Differential cell adhesion implemented by Drosophila Toll corrects local distortions of the anterior-posterior compartment boundary

Abstract

Maintaining lineage restriction boundaries in proliferating tissues is vital to animal development. A long-standing thermodynamics theory, the differential adhesion hypothesis, attributes cell sorting phenomena to differentially expressed adhesion molecules. However, the contribution of the differential adhesion system during tissue morphogenesis has been unsubstantiated despite substantial theoretical support. Here, we report that Toll-1, a transmembrane receptor protein, acts as a differentially expressed adhesion molecule that straightens the fluctuating anteroposterior compartment boundary in the abdominal epidermal epithelium of the Drosophila pupa. Toll-1 is expressed across the entire posterior compartment under the control of the selector gene engrailed and displays a sharp expression boundary that coincides with the compartment boundary. Toll-1 corrects local distortions of the boundary in the absence of cable-like Myosin II enrichment along the boundary. The reinforced adhesion of homotypic cell contacts, together with pulsed cell contraction, achieves a biased vertex sliding action by resisting the separation of homotypic cell contacts in boundary cells. This work reveals a self-organizing system that integrates a differential adhesion system with pulsed contraction of cells to maintain lineage restriction boundaries.

Fig. 1. Toll-1 is expressed in P histoblasts under the control of engrailed.

a Schematic depicting anterior (red) and posterior (light blue) histoblast nests in the epidermis of the Drosophila pupal abdomen. The boxed region highlights abdominal segment 2, and arrowheads point to the AP boundary. b Tissue wide expression pattern of Tl using fluorescently tagged endogenous Tl protein. Arrowheads point to the AP boundary. Anterior is to the left and posterior is to the right unless otherwise stated in this and all the following figures. Presented data are a representative image of n = 10 animals. Scale bar: 30 μm. c Tl protein localization visualized by Venus knock-in (green). Adherens junctions and P histoblasts were labeled with p120ctn::tagRFP (magenta) and hh::mTurquoise2 (cyan), respectively. Presented data are a representative image of n = 10 animals. Scale bar: 10 μm. d Mosaic analysis of Tl expression in engrailed (en) mutant cells. Tl expression was monitored with the Venus knock-in (green). Cells mutant for en were labeled with the loss of the marker expression, a monomeric red fluorescent protein with nuclear localization signal (mRFPnls) (magenta). Presented data is a representative image of n = 5 clones. Scale bar: 10 μm.

Fig. 2. Toll-1 is required for maintaining the sharp histoblast AP boundary.

a Schematic representation of boundary angle measurement. The boundary angle θ is defined as the angle made by a pair of neighboring cell junctions on the AP boundary. b The AP boundaries in control (left, 23 hAPF) and Tl knockdown (right, 23 hAPF) histoblasts by expressing a short hairpin RNA for Tl using a histoblast-specific GAL4 driver, esg-GAL4. Scale bar: 10 μm. c Quantification of the boundary angle in control, Tl knockdown, and Tl mutant tissues as a readout for the sharpness of the AP boundaries. n = 35, 29, 16, and 15 boundaries for Control, Tl^{RNAi #1}, Tl^{RNAi #2}, and Tl^r3/r4, respectively. Images were analyzed at the proliferation phase (22–24 hAPF). Statistical significances were evaluated using a Student’s t test (unpaired, two-sided). ***p < 0.001. d Diagram illustrating the measured boundary angle for the quantification of boundary angle change shown in (e–j). e–j Cellular configuration-dependent dynamics of the boundary angle change. Boundary angle (θ₁) change over 3 min for the pair of junctions that form the boundary angle (θ₀) less than 100˚ (e), between 100˚ and 160˚ (g), and more than 160˚ (i) 3 min before the measurement in the control and Tl knockdown tissues and the quantification of the boundary angle change (f, h, and j). Scale bars: 5 μm. Numbers of analyzed vertices (n) were indicated in graphs. Data were collected from two animals (22–24 hAPF, proliferation phase) for each experiment. Data are presented as mean values ± SEM. Statistical significance was evaluated using a Student’s t test (unpaired, two-sided). *p < 0.05, **p < 0.01, and ***p < 0.001.

