genetic evidence for a role of myosin activity previously provided (Barros et al., 2003). Our experiments allow narrowing down the function of myosin activity to either provide anchoring for Mira or contribution to configure the basal affinity zone. The integration of phosphoregulation and the regulation of the actomyosin network is key to understand how Mira localizes asymmetrically.

The letter further states, "We hope that the reviewer would agree that the current apical exclusion model does not leave much room to explain the role of myosin activity in Mira localization[…]". I strongly disagree with this statement. First, pretty much any model would allow for an indirect role for the cytoskeleton. But more importantly, the authors do not seem to understand the "current apical exclusion model", at least as it is articulated in Bailey and Prehoda, Dev Cell, 2015, which they have referenced. In this model, BH motifs cooperate with accessory interactions to mediate cortical recruitment of Par substrates (e.g. see Discussion section, "Multivalent Interactions Mediate Par Substrate Cortical Localization"), and that aPKC phosphorylation of the BH motif disrupts this interaction. In this model an "accessory interaction" could certainly require interactions with myosin – the key point being that any such interaction must not be sufficient for cortical targeting (otherwise BH phosphorylation wouldn't have an effect on cortical localization).

We thank the reviewer for pointing this out. The central idea of Bailey and Prehoda, 2015, that is based on experimental evidence, builds on experiments using overexpression of different versions of Mira (and other substrates) in S2 cells and in vitro assays to elegantly establish that phosphorylation of the BH motif by aPKC inhibits the ability of aPKC substrates to bind to the plasma membrane. If anything our data confirm this mechanism now in NBs using reporters and tools at endogenous levels of expression: Mira weakly associates with the PM in interphase, which i) depends on the BH motif, ii) is perturbed by introducing a negative charge at an aPKC phosphorylation site within the BH motif and iii) prevented by overexpressing constitutively active aPKC.

One testable interpretation of the data provided in (Atwood and Prehoda, 2009) is that controlling where aPKC is active in mitotic NBs, provides the spatial information where Mira can bind to the cortex. If this were the case, once aPKC is activated at prophase onset, Mira should be removed apically, but spared basally. This is not what we find. Moreover, the way Mira interacts with the cortex in mitosis should be the same regardless of the state of aPKC activation. This is again not what we find. Our data rather suggest that at least in the case of NBs additional processes occur during NB polarization which provide critical spatial information for Mira localization. Thus, our proposal of a basal affinity zone is in perfect agreement with the discussion about multivalent interactions in Bailey and Prehoda, 2015.

However, rather then being an indirect bystander, we propose that the actomyosin cortex directly provides spatial control of asymmetric Mira localization, which we believe is linked to control of asymmetric daughter cell size.

• The title of the paper is incorrect, according to the authors. The title states, "Switching the cortical binding mode is required for asymmetric fate determinant localization in Drosophila neuroblasts". However, the authors claim to have a Miranda allele that doesn't switch cortical binding modes and is still asymmetrically localized, directly refuting the statement made in the title. Stating one thing in the title and another in the text will confuse readers and make it very difficult for those in and outside the field to understand the impact of the paper. Furthermore, it suggests that the authors themselves do not understand the impact of their work.

We reveal that the cortical Mira binding mode can be distinguished in interphase and mitosis based on the actin dependence, altered cortical dynamics and sensitivity to introducing a negative charge into the BH motif. At one point a change in the mode retaining Mira must occur. We felt that “switch” described this well. However, this has apparently been misleading and perceived as if we were to exclude a role of the BH motif for localization in mitosis, which as stated already previously and above is not what we want to put forward. We have therefore taken the advice and changed the title to “aPKC-mediated displacement and actomyosin-mediated retention polarize Miranda in Drosophila neuroblasts”. If aPKC does not phosphorylate Mira, it remains uniformly bound to the PM. After NEB, Mira is however able to read spatial information provided by the actomyosin network to asymmetrically localize. Therefore aPKC-mediated cortical displacement is necessary but not sufficient to asymmetrically localize Mira asymmetrically.

[Editors' note: the author responses to the re-review follow.]

Reviewer #1:

This is a revised version of previously revised manuscript by Januschke and colleagues. They report the molecular mechanism of asymmetric Miranda localization in Drosophila neuroblast.

Same as the previous versions of this manuscript, I found that each experiment is well done and well documented. However, the writing and presentation again makes it difficult for me to assess general impact of this manuscript (thus I have to rely on other reviewers with more expertise in the field). The main difficulty in reading this manuscript for me was 1) sometimes it is not clear whether all of their conclusions are consistent among themselves, 2) sometimes it is not clear whether their claims are indeed consistent with previous findings especially when they argue against existing models. (examples will be provided below). Here, I am not trying to be nit-picky about writing, but the lack of clarity throughout the text has made it difficult for me (and probably for other reviewers in the previous rounds) to grasp the (potential) impact of this paper. With that being said, I am not saying that writing only can be a reason to accept or reject a paper.

In addition, in the course of revisions, the authors often responded to reviewer comments only in their rebuttal/response letters, without incorporating it into the main text. This has also made it difficult to relate how the concern was resolved in a way presentable to the future readers. Response letters should be in the format of 1. Reviewer comments, 2. response by authors, 3. Explanation how the authors incorporated reviewer comments and responses within the main text ("based on these.[…] we changed the main text as following […]"). In the previous versions, all the reviewers understood that the manuscript clearly demonstrates that the current model needs to be revised. However, the reviewers also commented that the authors' new model may not be consistent with the existing data. In response, the authors provided explanations only in the response letter, without changing much in the main text. The authors should keep in mind that the future readers may have the same questions as reviewers (very likely), and thus such questions from reviewers are better be addressed within the main text, not in the rebuttal letter.

We followed the advice of the editors and the reviewers to radically revise the manuscript.

Reviewer #2:

This manuscript contains a series of well-controlled and careful experiments that convincingly demonstrate that Miranda localises by different mechanisms in interphase and mitosis. The observations that Miranda is excluded from the cortex by aPKC and retained in a basal crescent by the acto-myosin cortex are not novel, but most previous work has presented these as competing models to explain how Miranda is asymmetrically localised. Thus, the main novelty in this manuscript is that both are true, but at different time points during the cell cycle. While I agree with the other referee's comments, I don't think that any of them challenge this basic conclusion, which in my opinion is sufficiently important to merit publication in eLife.

While I am very positive about these results and the quality of the data, I think that the presentation needs to be improved. My version was entirely lacking Figure 3 (which contains one of the key pieces of data showing that Miranda behaves differently in mitosis), and the videos were not labelled in the same way for downloading as in the text, which made it difficult to relate the two. The text also needs to be improved to highlight the main conclusions and discuss how these relate to previous work. To give just one example, the manuscript presents strong evidence that myosin activity is required for the basal recruitment of Miranda after NEBD (ML-7 treatment, rescue of this effect with SqhEE, low doses of Y-27632), but does not discuss these data in light of previous work from Barros et al. arguing that myosin excludes Miranda from the cortex or the argument against Barros et al. that high concentrations of Y-27632 inhibit aPKC. Januschke et al. seem to have cleared up all of this confusion by using a different drug, proving that it is specific for myosin activation and by separating the effects of Y-27632 on Rho kinase and aPKC at different doses, but they fail to put this all together into a single argument in the discussion that makes their contribution clear.

We have substantially changed the entire manuscript following the advise of the reviewer. To highlight our contribution we have specifically modified the first paragraph of the Discussion, which summarizes the relevance of the problem, our approach, and the main conclusion. Our discussion now dissects the precise contribution of aPKC and actomyosin along our data and the literature. The discussion of our data in light of the Barros et al., 2003 data and the Atwood and Prehoda, 2009 proposal is discussed in the Discussion section.

