A computational framework for predicting potential regulators of Fog signaling

Before we describe the computational methods we used to predict candidate Fog signaling regulators, we first built a comprehensive resource of Drosophila protein–protein interactions from the literature. This information can be represented as a protein–protein interactome, which describes a set of proteins and connections between those proteins. An interactome is mathematically formulated as a graph, where the nodes (proteins) are connected by edges (protein–protein interactions). We combined six existing Drosophila resources of protein–protein interaction data to build an interactome with 11,473 nodes and 233,054 undirected edges, ignoring self-loops and mapping all proteins to FlyBase identifiers (Table 1). Each edge in the interactome may be supported by multiple databases; while DroID contributes the largest amount of data, 10,654 edges (5%) are not supported by DroID and would be missed by a single-database analysis. Each edge in the interactome includes literature citations as well as sources of experimental or other evidence. In total, the interactions are supported by 21,065 pubmed-indexed literature citations and 58 PSI-MI evidence sources [24].

Table 1 Data sources used to build the fly interactome

Our computational goal was to identify candidates of NMII regulation by Fog signaling using this fly-specific interactome. We began by pooling three sets of proteins that are known to be relevant to Fog signaling [18], apical constriction (Gene Ontology 0003383), and gastrulation (Gene Ontology 0007369) to produce a list of 104 known protein regulators (called positives; Supplementary Table 2). Using the fly interactome, we applied three graph algorithms to identify protein candidates that are “near” the known protein regulators (Fig. 1). The Steiner Tree Approximation method aims to connect the positives with as few edges as possible; candidates are nodes that are used to connect the positives. The Paths to NMII method calculates the shortest path from each positive to Sqh; candidates are nodes that are on these paths. Finally, the Ranked Paths method calculates, for each unlabeled node, the shortest path from each positive to that node and assigns a score that is proportional to the average length of all shortest paths. All nodes in the Ranked Paths method are ranked; the nodes with a normalized score greater than 0.7 were selected as candidates (Fig. 1A). See the Methods for more details about the graph algorithms employed.

Fig. 1

Graph algorithms used to identify protein candidates. A Algorithms developed to identify candidate proteins from the interactome and the positive proteins. Brighter/lighter green indicates a higher Ranked Paths score. B Venn diagram of protein candidates from each method; see Supplementary Table 1 for candidate list

There is surprisingly little overlap among the candidates from the three methods (Fig. 1B). This is due to the differences in each method’s goal. For example, a Steiner tree will connect positives using the fewest possible non-positive nodes, so these predictions are often connected to many known positives (Supplementary Fig. 1). Calculating the paths from positives to NMII, on the other hand, may have multiple predictions on each path (though many of these paths overlap in practice, see Supplementary Fig. 2). Finally, the node score from the ranked paths approach indicates how close the node is to all positives, so the top predictions tend to be clustered together (Supplementary Fig. 3). A combined network of all predicted nodes, along with the adjacent positives, is shown in Supplementary Fig. 4.

Only four candidates, Cell division cycle 5 (Cdc5), Netrin-B (NetB), Spinophilin (Spn), and Ubiquitin-63E (Ubi-p63E), were identified by all three methods. NetB is not expressed in S2R + cells and thus was not pursued further [31]. Depletion of Spn is known to inhibit cellular contractility as assessed by a cellular contractility assay ([32], for details see below and Methods section) while depletion of Ubi-p63E and Cdc5 led to apoptosis preventing further assessment (unpublished data). Seventeen candidates were identified by two or more methods, and a total of 99 protein candidates were identified by any of the methods (Supplementary Table 2 and Supplementary Fig. 4). We excluded candidates that had already been shown to be a part of the Fog signaling pathway or were known Sqh interactors. As our contractility assay was performed in S2R + cells we further excluded candidates that were not expressed in these cells according to Harvard Fly RNAi database [17]. From this list of candidates 13 proteins were selected as promising for follow-up study (Table 2).