Fig. 3. Expression of Toll-1 makes cells adhere to each other.

a–c Mosaic clones ectopically expressing Tl in the A histoblast nest. Expression of full length and the extracellular domain of TI renders clone contours smooth (a). The boundary angle of the Tl expressing clones for each experiment (b). Schematics of Tl full length, extracellular domain alone, and intracellular domain alone (c). n = 21, 17, 13, and 31 boundaries for Control, Tl, Tl^Ex, and Tl^In overexpression clones, respectively. Statistical significances were evaluated using a Student’s t test (unpaired, two-sided). ***p < 0.001. Scale bar: 10 μm. d Overexpression of Tl results in cell aggregates. Drosophila S2 cells expressing E-cadherin (positive control), Tl full length (Tl), Tl extracellular domain (Tl^EX), and intracellular domain (Tl^IN). Scale bar: 100 μm. e Quantification of the aggregate formation. The fraction of the large aggregate was quantified. Data are presented as mean values ± SEM. Statistical significance was assessed using a Mann–Whitney U test (two-sided). n = 9 independent replicates for each experiment. *p < 0.05, **p < 0.01, and ***p < 0.001.

Fig. 4. Toll-1 does not affect junctional Myosin II localization and its localization is largely unaffected by the actomyosin cytoskeleton.

a, b Differential expression of Tl does not accumulate Myosin II at the interface between Tl expressing and non-expressing cells. While the expression of Ed in mosaic clones (magenta) accumulates junctional Myosin II (green) at clone edges, Tl expressing clones do not affect the junctional localization of Myosin II. Scale bar: 10 μm. Myosin II localization along the clone boundary (b). The signal intensity of endogenous Myosin II regulatory light chain tagged with the Venus fluorescent protein (Myosin II::Venus) on cell junctions was normalized to that of cytoplasm. 7, 5, and 9 clones were analyzed for control, ed, and Tl overexpressing clones, respectively. Data are presented as mean values ± SEM. Statistical significance was assessed with a Mann–Whitney U test (two-sided). *p < 0.05, **p < 0.01, and n.s. not significant. c Junctional Myosin II localization along the AP boundary when the boundary is straight or distorted. Myosin II::Venus (white) and hh::mTq2 (cyan). Myosin II enriched on cell junctions along the AP boundary when the boundary is straight (arrowheads) while the enrichment was not visible when the boundary was locally distorted in control experiments (arrows). Knockdown of Tl did not affect the enrichment of Myosin II::Venus when the boundary was straight. The AP boundary was visualized with the P histoblast-specific expression of hh::mTq2. Presented data are representative images of aligned junctions from n = 8 (control) and 6 (Tl RNAi) boundaries and misaligned junctions from n = 8 (control) and 5 (Tl RNAi) boundaries. Scale bar: 5 μm. d, e Schematic representation of aligned and misaligned junctions analyzed in (c, d). Three consecutive junctions (J1–J3) along the AP boundary (magenta) were analyzed for Myosin II signal intensity (see “Methods” for detail). Cases, where either of the boundary angles of the junction set was larger than or equal to 130˚, were considered as “aligned” and the smaller than 130˚ as “misaligned”. The relative intensity of Myosin II (normalized to that of non-boundary junctions) were plotted (e). The data were collected from staged pupae at the proliferation phase (21–24 hAPF). In total, 8 and 7 aligned junctions (from N = 8 and 6 boundaries) were analyzed for control and Tl RNAi, and 8 and 5 misaligned junctions (from N = 8 and 5 boundaries) were analyzed for control and Tl RNAi, respectively. Data are presented as mean values ± SEM. Statistical significance was assessed with a Mann–Whitney U test (two-sided). *p < 0.05, **p < 0.01, and n.s. not significant. f Relationship between the initial velocity after laser ablation of junctions on the AP boundary and boundary angle made between the ablated junction and its connected junction. Experiments were performed in the pupae at the proliferation phase (22–24 hAPF).