Reviewer #3:

This manuscript by Januschke's group proposed that the asymmetric Miranda localization in the asymmetric cell division of the Drosophila neuroblast is mediated by two mechanisms: cortical exclusion through the phospho-regulation by aPKC and cortical retention by actomyosin network. The concept that Mira utilizes two different mechanisms in different mitotic phases is very interesting but their data does not seem to fully support their idea and seem to leave some room for alternative interpretations.

First of all, there was no Figure 3 in this manuscript.

I think we cannot accept such incomplete manuscript for a peer-reviewing process, but I at least leave some comments.

In the Lgl3A over-expression experiments, as authors mentioned in the letter, the degree of the aPKC suppression seems to be varied; 7/25 seem to completely suppress aPKC and 4/25 seem to fail to suppress aPKC function. I think it is natural to treat the rest 14/25 as the intermediate between these two; a partial suppression of aPKC function. In such situation, " Mira initially asymmetrically localized, but LatA treatment cause Mira to become uniformly localized on the PM". This result is mostly same as the result when authors introduced MiraS96D in which " MiraS96D::mCherry always achieves asymmetric localization in mitosis…MiraS96D::mCherry relocalizes to the entire cortex upon LatA treatment".

Thus, it is highly possible that MiraS96D is a hypomorphic allele with regards to the phospho-regulation by aPKC.

In this point of view, I think these results would not support the authors' idea that " the BH motif mediates two different binding modes", and rather it is possible to interpret that Mira in mitosis would require both PM binding and other supportive interactions mediated by actomyosin for basal localization.

To improve focus, we removed the experiments combining Lgl3A and LatA as well as S96D and LatA and focus on the point that in interphase Serin96 is a critical residue involved in PM binding of Mira.

We now further include the phosphomutant S96A, which is not cleared, and even transiently localizes apically before NEB after which it localizes uniformly at the cortex in an F-actin independent manner (new Figure 4A,B). In the S96D mutant the Aspartic Acid at position 96 strongly removes uniform binding to the plasma membrane in interphase, driving Mira onto cortical microtubules (new Figure 4A,B). Therefore, this shows that S96 phosphorylation is critical for clearance from the PM.

As we stated before this does not exclude the possibility that the BH motif is required for both. Our new results treating colcemid arrested NBs with 200µM Y-27632 reveal that despite occurring after a significant delay (~50min) Mira accumulates apically when 200µM Y-27632 is added to colcemid arrested NBs (new Figure 5E). Therefore, while the precise contribution of aPKC after NEB, remains to be determined, it is possible that aPKC contributes to Mira asymmetry after NEB by displacing Mira from the apical cortex as suggested (Atwood and Prehoda, 2009). However, at this concentration other effects of Y-27632 cannot be ruled out. This is discussed in the Discussion section.

It appears to be clear that Mira localization occurs step-wise. It is first uniformly binding the plasma membrane, which is prevented by aPKC phosphorylation. Then after NEB (in agreement with other findings (Zhang et al., 2016). Mira reappears at the cortex. The BH motif is required for this to happen. We recently found that Mira interaction with its cognate mRNA is required to maintain its localization basally (Ramat et al. in press, uploaded for reviewers). These results support the notion, that the BH motif is required for Mira asymmetry establishment, which allows subsequent stabilization through a mechanism requiring actomyosin activity.

In the most of the figures, authors used Baz-GFP as a marker for the Par complex localization. Baz localization and aPKC localization are not always the same as shown in Figure 6F.

(This point was also raised by reviewer 4). The new Video S2 shows an aPKC::GFP Mira::mCherry expressing neuroblast, in which Mira dynamics in the transition from interphase to mitosis are similar if not identical to when Mira::mCherry is expressed alongside Baz::GFP. While the cytoplasmic levels of Baz::GFP and aPKC::GFP may vary, especially after incubation of NBs with multiple drugs, we have no evidence for differences in the cortical distribution of aPKC and Baz once NBs enter mitosis.

Thus, it would be important to examine the distribution of aPKC itself. Especially, it is important to examine the kinetics of aPKC in the presence of LatA (Figure 2C-E), because this is the only one result that distinguishing whether Mira was "cleared" by aPKC or "anchored" by actin.

We would like to point out, that regardless of the Par complex marker that is co-expressed, Mira crescents fall off homogenously from the cortex in colcemid arrested NBs when LatA is added. This is measured in Figure 2—figure supplement 1 of the revised manuscript: the ratio of fluorescence intensities found in a region of interest (ROI) at the extremity of the crescent and in a ROI in the centre remains constant. Taking the effect on uniform cortical Mira (displacement into the cytoplasm in an apico basal direction), which is likely to be driven by aPKC at the onset of mitosis as reference for cortical Mira behaviour in response to aPKC, the effect of LatA on Mira in colcemid arrested NBs is different as we do not observe directional (i.e. apical to basal) removal of Mira.

Nevertheless, following the advice of the reviewers 3 and 4, new Figure 2—figure supplement 1, now includes the kymograph analysis of the effect of LatA on colcemid arrested NBs that express Mira::mCherry and functional aPKC::GFP (Besson et al., 2015, new Video S4), which reveals the same picture as when using Baz::GFP as a marker for the Par complex. Plotting the cortical fluorescence intensities of both Mira and aPKC reporters revealed that Mira is lost 2.8min ± 1min (n=13, this information is found in the Results section), before aPKC levels rise above cytoplasmic levels at the basal pole.

Both these findings support the interpretation that anchoring to F-actin is critical to keep Mira basally rather than an indirect effect caused by cortical aPKC redistribution.

It is difficult to understand the different outcomes of the ML-7 treatment and Y-27362 treatment. Myosin inhibition resulted in the displacement of Mira from the cortex and Myosin/actin inhibition resulted in the expansion of Mira crescent, why? Authors should mention their interpretation about these results.

This is an interesting point to discuss. The regulation of myosin activity by phosphorylation of its regulatory light chain is complex. Studies in mammalian cells have revealed that phosphorylation of myosin regulatory light chain can be differently affected by Y-27632 and ML-7, which could result in different outcomes on myosin activity (Watanabe et al., 2007). We have added a section in the Discussion addressing this point.

Reviewer #4:

This manuscript proposes a mechanism for the polarization of the fate determinant Miranda in Drosophila neuroblasts. The current model for this process is that Miranda directly interacts with the plasma membrane and phosphorylation of the membrane-interacting peptide by aPKC displaces Miranda into the cytoplasm. The model proposed in the work under consideration is that Miranda polarization occurs via different mechanisms in interphase and mitosis, and the mitosic mechanism involves a "basal affinity zone" that somehow involves the cytoskeleton. In general, I feel that the authors have overinterpreted their data, which relies on drug treatments and in some cases, used flawed logic to draw their conclusions. Some of the main problems are summarized below:

• While the characteristics of interphase localization are interesting, they are not significant unless interphase localization is somehow important for Miranda function

Reviewer 1 has raised this point as well. New Figure 1—figure supplement 1 shows staining of larval NBs in interphase and mitosis for Mira and Pros. Pros, like Mira, is uniformly at the NB interphase cortex in the larval brain. We further stained mira mutant NBs in the same way and reveal that Pros is found in the nucleus. This is in agreement with previous findings in embryonic NBs (Matsuzaki et al., 1998). Therefore it is a strong possibility the PM-bound Mira in interphase is linked to regulating nuclear Pros (see main text: Introduction, Results and last paragraph of the Discussion).

• Drug treatments, such as LatA, lead to loss of Miranda cortical localization, which could result from drug-induced ectopic aPKC activity. The experiment the authors provide to discount this possibility ("anchoring vs. clearing") is not convincing for a number of reasons, including that Baz localization is used as a proxy for aPKC localization. The cortical localization of Miranda in LatA treated neuroblasts lacking aPKC function are consistent with a misinterpretation in this regard.

