Fig. 1 has/apkc function is important for restricting the length of microridges by preventing their precocious elongation. Immunocolocalization using antiaPKC antibody and phalloidin, which labels Factin, in wild type and has/apkc mutant at 48 hpf (a) followed by colocalization coefficient analysis by Pearson’s (PC) method and Manders’ overlap for aPKC with Factin (M1) and Factin with aPKC (M2) at 48 hpf in wildtype embryos (b). Graphical representation (c) of apical surface area and total cell surface area in wild type and has/apkc mutant at 48 hpf. Visualization of the distribution of ridge lengths and medians in wildtype and aPKCdeficient embryos at various developmental time points using bean plots (d). The frequency distribution of ridges (e) in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories during development. Quantification in d and e are based on phalloidin stainings performed in wildtype and apkc morphants (f). Timelapse imaging of microridges in clones expressing lifeActRFP under CMV promoter in wildtype and apkc morphants during 1922 hpf and in has/apkc siblings (wild type) and mutants during 2730 hpf (g). Error bars in c and e indicate s.d., whereas square brackets represent comparison. Probability values ‘P’ are for the Student’s ttest. Asterisks in the bean plot (d) represent significant difference in median values at P<0.05 as calculated by pairwise multiple comparison procedures using Dunn’s Method. Scale bars in a,f and g correspond to 10 µm. EXPRESSION / LABELING:
PHENOTYPE:

Fig. 2 The levels of apical Lgl increase in the apical domain of the head peridermal cells in has/apkc mutants.Immunostaining using antiLgl2 and Ecadherin antibodies (a) in wildtype and has/apkc mutant embryos. Immunocolocalization of Lgl and Factin in the apical domain of wildtype and has/apkc mutant embryos at 48 hpf (b). Colocalization coefficient analysis between Lgl and Factin by Pearson’s (PC) method and Manders’ overlap for Lgl with Factin (M1) and Factin with Lgl (M2) at 48 hpf in wildtype and has/apkc mutant embryos (c). Asterisks in c indicate significant difference at P<0.001 by Student’s ttest. Scale bar in a and b is equivalent to 10 µm. 
Fig. 3 Lgl functions in the elongation of microridges in the apical domain.Immunolocalization of Lgl and Factin in the apical domain of peridermal cells in given genetic conditions or morpholino injections at 48 hpf (a). Visualization of the distribution of ridge lengths and medians at 48 hpf, in various genetic conditions mentioned, using bean plots and horizontal lines, respectively (b) followed by estimation of percentage frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories (c). Estimation of mean and variance in ridge length for individual cells and comparison of the distributions of the means and variances across various genotypes using boxwhisker plots (d). In b and d the distributions denoted by same alphabet or distributions that share an alphabet in common (when denoted by multiple alphabets) do not differ significantly (Dunn’s multiple comparisons test, Pvalue<0.05; see Methods section for details). In c s.d. is shown by error bars. Scale bar in a is equivalent to 10 µm. MO, morphant/morpholino; Std, standard. 
Fig. 4 Lgl regulates the length of microridges by promoting their fusion.Live timelapse imaging of microridge elongation (a) in clones expressing lifeActRFP under CMV promoter in wildtype and lgl1 morphant embryos during 2330 hpf. Immunolocalization of Lgl and Factin at the basolateral domain in wildtype and pen/lgl2 mutant at 48 hpf (b). Confocal images of immunostaining using antiGFP antibody and phalloidin in wildtype embryosinjected with eGFPxLgl2 construct under CMV promoter at 24 hpf (c) and their orthogonal sections (d). Visualization of the distribution of the ridge lengths and medians in eGFPxLgl2 expressing clones and surrounding nonGFP cells using bean plots (e). The frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories (f). The boxwhisker plots (g) represent the distributions of means and variances per cell in eGFPxLgl2 expressing clones and surrounding nonGFP cells. Data presented in eg is based on ridgelength measurements done on phalloidin stainings performed in the eGFPxLgl2 injected embryos at 24 hpf. In bean plots (e) and boxwhisker plots (g) the alphabets ‘a’ and ‘b’ represent significant difference in median values at P<0.05 (pairwise multiple comparison using Dunn’s Method). Error bars in f represent the s.d. Scale bars in ad correspond to 10 µm. 