Table 2 Preliminary screen of computational identified candidatesPreliminary screen of computational putative targets

We initially screened all 13 candidates (Table 2) using a cellular contractility assay [16, 34], which capitalizes on the Fog-signaling pathway [35]. Briefly, this cell-based contractility assay uses exogenously expressed Fog-myc harvested from a stable S2:Fog-myc cell line, that, when applied to S2R + cells, leads to the phosphorylation of Sqh via Rho1 signaling, leading to the contraction of cells [18, 34]. Indicative of this contractility is the observation of phase-dark and phase-light ruffling or “bonneting” when imaged by phase-contrast microscopy (Fig. 2A, B). S2R + cells were treated with RNAi for seven days and then challenged to contract. Following this initial screen we identified two candidates that failed to contract following perfusion with Fog: the PP1 catalytic subunit Flapwing (Flw) and CG10347 which, according bioinformatic queries, is putatively a member of the heat shock protein (HSP) 20 family. Both Flw and CG10347 were predicted by the Steiner and Paths to NMII methods. A second round of computational predictions that used hits from Table 2 as positive proteins revealed CG11811 (hereafter we refer to as Oya [oya]) which, according to bioinformatic queries and previous studies, is a guanylate kinase [36, 37]. We have subsequently named CG11811 the putative Drosophila guanylate kinase, Oya after the Yoruba goddess of water who is associated with fertility and acts of creation. While Oya wasn’t predicted in the first round of candidates, it has been shown to directly interact with Sqh via coimmunoprecipitation [38] (Supplementary Fig. 5).

Fig. 2

Computationally identified candidates lead to the inhibition of NMII contractility. A-H Phase-contrast images of S2R + cells treated with control (A and B), CG10374 (C and D), Flapwing (E and F), or Oya (G and H) RNAi. Cells were either perfused with control cell media (A, C, E, G) or treated with Fog-conditioned media (B, D, F, H); yellow arrows indicate cells that have contracted following Fog perfusion. The white box in (A) denotes uncontracted cells shown at higher magnification while the yellow box in (B) denotes the contracted cells shown at higher magnification (right). Scale bars 10 µm. I-K Scatter plots quantifying the fraction of contracted cells with and without Fog perfusion following RNAi treatments with CG10274 RNAi (I), Flapwing RNAi (J), and Oya RNAi (K). I There was no statistically significant difference between control and CG10274 RNAi cells with and without Fog treatment, however in there was an inhibition of cellular contractility following RNAi treatment with Flapwing and Oya as compared to control RNAi treated cells (****p-values < 0.0001, Student's T-test, N = 3). (L & M) Results from RT-qPCR indicating the efficacy of RNAi treatments for cells treated with Flapwing RNAi (L) and Oya (M). There was a statistically significant reduction in Flw mRNA as compared to control treatments (****p-value > 0.0001, Student’s t-test, N = 3), similarly, we observed a statistically significant reduction in Oya mRNA following RNAi treatment as compared to controls (** p-value = 0.0024, ****p-value > 0.0001, one-way ANOVA, N = 5)

Using an identical protocol we performed a secondary screen focused on these three candidates (Flw, CG10347, and Oya). This screen revealed that cells depleted of CG10347 still contracted, with the fraction of contracted cells following the addition of Fog not statistically different from that of control RNAi treated cells (Fig. 2A-D & I), while cells depleted of Flw and Oya failed to substantially contract following perfusion with Fog (p value < 0.0001, Student’s t-test, N = 3, N = 54–129 image fields per condition) (Fig. 2E-H, J & K). Given that the RNAi depletion CG10347 failed to inhibit contractility in this assay we decided to no longer pursue this candidate.