Following the advice of reviewers 3 and 4, new Figure 2—figure supplement 1, now includes the kymograph analysis of the effect of LatA on colcemid arrested NBs that express Mira::mCherry and functional aPKC::GFP (Besson et al., Curr Biol 2015, and new Video S4), which reveals the same picture as when using Baz::GFP as a marker for the Par complex. Plotting the cortical fluorescence intensities of both Mira and aPKC reporters, further revealed that Mira is lost 2.8min ± 1min (n=13), before aPKC levels rise above cytoplasmic levels at the basal pole. This information is found in the Results section.

Furthermore, LatA treatment results in apical to basal redistribution of aPKC. Nevertheless, Mira crescents fall off homogenously from the cortex. This is measured in Figure 2—figure supplement 1: the ratio of fluorescence intensities found in a region of interest (ROI) at the extremity of the crescent and in a ROI in the centre remains constant. Taking the effect of aPKC on uniform cortical Mira, which is driven by aPKC phosphorylation in an apical to basal manner at the onset of mitosis as reference, the effect of LatA on Mira is different, as we do not observe “ends-on” reduction of crescents. Both these findings support the interpretation that anchoring to F-actin is critical to keep Mira anchored basally rather than an indirect effect caused by cortical aPKC redistribution.

The new Video S2 further shows an aPKC::GFP Mira::mCherry expressing neuroblast, in which Mira dynamics in the transition from interphase to mitosis are similar if not identical to when Mira::mCherry is expressed alongside Baz::GFP. While the cytoplasmic levels of Baz::GFP and aPKC::GFP may vary, we have no evidence for differences in the cortical distribution of aPKC and Baz once NBs enter mitosis.

• The authors over interpret the S96D mutant, which only partially abrogates plasma membrane binding. They state in the text that the localization of mutant is the same as WT but the videos they provided do not support this claim.

Following the advise of this reviewer, we revised this part of the manuscript. We now include the S96A mutant, which we find not to be cleared at prophase onset, but to transiently accumulate in the wrong place, i.e. apically before NEB, after which it localizes uniformly in an actin-independent manner in mitosis (new Figure 4A,B). These results show that mutating only one of the identified aPKC phosphorylation sites is sufficient to prevent cytoplasmic displacement of Mira at the onset of mitosis.

When this site is mutated to Aspartic Acid (S96D), unlike wild-type Mira, S96D localizes predominantly to cortical microtubules. However, in mitosis S96D, is always able to localize asymmetrically, which we fully agree may occur at reduced levels.

Following the advice of the reviewer we have revamped the paragraph and figure showing the S96D results, emphasizing the point that interphase and mitotic localization are differently affected by the S96D mutant, supporting the point that Mira binds to the plasma membrane in.

• Even if the previous problems are overlooked, the authors have not gone far enough to characterize the new interactions that they are proposing take place. The phenotype they observe when inactivating the cytoskeleton could arise from very indirect effects, yet they are proposing direct interactions with a "basal affinity zone". I do not believe that they have presented data supporting their claim that the zone is only basal, or that tell us anything about the interactions that would occur in the zone.

Following the advice of the reviewer we do not refer to basal affinity zone any more and more generally discuss the role of actomyosin remodeling.

• The text is poorly-written and convoluted. It would be very difficult to follow for someone outside the field.

We have radically revised the manuscript, summary, Results and Discussion, and removed data (Lgl3A overexpression in combination with LatA treatment) to improve focus and clarity.

[Editors' note: the author responses to the re-review follow.]

Reviewer #1:

[…]

However, the major weakness of this manuscript in writing remains. Aside from multiple clear grammatical mistakes, the lack of contexts when they describe their data really interferes readers' understanding (the most important question in the field that is addressed by each specific experiment is not explained). I listed up several specific places, where the way it's written confused me a lot. I recommend the authors to go through extensive editing focusing on 'reader friendly writing'. Given the history of this submission, we must insist that the next version will be the last. Please be certain that you have had others read and proof this so that the Board and reviewers will have an easier time deciding if you have made your essential points. Failing that, the paper will not be returned for another rounds of revisions.

We have revised the way the manuscript is written substantially following the reviewers advice. We have consulted a professional editing service to ensure grammatical correctness and to improve reader friendly writing. In addition, we had several scientific colleagues in the division and the school of life sciences distant to the field to read and comment on our revised version. We especially put an emphasis on explaining the rationale behind each experiment and hope that this has improved the clarity of the paper.

Reviewer #2:

This manuscript contains a series of well-controlled and careful experiments that present evidence that Miranda localises by different mechanisms in interphase and mitosis. The observations that Miranda is excluded from the cortex by aPKC and retained in a basal crescent by the acto-myosin cortex are not novel, but most previous work has presented these as competing models to explain how Miranda is asymmetrically localised. Thus, the main novelty in this manuscript is that both may be true, but at different time points during the cell cycle. The experiments are carefully quantified and informative, but more care is needed in the interpretations of the results. The authors should discuss alternative explanations for their data, even if they are not their preferred model. Another obvious weakness is the use of Y-27632 as an aPKC inhibitor when it also affects myosin activity through ROCK, although this experiment is still valuable and provides a useful comparison with previous work. The manuscript is an improvement over the previous version, but needs to be revised to include a more comprehensive discussion of the meaning of the results that considers alternative interpretations and that is more accessible to a general audience.

Following the advice of this and the other reviewers, we have completely restructured the Results section detailing the Y-27632 experiments. We have split this paragraph into two sections to improve clarity, especially to highlight the fact that while high doses of Y-27632 (200µM) inhibit aPKC, lower doses (25µM) of Y-27632 seem to affect Mira crescent size, which is likely to occur independently of aPKC inhibition. We have taken a new angle at the discussion following the advice discussing alternative interpretations to explain Mira localization. This includes discussing the role of known inhibitors (e.g. Lgl, Betschinger et al., 2003, Atwood and Prehoda, 2009), changes in the spatial regulation of aPKC based on recent observations made in the C.elegans zygote (Rodrigues et al., 2017, Wang et al., 2017) as well as general changes in Mira protein upon nuclear envelop breakdown such as dimerization (Yousef et al., 2008, Jia et al., 2015). We have expanded further on how we think actomyosin might affect Mira localization, to explain our findings better. We feel that the new angle the discussion now takes improves clarity and hope that it explains better the general relevance of our results to a broader readership.

Specific comments:

1) FRAP stands for Fluorescence Recovery after Photobleaching and not Fluorescence Redistribution.

We have change this as advised.

2) Intriguingly, such enlarged crescents were LatA and ML-7 sensitive, both indicators that enlarged Mira crescents were not due to lack of phosphorylation of Mira by aPKC. This needs more explanation and perhaps some discussion as to whether the expansion depends on aPKC phosphorylation of another substrate.

We have explained further this section of the Results, which now includes additional clarification: “Furthermore, the enlarged Mira crescents resulting from Y-27632 addition to cycling NBs were sensitive to LatA and ML-7 treatment, as were normal Mira crescents in controls. This suggests that also under this condition, actomyosin is important to retain Mira at the cortex mitosis, even when the size of the crescents is enlarged (Figure 5D; see Figure 5E for Mira crescent size quantification under the different conditions).”

Given that we do not have any data of other substrates of aPKC that are linked to actomyosin regulation, although a tantalizing possibility, we have decided not to expand the discussion on how that might affect altered Mira crescent size. However, aPKC is likely to contribute after NEB to Mira asymmetry by continuing to remove it from the plasma membrane apically. This was revealed with higher doses of Y-27632, which phenocopies Mira localization defects seen in apkc mutants, but given the nonspecific nature of Y-27632 this cannot be precisely determined in this way. This is stated in the Results section.

