Mechanical design in embryos: mechanical signalling, robustness and developmental defects

Abstract

Embryos are shaped by the precise application of force against the resistant structures of multicellular tissues. Forces may be generated, guided and resisted by cells, extracellular matrix, interstitial fluids, and how they are organized and bound within the tissue's architecture. In this review, we summarize our current thoughts on the multiple roles of mechanics in direct shaping, mechanical signalling and robustness of development. Genetic programmes of development interact with environmental cues to direct the composition of the early embryo and endow cells with active force production. Biophysical advances now provide experimental tools to measure mechanical resistance and collective forces during morphogenesis and are allowing integration of this field with studies of signalling and patterning during development. We focus this review on concepts that highlight this integration, and how the unique contributions of mechanical cues and gradients might be tested side by side with conventional signalling systems. We conclude with speculation on the integration of large-scale programmes of development, and how mechanical responses may ensure robust development and serve as constraints on programmes of tissue self-assembly.

This article is part of the themed issue ‘Systems morphodynamics: understanding the development of tissue hardware’.

1. Introduction

‘Embryology and morphology cannot proceed independently of all reference to the general laws of matter, to the laws of physics and of mechanics’.

—W. His in a letter to Professor Sir William Turner [1, p. 293]

‘There is, I think, but one way in which we may hope to find out what forces or energies are at work during development, and whether these forces are the same forces known to the chemist and physicist. Only by means of well-planned experiments can we expect by isolation and recombination to discover the forces at work’.

—T. H. Morgan on the ‘Biological Problems of To-day’. [2, p. 158]

2. Physical mechanics of development

Physical mechanics plays a key role in shaping tissues. During embryonic morphogenesis, the central role of mechanics is to direct large-scale movement of tissues, establish the basic body plan through the stages of gastrulation and neurulation, construct organ systems such as the heart and liver, and elaborate the body and organs with specific structures such as sensory organs and heart valves (figure 1, solid-line arrows). Additional roles for mechanics include providing cues to control cell behaviours and pattern cell identity (figure 1, dashed-line arrows). Metazoan embryos are typically small, their fundamental structures and physiological function established early, when embryonic tissues are compliant. Remarkably, these compliant structures continue to grow even as rigid structures, composed of dense collagen and bone, form and the embryo develops into the adult. As the adult form takes shape, these structures grow in size and their physiological roles may change. These roles can require tissues to carry out different tasks than they served in the embryo. For instance, early neural plate, pre-somitic mesoderm and notochord coordinate the anterior–posterior extension of the early embryo. Coincident with these movements they are also establishing scaffolds and laying down guidance cues for migratory cells. Later in development, cells from divergent origins integrate the physiological function across the embryo enabling whole-animal behaviours. For instance, neurons and sensory hairs in the epidermis, originally derived from ectoderm, connect to skeletal muscles to allow tadpoles to sense and activate escape behaviours from predators or turbulent waters.

Figure 1.

The roles of physical mechanics during development. The direct role of mechanics under genetic control (solid-line arrows). Genetic factors provided maternally and through pattern formation processes that establish the germ layers generate patterns of expression of morphogenetically ‘active’ proteins that regulate adhesion, the cytoskeleton, motors, ECM, etc. These protein complexes are responsible for establishing patterns of cell behaviours such as directed migration and apical contraction. Cell behaviours and their orientation are responsible for both the direction and magnitude of force generation in the embryo. These active proteins also establish the material properties of the embryo responsible for elastic, fluid and viscoelastic responses to internally or externally generated forces. Tissue movements, or deformations, are the net result of forces acting against the mechanical structures of the embryo. The rate and direction of these movements are controlled by the magnitude and direction of forces generated by multiple tissues and also by the anisotropic mechanical response of the tissue, including not only bulk elastic properties but how they adhere to one another and the geometry of their assembly. Mechanical signalling systems connect with conventional biochemical signalling systems through feedback to control molecular complexes and cell behaviours and to control gene regulatory networks and cell identity (dashed-line arrows). Feedback systems may involve mechanosensors that are integrated into the active protein complexes that regulate force production and material properties but may also reside in separate complexes that also detect environmental conditions.