Fig. 5 Microridge elongation in aPKCdeficient embryos occurs in an Lgldependent manner.Phalloidin stainings (a,d) at 48 hpf in genotypes mentioned. Ridge length estimation followed by visualization of the spread and frequency of their lengths and medians using bean plots (b,e). The frequency distribution of ridges (c,f) in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories across the above mentioned genotypes. In b and e the distributions sharing the same alphabet do not differ significantly (Dunn’s multiple comparisons test, Pvalue<0.05). Error bars in c and f represent the s.d. Scale bars in a and d corresponds to 10 µm. PHENOTYPE:

Fig. 6 Apically localized Lgl promotes an increase in the microridge length. Confocal sections at the apical (z=2) and basolateral level (z=7) of the peridermal cells stained (ac) for GFP and Factin at 30 hpf in wildtype embryos injected with eGFPmLgl1 (a) eGFPEzrin (b) and eGFPEzrinmLgl1 (c) under CMV promoter along with their corresponding orthogonal sections. Visualization of the distribution of ridge lengths and mediansestimated from clones expressing eGFPmLgl1 (d) eGFPEzrin (e) and eGFPEzrinmLgl1 (f) and their corresponding nonGFP controls using bean plots. Comparison between the ridge lengths exhibited by clones expressing eGFPEzrin, eGFPmLgl1 and eGFPEzrinmLgl1 (g). The frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories for clones expressing eGFPmLgl1 (d′), eGFPEzrin (e′) and eGFPEzrinmLgl1 (f′) along with their corresponding nonGFP controls. The comparison between frequency distributions observed in clones expressing eGFPEzrin, eGFPmLgl1 and eGFPEzrinmLgl1 (g′). Quantifications in (dg) and (d′g′) are based on phalloidin stainings performed at 30 hpf in the embryos injected with the above mentioned eGFP constructs. Note the minimal localization of eGFPEzrinmLgl1 and eGFP Ezrin to the basolateral cortex as compared with eGFPLgl1. The distributions represented by two different alphabets in dg show significant difference at P<0.05 (Dunn’s multiple comparisons test). Error bars in (d′g′) represent the s.d. Scale bars in ac correspond to 10 µm. 
Fig. 7 Activity of nonmuscle myosinII (NMII) is essential and sufficient for elongation of microridges. Immunocolocalization using anti pNMII antibody and phalloidin in wildtype and has/apkc mutant at 48 hpf (a). Wildtype and has/apkc mutant treated with 10 µM Blebbistatin at 48 hpf and stained using phalloidin to visualize Factin (b). Graphical representation of the distribution of ridge lengths and medians of Blebbistatin treated and control embryos at 48 hpf using bean plots (c). The same data is presented as percentage frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories (d). Distribution of means and variances of ridge lengths (e) for individual peridermal cells in wildtype sibling and has/apkc mutant treated with DMSO and Blebbistatin at 48 hpfpresented in boxwhisker plots. Immunocolocalization using anti pNMII antibody and phalloidin staining in wildtype embryos and embryos overexpressing constitutively active MLCK at 48 hpf (f). In c and e the distributions sharing the same alphabet do not differ significantly (Dunn’s multiple comparisons test, Pvalue<0.05). In ((e) top) the distributions for apkc sib and apkc sib+Blebbistatin are considered statistically significantly different at P=0.058. Error bars in d are for the s.d. Scale bars in a,b and f are equivalent to 10 µm. 
Fig. 8 Both Lgl and active MyosinII are required to build long ridges. Immunocolocalization using anti pNMII, antiLgl2 antibodies and phalloidin (a) in wild type and has/apkc mutant at 48 hpf. Graphical representation (b) of colocalization coefficients for pNMII and Lgl by Pearson’s method and Manders’ overlap of pNMII with Lgl (M1) and Lgl with pNMII (M2) in wildtype and has/apkc mutant embryos at 48 hpf. Phalloidin stainings (c) in embryos injected with control morpholino, MLCKCA+control morpholino, lgl1 morpholino and MLCKCA+lgl1 morpholino at 48 hpf, followed by ridge length measurements and visualization of distribution of ridge lengths and medians using bean plots (d). The same data is presented as percentage frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories (e). The boxwhisker plots (f) represent distributions of means and variances of ridge lengths for individual cells from various genetic conditions. In d and f, the distributions sharing the same alphabet do not differ significantly (Dunn’s multiple comparisons test, Pvalue<0.05). Error bars in e are for the s.d. Asterisks in b indicate significant difference at P<0.001 by Student’s ttest. Scale bars in a and c is equivalent to 10 µm. 