Reviewer #3:

This revision from Hannaford et al. remains seriously flawed and I do not recommend it for publication. First, many of the author's conclusions are contradicted by their own data. For example, a key conclusion is that Miranda is cleared from the cortex at prophase. It's difficult or impossible to make this assessment from the data the authors show because of Miranda signal from attached GMCs. Video S2 (included for a different reason) shows a beautiful division without any visible GMCs. This video supports a conclusion opposite of the author's: Miranda remains on the cortex throughout the cell cycle including in prophase.

We thank the reviewer for pointing this out. The key starting observation for our study relates indeed to the cortical dynamics of Mira as the NBs transits from interphase into mitosis. We find that just before NEB Mira has been removed from certain regions of the cortex, to which Mira returns after NEB to form the basal crescent. However, we did not and still do not claim that Mira is completely lost from the basal cortex just before NEB. We now made sure that this result and conclusion is clear in revised relevant sections. What we found is that there is a difference in quality and quantity of Mira localization at the basal NB pole just before and after NEB. This is quantified in Figure 1B’ and visible in Video S1 (compare frame 00:54 and 01:00), Video S2 (compare frame 01:04 and 01:12) as well as Video S8 (compare frame 00:39 and 00:45). If anything, Mira signal from GMCs would mask the extent to which Mira was removed from the NB cortex, bearing the risk to under- rather than overestimate the extent to which Mira is removed from the basal cortex just before nuclear envelope breakdown.

Another example: if Miranda is cleared all at once from the cortex upon LatA treatment, as the authors conclude, then why does the kymograph show the Miranda crescent shrinking? This effect is obfuscated by compressing the transition over a small number of pixels but Video S4 shows it beautifully. Scrubbing video S4 back and forth around the time that Miranda disappears clearly shows the crescent growing and shrinking, not disappearing all at once.

Fluorescent levels of Mira are low since we use the endogenous promoter to drive Mira::mCherry expression. Therefore, variations in signal-to-noise may lead to variations in detecting Mira between samples. To illustrate this point, we provide now additionally Video S5. In this experiment Baz::GFP and Mira::mCherry co-expressing colcemid-arrested metaphase NBs were treated with LatA and the fluorescence levels on the cortex are plotted for both markers. Like Video S4 this video also shows that Mira becomes cytoplasmic before changes in Baz (or aPKC) can be detected at the cortex. Importantly, in new Video S5, Mira crescents seem to broaden before homogenously fainting at the basal cortex (compare frame 00:00 to 05:45). In any case, we have not based our conclusion on individual samples, but measured the response of Mira cortical fluorescence to LatA treatment across several samples, the analysis of which was done on the raw datasets. In the kymographs and the fluorescence intensity measurements at the margins of Mira crescents and in the center are shown in Figure 2—figure supplement 1. Therefore, taken the behaviour of Mira on the interphase cortex in response to mitotic entry as a reference, when Mira is cleared in an apico-basal manner, our quantitative data indicates that Mira falls off homogenously from the cortex supporting a role for actomyosin in retaining it basally after nuclear envelope breakdown.

Another example: The MLCK inhibitor ML-7 causes cortical Miranda to be lost in metaphase, but the phenotype could be non-specific (SqhEE expression only causes delay of the phenotype), and even if it isn't this one result doesn't prove that Miranda loss is due to a direct effect. Thus, close inspection of the author's data does not support clearing of Miranda at prophase or loss of anchoring by cytoskeletal poisons (in fact, for two experiments the opposite conclusion is supported).

We acknowledge that small molecule inhibitors can have off target effects, but these can be mitigated to some extent by using several that act in different ways to perturb a pathway (i.e. LatA and ML7) and assessing if both have the same effect. Rescuing the effect Y-27632 or ML-7 by expressing a phosphomimetic version of the substrate is a further well-established approach (e.g. Das and Storey, 2014, Barros et al., 2003 and Bertet et al., 2004) to test the specificity of effects. Why we only see a significant delay with this approach and not complete suppression, might be due to the levels of the phosphomimetic protein supplied, which are difficult to quantitatively control. We agree that this experiment does not show a direct effect on Mira, neither does it rule that out, it could well be caused indirectly through changes in the actomyosin network. We have elaborated on how we think actomyosin activity might contribute to stabilize Mira localization in the discussion, which accommodate also indirect effects, such as changes in the local tension landscape. This can be found in the new Discussion.

The paper contains a fair amount of data, such as the difference in photobleaching dynamics of Miranda in interphase and mitosis, and the examination of Y-27632 treated neuroblasts, which appear to be superfluous without a more clear explanation.

The photobleaching experiments are important as they lend strong support to the idea, that the way Mira behaves at the cortex, i.e. its turnover in interphase and mitosis are different. The results obtained by measuring cortical Mira dynamics by FRAP, therefore support our idea that Mira retention at the cortex has at least two states: BH motif mediate retention at the plasma membrane (interphase) and actomyosin-dependent Mira retention/stabilization at the basal cortex after nuclear envelope breakdown. We have revised this section and emphasized this rational before detailing the Results (“Second, the turnover of Mira at the cortex when bound to the PM in interphase and when interacting with F-actin after NEB should be different.”).

As stated in the replies to the other reviewers comments, we have revised the section describing the Y-2763 experiments to improve clarity and provide more information and elaborated on the interpretation of these results in the Discussion as mentioned above.

Considering that the paper's foundational conclusions are flawed, it is not surprising that the strongest tests of the resulting model (that Miranda is cleared from the cortex by aPKC phosphorylation of its BH motif during prophase and anchored by actomyosin during metaphase) fail miserably. These tests come in three experiments. First, Miranda∆BH – if the BH is only required for interphase localization, then Miranda∆BH should be cytoplasmic and interphase and polarized during metaphase. However, the authors find that it's cytoplasmic in metaphase too. Expression of constitutively active aPKC (aPKC ∆N) is also a great test of the model. If aPKC activity is just required to clear interphase cortical localization, constitutively active aPKC should result in cytoplasmic Miranda in interphase and polarized Miranda in metaphase. The result again contradicts the model's prediction: Miranda is cytoplasmic in both.

We would like to point out, that the mechanism that we are proposing is compatible with a potential role for the BH motif in interphase and in mitosis. The point that we are trying to make is that while plasma membrane binding is sufficient to retain Mira at the membrane in interphase, the BH motif dependent mechanism plus an additional actomyosin dependent process is required to stabilize or retain Mira at the cortex after nuclear envelope breakdown. This view is consistent with propositions found in the literature (e.g. Bailey and Prehoda, 2015).

An interesting side observation is further that when the BH motif is deleted, Mira is not found on mitotic microtubules (Figure 3A) as wild type Mira protein typically is in other mutant contexts (Albertson and Doe, 2003; Barros et al., 2003; Rolls et al., 2003; Slack et al., 2007). This may be irrelevant, but it may suggest that deletion of the BH motif affects other aspects of Mira behaviour and not only plasma membrane binding. Prompted by the reviewer’s comment, we have added this to the Discussion.

Regarding aPKC∆N overexpression, we agree that this could be a great test, however, it is very difficult to predict the precise consequences of overexpressing a deregulated kinase. Nonetheless, we found that Mira localization is telophase rescued (Peng et al., 2000) when aPKC∆N is overexpressed. We observed this by antibody staining against Mira and in living NBs monitoring Mira::mCherry localization. We have added this information into Figure 4—figure supplement 1A and A’. These results suggest that loss of the BH motif and the effect of expressing constitutively active aPKC are not exactly the same as Mira retains the ability to engage with the cortex at least in telophase in the aPKC∆N context. aPKC∆N overexpression may bring about additional changes in the NBs cortex that prevent normal Mira polarization making it difficult to separate direct and indirect effects, which may be an alternative explanation why Mira is not localized directly after NEB in this context. We have added a section to the Discussion regarding this point.

