(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



A mechanical transition from tension to buckling underlies the jigsaw puzzle shape morphogenesis of histoblasts in the Drosophila epidermis [1]

['Annafrancesca Rigato', 'Aix Marseille Univ', 'Cnrs', 'Centrale Marseille', 'Institut Fresnel', 'Turing Center For Living Systems', 'Marseille', 'Ibdm', 'Huicheng Meng', 'Claire Chardes']

Date: 2024-06 

The polygonal shape of cells in proliferating epithelia is a result of the tensile forces of the cytoskeletal cortex and packing geometry set by the cell cycle. In the larval Drosophila epidermis, two cell populations, histoblasts and larval epithelial cells, compete for space as they grow on a limited body surface. They do so in the absence of cell divisions. We report a striking morphological transition of histoblasts during larval development, where they change from a tensed network configuration with straight cell outlines at the level of adherens junctions to a highly folded morphology. The apical surface of histoblasts shrinks while their growing adherens junctions fold, forming deep lobules. Volume increase of growing histoblasts is accommodated basally, compensating for the shrinking apical area. The folded geometry of apical junctions resembles elastic buckling, and we show that the imbalance between the shrinkage of the apical domain of histoblasts and the continuous growth of junctions triggers buckling. Our model is supported by laser dissections and optical tweezer experiments together with computer simulations. Our analysis pinpoints the ability of histoblasts to store mechanical energy to a much greater extent than most other epithelial cell types investigated so far, while retaining the ability to dissipate stress on the hours time scale. Finally, we propose a possible mechanism for size regulation of histoblast apical size through the lateral pressure of the epidermis, driven by the growth of cells on a limited surface. Buckling effectively compacts histoblasts at their apical plane and may serve to avoid physical harm to these adult epidermis precursors during larval life. Our work indicates that in growing nondividing cells, compressive forces, instead of tension, may drive cell morphology.

Funding: Agence Nationale de la Recherche (ANR-17-CE30-0007, ANR-18-CE13-028, ANR-22-CE42-0010, ANR-22-CE13-0039); Excellence Initiative of Aix-Marseille University - A*Midex (capostromex), a French Investissements d'Avenir programme. This project is funded by the « France 2030 » investment plan managed by the French National Research Agency (ANR-16-CONV-0001, ANR-21-ESRE-0002), and from Excellence Initiative of Aix-Marseille University - A*MIDEX. RSS was supported by a Biotechnological and Biological Sciences Research Council (BBSRC) Institute Strategic Programme Grant to the John Innes Centre (BB/X01102X/1). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Here, we investigate the shape of developing histoblasts. We developed a protocol for time-lapse imaging of individual cells throughout larval stages. We observed that histoblasts go through a considerable morphological transition between 90 h and 110 h AEL, changing from a tensed network configuration with straight cell outlines to a highly folded morphology of cell shapes, suggestive of compressive forces. We show that the formation of folded junctions arises from the frustrated growth of adherens junctions of histoblasts in the spatially constrained apical domain of these cells. The imbalance between growth of junctions and shrinkage of the apical domain of histoblasts compresses the junctions and triggers a buckling instability. We also show that growth of cells on the epidermis creates a packing that affects the apical size and shape of histoblasts. We thus propose that morphogenesis of histoblasts in the Drosophila larva is compression-driven.

The epidermis of the Drosophila larva consists of two cell populations: the larval epithelial cells (LECs), which are large polyploid cells, and the histoblasts, which are the precursors of the adult epidermis. These two cell populations form a continuous cell monolayer [ 26 ]. Histoblasts are clustered in nests of a fixed cell number (5 to 17 cells per nest) surrounded by LECs. Growth of both histoblasts and LECs happens without cell division over a large span of larval development, from 4 h after hatching until the pupal stage [ 26 , 27 ]. Histoblasts do not exchange neighbors either during this period [ 28 ]. Thus, there is a complete absence of rearrangements of the cellular lattice. The growth rates of histoblasts and LECs are different, with LECs increasing in volume about 150-fold during larval life, and histoblasts 60-fold [ 26 ]. The larval body stops growing around 90 h after egg laying (h AEL) [ 26 , 29 ], while the epidermal cells continue to grow. Thus, this binary cellular system, where two cell populations grow and compete for space in the absence of stress-releasing topological transitions is likely to present a mechanical regime yet unexplored by other epithelial model systems.

