The roles of inter-tissue adhesion in development and morphological evolution

ABSTRACT

The study of how neighboring tissues physically interact with each other, inter-tissue adhesion, is an emerging field at the interface of cell biology, biophysics and developmental biology. Inter-tissue adhesion can be mediated by either cell–extracellular matrix adhesion or cell–cell adhesion, and both the mechanisms and consequences of inter-tissue adhesion have been studied in vivo in numerous vertebrate and invertebrate species. In this Review, we discuss recent progress in understanding the many functions of inter-tissue adhesion in development and evolution. Inter-tissue adhesion can couple the motion of adjacent tissues, be the source of mechanical resistance that constrains morphogenesis, and transmit tension required for normal development. Tissue–tissue adhesion can also create mechanical instability that leads to tissue folding or looping. Transient inter-tissue adhesion can facilitate tissue invasion, and weak tissue adhesion can generate friction that shapes and positions tissues within the embryo. Lastly, we review studies that reveal how inter-tissue adhesion contributes to the diversification of animal morphologies.

Summary: The study of how neighboring tissues physically interact with each other, inter-tissue adhesion, is an emerging field at the interface of cell biology, biophysics and developmental biology.

Introduction

Embryonic development is an amazingly complex process that is nonetheless remarkably reproducible. In a simplistic view, development starts with the precise spatio-temporal deployment of cell-type-specific gene regulatory networks (GRNs). These GRNs regulate the expression of genes that produce proteins responsible for altering the behavior of individual cells. The behavior of individual cells can influence the behavior of neighboring cells and give rise to coordinated tissue-level changes. Technological advances in the past two decades are beginning to reveal the role mechanical forces have in coordination across length scales (Shawky and Davidson, 2015). A well-characterized example is the apical constriction of a large group of cells resulting in tissue invagination (Heer and Martin, 2017; Kasza and Zallen, 2011). These mechanical processes can also feed back and regulate gene expression (Smith et al., 2018; Heller and Fuchs, 2015). At larger length scales, tissue behavior can influence neighboring tissues through inter-tissue adhesion. For this Review, we define a single tissue as a group of cells that form a functional unit and give rise to part of an organ or structure. Most tissues we cover are separated by an extracellular matrix (ECM) or basement membrane (BM), but there are exceptions.

Our understanding of how inter-tissue adhesion impacts embryonic development remains limited. In this Review, we discuss how inter-tissue adhesion affects the morphogenesis of neighboring tissues. It should be noted that the ECM between two adhered tissues can store or alter the diffusion of secreted signaling proteins and affect gene expression. However, we focus here on the mechanics of inter-tissue adhesion and will examine this process across many species, from Caenorhabditis elegans to humans. We highlight six mechanisms in which inter-tissue adhesion functions: tissue coupling, creating mechanical resistance, transmitting tension, mechanical instability, transient adhesion and generating friction. We will then discuss cases where inter-tissue adhesion may contribute to the diversification of morphologies throughout the course of evolution.

How do tissues adhere?

There are two primary mechanisms by which inter-tissue adhesion connects neighboring tissues. The first is through a shared ECM. Tissues can bind to a shared ECM, or the ECMs attached to individual tissues can physically merge. The ECM is a fibrous matrix composed of proteins including collagens, fibronectin, elastin and proteoglycans (Walma and Yamada, 2020). During development, the ECM serves many diverse roles, including regulating cell proliferation, controlling stem cell differentiation, contributing to tissue motion and acting as a substrate for cell migration (Dzamba and DeSimone, 2018; Bonnans et al., 2014). When cells in separate tissues bind to the same ECM, they become mechanically linked. In addition to the fibrous ECM network, there is a specialized ECM known as the BM. The BM is a thin, dense, sheet-like ECM primarily composed of laminin, collagen IV, perlecan and nidogen (Keeley and Sherwood, 2019; Walma and Yamada, 2020). The primary function of the BM during development is to separate epithelial tissues from the underlying connective tissue, allowing tissues to move relative to each other. However, two BMs can become physically linked and mediate inter-tissue adhesion, but the mechanism of how BM–BM linkage occurs is not well understood (Keeley and Sherwood, 2019).

The second mechanism of tissue–tissue coupling is through cell–cell adhesion that utilizes cell-surface adhesion molecules. Cell–cell adhesion is vital for proper tissue organization, boundary formation and collective cell motion (Gumbiner, 1996; Halbleib and Nelson, 2006). Several proteins mediate cell–cell adhesion, such as cadherins, selectins and proteins of the immunoglobulin superfamily (Kashef and Franz, 2015). Cell–cell adhesion also creates cohesion within a tissue, which can be critical for robust inter-tissue adhesion.