Remarkably the authors don't even discuss the metaphase result in the paper! Finally, Miranda S96D (reduces but does not ablate BH interactions with the membrane) should have reduced cortical interactions in interphase but polarized normally in metaphase. Here again the result, reduced cortical signal in both interphase and metaphase, is not consistent with the author's proposed model.

Remarkably, these results directly contradict the author's proposed model but are precisely as predicted from a simpler model in which aPKC regulated BH interactions with the membrane control Miranda cortical association throughout the cell cycle. Of course the more complex model could be saved by making it even more complex – perhaps the BH is also involved in the actomyosin interaction, perhaps aPKC also regulates the actomyosin anchor, perhaps a single point mutant that partially disrupts the BH lipid interaction also partially disrupts the actomyosin interaction – and the authors make a half-hearted attempt to do so (at least for the first experiment), but more data would need to be provided to support a more complex model (e.g. how aPKC might regulate the anchor).

We favor an alternative interpretation to the reviewer’s suggestion. The S96D mutation, by possibly bringing a negative charge to the BH motif, weakens the interaction the mutant protein with the plasma membrane. This has strong effects on S96D mutant localization in interphase, as it is cytoplasmic in the large majority of NBs analysed (Figure 3A). This argues that in interphase Mira critically depends on plasma membrane binding for its cortical retention. In contrast, in all NBs analysed the S96D mutant after NEB shows cortical localization. This is perhaps weakened compared to controls, but occurs even with an asymmetric bias towards the basal pole. In this view, an additional mechanism that we propose to be mediated by actomyosin-dependent processes, overcomes the requirement for BH motif mediate maintenance of Mira after nuclear envelope breakdown. Therefore, these experiments are rather informative and support our idea that BH motif mediated retention of Mira after NEB is not the only mechanism that keeps Mira at the basal cortex.

I believe the problems outlined above represent deep and fundamental flaws and therefore preclude publication in its current form. I also note that the authors removed the model figure from the paper. In my opinion this is moving in the wrong direction and encourage the authors to revise the paper so it is more clear, not less.

We appreciate that the reviewer continues to provide constructive feedback, despite his/her contrary views on the matter. We agree that a graphical summary might improve reader friendliness and have added a model in new Figure 6. We focus our discussion also around the BH motif and have elaborated on our views, how actomyosin activity may contribute to Mira basal retention after nuclear envelope breakdown, although this remains to be determined in detail. This can be found in the Discussion section.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The core discovery of this manuscript 'Miranda localization is regulated differently during interphase and mitosis' is interesting and worth publishing. Knowing that Miranda localization switches the mode (plasma membrane-bound in interphase, and anchored via actin cytoskeleton in mitosis) is an important stepping stone to fully understand how Mira localization is regulated to achieve asymmetric neuroblast division.

The reviewers had a lengthy discussion as to how to proceed with this manuscript. First, we agreed that the core discovery described above is worth publishing if accurately stated: that Miranda localization have two modes in interphase and mitosis. However, while several reviewers agreed that the data in the manuscript sufficiently supported their model, we were not able to reach a consensus on this point. In the end, the reviewers agreed to recommend publication provided that the authors clearly and accurately state their results in a manner that allows critical assessment of all the data, such that all the readers can judge on their own.

We mention the differences of Mira’s interaction with the cortex in the Abstract:

“We reveal a step-wise polarization of Miranda in larval neuroblasts and find that Miranda’s dynamics and cortical association are differently regulated between interphase and mitosis. In interphase Miranda binds to the plasma membrane. Then, before nuclear envelope breakdown, Miranda is phosphorylated by aPKC and displaced into the cytoplasm. This clearance is necessary for the subsequent establishment of asymmetric Miranda localization. After nuclear envelope breakdown, actomyosin activity is required to maintain Miranda asymmetry.”

We state that Miranda localization has two modes in interphase and mitosis in the last paragraph of the Introduction that summarizes our results:

“We reveal that Mira uses two modes to interact with the cortex: in interphase, to retain Mira uniformly at the cortex direct interaction of Mira’s BH motif with phospholipids of the PM are necessary and likely sufficient. This interaction is inhibited by aPKC-dependent phosphorylation of the BH motif at prophase. After nuclear envelope breakdown Mira requires BH motif and actomyosin dependent processes for asymmetric retention at the cortex. Therefore, we propose that Mira binds to the PM in interphase and to the actomyosin cortex in mitosis, both of which appear BH motif dependent.”

This is also stated in the third paragraph of the Discussion:

“We propose that Mira has two different modes by which it can be retained at the cortex (Figure 6). In interphase, Mira localizes uniformly to the cortex via direct interactions with the PM for which its BH motif is necessary and likely to be sufficient and which occurs independently of an intact F-actin cortex (Figure 2A). After NEB, Mira still relies on the BH motif to localize in a basal crescent, but at this stage of the cell cycle actomyosin-dependent processes retain Mira basally (Figure 3A, Figure 2A-D). The transition between these localizations depends on phosphorylation by aPKC (Figure 3A, Figure 4A).”

All reviewers noted concerns regarding the accuracy of descriptions. More specific comments will follow below, but the collective result of inaccurate statements/descriptions is that 1) the real contribution of the manuscript is blurred,

We have added a new summary statement at the end of the Introduction, to better summarize our contribution:

“We reveal that Mira uses two modes to interact with the cortex: in interphase, to retain Mira uniformly at the cortex direct interaction of Mira’s BH motif with phospholipids of the PM are necessary and likely sufficient. This interaction is inhibited by aPKC-dependent phosphorylation of the BH motif at prophase. After nuclear envelope breakdown Mira requires BH motif and actomyosin dependent processes for asymmetric retention at the cortex. Therefore, we propose that Mira binds to the PM in interphase and to the actomyosin cortex in mitosis, both of which appear BH motif dependent.”

We also have changed the beginning of the Discussion to better summarize our approach and contribution:

“To address this apparent inconsistency, we reassessed in vivo the relative contribution of aPKC and actomyosin throughout the cell cycle analysing Mira localization using endogenously expressed reporters in living NBs. This has allowed us to resolve this problem as we find that asymmetric Mira localization is established stepwise and involves both aPKC-dependent phosphorylation and actomyosin-dependent anchoring, which are required at different time points in mitosis.

We propose that Mira has two different modes by which it can be retained at the cortex (Figure 6). In interphase, Mira localizes uniformly to the cortex via direct interactions with the PM for which its BH motif is necessary and likely to be sufficient and which occurs independently of an intact F-actin cortex (Figure 2A). After NEB, Mira still relies on the BH motif to localize in a basal crescent, but at this stage of the cell cycle it might be required to mediate actomyosin-dependent basal retention of Mira (Figure 3A, Figure 2A-D). The transition between these localizations depends on phosphorylation by aPKC (Figure 3A, Figure 4A).”

2) The manuscript reads as if the authors are proposing an alternative model which they claim to be better than the existing model, but the new model also leaves many unexplained observations, making readers wonder whether the new model is a real improvement or only confusing the field.

Therefore, we ask you to edit the text and submit the revised version. We decided to allow you one more round of revision because it will only require textual revision. With that said, all the reviewers expressed strong concerns that the current writing is not accurate enough, and if it is not fully taken care of in the next round, we must reject your manuscript, and no further revision will be allowed. In revising, we ask you to ensure that the revised main text is self-sufficient in conveying all the messages accurately and clearly, and not to utilize the rebuttal/response letter as a platform to explain things that are not entirely consistent with the main text.

We hope that the changes detailed above and below satisfactorily address your concerns.

Guidelines for revision:

Each Results section should clearly indicate how the data contributes (or doesn't) to the model.