To determine the efficacy of our RNAi depletion, we performed three independent rounds of reverse transcription quantitative PCR (RT-qPCR) to determine the transcriptional abundance of either flw or oya in our RNAi treated cells. Cells treated with either flw, oya, or oya 5’-untranslated region (UTR) RNAi were compared against control treated cells using elongation factor-1 (ef1) as an internal reference (Fig. 2L, M). The RT-qPCR data showed a statistically significant decrease in flw mRNA as compared to control RNAi treated samples (p-value < 0.001, Student’s t-test, N = 3) (Fig. 1L). Similarly, in cells treated with dsRNA directed against either coding region or 5’ UTR of oya we observed a statistically significant decrease in oya mRNA as compared to control RNAi treated samples albeit, the dsRNA targeting the coding region showed a far more substantial decrease (p-value = 0.0024, and p-value > 0.0001 for 5’UTR and the coding region dsRNA respectively, ANOVA, N = 5). Given the results of initial contractility assay and efficacy of our RNAi we decided to explore further how the depletion of Flw and Oya can lead to a decrease in NMII contractility.

Oya depletion affects the abundance and spatial distribution of phosphomyosin

In order to further probe the hypocontractility observed upon depletion of CG1181, we investigated both the abundance and spatial distribution of phosphomyosin in CG1181 depleted cells. To this end, cells were treated with either control, Oya, or Sqh RNAi for seven days and then challenged to contract by perfusion of Fog-enriched media. We then immunostained the cells with an antibody raised against a synthetic phosphopeptide mimicking phosphorylated NMII [39] as well as fluorescently-labeled phalloidin to visualize actin filaments (Fig. 3A-C). Imaging the cells by epifluorescence microscopy, we quantified the phosphomyosin staining following the Fog and RNAi treatments (Fig. 3D, E). This analysis revealed that Oya depleted cells have levels of phosphomyosin that are significantly lower than those of control treated cells but greater than that of Sqh RNAi treated cells (p-value < 0.00062, ordinary one-way ANOVA, N = 3, N = 289–510 cells) (Fig. 3D). Taking the fluorescence as an assay of phosphomyosin abundance, such a result suggests that, upon Fog induction, the amount of active, phosphorylated myosin is reduced in Oya depleted cells, likely contributing to cellular hypocontractility. However, Oya depletion does not completely inhibit the phosphorylation of the regulatory light chain, as evidenced by the significantly higher fluorescence values compared to Sqh depleted cells.

Fig. 3

RNAi depletion of Oya disrupts the localization of phosphorylated non-muscle myosin II. A-C Drosophila S2R + cells stained for actin (right panel, cyan in merged image) and phosphomyosin (middle panel, red in merged image) treated with (A) control, (B) Oya, or (C) Sqh RNAi. Yellow arrowheads denote coalesced phosphomyosin contractile network while cyan arrowheads denote a dispersed network. Scale bar is 10 µm. D The quantification of the ratio of the fluorescent intensities of phoshomyosin to actin following RNAi treatments. The mean (± SEM) phosphomyosin:actin ratio in Oya depleted cells (magenta circles) was statistically significantly lower than that of control treated cells (yellow circles) but greater than that of Sqh depleted cells (blue circles) (**** p < 0.0001, one-way ANOVA with Tukey’s post-hoc analysis). (E) Quantification of the coalescence index measuring the degree of phosphomyosin coalescence in cells treated with control (yellow circles), Oya (magenta circles), and Sqh RNAi (blue circles). The mean (± SEM) coalescence in Oya depleted cells is less than control treated cells yet higher than Sqh depleted cells (** p-value = 0.00295, **** p-value < 0.0001, one-way ANOVA with Tukey’s post-hoc analysis N = 3)