Fig. 5. Trans interaction of Toll-1 between P histoblasts increases homotypic cell contacts by biasing the vertex sliding.

a The reduction of cell mixing index γ at the locally distorted AP boundary is dependent on Tl. Cell mixing index γ is the fraction of the adherent’s junctional length of cells that are in contact with cells of the adjacent compartment. Change in γ between two frames (3 min) was measured for individual cells categorized by the value for γ at the first frame. Totally, 1220 cells (107, 240, 346, 251, 211, and 65 cells for >0, >0.1, >0.2, >0.3, >0.4, and >0.5 bins, respectively) from 6 boundaries for control and 964 cells (91, 176, 204, 241, 189, 63 cells for >0, >0.1, >0.2, >0.3, >0.4, >0.5 bins, respectively) from 5 boundaries at the proliferation phase (21–23 hAPF) were analyzed. Data are presented as mean values ± SEM. p Values were from the Student’s t test (unpaired, two-sided). *p < 0.05. b Cells with higher area fluctuation reduce γ efficiently in control but not in Tl knockdown tissue. The analysis was performed for A cells having γ > 0.5 (10 animals for control and 15 animals for Tl knockdown at the proliferation phase (21–23 hAPF)). Cells were categorized as low or high fluctuation based on the degree of area fluctuation (see “Methods”) and were analyzed for γ change separately. Statistical analysis was performed using the lm function in R (version 4.0.0). The number of analyzed cells were indicated as n. Data are presented as mean values ± SEM. Statistical significance was determined using a Mann–Whitney U test (two-sided). *p < 0.05, ***p < 0.001, and n.s. not significant. c The cell mixing index γ declines as the area of the cell fluctuate. The cross-sectional area and γ for an A cell that has initially high γ were plotted as a function of time. d Measurement of vertex sliding productivity at the AP boundary during contraction and expansion. The two possible situations that would result in the straightening of the AP boundary are illustrated. The productivity of vertex displacement was evaluated by comparing cells’ displacement behavior to the hypothetical vertex displacement that they would undergo if the cell exhibited isotropic cell shape change with the same degree of area change (Δd). Δd is calculated by subtracting the calculated displacement of the vertex (d^iso) when the isotropic contraction was supposed from the actual displacement (d) of the vertex on the AP boundary for A cells for the expansion phase and contraction phase, separately. e Two models illustrating the homophilic interaction of Tl regulating vertex sliding at the interface between A and P histoblasts. The analyzed cell configuration is shown on the left (one A cell and two P cells, highlighted with red dashed line). Large red arrows indicate the direction of vertex sliding; small red arrows indicate the direction of the forces exerted on the cell–cell contact between P cells, which result in either separation (top) or closure (bottom) of homotypic cell–cell contacts between P cells. Blue arrows represent the recruitment of more Tl at the leading edge of the cell–cell contact between P cells. f Tl was required for the biased outward vertex sliding for A cells at the AP boundary. Δd was plotted for cells that initially had a high γ value (γ > 0.55) at each area expansion and contraction phase separately (three cells each for control and Tl knockdown) from the pupae at the proliferation phase (21–23 hAPF). Outward sliding was less effective than expected from hypothetical isotropic vertex displacement in controls. Inward sliding was comparable to the hypothetical isotropic vertex displacement. The outward vertex sliding became unbiased in the Tl knockdown tissue. The numbers of analyzed phases (contraction or expansion) were indicated as n. Statistical significance was determined using a Student’s t test (unpaired, two-sided). **p < 0.01, and ***p < 0.001.