Other than tensile forces, compressive forces are also important shape generators. For example, an elastic body under compressive forces can go through buckling instability [ 18 – 20 ], a process at play in gut vilification [ 21 , 22 ], in gastrulating embryos [ 23 ], in the formation of brain cortical lobules by differential growth of apposed cortical cell layers [ 24 ], or in some cell-ECM systems [ 25 ]. While these examples are taken from elasticity operating at a large scale, in this work, we find evidences that similar effects exist at the level of individual cells. In particular, we show that this happens during epidermal morphogenesis in Drosophila.

Because of their biological relevance and ease of imaging, epithelia have been particularly well characterized in terms of mechanics. The combination of modeling, analysis of cell shape, and mechanical perturbations has led to the following understanding of epithelial mechanics: (i) the tissue is in a tensed state; (ii) cell growth is usually balanced over the cell cycle (cells double in volume from beginning to end of the cell cycle); and (iii) stress can be released through topological transitions such as cell neighbor exchange and oriented cell divisions [ 13 ]. With these elements, models were developed for epithelial tissues, able to capture many features of morphogenesis [ 14 , 15 ]. Epithelial monolayers are then described as tensed networks, formed by polygonal-shaped cells with straight borders [ 16 , 17 ]. Nevertheless, tensed epithelia that respect the characteristics above are only a portion of the complex scenario of morphogenesis.

Morphogenesis proceeds through the mechanical interaction of cells in order to shape tissues. Our understanding of the cellular processes that power shape changes has considerably improved in recent year [ 1 – 4 ]. At the single cell scale, the role of internally generated active forces from the cytoskeleton in setting morphological changes has been well established [ 2 , 5 , 6 ]. But the shape of a cell is also determined by environmental constraints, as adhesion links cells to their neighbors and to the extracellular matrix [ 7 ]. Growth can act as a potent environmental constraint to shape cells and tissues. Spatial variations in the orientation or the rate of growth lead to mechanical prestress impinging upon cells [ 4 , 8 – 11 ]. For example, if one or a patch of cells grows more than its surroundings, the overgrowing patch will be compressed and the surroundings will be stretched [ 12 ].