We then turned to probing the effects of Oya knockdown on the spatial distribution of phosphomyosin. Following Fog treatment in S2R + cells the NMII network reorganizes assembling into ordered peri-nuclear structures such as rings which are highly reminiscent of medioapical polarization observed in epithelial cells in vivo [40]. In order to quantify NMII distribution we used a previously established coalescence index [39, 41]. The higher coalescence index value the more organized (or less diffuse) the actomyosin network. This analysis reveals that Oya depletion results in a phosphomyosin distribution pattern significantly more diffuse than control RNAi treated cells yet more ordered than Sqh RNAi treated cells, mirroring the same pattern we observed in quantifying the amount of phosphomyosin (p-value = 0.002947, Ordinary one-way ANOVA, N = 3, N = 135–161 cells) (Fig. 3E). Thus, phosphomyosin distribution of Oya treated cells presented an “intermediate” phenotype where certain cells (Fig. 3B, yellow arrows) managed to recruit the ring of phosphomyosin seen in control treated cells while others displayed diffuse patterns of phosphomyosin (Fig. 3B, red arrows) similar to Sqh dsRNA treated cells. It also should be noted that depletion of Sqh leads to failed cytokinesis and subsequently larger cells. In addition we also imaged the distribution of EGFP-tagged Sqh in live cells following control, Oya, or Sqh dsRNA treatment (Supplemental Fig. 6A-D). Similar to our phosphmyosin staining, we observed an intermediate, diffuse pattern of Sqh-EGFP following Oya depletion whereas a clear perinuclear organization of Sqh can be observed in control depleted cells (Supplementary Fig. 6A, C and D). Depletion of Sqh led to a highly diffuse NMII pattern as expected (Supplementary Fig. 6B). Together, Oya knockdown appears to have a moderate effect on the recruitment and organization of NMII filaments, reducing the abundance of phosphomyosin while inhibiting the assembly of higher-order myosin structures which is likely contributing to the hypocontractility phenotype we observed in contractility assay.

Depletion of Oya may alter cytoplasmic concentrations GTP

Previous studies suggested Oya is a putative guanylate kinase due to its sequence similarity with mammalian guanylate kinases as well as its ability to phosphorylate GMP and dGMP using ATP as a phosphate donor [42]. Considering that depletion of guanylate kinases has been demonstrated to decrease intracellular concentrations of both GDP and GTP [43] as well as the centrality of GTP-mediated signaling in the Fog pathway via the Rho family of GTPases, we hypothesized that Oya’s regulation of the pathway could be due its putative role in regulating intracellular GTP levels. To test this hypothesis, we treated cells with control or Oya RNAi, or treated them with Mizoribine selective inhibitor of inosine-5'-monophosphate dehydrogenase (IMPDH) and guanosine monophosphate synthetase as a control and then fixed and stained the cells with an anti-tubulin antibody (Supplemental Fig. 7). Given the importance of GTP-tubulin to the formation of microtubules, changes to microtubule fluorescence intensity would be an indicator of cytoplasmic GTP levels. We quantified anti-tubulin fluorescence intensity and found that both treatment with Oya RNAi and Mizoribine led to a statistically significant decrease in tubulin staining as compared to control treated samples (p-value = 0.0001 One-way ANOVA, N = 38–84 cells) (Supplemental Fig. 7). Further, Oya RNAi and Mizoribine treated cells were statistically indistinguishable suggesting that RNAi Oya affects the incorporation of tubulin subunits into microtubules possibly through lowering cytoplasmic GTP levels. In terms of the NMII contractility and the Fog pathway, decreased cytoplasmic levels could affect the function of Rho family GTPases. This possibility warrants further exploration but is beyond the scope of this study. For the remainder of this study we shifted our focus to the PP1 complex component Flw.