The role of inter-tissue adhesion in shaping development

Regardless of the adhesion mechanism, once tissues adhere, there are multiple ways inter-tissue adhesion can impact development. However, the role of inter-tissue adhesion is relatively understudied compared to other areas of morphogenesis, including cell motion and cell shape. Nevertheless, a number of recent reports provide insights into how inter-tissue adhesion affects development. Below, we identify six predominant functions of inter-tissue adhesion throughout development: tissue coupling, generating mechanical resistance, producing friction, transient adhesion, creating mechanical instability and transmitting tension. We highlight key studies that illustrate each function.

Tissue coupling and its effect on cell and tissue motion

A straightforward mechanism by which inter-tissue adhesion can influence morphogenesis is the coordination of motion between tissues. For instance, consider two flat tissues adhered together. If cells in the top tissue move, this movement can influence the motion of cells in the underlying tissue. The result of coupling would be the coordination of cell motion between the two tissues. For instance, there is tightly correlated cell motion between the zebrafish neural plate and the underlying mesoderm, which are adhered together via a shared ECM (Araya et al., 2016, 2014). Furthermore, in the beetle Tribolium castaneum, removal of the bilayered extraembryonic tissue requires the two tissues to adhere. In this case, one extraembryonic layer ruptures the tissue, and the other layer drives its removal (Hilbrant et al., 2016).

A well-studied example of how tissue coupling influences morphogenesis is the periodic pattern of chicken feather buds. Turing's reaction–diffusion equations have been used to model this patterning process (Fig. 1A). Reaction–diffusion relies on a system with a minimum of two diffusible components: a slow-diffusing activator and a fast-diffusing inhibitor (Landge et al., 2020). The activator induces its own expression in a positive feedback loop and stimulates the expression of the inhibitor. The inhibitor blocks signaling by the activator, forming a negative feedback loop. Since the inhibitor diffuses faster than the activator, this limits the expansion of the activator domain and creates a periodic pattern. In the case of the chicken feather bud, mechanics also appear to contribute to this patterning process (Fig. 1B). Throughout the development of feather bud primordia, the epidermis and dermis adhere together via a shared BM (Fig. 1C). The first perceptible signs of feather follicle formation emerge simultaneously: (1) expression of fibroblast growth factor 20 (FGF20) in the epidermis and (2) aggregation of cells in the dermis (Ho et al., 2019). Secretion of FGF20 by the epidermal cells acts as a chemoattractant, stimulating aggregation of dermal cells. When the dermal cells aggregate, they activate FGF20 expression in the epidermis, producing a positive feedback loop (Ho et al., 2019). FGF20 activation by the dermal cell probably occurs in part due to inter-tissue adhesion. As the dermal cells aggregate, the overlying epidermal cells bunch together, presumedly due to their shared adhesion to the BM (Shyer et al., 2017). In response to dermal aggregation, β-catenin in the epidermal cells moves to the nucleus where it binds to T-cell factor/lymphoid enhancer binding factor-1 (TCF/Lef-1)-family members and acts as a transcriptional activator, leading to expression of the feather bud GRN, including FGF20 (Ho et al., 2019) and bone morphogenetic protein 2 (BMP2) (Noramly et al., 1999). Secretion of BMP2 inhibits the expression of FGF20, which limits the size of the feather bud primordia (Ho et al., 2019). Moreover, the size and spacing of the feather bud primordia are, in part, determined by the contractility of the dermal cells and the stiffness of the environment, which provides resistance to cell contraction (Shyer et al., 2017). This developmental process illustrates the role inter-tissue adhesion plays in coordinating motion in two tissues, which in conjunction with diffusible cell signaling proteins, results in periodic pattern formation.

Fig. 1.

Coordinating and resisting tissue motion: two opposing roles of inter-tissue adhesion. (A) A reaction–diffusion model predicts that different diffusion rates of activators and inhibitors create the periodic pattern and size of the developing feather buds in chicken. (B) Illustration of periodic feather bud patterning on the dorsal epidermis of the developing chicken. (C) At the beginning of chicken feather bud development, epidermal cells (top layer of cells) secrete FGF20, which attracts dermal cells (dark purple). Simultaneously, the dermal cells aggregate, leading to bunching of the overlying epidermal cells, causing β-catenin (β-cat) to move to the nucleus of epidermal cells and activate expression of BMP2 and FGF20. This pathway of chemical signals (BMP2 and FGF20) and mechanical signals (dermal cells) functions to create a periodic pattern similar to that created by the reaction–diffusion model with the addition of inter-tissue adhesion-based mechanics. (D) During zebrafish development, neural tube convergence is resisted by inter-tissue adhesion to the neighboring PSM. NC, notochord. The neural tube and PSM are adhered by a gradient of Fibronectin, which is enriched laterally (dark orange) and is less concentrated medially (light orange). (E) Formation of this dynamic gradient of Fibronectin matrix depends upon shear stress at the interface of the neural tube and PSM, as well as activated Integrin α5β1.