We have incorporated these changes into the main manuscript. We first introduced a statement to make clear what we test our results:

“According to this model, Mira retention at the cortex is primarily mediated through direct interaction with the PM mediated by its BH motif.”

The first section ends with the proposition that – based on the differences in intensities, Mira binding to the cortex might have occur through different binding modes:

“In conclusion, Miranda transitions from a uniformly cortical localization with low intensity levels in interphase, to a basal localization with high intensity levels in metaphase (Figure 1B, B’, C). These cell-cycle dependent differences in cortical Mira intensities prompted the idea that Mira might use different modes of binding to the cortex in interphase versus mitosis. Therefore, we assayed for potential differences in cortical binding of Mira in interphase versus mitosis to address whether Mira is retained at the cortex primarily by BH motif interaction with the PM, or whether other modes of cortical retention contribute.”

The second section describing the results of LatA and ML-7 treatment, is in support of the model that Mira uses different modes to bind to the cortex in interphase and in mitosis. This can be found here):

“In summary, these results support the notion that Mira interacts differently with the cortex in interphase and in mitosis since F-actin and myosin activity contribute to establish and maintain asymmetric Mira crescents at the basal cortex following NEB, but they are not essential for uniform cortical localization of Mira in interphase nor for Mira clearance during prophase.”

The third section addressing the role of the BH motif describes data that support the model that Mira only uses one mode to bind to the cortex, regardless of the cell cycle stage, which is BH motif mediated interactions with the PM. This is stated here:

“These findings support the idea that, in interphase, the BH motif is necessary and likely to be sufficient to mediate interactions with phospholipids of the PM leading to uniform cortical localization of Mira. These findings also show that phospho-regulation of the BH motif affects Mira localization in both phases of the cell cycle. Therefore, these observations support the model that Mira uses only one mode, BH motif mediated PM interactions, for cortical association throughout the cell cycle. However, this model does not readily explain differences in the response of Mira to LatA and ML-7 in interphase versus mitosis (Figure 2).”

The fourth section describing the FRAP data supports the model that Mira has two different modes to interact with the cortex. This is stated:

“In conclusion, these results show that, in unperturbed NBs, Mira turnover at the PM in interphase and at the basal cortex in mitosis are different, supporting the notion that Mira has different binding modes in interphase versus mitosis. Furthermore, in apkc mutant NBs, instead of being cleared, Mira may persist throughout mitosis with the same actin-insensitive uniform localization, and similar turnover, as in interphase.”

The section detailing the effects of low concentrations of Y-27632 on Mira localization is compatible with Mira using two different modes for cortical retention in interphase and in mitosis. This statement can be found here:

“In conclusion, while Y-27632 at higher concentration, can indeed mimic the effect of apkc mutation on Mira, these results suggest that when NBs polarize in the presence of low concentrations of Y-27632, Mira crescent size is affected, which is likely to occur independently of aPKC inhibition. These results suggest that Mira cortical retention has different mechanisms of regulation in interphase and in mitosis. They also hint at an additional, Y-27632 sensitive layer of regulation controlling basal Mira crescent size.”

We also ask that in the discussion the authors recognize that there are observations they have made that are inconsistent with this model. They can do their best to try and explain them away, but at least readers will be able to more readily appreciate these inconsistencies and judge the author's explanations for themselves. Please ensure that the revised manuscript is self-sufficient in conveying these points, without the need of relying on response letter.

We have modified the Discussion highlighting the data we and other obtained that are consistent with the BH motif mediated PM interactions being the primary mode of Mira cortical retention throughout the cell cycle. This can be found in the Discussion:

“We observe that deletion of the BH motif as well as overexpression of aPKCΔN disrupt cortical localization of Mira in interphase and mitosis (Figure 3A and Figure 3—figure supplement 1) and that the phosphomimetic S96D mutation reduces Mira localization in interphase as well as in mitosis (Figure 3A). These findings argue for the model that throughout the cell cycle Mira cortical association depends solely on BH motif mediated interaction with the PM, that is negatively regulated by locally controlled aPKC phosphorylation (Atwood et al., 2009; Bailey et al., 2015). “

and: “What could be the role of F-Actin for Mira localization in this model? F-Actin clearly contributes to aPKC regulation of Miranda localization by restricting the localization of the Par complex to the apical pole as LatA addition changes the distribution of aPKC and Baz (Figure 2A).”

as well as: “We propose that the BH motif may mediate Mira’s interaction with actomyosin, which remains to be tested.”

The reviewers consider that the most reasonable model to be put forward in this manuscript is "either Miranda BH-phospholipid or BH-actomyosin interactions are sufficient for cortical localization. In interphase BH-phospholipid interactions mediate uniform cortical association but early in prophase they start becoming inactivated by aPKC in an apical to basal fashion. By metaphase all BH-phospholipid interactions are inactivated and BH-actomyosin interactions take over (BH-phospholipid interaction might have supportive role in mitosis, too).

We have followed the advice of the reviewers and discuss this possibility. This can be found in the Discussion:

“We propose that Mira has two different modes by which it can be retained at the cortex (Figure 6). In interphase, Mira localizes uniformly to the cortex via direct interactions with the PM for which its BH motif is necessary and likely to be sufficient and which occurs independently of an intact F-actin cortex (Figure 2A). After NEB, Mira still relies on the BH motif to localize in a basal crescent, but at this stage of the cell cycle it might be required to mediated actomyosin-dependent processes Mira basal retention (Figure 3A, Figure 2A-D). The transition between these localizations depends on phosphorylation by aPKC (Figure 3A, Figure 4A).”

and in the new Figure 6 and in its legend:

Figure 6. Model. Mira associates with the cortex using two different modes, which characteristics are detailed in the bottom row. During interphase, Mira directly binds to the phospholipids of the PM via its BH motif (black double arrow). During prophase, aPKC-dependent phosphorylation of this motif abolishes this interaction, resulting in the progressive clearance of Mira from the cortex, in an apical-to-basal manner driven Mira into the cytoplasm. This clearance in prophase is necessary for Mira to associate with the basal cortex after NEB, via Actomyosin-dependent retention. Both the precise phosphoregulation and molecular characteristics of this mode remain to be determined. The BH motif, also required at this step, may directly or indirectly mediate interactions between Mira and actomyosin (green double arrow). PM interactions via its BH motif (black double arrow) may still contribute, but are not sufficient to mediate Mira basal retention after NEB.”

Overall, the reviewers agreed that there are many observations that do not exactly fit to the authors' model (e.g. overexpression of aPKC-deltaN). Some experiments were suggested to test authors' model, which resulted in observations that are inconsistent with the authors model. Each time, the authors provided 'possible explanations why the test did not support their model', yet remained that 'the model is still correct'. The authors should put more effort into explaining these inconsistent results (instead of 'explaining away' inconvenient results, trying to reach better explanation that makes sense as a whole). The reviewers do not expect the authors to figure out the mechanism of Miranda localization entirely. The reviewers appreciate that knowing that Miranda localization likely has two modes is sufficiently important progress, but writing is not conveying that this is the core message of the manuscript, and instead it claims more than the experimental data can support.

As mentioned above we have revised the discussion according to the reviewer’s advice.

We have modified the discussion highlighting the data we and other obtained that are consistent with the BH motif being the primary mode of Mira cortical retention throughout the cell cycle. This can be found in the Discussion:

“We observe that deletion of the BH motif as well as overexpression of aPKCΔN disrupt cortical localization of Mira in interphase and mitosis (Figure 3A and Figure 3—figure supplement 1) and that the phosphomimetic S96D mutation reduces Mira localization in interphase as well as in mitosis (Figure 3A). These findings by themselves argue for the model that throughout the cell cycle Mira cortical association depends solely on BH motif mediated interaction with the PM, that is negatively regulated by locally controlled aPKC phosphorylation (Atwood et al., 2009; Bailey et al., 2015). “

and:

“What could be the role of F-Actin for Mira localization in this model? F-Actin clearly contributes to aPKC regulation of Miranda localization by restricting the localization of the Par complex to the apical pole as LatA addition changes the distribution of aPKC and Baz (Figure 2A).”

