Plexin/Semaphorin antagonism orchestrates collective cell migration and organ sculpting by regulating epithelial-mesenchymal balance

Abstract

Cell behavior emerges from the intracellular distribution of properties such as protrusion, contractility, and adhesion. Thus, characteristic emergent rules of collective migration can arise from cell-cell contacts locally tweaking architecture, orchestrating self-regulation during development, wound healing, and cancer progression. The Drosophila testis-nascent-myotube system allows dissection of contact-dependent migration in vivo at high resolution. Here, we describe a role for the axon guidance factor Plexin A in collective cell migration: maintaining cell-cell interfaces at a precise point on the mesenchymal-to-epithelial continuum. This is crucial for testis myotubes to migrate as a continuous sheet, allowing normal sculpting-morphogenesis. Cells must maintain filopodial N-cadherin-based junctions and remain ECM-tethered near cell-cell contacts to spread while collectively moving. Our data further suggest Semaphorin 1b is a Plexin A antagonist, fine-tuning activation. This reveals a contact-dependent mechanism to maintain sheet integrity during migration, driving organ morphogenesis. This is relevant for mesenchymal organ sculpting in other migratory contexts such as angiogenesis.

Fig. 1.. plexA and sema1B knockdown cause gap formation defects in the testis musculature suggesting migration defects.

(A) Schematics of TNM collective cell migration during pupal development. Below: Steps of muscle-dependent testis morphogenesis after migration at 26.5°C. Bottom: Schematic of types of adult muscle-dependent morphogenetic testis defects. (B) Live imaging of testes ex vivo, with migrating TNMs expressing Lifeact::eGFP and Cherry^nls under the control of dMef2-Gal4. Scale bar, 100 μm. (C to E) Adult testis with musculature stained with phalloidin. Below: Magnification of the hub region. (C) Wild type (dMef-Gal4 crossed with w¹¹¹⁸ flies). (D) plexA^RNAi and (E) sema1b^RNAi. (F and G) Quantification of adult gap defects. (F) Manually traced gaps in plexA^RNAi as an example. (Fa) Quantification of the percentage of uncovered area after different plexA and sema1b genetic perturbations. (G) Computationally unrolled testis from (F), with a histogram curve displaying grey values of gap data proportional to gap size. Bottom: These values plotted as a heatmap for multiple replicates for two different plexA^RNAi and sema1b^RNAi lines. n = 8 for each genotype (related to fig. S1).

Fig. 2.. Depleting plexA or sema1b has very different effects on collective cell behavior.

(A to F) Ex vivo culture of 30-hour testis. TNMs express Lifeact::eGFP and Cherry^nls under the control of dMef2-Gal4. [(A) and (B)] Wild type, [(C) and (D)] plexA^RNAi, and [(E) and (F)] sema1b^RNAi. (A) Yellow arrows: Cells are well spread and maintain equal cell-cell distances. (C) White arrows denote broad cell processes well ahead of the cell body, and yellow arrows denote interconnecting cell processes, and circles denote dense cell clusters. (E) Yellow arrows denote cells no longer connected to proximal neighbors. [(B), (D), and (F)] High-magnification live-cell imaging. Top right: Rear cells. Bottom right: Magnification of boxed regions in front cells. Yellow arrows: Interdigitating filopodia. Rear cells are from the same videos. Scale bars, 20 μm (related to fig. S3).

Fig. 3.. plexA^RNAi and sema1b^RNAi alter cell spacing, cell area, and gap size in different ways.