Results

As their apical surface shrinks, histoblasts grow basally To our knowledge, the only available data about epidermal growth in this system was obtained by electron microscopy and estimated from 2D data by Madhavan and colleagues [26], who reported a 150-fold volume increase of LECs and 60-fold of histoblasts. While an increasing perimeter at the adherens plane seems compatible with growth of cells, the decrease in apical area is not. To further investigate this apparent inconsistency, we analysed histoblasts growth in 3 dimensions, from the beginning of lobules formation to the wandering stage. We analyzed the 3D shape of histoblasts in the course of the morphological transition by imaging their basolateral membrane with a src:GFP fusion and adherens junctions with an E-cad:mKate fusion (Fig 2). Before the lobules appear, cell borders have the same straight shape at the adherens and basal planes (Fig 2A–2C). From the orthogonal projections in Fig 2C (ZX and ZY), the distance between the basal and apical surfaces of histoblasts is about 6 μm, which is comparable to the thickness of LECs [34]. Through the transition, and while cells shrink apically, the basal side of histoblasts expands in a rounded bulb-like shape (Fig 2D–2F). As shown in Fig 2D, the nest is much larger basally than apically, and no fold is visible basally. The difference in morphology of the apical and basal sides are well apparent on the overlay of the basal and apical membranes of Fig 2G. Additionally, cell thickness is greatly increased in the course of the transition, with the apical and basal plane now being well separated, as shown from the orthogonal projections (ZX, ZY) in Fig 2F. The increase of cell thickness correlates well with the formation of lobules, as demonstrated by plotting the thickness against cell circularities (Fig 2I). The average cell thickness increases from about 6 μm to 10 μm in the course of the morphological transition (Fig 2I). As the apical area decreases while the height increases, we segmented histoblast nests after expression of a cytosolic GFP with a histoblast driver (esg-Gal4) to estimate their volume in the course of the transition. Because the cell number is constant in the histoblasts at this stage, the measurement gives us the cell-volume increase averaged over the nest. The measurements were done on individual nests, tracked over time through chronic imaging at 3 different time points. As reported in S4 Fig, cell volume increases 2-fold between 90 and 115 h AEL. Hence, histoblasts do grow during the last larval stage, despite the apical shrinkage. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. 3D characterization of histoblast growth and junctional buckling. (A-F) Live imaging of cadherin junctions (E-cad:mKate) and cell membrane (src:GFP) before (A-C) and after (D-F) apical junction remodeling. (G, H) Superposed projections of the apical and basal plane of a histoblast nest at 110 h AEL. While the apical side presents junctional lobules, the basal side is characterized by large and bulb-shaped cells. (I) Cell thickness plot as a function of cell circularity. Light blue line represent the means, error bars the standard deviation. Thirty different nests were analyzed, from 10 larvae. (J) Schematics of cell growth below the apical surface as junction remodeling occurs. At the top, section of histoblast and LECs showing how histoblast develop below the epithelial surface. Red dashed lines represent the apical and largest basal plane corresponding to images A-F. The yellow dashed line represents the plane of septate junctions, corresponding to images K and L. (K, L) Merged projection of apical junctions and septate junctions at 90 h AEL (K) and 110 AEL (L). Before junctional buckling, apical and septate junctions have similar, superimposing structures. When apical junctions fold, the deformation is partially lost at the plane of septate junctions. White dashed circle: example of straight septate junction. Yellow dashed circle: example of septate junction not following the corresponding apical junction. The data underlying the graphs shown in the figure can be found in S1 Data. Scale bar = 10 μm. https://doi.org/10.1371/journal.pbio.3002662.g002 As shown in Fig 2G, histoblasts are round-shaped at the basal side. Hence, the lobules observed at the level of adherens junctions are not a characteristic of the entire lateral cell membrane. To investigate whether the lobules are only localized at the apical plane and lost immediately below, or whether they are gradually lost, we imaged septate junctions with a disc large protein fusion (dlg:GFP) together with E-cadherins (cad:mkate). Septate junctions are localized just below adherens junctions. Before the morphological transition, septate junctions have the same shape as adherens junctions—their projections superpose (Fig 2K). When lobules appear, septate junctions follow only partially the shape of adherens junctions despite being localised very close to them (Fig 2L). Thus cell border at the level of septate junctions are less folded than at the level of the adherens junctions but more than at the basal level. In summary, over the time window from 90 h to 110 h AEL, histoblasts increase their volume 2-fold while their adherens junctional material also increase by a factor of 1.7. However, in that same time window, their apical area decreases. The volume increases at the basal side of cells. We formulated a working hypothesis summarized in Fig 2H, whereby junctional folding in histoblasts is generated by the continuous growth of junctions, which is not balanced by a concomitant expansion of the cell population at their adherens plane, like in most epithelia studied currently. Growth of a slender elastic structure in a constrained environment leads to its compression and eventual buckling [35]. We thus propose that the lateral constraints that lead to shrinking of the apical surface of histoblasts could also influence the mechanical buckling of the growing junctions.