Mechanical resistance generated by inter-tissue adhesion

In the development of feather bud primordia, inter-tissue adhesion couples the motion of cells. However, in some instances of inter-tissue adhesion, one tissue resists the movement of the neighboring tissue, thereby constraining its range of motion. Precisely tuned mechanical resistance can be a requirement for the symmetry or extension of a tissue. For instance, during anterior primary neural development in chicken, anterior–posterior growth of the neural tube is restricted due to ECM-mediated adhesion to the surrounding tissues (Danesin et al., 2021).

During post-gastrulation zebrafish neural tube morphogenesis, inter-tissue adhesion plays a role in establishing bilateral symmetry by creating resistance to tissue convergence (Guillon et al., 2020). The posterior neural tube is neighbored on the left and right sides by the presomitic mesoderm (PSM) (Fig. 1D). The neural tube and PSM connect via a shared Fibronectin matrix. Two components are required to form this Fibronectin matrix: soluble Fibronectin and active Integrins. Fibronectin is secreted into the extracellular space as a compact disulfide-linked dimer (Schwarzbauer and DeSimone, 2011). The dimer conformation prevents fibril formation until Fibronectin binds to Integrin heterodimers on the surface of cells. Binding induces Integrin clustering, and actomyosin-generated tension unfolds the Fibronectin dimers. Fibronectin then non-covalently crosslinks to other Fibronectin molecules, producing fibrils that thicken and lengthen (Schwarzbauer and DeSimone, 2011). Integrin α5β1, one of the primary Fibronectin receptors, is necessary for this inter-tissue adhesion and Fibronectin matrix assembly during neural tube morphogenesis (Dray et al., 2013). Active Integrin α5β1 heterodimers are mainly present on the surface of the PSM where cells have little Cadherin 2. In contrast, Cadherin 2 prevents Integrin activation on internal mesenchymal PSM cells by stabilizing their inactive conformation (Julich et al., 2015; McMillen et al., 2016).

Throughout neural tube morphogenesis, neural tube cells converge to the dorsal midline, whereas Fibronectin-mediated inter-tissue adhesion to the PSM resists this convergence (Fig. 1D) (Guillon et al., 2020). Maternal zygotic mutants for Integrin α5 show disrupted cell–Fibronectin adhesion, resulting in precocious neural tube convergence and loss of bilateral symmetry in the mesoderm. Disruption of bilateral symmetry is observable by measuring the differences in the size of the PSM. This asymmetry likely arises due to a mechanical instability that creates gaps between the PSM and neural tube, disrupting the balance of inter-tissue adhesion and tissue surface tension between the tissues (Guillon et al., 2020).

Mechanically, the neural tube and PSM interface behaves like an engineered adhesive lap joint. A medial-to-lateral shear stress gradient inherent to lap joints can be counterbalanced by a gradient of adhesive (Carbas et al., 2014). Here, as the neural tube converges, shear stress is generated along the medial-to-lateral axis of the neural tube–PSM interface (Guillon et al., 2020). Fibronectin is increasingly assembled into large fibers at regions of high shear stress along the lateral boundary of the neural tube–mesoderm interface (Fig. 1E) and colocalizes with high levels of non-muscle myosin and filamentous actin. Over time, this remodeling of the ECM creates a medial–lateral Fibronectin gradient (Fig. 1E). Thus, Fibronectin acts as a ‘smart’ adhesive in the lap joint as it continually remodels to where it is most needed (Guillon et al., 2020). This illustrates how cells must continuously tune ECM deposition, cell–cell adhesion and cell–ECM connections during development to maintain symmetry. Furthermore, this exemplifies how inter-tissue adhesion creates resistance of tissue motion at tissue–tissue boundaries.

Tissue adhesion and friction

Friction is a measure of the force that resists the relative displacement of juxtaposed tissues (Wu et al., 2015). Given the diversity of molecular and cellular mechanisms by which tissues adhere to one another, a logical conjecture is that there are corresponding differences in inter-tissue friction during development. For instance, in the developing zebrafish myotome (skeletal muscle), differential inter-tissue friction, caused by differences in ECM-mediated inter-tissue adhesion, is required to form the myotome chevron shape (Tlili et al., 2019).