The laws of physics require a direct role for mechanics in carrying out developmental programmes of morphogenesis. Genes and cues from the environment establish tissue mechanical properties, including viscoelasticity, plasticity and strength, e.g. a tissue's ability to resist fracture or delamination. These programmes further serve to activate force production, both guiding and limiting the spread of mechanical stress through the embryo. Furthermore, genetic responses to environmental conditions orchestrate the temporal aspects of morphogenesis including their start- and endpoint so tissue movements synchronize with other contemporaneous biochemical and physical processes. To drive the large-scale tissue movements of morphogenesis, cell biological targets of genetic programmes together with environmental cues must activate local cell mechanics, ECM assembly, establishment of tissue interfaces and the initiation of planar polarity [3–5].

The central role for mechanics, and even some cell mechanical processes, such as apical constriction or gel swelling, was self-evident to early biologists and embryologists even before modern concepts of gene regulation or protein-molecular complexes were known [6–8]. As advanced tools have emerged, we have discovered cell and molecular factors that control mechanics and are beginning to expose the physical principles that allow molecular or cellular mechanisms to direct the mechanical tasks of morphogenesis. We refer interested readers to some of the many recent reviews on methods to measure force production and mechanical properties within embryos [9,10] and conceptual frameworks to interpret the role of these physical factors on morphogenesis [11–15].

Experimental biomechanical approaches combined with a solid conceptual framework are critically required to disentangle the many roles of physical mechanics responsible for morphogenesis. Before we discuss the broader complexities of these roles we will highlight two studies from our own group that have sought to sort out these roles.

3. Example: the role of mechanical design in directing epithelial folding

Quantitative studies of tissue mechanics reveal key principles of mechanical design in embryos. Co-localization studies frequently describe actomyosin accumulation on the concave surfaces of folding epithelial sheets during development [16,17], but many other mechanical processes may drive folding [18]. For instance, several hypotheses had been proposed to account for primary invagination in sea urchin (figure 2a) which involved buckling or folding of an epithelial monolayer in the late mesenchyme blastula. Computational or in silico modelling of primary invagination in the sea urchin revealed many different mechanisms could plausibly generate folding, but each required a distinctive tissue microenvironment, e.g. actomyosin contractility required a compliant apical extracellular matrix (ECM; figure 2b; [19]). This theoretical study identified the apical ECM as a key mechanical feature of the sea urchin embryo that once measured directly (figure 2c; [20]) ruled out a role for apical actomyosin contractility in the real embryo (figure 2d). Similar approaches, combining computational modelling and experimental testing, have been used to rule out proposed biomechanical processes, e.g. patterns of contractility needed for head fold formation in the developing chick embryo [21].

Figure 2.

Quantitative biomechanical analysis of sea urchin embryo mechanical design reveals the contribution of stiff ECM to epithelial folding, ruling out a role for actomyosin-driven apical constriction. (a) The start of sea urchin gastrulation involves the folding of a single cell–layered epithelium to form the primitive gut tube. The basal surface of the epithelium of the near spherical embryo faces the fluidic-filled blastocoel while the apical surface is attached to a complex ECM. Folding of the columnar epithelium of the vegetal plate produces a short cylindrical archenteron composed of endoderm and secondary mesenchyme (primary mesenchyme cells have already ingressed to begin spiculogenesis within the blastocoel). (δ, depth of invagination) (b) Fully three-dimensional computational models of sea urchin gastrulation were constructed based on the sampled geometry of a living embryo with layers representing two layers of ECM and a single cell layer. These models demonstrated that several hypotheses were able to generate plausible invaginations [19] but that each placed specific constraints on the embryo's mechanical design (‘topo-map’ shows dependence of simulated depth on modulus of two ECM layers with a fixed cell modulus). Apical constriction, shown here, required a relatively soft apical ECM in order to produce a successful invagination (more than 12 µm). (c) In order to identify the mechanical regime of the living embryo, biomaterials testing with parallel plate compression were combined with ECM and cytoskeleton disruption. The measured modulus under different disruptions revealed the apical ECM was considerably stiffer (more than 4.5 kPa) than cells (0.7 kPa). (d) The measured mechanical properties of the embryo ruled out the role of apical constriction in the physical mechanics of invagination [20].