(A to E) Distance to closest neighbor quantification for wild type (A), plexA^RNAi (B), and sema1b^RNAi (D). Performed at 200 to 400 min when quantified cells had left the seminal vesicle. Overlay of micrographs with color-coded dots, indicating the distance to closest neighbor. [(C) and (E)] Quantification of percentage of cells at given distance to a neighbor. Curve is the mean of 8 curves for eight testes with ~60 to 100 quantified cells per testis. Light-colored area = SD. (F to J) Cell distribution spatial statistics. [(F), (G), and (I)] Kernel density maps showing local cell density of (F) wild type and (G) plexA^RNAi. Yellow arrows denote regions of cell clustering, and blue arrows denote gaps (I) of sema1b^RNAi. [(H) and (J)] Ripley’s H function. (H) The yellow arrow in plexA^RNAi marks the part of the graph where cell density increases locally. (J) The yellow arrow in sema1b^RNAi marks the part of the graph where cells become globally more homogeneous (detailed explanation in fig. S5B). Mean of n = 8 testes. Light-colored area = range. (K to M) Machine learning–based segmentation of videos of the three genotypes at two time points. Green denotes area covered by cells, and red denotes gaps. (N and O) Ratio of gap area to covered area at the early and late time points. n = 8 for each genotype. Statistical test: Ordinary one-way analysis of variance (ANOVA), Šídák’s multiple comparison test. P values: (N) (from left to right): 0.2282, 0.0117, (O): 0.0179, 0.2504. (P) Increase of ratio of gap area to covered area for individual videos from t = 0 to t = 425 min. (k-k‴). Ordinary one-way ANOVA, Šídák’s multiple comparison test. P values (from left to right): 0.1702, 0.0005, 0.4175. (Q to S) Images of testes at 40-hour APF when migration is completed and testis elongation is beginning. Insets denote boxed areas in proximal testis (related to fig. S5). ns, P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Fig. 4.. Overexpressing or activating PlexA causes strong behavior and cell-cell adhesion defects during collective cell migration.

(A to C) Adult testis defects caused by overexpressing wild-type PlexA (A), PlexA^Δcyto (B), and non-inactivatable PlexA^SA (C). Scale bars, 100 μm. (D) Quantification of uncovered area. Boxplot: Line denotes mean, and quartiles denote box and range. (E to H) Ex vivo culture of 30-hour testis. TNMs express Lifeact::eGFP and Cherry^nls under the control of dMef2-Gal4. [(E) and (G)] Wild type. (F) Wild-type PlexA overexpression. (H) PlexA^SA overexpression. Yellow arrows denote dispersed cells. (I to K) Distance to closest neighbor quantification. Performed at 200 to 400 min, when quantified cells had left the seminal vesicle. [(I) and (J)] Overlay of micrographs with color-coded dots, indicating distance to the closest neighbors. Scale bar, 100 μm. (K) Quantification as a distribution curve of the mean of 8 curves for eight testes with ~60 to 100 quantified cells per testis. Light-colored area = SD. (L) Ripley’s H function (detailed explanation in fig. S5B). Mean of n = 8 testes. Light-colored area denotes range. Right: Ripley’s H function graphs from Fig. 3 for direct comparison.

Fig. 5.. PlexA regulates N-cadherin–based cell-cell contacts, whereas Sema1b’s influence on sheet cohesion seems independent of N-cadherin.

(A to D) Airyscan 2 micrographs of fixed/mounted testis (30 h APF) stained with Phalloidin (magenta), 4′,6-diamidino-2-phenylindole (DAPI) (blue) and N-cadherin antibody (green). [(A), (Aa)] Wildtype. [(B), (Ba), (C), (Ca)] plexA^RNAi. [(D), (Da)] sema1b^RNAi. Magnifications below represent the yellow boxes above. Axon bundle-like structures in plexA^RNAi are depicted in (C) and (Ca) with a graphical representation in (Ca). Yellow arrows denote N-cadherin–based cell-cell contacts. (E to N) Genetic interaction experiments with plexA^RNAi, sema1b^RNAi, PlexA^SA, and NCad^RNAi. [(E) to (K)] Adult testis with different morphological defects. Musculature stained with phalloidin. Distal, medial, and proximal portions of the testis marked. Yellow arrows denote gaps in the muscle sheet. Scale bars, 100 μm. (L) Gap quantification after computational unrolling (compare Fig. 1G). (M) Sum of gaps in the proximal region (0 to 40%). (N) Sum of gaps in the distal region (41 to 100%). n = 8 testes per condition. Statistical test: Ordinary one-way ANOVA, Šídák’s multiple comparison test. P values: (l′, from top to bottom): 0.1739, 0.0877, and 0.0104. (l″, all values): < 0.0001 (related to figs. S7 and S8). ns, P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Fig. 6.. plexA^RNAi reduces cell-cell contact adjacent matrix adhesions, while plexA overexpression or sema1b^RNAi do not.