Friction generated by inter-tissue adhesion can help position organs within an organism. During zebrafish gastrulation, the prechordal plate (PPL) moves towards the animal pole, and the neighboring neurectoderm slides against the PPL in the opposite direction towards the vegetal pole (Fig. 2A) (Labernadie and Trepat, 2018). As the neurectoderm slides, the anterior cells reorient towards the animal pole, which establishes the position of the neural anlage (Smutny et al., 2017). Slowing down PPL cell motion by mutating wingless-type 11 family member 2 (wnt11f2) (Ulrich et al., 2003) causes a sizable reduction in anterior neurectoderm cell rearrangements. Additionally, the neural anlage is mislocalized towards the vegetal pole in these mutants, indicating a role for PPL motion in positioning the neural anlage (Smutny et al., 2017). Examination of ECM components (i.e. Fibronectin) found that there is no deposition at the boundary between the PPL and neurectoderm (Smutny et al., 2017); however high levels of E-cadherin (Cadherin 1) are observed at the boundary, and E-cadherin is required for adhesion between the PPL and neurectoderm (Montero et al., 2005). Cadherins are transmembrane proteins and are the main adhesion proteins responsible for cell–cell adhesion during embryonic development (Leckband and de Rooij, 2014; Arslan et al., 2021). Accordingly, reducing E-cadherin-dependent friction abrogates cell rearrangement in the neurectoderm (Smutny et al., 2017). Diminished friction also causes mislocalization of the neural anlage towards the vegetal pole (Fig. 2B). Mathematical modeling of the neurectoderm as a viscous compressible fluid subject to external friction from the PPL recapitulates the experimental cell-flow patterns observed in both wild-type and wnt11f2-mutant embryos. Furthermore, an ex vivo experiment in which E-cadherin-coated beads were moved on top of cultured ectodermal progenitor cells, mimicking the motion of PPL cells, resulted in the rearrangement of the ectoderm cells. Overall, this study found that the speed at which both tissues move determines the magnitude of the friction and placement of the neural anlage, meaning that friction is precisely tuned to properly localize the neural anlage (Smutny et al., 2017). This highlights the importance of inter-tissue adhesion mediated by cell–cell adhesion and demonstrates how inter-tissue friction helps position cells within an embryo.

Fig. 2.

Inter-tissue adhesion-generated friction and transient adhesion, and their role in positioning tissues. (A) In the gastrulating zebrafish, the neurectoderm moves to the vegetal pole (VP), and the prechordal plate (PPL) moves to the animal pole (AP). Adhesions between the two tissues are formed by E-cadherin–E-cadherin interactions, which generates friction as the tissues move relative to each other. D, dorsal; V, ventral. (B) Knockdown of E-cadherin results in mislocalization of the neural anlage, which can be visualized by in situ hybridization for otx2. Brackets indicate distance to the anterior edge of the neural anlage. WT, wild type. (C) During mammalian embryo implantation, the blastocyst positions itself using L-selectins that bind to oligosaccharides at the endometrium. MUC-1 is a glycoprotein that prevents binding and is removed just before implantation. Once in position, a strong attachment is made that includes interactions with integrins, trophinin and dystroglycan 1, leading to invasion of the endometrium.

Transient inter-tissue adhesion and invasion

The classical view of the BM is that it functions to separate adjacent tissues (Keeley and Sherwood, 2019). However, there are processes in which BMs transiently adhere two tissues, subsequently followed by invasion of one tissue by the other. For example, in C. elegans, the uterus and vulva move independently of each other during development but must align prior to uterine invasion of the vulval tissue (Morrissey et al., 2014). To properly align, the BMs transiently adhere using a BM-adhesion system called B-LINK. Similarly, in Drosophila, the wing disc must be in juxtaposition with the larval epidermis before invading the epidermis and everting, but if and how they adhere is not fully understood (Pastor-Pareja et al., 2004).

Transient inter-tissue adhesion is critical for implantation of the mammalian embryo. Here, we will primarily focus on humans, but it should be noted that a plethora of research is available describing implantation in other mammals. Implantation is quite challenging as there is only a small window of time within the menstrual cycle when the embryo can implant. The endometrium determines this temporal constraint and must undergo several modifications to become receptive, including loss of cell polarity (Whitby et al., 2020). It has been proposed that the loss of cell polarity may prevent the polarized embryo from repulsing the polarized luminal endometrium. Selectins (carbohydrate-binding proteins) on the blastocyst form weak attachments to L-selectin ligands on the endometrium (Fig. 2C) (Genbacev et al., 2003; Feng et al., 2017). Weak binding enables the embryo to roll along the uterine wall, allowing it to find a favorable site on the uterus (Ashary et al., 2018). This site should be firm enough to anchor onto (Ashary et al., 2018) and allow for the correct orientation of the inner cell mass (Ochoa-Bernal and Fazleabas, 2020). Mucin-1 (MUC-1) is an anti-adhesion molecule thought to prevent binding in unfavorable locations by countering the weak adhesion of L-selectin (Ochoa-Bernal and Fazleabas, 2020). Removal of MUC-1 by cleavage at the site of implantation enables strong adhesion. Additionally, integrins, dystroglycan 1 (DAG1) and trophinin mediate robust adhesion to the endometrium (Fig. 2C) (Aplin and Ruane, 2017). Once strongly attached, the embryo can breach the luminal endometrium and invade for successful implantation (Ochoa-Bernal and Fazleabas, 2020).