Depletion of other PP1 components does not phenocopy depletion of Flapwing

Nurse cells mutant for flw exhibited hyper-contracted ring canals [44], and thus our results from the cellular contractility assay (Fig. 1E, F & J) are seemingly contradictory given that PP1 complex is commonly understood to function as phosphatase for targets such as the regulatory light chain of NMII. In order to further interrogate role of the PP1 complex in cellular contractility, we turned our attention to myosin binding subunit (MBS) and myosin phosphatase targeting subunit 75D (MYPT-75D) which are both involved in the targeting of the PP1 complex [43,44,45]. MYPT-75D contains a prenylation motif suggesting that it may be involved in membrane targeting while MBS lacks this prenylation [44,45,46]. We depleted MBS, MYPT-75D, and Flw, as well as Flw and MBS, and Flw and MYPT-75D in combination, and following our cellular contractility assay, challenged the cells to contract upon the addition of Fog (Fig. 4A-G). We again observed a statistically significant inhibition of cellular contractility following the addition of Fog in cells depleted of Flw as compared to control RNAi treated cells (p-value < 0.0001, ANOVA, N = 3–4, N = 50–75 cells) however, depletion of MBS failed to inhibit contractility and was no different from control RNAi treated samples (Fig. 4A-C & H). Depletion of MYPT-75D yielded mixed phenotype, with the fraction of cells undergoing contractility following the addition of Fog being statistically different from both control RNAi treated cells and that of Flw RNAi treated cells (p-values = 0.0389 and < 0.0001 respectively, ANOVA N = 3, N = 25–40 image fields) (Fig. 4A-E & H). Double depletion of Flw and MBS and Flw and MYPT-75D did lead to a slight rescue, with the fraction of contracted cells following Fog perfusion being statistically significant from both Flw RNAi (p-value = 0.004 and 0.0017, respectively ANOVA, N = 3, N = 25–40 image fields), but this rescue was incomplete as these RNAi conditions were as also statistically different from control RNAi treated cells (p-value < 0.0001, ANOVA, N = 3, N = 25–40 image fields) (Fig. 4A-H). These results indicate that despite potentially being a part of the same complex, the targeting subunits, MBS and MYPT-75D, and the catalytic subunit, Flw play differential roles in regulating the contractility of cells.

Fig. 4

Depletion of PP1 complex has distinct effects S2R + cellular contractility. A-G Phase-contrast imaging of S2R + cells from the cellular contractility assay (A) in the absence of Fog or (B-G) following the perfusion of Fog. Cells were treated with (A and B) control, (C) MBS, (D) Flw, (E) MYPT-75D RNAi or double-depleted with (F) Flw and MBS, or (G) Flw and MYPT-75D RNAi. The white box in (A) denotes an uncontracted cell shown at higher magnification while the yellow box in (B) denotes a contracted cell shown at higher magnification.Yellow arrowheads indicate contracted cells, cyan arrowheads indicate rounded cells. Scale bars 10 µm. H and I Quantification of the mean (± SEM) fraction of contracted cells and fraction of rounded cells following treatment with control (yellow circles), Flw (blue circles), MBS (green circles), MYPT-75D (peach circles), Flw and MBS (magenta circles), and Flw and MYPT-75D RNAi (purple circles). The fraction of Flw depleted cells was statistically significantly lower than all other conditions, while depletion of MBS was no different than control RNAi treated samples. Depletion of MYPT-75D was also statistically significant from both control RNAi and Flw RNAi treated cells showing an intermediate phenotype. Double depletion of Flw and MBS and Flw and MYPT-75D also showed an intermediate phenotype, being statistically significantly different from control RNAi as well as single RNAi treatments of Flw, MBS, and MYPT-75D (**p-value = 0.0090,***p-value = 0.0004, ****p-value < 0.0001, one-way ANOVA with Kruskal–Wallis test, N = 3). (I) We observed a statistically significant increase in the fraction of rounded cells following Flw RNAi as compared to all other conditions. The fraction of rounded cells following MBS or MYPT-75D depletion was no different than control RNAi treated cells, while double depletion led to an intermediate cell rounding phenotype statistically significantly different from that of Flw, MBS, or MYPT-75D RNAi treated cell, as well as control cells (*p-value = 0.0417, ***p-value = 0.002, ****p-value < 0.0001, one-way ANOVA with Kruskal–Wallis test, N = 3)

and I). Double depletion of Flw and MBS and Flw and MYPT-75D increased the number of rounded cells, indicative of Flw’s penetrance (Fig. 4F, G and I). These results closely mirror that of our contractility assay and again point to differences in the regulation of cellular contractility between PP1 components.