(A and B) High-resolution live-cell imaging of ex vivo cultured (30-hour APF) testis for quantification of integrin adhesion distribution in wild type [(A) and (Ab)] or plexA^RNAi [(B) and (Bb)]. TNMs express focal adhesion targeting domain (fat)::eGFP under control of dMef2-Gal4. Details of the method are in fig. S10. [(A) and (B)] Left: Matrix adhesions. Middle left: myr-RFP helping define cell edges. Middle right: Lines displaying distances from matrix adhesions to the nearest cell edge. Right: Lines displaying distances from random points to the nearest cell edge. [(Aa), (Ab), (Ba), and (Bb)] Close-ups of yellow boxes in (A) and (B) (left) showing either cell-cell contacts or front cells. Yellow arrows denote matrix adhesions near cell-cell contacts. Scale bars (overview), 20 μm and (magnification) 5 μm. (C and D) Quantification of matrix adhesions near (C) or far (D) from cell-cell contacts. Statistical test: Mixed effect analysis, Šídák’s multiple comparison test. P values (from top to bottom): 0.6454, 0.8397, 0.0201, 0.9750, 0.8481, and 0.0003 (related to fig. S10). WT, wild type. ns, P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Fig. 7.. plexA^RNAi increases cell height while plexA overexpression or sema1b^RNAi do not.

(A to F) Quantification of cell height. Wild type [(A) and (C)], plexA^RNAi [(B) and (D)], and sema1b^RNAi (E). [(A) and (B)] Right: xz and yz sections through 3D micrographs. Scale bars, 20 μm. (C to F). Sheet thickness quantified with the Imaris XTension Biofilm Analysis. (F) Mean height of the cell sheet. n = 25 (wild type), 5 (plexA^RNAi), and 8 (sema1b^RNAi). Statistical test: Kruskal-Wallis, Dunn’s multiple comparison test. Adjusted P values (from top to bottom): >0.9999 and 0.0008. (G and H). Airyscan2 imaging. (I, Ia, J, and Ja) Individual cells with cell outlines coded by depth. (K) Multiple examples from each genotype. (L) Plot of xy area versus z-axis height. WT: n = 7, plexA^RNAi: n = 6. (M and N) Quantification of cell area and cell height. (M) Kolmogorov-Smirnov normality test passed: P values (from left to right): (i): >0.1 and 0.0962. Student’s t test, P value: 0.0434. (N) Kolmogorov-Smirnov normality test not passed. P values (from left to right): 0.3601 (not passed) and >0.1. Kolmogorov-Smirnov test, P value: 0.0484. (O and P) Schematic of distinctive cell architecture defects after plexA^RNAi and sema1b^RNAi. ns, P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.

Fig. 8.. Genetic interactions suggest that PlexA and Sema1B are antagonists and PlexA acts as a Rap2L/Ras2-GAP.

(A to G) Genetic interaction experiments with plexA^RNAi and sema1b^RNAi. [(A) to (D)] Adult testis with different morphological defects. Musculature stained with phalloidin. Yellow arrows denote medial gaps. Below: Magnifications of the distal regions. Scale bar, 100 μm. (E) Quantification of area uncovered. plexA^RNAi: n = 8, sema1b^RNAi and plexA^RNAi + sema1b^RNAi: n = 12. Statistical test: Ordinary one-way ANOVA, Šídák’s multiple comparison test. P value = 0.5562. (F) Gap quantification after computational unrolling. (G) Muscle gaps in the proximal region (0 to 70%). Statistical test: Ordinary one-way ANOVA, Šídák’s multiple comparison test. P values (from top to bottom): 0.9861 and 0.0044. (H to P) Genetic interaction experiments with plexA^RNAi, Rap1^RNAi, Rap2L^RNAi, Ras1^RNAi, and Ras2^RNAi. [(H) to (O)] Adult testis with different morphological defects in. Below: Magnifications of the distal regions. Arrows denote gaps. (P) Quantification of area uncovered. Rap1^RNAi, plexA^RNAi + Rap1^RNAi: n = 8 and others: n = 16. Statistical test: Ordinary one-way ANOVA, Šídák’s multiple comparison test. P values (from top to bottom): 0.0037, 0.0002, 0.0828, and <0.0001. (Q) Illustrations of how signaling, cell architecture, and collective cell migration are altered upon knockdown of plexA and sema1b and (R) schematic of a proposed PlexA/Sema1b pathway. ns, P > 0.05; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001; ****P ≤ 0.0001.