Mechanical instability due to differential growth between adhered tissues

In the context of inter-tissue adhesion, differential growth between the two adhered tissues can generate in-plane stress. For instance, a fast-growing tissue can be compressed by a slow-growing tissue when they adhere together. This in-plane stress can be relieved by bending the tissues (Fig. 3A) (Collinet and Lecuit, 2021; Nelson, 2016), and such a mechanical instability can create self-organized folds within tissues. For instance, mechanical instability may underlie fold development in the erectile ruff of the frilled dragon Chlamydosaurus kingi. On the neck of C. kingi is a disc-shaped ruff that forms three stereotypical folds (Montandon et al., 2019). Ruff attachment to the neck constrains the growing ruff, generating the observed folds.

Fig. 3.

Inter-tissue-mediated mechanical instability during development of cortical folds. (A) When two adhered tissues grow at different rates (top), stress is generated, which can be relieved by buckling (bottom). (B) The human cerebral cortex exhibits a complex folding pattern, with gray matter overlaying white matter. (C) Differential growth of the future gray matter compared to the slower growing white matter may contribute to buckling in the developing brain. The white matter is primarily composed of apical radial glia cells (aRGCs) and basal radial glia cells (bRGCs) which produce the neurons that migrate to the future gray matter. The gray matter then expands by neuron maturation.

During development, the cerebral cortex of the human brain transforms from a smooth tissue into a complex folded tissue in a process called gyrification (Fig. 3B) (Llinares-Benadero and Borrell, 2019). Cortical gyrification is specific to the class Mammalia but is not present in most smaller mammals like rodents (Garcia et al., 2018). Folding enables the large surface area of the brain to fit into the cranium (Del-Valle-Anton and Borrell, 2022). In addition, folding may position axons in close proximity within a minimal volume, establishing a functional circuit (Klyachko and Stevens, 2003).

One potential mechanism driving gyrification relies upon differential growth that produces mechanical instability (Richman et al., 1975). The developing gray matter, which consists of neuronal cell bodies, grows rapidly, while the white matter, which consists of axonal projections and radial glial cells, grows slower (Fig. 3B) (Del-Valle-Anton and Borrell, 2022). This hypothesis has been modeled and tested experimentally in simulacra using gels. In one experiment, a bilayered gel was used, in which one layer expands when exposed to a solvent, mimicking differential growth in the brain (Tallinen et al., 2014, 2016). Designing the gel to have the same 3D shape and size as the developing human brain results in a fold pattern remarkably similar to that of the human cortex (Tallinen et al., 2016). In addition to differential growth, mechanical properties of the tissues, the shape of the smooth cortex before fold formation, and the density of neurons may also contribute to brain folding (Llinares-Benadero and Borrell, 2019). The exact mechanism of gray matter expansion is unknown, but it is suspected that neuronal migration sets the stage for folding by populating the future gray matter with neurons that then mature by developing dendrites, contributing to expansion (Fig. 3C) (Garcia et al., 2018).

The ECM appears to play an important role during gyrification. Multiple transcriptomic studies show that numerous genes encoding integrins and ECM proteins, such as collagens and laminins, are differentially expressed between the layers of the cortex (Amin and Borrell, 2020). The ECM may be involved in gyrification by promoting cell proliferation in the gray matter. Moreover, the ECM may aid in neuronal migration in the gray matter, contributing to its rapid expansion (Del-Valle-Anton and Borrell, 2022). For instance, the transmembrane cell adhesion molecules fibronectin leucine rich-repeat transmembrane protein 1 (FLRT1) and FLRT3 are required for cortical neuron migration (Del Toro et al., 2017). The endogenously smooth mouse cortex develops cortical folds in FLRT1 and FLRT3 double-mutant mice without increasing cell proliferation. This phenotype manifests because of reduced intercellular adhesion that promotes migration. In addition, FLRT1 and FLRT3 have low expression levels in the developing folded cortices of ferrets and humans, suggesting that their levels must remain low to enable neuronal migration (Del Toro et al., 2017).