Flapwing, MYPT-75D, and MBS have distinct localization patterns and differentially regulate phosphomyosin distribution

NMII generated contractility relies on the concerted phosphorylation of the regulatory light chain, which results in a conformational change to the overall NMII holoenzyme allowing for oligomerization and the subsequent binding and contraction of actin filaments. Traditionally, the activity of phosphatases such as the PP1 complex, opposes this phosphorylation and maintains the NMII holoenzyme in the “closed” conformation. Previously it has been reported that MYPT-75D localizes to the cell membrane [47] while MBS can be found throughout the cytoplasm [46]. As it is likely that these spatial differences add yet another level of regulation, we sought to determine if these localization patterns are consistent in S2R + cells (Fig. 5). Consistently, we observed Flw forming a ring inside the peri-nuclear network of NMII in the center of these cells (Fig. 5A). This is in contrast to MBS, which rather than forming a distinct ring, could be found in the center of the this peri-nuclear NMII ring as an amorphous cloud (Fig. 5B), while MYPT-75D had a more global localization pattern forming a haze throughout the cytoplasm (Fig. 5C). Thus, it appears that there are differences in localization between these components of the PP1 complex which may translate to differences in the regulation of NMII contractility.

Fig. 5

PP1 complex members have distinct localization patterns. A-C S2R + cells co-expressing TagRFP-tagged (A) Flapwing (left panel, cyan in merge), (B) MBS (left panel, cyan in merge), or (C) MYPT-75D (left panel, cyan in merge) with SQH-EGFP (middle panel, red in merge) imaged by TIRF microscopy. Yellow lines indicate a representative region of interest where line scans were taken and graphically represented in D-F. Scale bar 10 µm. D-F Line scans from cells co-expressing TagRFP-tagged (D) Flw, (E) MBS, and (F) MYPT-75D (all in cyan), with EGFP-Sqh (red). Graphs represent the Normalized Fluorescent Intensity of 7–10 cells per condition

The distinct localization of PP1 complex proteins likely influences the spatiotemporal distribution of phosphorylated NMII regulatory light chain, and thus cellular contractility. To interrogate the distribution of phosphorylated NMII regulatory light chain we depleted S2R + cells of Flw, MBS, MYPT-75D or treated them with control RNAi, and then immunostained them using an anti-phosphomyosin antibody we previously used (Fig. 6A-D). Following fixation and immunostaining we imaged and quantified the mean fluorescence pixel intensity of cells depleted of Flw, MBS, MYPT-75D or treated control RNAi in the absence of Fog in order to obtain a basal phosphorylation rate (Fig. 6A-E). We found that depletion of both Flw and MYPT-75D lead to a statistically significant increase in phosphomyosin staining as compared to control or MBS RNAi conditions (p-value < 0.0001, ANOVA, N = 30–57 cells) (Fig. 6E). The observed hypocontractility, when viewed in conjunction with the increased abundance of phosphorylated myosin, suggests that Flw depleted cells might display defects in spatial or temporal regulation NMII. In order to explore this latter possibility we decided to quantify the spatial distribution of the phosphorylated regulatory light chain using our coalescence index [39, 41]. Again, the greater the coalescence index the more punctate the distribution which in turn indicates a more organized phosphomyosin network. This analysis revealed that while being statistically indistinguishable between one another, the phosphomyosin distribution of Flw or MYPT-75D was significantly more diffuse than MBS or control RNAi treated cells (p-value < 0.0001, ANOVA, N = 25–108 cells) both with and without the perfusion of Fog (Fig. 6G and H). The Fog pathway has been previously investigated in S2R + cells, where it was observed that the addition of Fog and the subsequent contractility coincided with a "purse string" structure of NMII circling the organelle-rich center domain of the cell. The contraction of this circular structure resulted in the "bonneted" morphology of S2R + cells plated on con A (Fig. 2B) [12]. Another proxy for the coalescence index is the prevalence of this phosphomyosin ring. In control treated cells, upon addition of Fog, we observed defined rings of phosphorylated myosin surrounding the central region of the cell (Fig. 6A), consistent with the literature. However, in Flw depleted cells, we saw a significant decrease in the proportion of cells displaying defined rings of phosphomyosin following Fog addition as compared to control, MBS, and MYPT-75D RNAi treated cells (p-value < 0.0001, one-way ANOVA, N = 29–89 image fields) (Fig. 6B and F). Thus, the depletion of either Flw or MYPT-75D results in an increase in the phosphorylation state of the regulatory light chain network but a decrease in its organization. Further, given some of the overlap in phenotypes, it may suggest a larger role for MYPT-75D in targeting Flw over MBS in these cells. These results also indicate that increased phosphorylation of the regulatory light chain is not enough to induce cellular contractility; it must also be spatially organized as well.