Fig. 9 The loss of aPKC function and overexpression of MLCKCA results in formation of Latrunculin resistant stable microridges. Phalloidin stainings of has/apkc sibling and mutant embryos (a) and wild type and MLCKCA overexpression embryos (e) treated with Latrunculin A. Visualization of the spread and frequency of ridge lengths and medians from the given genotypes and treatments using bean plots (b,f). Estimation of percentage frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories (c,g). Estimation of cell wise means and variances of ridge lengths based on phalloidin stainings (d,h). In b,d,f and h, the distributions sharing the same alphabet do not differ significantly (Dunn’s multiple comparisons test, Pvalue<0.05). The two distributions in ((h) top), WT DMSO and MLCKCA+LatA, are different at P=0.061. Error bars in c and g are for the s.d. Scale bars in a and e are equivalent to 10 µm. PHENOTYPE:

Fig. S1 aPKC localises to the apical domain in peridermal cells but does not play a role in formation of cellular junctions. Immunostaining using antiaPKC and Ecadherin antibodies (a) in wildtype embryos at 48hpf. Z=2 and Z=6 indicate the 2^{nd} (apical) and the 6^{th} (basolateral) confocal sections, respectively, from the apical side. Boxwhisker plot for colocalisation coefficients (b; top) between Ecadherin and Factin by Pearson’s method (PC) and Manders’ overlap of Ecadherin with Factin (M1) and Factin with Ecadherin (M2). Similar Pearson’s analysis (PC) between Ezrin and Factin (b; bottom) and Manders’ overlap of Ezrin with Factin (M1) and Factin with Ezrin (M2) at 48hpf in wildtype embryos. Immunolocalisation using antiaPKC antibody and phalloidin (c) in wildtype and apkc morphants at 48hpf. Immunolocalisation using antiEzrin antibody and phalloidin (d) in wildtype and has/apkc mutants at 48hpf. Quantification of total number of ridges (e) per peridermal cell during various developmental time points, based on phalloidin stainings performed in wildtype and apkc morphant embryos. Confocal images of stainings for ZO1 and Ecadherin in wildtype embryos and has/apkc mutants at 48hpf (f). XZ are orthogonal sections. Basolateral marker Ecadherin does not localise to apical microridges (b) and serves as a negative control with Pearson coefficient=0.26+0.65. The Ecadherin antibody generates a few background speckles in the apical domain. Since some of these speckles fall on Factin, it yields high Manders overlap coefficient of M1=0.413+0.14, which is misleading. However, Factin shows very low overlap with Ecadherin background speckles yielding a very low Manders’ coefficient of M2=0.005+0.004. Ezrin serves as a positive control for correlation coefficient analysis. A major fraction of Ezrin localises with actin (b,d) giving rise to Pearson coefficient of 0.813+0.04. Similarly, Manders’ coefficients of Ezrin with Factin (M1=0.73+0.09) and Factin with Ezrin (M2=0.68+0.08) are high, reflecting significant overlap between Ezrin and Factin (b,d). Scale bar = 10 µm. Abbreviations: PC Pearson’s coefficient; M1 and M2 Manders overlap coefficients. 
Fig. S2 Lgl colocalises with Factin at microridges and exhibits temporal increase in its levels in aPKC deficient embryos. Lgl1 deficiency or Lgl2 overexpression does not perturb the overall polarity in the peridermal cells. Immunocolocalisation using anti Lgl2 antibody and phalloidin (a) in wildtype and aPKC deficient embryos at 24hpf. Lgl2 staining in wildtype sibling and in has/apkc mutants at 32 and 48hpf (b). Confocal images along with orthogonal (XZ) sections of ZO1 and Ecadherin stainings in wildtype embryos and lgl1 morphants at 48hpf (c). Immunolocalisation using anti aPKC and anti Ecadherin antibodies in clones expressing cmv:egfpxlgl2 construct (d) at 30hpf. Scale bar = 10µm. 
Fig. S4 Overexpression of heavy and light chains of MyosinII does not cause significant changes in ridge lengths. Confocal images of wildtype embryos injected with cmv:mhcIIgfp or cmv:mlcIIgfp and stained for GFP and Factin at 30 hpf (a). Visualisation of the distribution of the ridge lengths and medians of the clones overexpressing cmv:mhcIIgfp or cmv:mlcIIgfp and the neighbouring nonGFP cells using bean plots (b). The percentage frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories (c), as well as estimation of mean and variance in ridge length for individual cells and comparison of their distributions (d) using boxwhisker plots. Quantifications in (b), (c) and (d) are based on embryos injected with cmv:mhcIIgfp or cmv:mlcIIgfp and stained for GFP and Factin at 30 hpf. Distributions represented with different alphabets show significant difference in median values at p<0.05 by pairwise multiple comparison procedures using Dunn′s Method. Error bars in (c) are for the standard deviation. Scale bar =10 µm. Abbreviations: MHCII: MyosinII heavy chain; MLCII: MyosinII Light chain. 