In addition to affecting cell behavior, the ECM impacts the mechanical stiffness of the developing cortex, which can then influence folding (Llinares-Benadero and Borrell, 2019). Varied combinations of hyaluronan and proteoglycan link protein 1 (HAPLN1), collagen I and lumican (HCL cocktail) were found to induce folding when added to human fetal neocortex explants (Long et al., 2018). Proliferation did not increase upon treatment with the HCL cocktail, indicating that these ECM components must cause folding via a different mechanism. Using atomic force microscopy, the authors found that adding all three ECM components induces a stiffness pattern that mirrors the folding pattern (Long et al., 2018). Taken together, these results demonstrate how rapid expansion of one tissue can produce folding and highlights a role that the ECM may play in tissue expansion. However, more research is needed to understand the extent to which mechanical stability is involved in brain folding and its role in determining the spatial pattern of brain folds.

The effect of tension generation in neighboring tissues

Tension (force per unit length) can be transmitted through a tissue via inter-tissue adhesion. One tissue that specifically evolved to generate tension is muscle. In the developing flight muscle in Drosophila, ECM components build up between the myotube and tendon as tension in the myotube increases, enabling the connection between the myotube and tendon to be maintained during growth (Lemke and Schnorrer, 2017).

A well-studied example of tension in inter-tissue adhesion is the elongating C. elegans embryo. In the lateral epidermal cells of the embryo, the PAR module (comprising the polarity proteins PAR-3, PAR-6, and PKC-3) localizes to the lateral boundary (Fig. 4A,B) (Gillard et al., 2019). PAR polarization is required to orient the actin cytoskeleton for anisotropic elongation of cells in the correct direction (Lardennois et al., 2019). This planar polarization correlates with the onset of muscle contraction, but there is no direct connection between the muscle and lateral epidermis (Fig. 4B,C) (Gillard et al., 2019). Instead, the dorsal-ventral epidermis may act as a ‘middleman’ to transmit tension from the muscle to the lateral epidermis. The muscles and dorsal-ventral epidermis adhere to a shared ECM. The muscles bind to this ECM utilizing integrins, whereas the dorsal-ventral epidermis attaches to the ECM using C. elegans hemidesmosome (CeHD)-like junctions (Fig. 4D). The basal side of CeHD-like junctions is composed of the transmembrane basal receptor LET-805 bound to VAB-10, a spectraplakin homolog (Fig. 4D) (Bosher et al., 2003). Intermediate filaments link the basal VAB-10 to the apical receptors MUP-4 and MUA-3 (Zhang and Labouesse, 2010). Once muscle contraction begins, G-protein-coupled receptor kinase interactor 1 (GIT-1) is recruited to VAB-10, potentially due to conformational changes that expose the SH3 domain on VAB-10 (Suman et al., 2019). The signaling module composed of GIT-1, PIX-1 and PAK-1 phosphorylates the intermediate filaments, maturing the CeHD and forming a ‘molecular tendon’ that can transmit force (Zhang et al., 2011). The ‘molecular tendon’ enables tension generated by muscle contraction to be transmitted through the ECM to the neighboring dorsal-ventral epidermal tissue (Gillard et al., 2019). The mature CeHD may then transmit mechanical forces to the lateral epidermis, likely through apical adherens junctions that connect the dorsal-ventral epidermis to the lateral epidermis (Fig. 4C). This highlights the role of tension in the development of inter-tissue adhesion, as well as the role of indirect tissue adhesion in transmitting tension between tissues that are not in direct contact.

Fig. 4.

Distribution of tension by inter-tissue adhesion in the elongating C. elegans embryo. (A) C. elegans muscle (yellow) is internally attached to the dorsal-ventral epithelium (DVE; cyan), and the DVE is attached to the lateral epithelium (LE) by adherens junctions (AJs), as illustrated in detail in the insets (B and C). (B) The PAR module (PAR), consisting of PAR-3, PAR-6 and PKC-3, localizes to the lateral boundary of the LE where it helps orient the actin cytoskeleton. (C) When the muscle contacts (step 1), the DVE acts as a ‘middleman’ to transmit tension to the LE. Transmission of tension is accomplished with the help of C. elegans hemidesmosomes (CeHD-like junctions) within the DVE. As the muscle contracts, a GIT-1–PIX-1–PAK-1 complex phosphorylates and matures intermediate filaments (IFs) (step 2). The mature CeHDs in the DVE can then transmit force to the LE through apical AJs (step 3). (D) The muscle and CeHD are linked by a shared ECM. The muscle binds to the shared ECM using integrins, while the CeHD uses LET-805 (LET) at the basal side of the DVE. MUP-4 (MUP) and MUA-3 (MUA) located at the apical side are linked to LET by IFs. GIT, GIT-1; P, phosphorylation; PAK, PAK-1; PIX, PIX-1; VAB, VAB-10.