Manipulation of junctional growth affects junctional folding In our hypothesis, histoblasts’ buckling is the result of the concomitant growth of their junctions and the shrinkage of the apical domain of the histoblast nest. To test the contribution of junctional growth, we specifically altered the growth of adherens junctions in histoblasts by genetic means. By reducing the available junctional material, we expect a reduction of folding. First, we impaired the activity of Rab11, known for its role in cadherin recycling. Rab11 is responsible for the transport of newly synthesized cadherin as well as recycled cadherin and other proteins to the cell junctions [36]. We induced the overexpression of a Rab11-dominant negative (Rab11-DN) specifically in the histoblasts. In Fig 3A and 3B, wild-type (WT) histoblasts have long and folded junctions, while Rab11-DN have shorter and straighter junctions. We measured a circularity of 0.74 for Rab11-DN against 0.17 in WT (p-value = 4.10−38), an average junction length of 30 μm in Rab11-DN against 100 μm in WT, (p-value = 3.10−32). PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Genetic perturbation of histoblasts junctions and cell cycle. (A) Live image of junctions (E-cad:mKate) of WT Drosophila larva, at wandering stage (110 h AEL). (B) Live image of junctions (E-cad:GFP) of Drosophila larva, at wandering stage (110 h AEL), upon E-cadherin recycling impairment in histoblasts (esg>Rab11-DN). Histoblasts show shorter, straight junctions and reduced apical area. The diffused cytosolic signal is due to a YFP linked to Rab11-DN, partially emitting in the GFP spectrum. (C) Same, upon expression in histoblast of Cdc25/Stg (esg>Stg). Histoblasts are more numerous, smaller, buckling is abolished, and the nest is larger. (D) Live imaging of histoblast membranes (esg>src:GFP) in a WT larva, at the wandering stage. The z-projection shows both the apical buckled junctions, and the basal round shape of histoblasts. (E) Same as D, upon down-regulation of E-cadherin in histoblasts (esg>Cad-RNAi). Histoblasts’ junctions appear straight, and the basal side as round and wide as in the WT larva. (F-I) Morphometric quantifications. (E) Scatter plot and density plot of perimeter vs. area of WT (pink), esg>Rab11-DN (green) and esg>Stg (blue) histoblasts. (F-H) Box plots of essential morphological descriptors: perimeter (F), cell area (G), and circularity (H). Details on statistical analysis can be found in the Materials and methods section and Table C in S1 Text. Scale bar = 20 μm. WT = wild type, N cells = 352 (WT), 49 (Cad-RNAi), 82 (Rab11-DN), 880 (Stg+). Average cell number per nest = 15 ± 2 (WT, ±STD), 12±2 (Rab11-DN), 108±22 (Stg+), 14±2 (Cad-RNAi). The data underlying the graphs shown in the figure can be found in S1 Data. https://doi.org/10.1371/journal.pbio.3002662.g003 Next, we directly impaired Cadherin levels by silencing its gene (shotgun) with RNA interference. We expressed Cad-RNAi in the histoblasts (esg > Cad-RNAi) of larvae carrying src:GFP as a membrane marker to compensate for the absence of an adherens marker. While junctional folds are visible in WT membrane-labeled histoblasts (Fig 3D), cadherin-depleted junctions do not develop folds at the wandering stage and have polygonal-shaped cells (Fig 3E). The measurements of cell morphological parameters (Fig 3G–3I) reveal no significant change in cell area, but a 2-fold decrease of cell perimeter (median value 52μm versus 100 μm, p-value 10−19 according to Mann–Whitney test), with a consequent increase of circularity from 0.2 to 0.7 (p-value = 5.10−25, Mann–Whitney test). The scatter and density plots in Fig 3F confirm that Cad-RNAi and Rab11-DN (resp. green and cyan) cells are found in a different region than WT cells (pink) in the perimeter-area morphometric space. Combined, these experiments provide evidence that the morphological transition that we observe depends on the addition of junctional material to the adherens belt of histoblasts.

Reintroduction of cell division abolishes junctional folding In the context of a buckling instability, the length of junctions is an important intrinsic factor that controls the critical compressive load at which buckling proceeds. Indeed, for an elastic beam, this critical load scales as ~L−2, where L is the characteristic length of the beam [18]. By depleting junctions from their E-cadherin pool, the previous cad-RNAi and Rab11-DN experiments may have changed the material properties of junctions. One means to change the length of junctions, without altering the rate of junctional material addition, is to reintroduce cell divisions in histoblasts. The net result is a denser network of shorter junctions, while maintaining the external mechanical constraints on the nest unchanged, as well as the material properties of junctions such as the density in E-cadherins or their elasticity. We forced histoblasts to divide by overexpressing the mitotic controller cdc25 –string (stg) in Drosophila [37]. This perturbation only impacts the cell cycle and not growth [38]. As a consequence, histoblasts were more numerous in the histoblasts nests (around 100). As shown in Fig 3D, individual histoblasts are consequently smaller and have straight junctions. This experiment further supports the model whereby buckling of the junctions proceeds because of their excessive lengthening while the apical area is constrained. It also highlights the essential role played by stalling of the cell cycle. Would the histoblasts divide like, for example, cells of the imaginal discs, the deep junctional lobules would not occur. Besides the direct effect on cell junctions, the whole histoblast nest is also larger than in the control (Fig 3C). A possible interpretation for this experiment is that the shortened length of junctions prevents the onset of buckling and thus improves their ability to withstand mechanical stress. As the junctions are less prone to buckling, the overall nest becomes stiffer and gets squeezed to a lesser degree. Thus, forcing cell divisions in histoblasts abolishes buckling and reduces compaction of the histoblast nest.