Fig. 6

Depletion of Flw increases the amount of phosphomyosin in S2R + cells, but leads to a less organized contractile network. A-D) S2R + cells fixed and stained for phosphomyosin in the absence of Fog (left panels), or after the perfusion of Fog (right panels) following treatment with (A) control, (B) Flw, (C) MBS, or MYPT-75D RNAi. Yellow arrowheads denote cells with a coalesced phosphomyosin contractile network in the form of peri-nuclear rings, while cyan arrowheads indicate a more diffuse phosphomyosin network. Yellow boxes denote cells with peri-nuclear rings shown at higher magnification, while cyan boxes denote cells with a diffuse phosphomyosin network shown at higher magnification. Scale bars 10 µm. E–H Quantification of the mean (± SEM) (E) Normalized Fluorescence Intensity, (F & G) Coalesce Index, (H) Number of cells with defined rings for cells following treatment with control (yellow circles), Flw (blue circles), MBS (green circles), and MYPT-75D (peach circles). E Depletion of Flw and MYPT-75D led to a statistically significant increase in normalized mean phosphomyosin fluorescence intensity while the depletion of MBS was no different than that of control RNAi treated samples. F The coalescence index indicated that depletion of Flw and MYPT-75D were statistically indistinguishable from one another but were statistically less organized as compared to MBS and control RNAi treated samples. G In comparing the number of defined rings, Flw depletion led to a lower number of cells with rings as compared to all other RNAi conditions. (**p-value = ,***p-value = , ****p-value < 0.0001, one-way ANOVA with Tukey’s post-hoc analysis, N = 3, N = 29–89 image fields)

Loss of contractility following flapwing depletion is partially mediated through moesin activity

Flw is known to be able to dephosphorylate and thus inactivate the Ezrin-Radixin-Moesin (ERM) protein Moesin [48]. Considering that Moesin has been implicated in driving cellular rounding [49], a phenotype observed in Flw depleted cells, we hypothesized that the phenotypes observed upon Flw knockdown could be due to an increased abundance of phosphorylated, active, Moesin. In order to test this hypothesis, S2R + cells were depleted of Moesin, Flw, and Flw and Moesin in combination or were treated with control RNAi and then were challenged to contract following perfusion of Fog (Fig. 7). Depletion of Moesin led to a hypercontractile phenotype, with a larger fraction of cells undergoing Fog-induced contractility than control RNAi treated cells (p-value = 0.00230, One-way ANOVA, N = 3). This finding may be indicative of less resistance to NMII contractility as depletion of Moesin is likely leading to inability of the cortical actin network to maintain tension. Additionally, the hypocontractile phenotype in Flw depleted cells was partially rescued by the double depletion of Flw and Moesin (p-value = 0.001634, One-way ANOVA N = 3). We also quantified the fraction of rounded cells following treatment with control, Moesin, Flw, and Moesin and Flw RNAi and observed a trend similar to our contractility assay. Moesin depleted cells were no different than control RNA depleted cells, and Flw RNAi led to stark increase in the fraction of rounded cells which was as compared to Moesin and control RNAi treated cells (p-value < 0.0001, One-way ANOVA, N = 3). However, the fraction of rounded cells failed to be rescued by depletion of Moesin and Flw in tandem. Collectively, these results indicate an antagonistic relationship between Moesin and Flw and may suggest that the hypocontractility we observed following Flw may be the result of increased cortical tension, a consequence of hyperphosphorylated Moesin. This increased tension and the loss of spatial resolution of the phosphorylated regulatory light chain of NMII is enough to inhibit contractility in our assay following Flw RNAi.

Fig. 7