The influence of inter-tissue adhesion on morphological evolution

The field of evolutionary developmental biology (evo-devo) has made substantial progress in understanding the genetic modifications that underlie the evolution and diversification of morphological structures (Rebeiz and Tsiantis, 2017; Smith et al., 2018; Carroll, 2006). However, there is a limited understanding of the various mechanics utilized at the tissue and cellular level to diversify morphologies over the course of evolution. Recent work has begun to highlight differences in cell shape (Green et al., 2019), cell rearrangement (Osterfield et al., 2017, 2015) or ECM deposition (Smith et al., 2020) in different species of Drosophila that contribute to the development of diverse morphologies or the evolution of new structures. Here, we review model systems that have the potential to shed light on the varied ways that inter-tissue adhesion diversifies shape over the course of evolution. It is important to note that it is difficult to decipher the stepwise modifications that produce differences in inter-tissue adhesion over the course of evolution because we cannot step back in time. Instead, research focused on understanding morphological diversity can examine how aspects of inter-tissue adhesion, such as the localization of certain proteins, vary between different species and how these variations correlate with shape differences. Future research can identify the genetic changes that underlie these modifications.

Altering mechanical instability and the evolution of folding patterns

During evolution, differences in the relative rate of growth between tissues (e.g. allometry) can generate mechanical instability and buckling in the presence of inter-tissue adhesion. Differences in the mechanics of these inter-tissue adhesions can result in quantifiable differences in shape. For instance, there is a large diversity in cerebral cortex folding patterns across species that may have evolved to accommodate a larger cortical surface area (Garcia et al., 2018). Variation in brain size, stiffness, cortical growth and initial brain shape could all be involved in the evolution of cerebral cortex folding by altering how inter-tissue adhesion influences brain folding (Budday et al., 2014; Kroenke and Bayly, 2018).

An excellent example of how differences in mechanical instability bring about differences in shape across species is the looped morphology of the vertebrate gut tube. The gut tube has a looped morphology with a consistent size, shape and number of loops that are unique to each species (Fig. 5A) (Savin et al., 2011). Amid development of these loops, the gut adheres to another tissue called the dorsal mesentery. In chicken, cell proliferation in the gut tube is uniformly higher than in the dorsal mesentery, creating differential growth between the two adhered tissues. To relieve the stress, the gut tube develops characteristic loops. Physical measures of differential growth rates, tissue geometry and elasticity accurately predict looping patterns across several species. For instance, a softer gut tube and mismatched strain relative to the dorsal mesentery correlate with the tightly coiled loops specific to the mouse gut (Fig. 5A) (Savin et al., 2011). These differences in looping patterns, including the number or size of loops, are likely to be physiologically relevant, helping species adapt to new diets (Beddard, 1898). How alterations in gut looping arise from genetic variation is not understood, but changes in BMP signaling have been implicated. In the zebra finch, the gut tube has few loops and a smaller looping radius compared to the gut tube of chicken (Savin et al., 2011). Differential growth between the gut tube and dorsal mesentery is two times higher in the chicken compared to the zebra finch (Savin et al., 2011), and BMP signaling is higher in the chick dorsal mesentery (Nerurkar et al., 2017). Manipulating levels of BMP2 in the chicken suggests that BMP2 suppresses dorsal mesentery elongation, changing the degree of differential growth, and generating alterations in the number and radius of the gut loops. Thus, mutations affecting the BMP signaling pathway could underlie variation in vertebrate gut looping (Nerurkar et al., 2017).

Fig. 5.

Diversifying anatomical shape and the role of inter-tissue adhesion. (A) There is great diversity among vertebrates in the gut looping patterns and loop size, which form due to differential growth between the gut tube (pictured) and dorsal mesentery (not pictured). Representative loop radii (r) are indicated. (B) During Drosophila pupal wing development (top), the wing adheres to the surrounding cuticle (gray) using the apical ECM protein Dumpy (orange). As the hinge contracts (dark green) tension is generated in the wing (light green). Wing shape of the adult fly (bottom). The pattern of anchorage to the cuticle is hypothesized to contribute to wing shape, as tissue-specific RNAi knockdown of dumpy (purple) in the pupal wing (see top right) leads to predictable deformations to wing shape, as observed by an indentation in the wing (position marked with *). (C) In the scuttle fly, the serosa (purple, extraembryonic tissue) surrounds the entire embryo during late embryogenesis. The spreading of the serosa in the scuttle fly depends upon de-adhesion of the serosa from the yolk. (D) In contrast, fruit flies develop a combined amnion and serosa called the amnioserosa, which remains in contact with the yolk and does not spread to encapsulate the entire embryo. It is hypothesized that secretion of Mmp1 in the scuttle fly, but not the fruit fly, decouples the amnion and serosa and enables spreading of the serosa in the scuttle fly. Dashed lines indicate the position of cross sections.