Fig. S5 MyosinII activity (pNMII levels) is higher in aPKC deficient embryos and is required for elongation of ridges in wildtype as well as in apkc morphants. pNMII staining (a) in wildtype and apkc morphant embryos at 20, 24 and 30hpf and in has/apkc mutants at 36hpf. Boxwhisker plot for colocalisation coefficients (b) of Factin and pNMII by Pearson’s methods (PC) and Manders’ overlap coefficient of pNMII with Factin (M1) and Factin with pNMII (M2) at 36 hpf in wildtype and has/apkc mutant embryos. Wildtype and apkc morphants (apkc MO) treated with 10µM Blebbistatin for 1.5h, fixed at 21hpf (c) and at 30hpf (d) followed by phalloidin staining to visualise Factin. Probability values “P” are for the student’s ttest. Asterisk in (b) indicate significant difference at p<0.001. Scale bars correspond to 10 µm. Abbreviation: pNMII phospho Non muscleMyosinII. 
Fig. S6 Overexpression of MLCKCA does not give rise to longer ridges in pen/lgl2 mutants. Confocal microscopy analysis of the peridermal cells in given genotypes in either uninjected (control) or MLCKCA injected embryos stained with phalloidin (a). Graphical representation of the distribution of ridge lengths and corresponding medians in various genotypes using bean plot (b). A bar graph showing percentage frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories (c). Representation of cell wise means and variances of ridge lengths in various genotypes using boxwhisker plot (d). In (b) and (d) the distributions sharing the same alphabet do not differ significantly (Dunn’s multiple comparisons test, pvalue < 0.05). Error bars in (c) are for the standard deviation. Scale bar in (a) is equivalent to 10 µm. 
Fig. S7 Lgl overexpression results in increased pNMII levels; MyosinII activity is essential for the manifestation of long ridge phenotype in cells overexpressing Lgl. Confocal images of wildtype embryos injected with cmv:egfpxlgl2 and stained for GFP and pNMII at 30 hpf (a). eGFPxLgl2 clones (b,c) treated with 10µM Blebbistatin for 1.5h, fixed at (b) 24hpf and (c) 30hpf and stained using phalloidin to visualise Factin. Scale bar corresponds to 10 µm. 
Fig. S8 Factin polymerisation is required during ridge elongation phases in various genetic backgrounds. Peridermal cells in apkc morphant at 20 and 24hpf; MLCKCA injected and wildtype embryos at 29hpf (a), eGFPxLgl2 clones at 30hpf and wildtype larvae at 5dpf (b), treated with 2µM Latrunculin A for 30 minutes, fixed and stained using phalloidin to visualise F actin. Scale bar corresponds to 10 µm. 
Fig. S9 Assessment of the specificity of lgl1 morpholinos used in this study. Confocal images of microridge Factin stained by phalloidin in uninjected embryos and embryos injected with standard control morpholino (Std control MO), lgl1ATG morpholino (lgl1ATGMO) and lgl1UTR morpholino (lgl1UTRMO) at 48hpf (a). Analysis of ridge lengths by Factin staining in peridermal clones expressing eGFPmLgl1 in embryos coinjected with the mentioned morpholino (or no morpholino) and a vector driving expression of egfpmlgl1 under CMV promoter (b). Quantification of ridge lengths in the clones and visualisation of the length distributions and the respective medians in various genetic conditions by bean plot (c). The percentage frequency distribution of ridges in short (05 µm), intermediate (520 µm), long (20100 µm) and very long (>100 µm) categories for the given genetic conditions (d). Estimation of cell wise means and variances of ridge lengths for each genetic condition and their representation by boxwhisker plot (e). In (c) and (e), the horizontal black lines and associated probability values (P) represent the comparison by Dunn’s multiple comparisons. P<0.05 is considered statistically significant. In (d) Standard deviation is shown by error bars. Scale bars in (a,b) are equivalent to 10µm. The asterisks in (b) depict the cells that do not express eGFPmLgl1. Abbreviation: Std: standard; MO: morphant/morpholino, mLgl1: mouse Lgl1 . 
Acknowledgments: 

ZFIN wishes to thank the journal Nature communications for permission to reproduce figures from this article. Please note that this material may be protected by copyright. Full text @ Nat. Commun. 