As an example of inaccurate statement, in the first section of the results, the authors start out by stating Miranda is 'cleared' and 'reappears' in mitosis, clearly leaving the impression that the previous model of expanding apical Par complex gradually displacing Miranda is wrong. However, as the video shows, Miranda 'clearance' occurs from the apical to basal. Although the authors explain that this clearing happens from apical to basal, but in subsequent sections, they do not come back to this fact and stick to the expression of 'clearance' and 'reappearance'. Here, the emphasis should be the fact that mitotic Miranda crescent is much stronger, prompting the investigation of the mechanism by which Miranda anchoring is promoted during mitosis (which they show to be actin dependent).

We follow the advice of the reviewers and use differences in Mira cortical intensities as an argument to further investigate if differences in Mira binding to the cortex in interphase and mitosis exist. This can be found in the Results section:

“In conclusion, Miranda transitions from a uniformly cortical localization with low intensity levels in interphase, to a basal localization with high intensity levels in metaphase (Figure 1B, B’, C). These cell-cycle dependent differences in cortical Mira intensities prompted the idea that Mira might use different modes of binding to the cortex in interphase versus mitosis. Therefore, we assayed for potential differences in cortical binding of Mira in interphase versus mitosis.”

The weakest explanations in the current manuscript are the following:

1) If the authors' model is entirely correct, aPKC-deltaN expression should result in delocalization of Miranda in interphase, but normal, crescent localization in mitosis. But the actual observation is that aPKC-deltaN results in the delocalization of Miranda in interphase and mitosis. In the main text, the authors do not mention that Miranda fails to localize even in mitosis upon aPKC-deltaN expression, and conclude that their data is consistent with their model. In explaining this inconsistency, they state 'it is very difficult to predict the precise consequences of overexpressing a deregulated kinase'. A better explanation is required for the authors to be able to propose that their new model fits better than the existing model with all experimental results.

We mention now in the main text that aPKC-deltaN results in the delocalization of Miranda in interphase and mitosis. This can be found in the Results section:

“Finally, ectopic activation of aPKC by overexpression of constitutively active aPKCΔN (Betschinger et al., 2003b) also prevented Mira cortical localization in interphase and most of mitosis.”

We also provide a better explanation. i.e. that control of the levels of Mira phosphorylation might be important, which can be found in the Discussion:

“We observe that deletion of the BH motif as well as overexpression of aPKCΔN disrupt cortical localization of Mira in interphase and mitosis (Figure 3A and Figure 3—figure supplement 1) and that the phosphomimetic S96D mutation reduces Mira localization in interphase as well as in mitosis (Figure 3A). These findings argue for the model that throughout the cell cycle Mira cortical association depends solely on BH motif mediated interaction with the PM, that is negatively regulated by locally controlled aPKC phosphorylation (Atwood et al., 2009; Bailey et al., 2015).”

MiraS96D: the model predicts that MiraS96D should be cortically polarized in metaphase but it's partially cytoplasmic (S96D only partially destabilizes the BH-phospholipid interaction). The author's explanation is that cortical localization is affected more in interphase, which is not a thorough discussion.

We discuss now two possibilities. The S96D results are consistent with the model that BH motif mediated interactions with the plasma membrane retaining Mira at the cortex in interphase and mitosis, as Mira localization is reduced in both cases. This can be found in the Discussion:

“We observe that deletion of the BH motif as well as overexpression of aPKCΔN disrupt cortical localization of Mira in interphase and mitosis (Figure 3A and Figure 3—figure supplement 1) and that the phosphomimetic S96D mutation reduces Mira localization in interphase as well as in mitosis (Figure 3A). These findings argue for the model that throughout the cell cycle Mira cortical association depends solely on BH motif mediated interaction with the PM, that is negatively regulated by locally controlled aPKC phosphorylation (Atwood et al., 2009; Bailey et al., 2015).”

However, this does not take into account the observation that the S96D mutation affects interphase localization of Mira stronger than its mitotic localization (In ~70% of interphase NBs S96D mutant Mira does not localize to the cortex, while S96D mutant Mira always localizes to the cortex of mitotic NBs). These results argue that the effect of the Aspartate on Mira localization is less strong in mitosis, which is the basis for the interpretation that basal Mira is stabilized by additional interactions after NEB.

This can be found in the Discussion:

“The phosphomimetic S96D mutation strongly disrupts uniform localization to the PM in interphase, but localizes at the basal cortex in mitosis, albeit at reduced levels (Figure 3A). This is consistent with the existence of Mira stabilizing interactions in mitosis, that are not present in interphase, which reduce the effect of a negative charge provided by the Aspartate in the phosphomimetic mutant on cortical Mira localization.”

Unfortunately, readers will be wondering why Miranda5D (which completely abolishes BH-phospholipid interactions) wasn't tested because it would very easily resolve the explanations provided by the author for both of these inconsistencies (especially because the explanations are so poor).

We agree that this would be interesting to test. In this context, it would be important to determine which sites in Mira are phosphorylated in vivo, too (e.g. pulling down endogenous Mira from control and aPKC mutant samples followed by mass spectrometry approaches).

2) According to the authors' model, Mira-deltaBH should be cytoplasmic in interphase, yet should localize to the basal crescent in mitosis. However, they find that Mira-deltaBH fails to localize to the crescent even in mitosis. In the main text, they simply conclude that 'this is consistent with Mira being localized to the cortex via interaction with plasma membrane'. Then, in response to the reviewers' comment that deltaBH shouldn't be cytoplasmic in mitosis, only in response letter, they point out that mitotic Miranda localization may require BH domain-mediated plasma membrane localization in addition to actomyosin. The authors should avoid any discrepancies between the main text and the response letter, and a cohesive story must be presented within the main text (and the observations that do not fit to the model must be clearly presented and acknowledged).

We have updated our model following the reviewer’s advice stating that the BH motif may mediate also interaction with actomyosin. This statement can be found in our summarizing statement of the Introduction:

“Therefore, we propose that Mira binds to the PM in interphase and to the actomyosin cortex in mitosis, both of which appear BH motif dependent.”

This statement is further found in the Discussion:

“A surprising finding is that the BH motif is essential for Mira localization in interphase and in mitosis. In mitosis, BH motif mediated PM binding is no longer sufficient to localize Miranda to the basal pole. This is indicated by the requirement of the actomyosin cytoskeleton after NEB (Figure 2C,D). However, the BH motif is still necessary for Mira localization after NEB. It is possible that the BH-phospholipid interactions still play a role in mitosis. Deletion of the BH motif could also cause more indirect effects. For example, Mira∆BH is not found on mitotic microtubules, where Mira is typically observed in conditions where it is unable to localize correctly (Albertson and Doe, 2003; Barros et al., 2003; Rolls et al., 2003; Slack et al., 2007). We propose that the BH motif may mediate Mira’s interaction with actomyosin, which this remains to be tested.”

Also, the reviewers do not agree that the data shows that Miranda disappears all at once upon LatA treatment. The revision should remove this argument and focus on the argument that Miranda disappears before aPKC arrives. Of course, we do not ask the authors to figure out everything about Mira localization, but they should explain 'why a particular observation betrays their prediction based on the model' in a cohesive manner, and more sincerely (i.e. do not bury important discussions within the response letter, which become inconsistent when placed in the main text).