Tension in the developing Drosophila wing and its potential role in the evolution of morphology

During the late stages of pupal wing development in Drosophila, contractions from the neighboring hinge pull on the wing. The wing resists this pulling by anchoring onto the surrounding cuticle with the ECM protein Dumpy (Etournay et al., 2015; Ray et al., 2015). This configuration generates tension along the proximal–distal axis of the wing (Fig. 5B top left) (Etournay et al., 2015). Cells in the wing respond to this tension by oriented cell division and cell rearrangement to elongate the wing in the direction of tension. Tissue-specific reduction in expression of dumpy in regions of the wing leads to predictable deformations in wing shape (Fig. 5B top right) (Ray et al., 2015). These changes in wing shape are observable as indentations in areas of the wing that no longer anchor to the surrounding cuticle and arise because the cells no longer resist the contractions from the hinge (Fig. 5B bottom). These observations suggest that evolution of wing shape could arise via altered patterns of wing anchorage during development (Ray et al., 2015). Interestingly, some patterning genes are expressed around the periphery of the developing wing in different species. In Nasonia, a genus of parasitic wasps, the expression of the signaling molecule unpaired-like at the periphery of the lateral wing tip correlates with wing width (Loehlin and Werren, 2012). In Lepidoptera, which includes butterflies and moths, the transcription factor cut is expressed on the periphery of the wing in a pattern that correlates with the complex contours of the wing, such as the wing tail (Macdonald et al., 2010). Currently, there is no conclusive evidence demonstrating a role for unpaired-like and cut in the evolution of wing shape. However, these genes could regulate anchorage to the cuticle and thus contribute to the evolution of varied wing morphologies by means of altered mechanics at the site of inter-tissue adhesion (Ray et al., 2015).

Temporal shifts of inter-tissue adhesion and a potential role in the evolution of tissue spreading

In many insects, extraembryonic tissues spread and encapsulate the embryo. However, there is a great diversity in the arrangement of these extraembryonic tissues across species (Schmidt-Ott and Kwan, 2016). In Drosophila, this spreading involves changes in cell shape from tall columnar cells to short squamous cells to increase the surface area of the tissue (Pope and Harris, 2008). However, Drosophila differs from many other species as their extraembryonic tissues do not expand past the yolk and do not encapsulate the entire embryo (Fig. 5D) (Schmidt-Ott and Kwan, 2016). By contrast, complete encapsulation does occur in the scuttle fly Megaselia abdita (Fig. 5C) (Schmidt-Ott and Kwan, 2016). In wild-type M. abdita embryos, there are two phases of serosa (extraembryonic tissue) spreading (Caroti et al., 2018). During the first phase, the movement of the serosa and yolk are tightly coupled, followed by a pause in serosa spreading. In the second phase, the serosa begins to spread again, but the coupling between the serosa and yolk motion is lost. Inhibition of Matrix metalloproteinase 1 (Mmp1), which breaks down ECM, causes persistent coupling of the serosa and yolk sac into the second phase. In addition, there is a delay in the closure of the serosa in Mmp1-knockdown embryos, suggesting that uncoupling of the serosa and yolk sac in the second phase enables fast spreading of the serosa in wild-type M. abdita embryos. These observations suggest that changes in Mmp1 expression could be a mechanism for the evolution of the non-encapsulating extraembryonic tissue in Drosophila (Caroti et al., 2018). Indeed, Mmp1 is not expressed in the Drosophila yolk sac (Page-McCaw et al., 2003), consequently, the extraembryonic tissue in Drosophila may remain coupled throughout development (Fig. 5C,D) (Caroti et al., 2018).

Conclusions

Multiple lines of evidence across species and developmental processes indicate that inter-tissue adhesion plays a significant mechanical role in embryonic morphogenesis. However, this is still an emerging field in developmental biology, and new tools are needed to quantify and manipulate inter-tissue adhesion with spatio-temporal control. The growing use of CRISPR-based gene editing can help solve this problem by enabling more precise spatial and temporal shifts in gene expression. Additional techniques, including optogenetic manipulation of cell mechanics or FRET-based sensors, will aid in advancing studies of inter-tissue adhesion. Moreover, comparative evo-devo studies can provide insights into how diverse shapes develop through evolutionary tinkering with the mechanics of inter-tissue adhesion. Here, vertebrate gut looping provides an excellent example of how a comparative approach among species can lead to insights into the mechanics of inter-tissue adhesion. Variations in ECM composition, cell–ECM binding, cell–cell adhesion, tissue geometry, growth rates or tissue stiffness can influence the mechanics of inter-tissue interfaces throughout development and underlie evolutionary modifications in morphology. It will be fascinating to learn which exact processes have been altered over the course of evolution and how such changes shape an organism.