Computer simulations of buckling junctions It stems from our analysis that junctions are elastic structures, which evolve quasi-statically as they grow. A first qualitative understanding for junctional buckling is rooted in the mechanics of elastic beams under compression [18]. As junctions grow, they are compressed because the apical domain of histoblast remains narrow—and even shrinks. Note that shrinkage of the apical surface is not a strict requirement. As long as growth of junctions is not properly balanced by a corresponding increase in apical surface, they will experience compression and will be prone to buckling. Two observations, however, discriminate lobules formation in histoblasts from the simple buckling of a beam (Fig 5A and 5B). First, an elastic beam under compression buckles through its lowest mode of deformation (Fig 5A, top), while histoblast junctions deform through high modes—several undulations are often observed for a single junction (Fig 5B, top). Second, if a buckling beam is suddenly cut at one point, it should relax to its resting shape (Fig 5A, bottom). Instead, histoblast junctions hardly relax after laser ablation (Fig 5B, bottom), as shown in Fig 4. PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Computer simulation of buckling. (A, B) Some features of histoblast buckling are incompatible with simple elastic beam buckling. (A, top) An elastic beam under compressive stress at the extremities buckles through low mode deformation. (A, bottom) The stored bending energy is released upon breaking the beam, which results in a recoil of the beam. (B, top) Histoblasts buckle at small wavelengths (high mode). (B, bottom) Upon laser dissection, junctions hardly relax. (C) Computer simulations of histoblast mechanics. Cell boundaries are represented as point masses (vertices) connected by springs (edges). The elastic energy of the boundary includes stretching and bending terms. Connection to the elastic environment is simulated with springs that pull vertices back to their position. (D-F) Snapshots of the computer simulation of an histoblast nest going through the morphological transition. In the initial configuration, cells are polygonal (D). Early on in the transition, isolated lobules appear along the cell–cell interfaces (arrowheads in E). Lobules have fully developed in the final state. https://doi.org/10.1371/journal.pbio.3002662.g005 High mode buckling can easily be rationalized and should, in fact, be the expected outcome based on simple elastic considerations. Indeed, lateral junctions are not mechanically isolated structures. They are embedded in their environment through connections to the cytoskeleton and the apical cell domain. While a molecular description of such a connection is beyond the scope of this study, our optical tweezer experiments demonstrated the interconnection of large domains in histoblast, most likely transmitted through the cell cortex. On simple elastic terms, one can view the junction as a slender beam connected to an elastic substrate or “foundation.” Such an elastic foundation precludes low mode buckling because their large amplitude are too costly in elastic energy. High-mode buckling has, for example, been observed in compressed microtubules, when coupled to the surrounding cytoskeleton or to immobilized molecular motors in a motility assay [43,44]. At a larger scale, high mode buckling is also observed in morphological instabilities of soft materials [45]. The absence of relaxation upon ablation suggests that a process is at play in the course of the morphological transition to dissipates the compressive stress born by junctions. The most likely underlying mechanism would be a reorganization of the cytoskeletal cortex. It is not clear at this point whether this dissipation is triggered by the high stress experienced by buckled junctions—a characteristic of plastic processes—or whether time plays a critical role. In any case, to the best of our knowledge, no work has addressed the possibility of a mechanical buckling process in the presence of a stress-dissipating adaptation to stabilize previous deformations. We designed a computer simulation of tissue mechanics that encapsulates the essential elements at play in the course of the morphological transition. This simulation was implemented in MorphoDynamX using a similar approach to that published in Sapala and colleagues’ study [46], which is well adapted to describe cell assemblies under either a tensile or compressive stress. Cell boundaries are represented as point masses (vertices) connected by springs (edges, Fig 5C). One cell boundary is constituted of many such edges (approximately 50 to 100) to provide a continuum description. The elastic energy of the boundaries includes stretching and bending terms (inset in Fig 5C). At each iteration loop, the length of edges and the outer boundaries of the entire nest are assigned new values that reflect the growth of cell–cell junctions and shrinkage of the nest. Mechanical equilibrium is then found using the backward Euler implicit solver. The simulation includes a plastic deformation leading to stress dissipation at each iteration loop and an elastic recall that pulls vertices back to their position, to account for the elastic embedding of the cell–cell junction in the surrounding cell medium—thus introducing the elastic foundation. We start the simulation with a polygonal shape (Fig 5D). The shrinkage of the nest and lengthening of the cell–cell junctions is then induced. Lobules, associated with buckling, first appear in an isolated fashion along cell junctions (black arrowheads in Fig 5E). Note that similar isolated irregularities are observed at the onset of the transition in live samples (Fig 1B). As the simulation proceeds, the lobules become more pronounced, eventually giving rise to a final state highly reminiscent of the experimental observations (8F). These simulations show that a simple mechanical model of junction growth in a constrained shrinking space leads to buckled junctions very reminiscent of the ones we observe experimentally. Notably, a nontrivial outcomes is that elastic buckling can proceed even when a slow process allows to dissipates mechanical stress through a plastic process.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002662

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/