We have revised this Results section and removed the data, that we feel shows that Mira levels decline at similar rates at the extremities of the crescent and in its center. We now use the timing of changes in Mira localization versus changes in Par complex distribution in response to LatA as an argument as advised by the reviewers. This can be found in the Results section:

“However, LatA treatment after NEB also led to the redistribution of Baz/Par3 and aPKC to the entire NB periphery. Therefore, the observed effect on Mira could be indirect, caused by changes in aPKC localization when F-actin is compromised. Assuming that aPKC activity is restricted to the cortex (Atwood et al., 2007; Rodriguez et al., 2017) we sought to distinguish between direct and indirect effects on Mira by determining if Mira loss preceded (indicative of direct effect of loss of actin) or followed (indicative of an indirect effect caused by changes in aPKC localization) changes in Par complex distribution in colcemid arrested NBs upon LatA treatment. We found that Mira loss preceded changes in cortical aPKC/Baz localization in response to LatA. This occurred about 2.8 ± 1min (n=13) before aPKC (Video S4) or Baz (Video S5) became detectable at the basal cortex (Figure 2—figure supplement 1).”

We would like to provide an example of how reviewers' discussion proceeded. One reviewer noted: "I prefer an alternative interpretation, in which all of the localization after NEB is actomyosin dependent (ML-7 abolishes the crescent) and that this is mediated by the BH domain.": this clearly suggests that this reviewer appreciated the authors' discovery but still required to introduce his/her own interpretation to support authors' model. In response, another reviewer asked, "Then why is Miranda cortical in metaphase neuroblasts treated with LatA and lacking aPKC function (Figure 4—figure supplement 1 panel B)? If localization after NEB is actomyosin dependent, then surely it shouldn't localize after treatment with LatA." In response to this question, the first reviewer responded, "My interpretation of this experiment is that it indicates that aPKC-dependent clearing is an essential prerequisite for the formation of the basal crescent. In the absence of aPKC, Miranda remains in the interphase state where it binds to phospholipids. I could be wrong, as this result is entirely consistent with the simple (existing) model, but the latter doesn't explain the ML-7, Y26732 and FRAP data. What I find harder to understand is why the basal crescent doesn't form in the presence of constitutively-active aPKC, which is why the results aren't clear cut. It seems that Miranda needs to be phosphorylated by aPKC before NEB to form the basal crescent, but that too much aPKC activity or activity after NEB inhibits this. It is a shame that the manuscript didn't really get to grips with these questions."

In our opinion, a manuscript should not leave this much room of interpretation to the readers, and the authors must clearly present their data (or if that is not sufficient, more data would be clearly required. But the reviewers are not asking more experiments here).

We feel that the advice to clarify what is tested (PM binding explains Mira cortical localization throughout the cell cycle or whether different binding modes exist) and to state at the end of each Results section as well as in the Discussion which data fit one primary binding mode and which support different modes has helped to address these points. (See above).

The Abstract states:This clearance is necessary for the subsequent establishment of asymmetric Miranda localization.”

We discuss the possibility that aPKC phosphorylation at prophase is important for basal Mira crescent formation after NEB:

“This aPKC-dependent step in prophase might be a prerequisite for basal crescent formation in metaphase. One possibility is that phosphorylation of the BH motif might potentiate Mira’s ability to engage with actomyosin for basal retention after NEB. Mira phosphorylation might need to be properly balanced and locally controlled to allow for Mira asymmetric localization. This could explain why the phosphomimetic S96D mutant, displays reduced basal localization in metaphase and that upon overexpression of aPKCΔN Mira does not form basal crescents in metaphase (Figure 3—figure supplement 1).”

Additionally, the model figure is currently extremely vague because it is hard to see what the authors are implying is taking place as far as Miranda-cortex interactions. Please revise the model figure to describe your model more clearly.

We have revised Figure 6 to include more graphical details and added a legend describing that Mira has two modes to bind to the cortex that are different in interphase and in mitosis. The new legend can be found here:

Figure 6. Model. Mira associates with the cortex using two different modes, which characteristics are detailed in the bottom row. During interphase, Mira directly binds to the phospholipids of the PM via its BH motif (black double arrow). During prophase, aPKC-dependent phosphorylation of this motif abolishes this interaction, resulting in the progressive clearance of Mira from the cortex, in an apical-to-basal manner driven Mira into the cytoplasm. This clearance in prophase is necessary for Mira to associate with the basal cortex after NEB, via Actomyosin-dependent retention. Both the precise phosphoregulation and molecular characteristics of this mode remain to be determined. The BH motif, also required at this step, may directly or indirectly mediate interactions between Mira and actomyosin (green double arrow). PM interactions via its BH motif (black double arrow) may still contribute, but are not sufficient to mediate Mira basal retention after NEB.”

https://doi.org/10.7554/eLife.29939.033

Article and author information

Author details

  1. Matthew Robert Hannaford

    Cell and Developmental Biology, School of Life Sciences, University of Dundee, Dundee, United Kingdom
    Contribution
    Conceptualization, Data curation, Formal analysis, Writing—original draft
    Competing interests
    No competing interests declared
    ORCID icon 0000-0001-7772-0450
  2. Anne Ramat

    Cell and Developmental Biology, School of Life Sciences, University of Dundee, Dundee, United Kingdom
    Contribution
    Data curation
    Competing interests
    No competing interests declared
    ORCID icon 0000-0002-2103-9583
  3. Nicolas Loyer

    Cell and Developmental Biology, School of Life Sciences, University of Dundee, Dundee, United Kingdom
    Contribution
    Data curation, Formal analysis, Writing—original draft
    Competing interests
    No competing interests declared
    ORCID icon 0000-0001-5010-2564
  4. Jens Januschke

    Cell and Developmental Biology, School of Life Sciences, University of Dundee, Dundee, United Kingdom
    Contribution
    Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Writing—original draft, Project administration, Writing—review and editing
    For correspondence
    j.januschke@dundee.ac.uk
    Competing interests
    No competing interests declared
    ORCID icon 0000-0001-8985-2717

Funding

Medical Research Council (G1000386/1)

  • Matthew Robert Hannaford

Medical Research Council (MR/J50046X/1)

  • Matthew Robert Hannaford

Medical Research Council (MR/K500896/1)

  • Matthew Robert Hannaford

Medical Research Council (MR/K501384/1)

  • Matthew Robert Hannaford

Wellcome Trust (100031Z/12/Z)

  • Anne Ramat
  • Nicolas Loyer
  • Jens Januschke

Royal Society (100031Z/12/Z)

  • Jens Januschke

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank C Doe, J Knoblich, F Schweisguth, D StJohnston, F Matsuzaki, C Gonzalez, J Skeath, A Wodarz, M Gho and the Kyoto and Bloomington stock centers for reagents and/or protocols. We thank A Müller, C Weijer, M Gonzalez-Gaitan, T Tanaka and K Storey for critical reading. MRH is supported by an MRC PhD studentship. We thank the Dundee imaging facility for excellent support. Work in JJ’s laboratory is supported by Wellcome and the Royal Society Sir Henry Dale fellowship 100031Z/12/Z. MRH is supported by an MRC studentship funded by these grants: G1000386/1, MR/J50046X/1, MR/K500896/1, MR/K501384/1. The tissue imaging facility is supported by the grant WT101468 from Wellcome.

Reviewing Editor

  1. Yukiko M Yamashita, Reviewing Editor, University of Michigan, Ann Arbor, HHMI, United States

Publication history

  1. Received: June 27, 2017
  2. Accepted: January 14, 2018
  3. Version of Record published: January 24, 2018 (version 1)

Copyright

© 2018, Hannaford 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

  • 1,037
    Page views
  • 123
    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)

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

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

Further reading

    1. Cell Biology
    2. Developmental Biology and Stem Cells
    Takuya Akiyama et al.
    Short Report Updated
    1. Developmental Biology and Stem Cells
    Laura Lleras Forero et al.
    Research Article Updated