4. Example: disentangling the role of protein complexes that serve both as the source of viscoelastic material properties and as the source of force production

More recently, quantitative measurements of developmental rates, tissue stiffness and bulk force production during Xenopus gastrulation revealed a cryptic role for actomyosin. Typically, the role of a pathway or gene in a developmental process can be ruled out if inhibition or knock down leaves no phenotypic changes. For instance, Xenopus embryos and explants cultured in a Rho kinase inhibitor, Y-27632, showed no effect on either the rate of gastrulation or the lengthening of dorsal tissue isolates (figure 3a; [22]). However, quantitative measurements of mechanical properties revealed strong defects in tissue compliance [23]. The two observations appeared contradictory as other studies revealed that myosin II, a downstream target of Rho kinase was absolutely required for gastrulation [25,26]. One interpretation, based on the logic of genetic pathways, was that Rho kinase was not needed during gastrulation. However, as Rho kinase is ubiquitously involved in cell force generation we considered two alternative hypotheses. The first was that Rho kinase has a role in controlling both force production and tissue compliance; this implied that Rho kinase might be acting cryptically, its impact on mechanical resistance and force generation balanced to produce no overt phenotype. A second alternative was that cell contractility was maintained through compensatory changes in other pathways regulating actomyosin function in the embryo [27]. As we had observed Rho kinase mediated much of the intracellular actomyosin dynamics at these stages (figure 3b) [24] we reasoned we could test the cryptic role of Rho kinase by adding an artificial load to the tissue during elongation. If significant compensation of myosin II contractility were responsible for maintaining rates of morphogenesis we would expect control levels of force production. Use of a novel gel force reporter (figure 3c; [22]) revealed that Rho kinase-inhibited explants simply generated less force (figure 3d; [22]), and were thus able to elongate normally because Rho kinase served a dual function regulating both tissue compliance and bulk force production.

Figure 3.

Biomechanical analysis of Xenopus convergent extension reveals cryptic features of mechanical design and function of Rho kinase. Dorsal convergence and extension (CE) is one of the central tissue-scale movements driving gastrulation and anterior–posterior elongation of the vertebrate body plan. The process of CE can be separated from other tissue movements and studied within explants (dorsal isolates) composed of the dorsal-most tissues of early gastrula stage embryos. The rate of extension of dorsal isolates is the same as the extension of the same tissues within intact embryos. Isolates, by contrast to whole embryos, allow more experimental control over their biochemical and mechanical microenvironment (e.g. boundary conditions, surrounding materials) and provide a more regular geometry that allows easily interpretable biomechanical testing. (a) Dorsal isolates incubated in Y-27632 extend at the same rate (strain per hour) as control tissues [22] but exhibit reduced stiffness [23], which is also reflected in reduced actomyosin contractility and formation of punctuated actin ‘asters’ (b; [24]). (c) The reduced contractility without observable changes in deformation rates was paradoxical. However, a novel force-sensing gel (c) revealed that force production after Rho kinase inhibition was reduced by approximately 70% (d; [22]). Biomechanical analysis demonstrated Rho kinase controlled both stiffness and force production to a similar degree that had minimal impact on morphogenetic movements.

These two examples serve to illustrate some of the complexities of mechanical mechanisms that drive morphogenesis. Efforts are underway in a variety of developmental model systems ranging from studies of epiboly in zebrafish [28] to epithelial morphogenesis in Drosophila [29] that are seeking to tease apart the direct role of mechanics in shaping tissues from the signalling role of mechanics in programmes of self-assembly. In the remainder of this review, we will discuss the alternative roles of physical mechanics in morphogenesis, and how developmental defects, distinct from those that arise from lesions in direct mechanical mechanisms, might arise from their dysfunction.

5. Mechanical signalling during morphogenesis

Recent studies of mechanobiology of monocultures of immortalized cells suggest that signals from physical mechanical cues can pattern tissues and cells in a manner similar to more traditional cues from active ligands and chemokines. The effects of signalling through mechanical cues can be observed readily in cultured cells under a diverse range of conditions [30]. The plethora of responses of cultured cells to mechanical cues may reflect a universal response, or alternatively, reflect special programmes that enable physiological adaptation of somatic adult cells to unique environmental cues. Clear cases of embryonic cells or tissues responding to mechanical cues are far less common; such rarity may be due to infrequent reliance on mechanical cues or may simply reflect the limitation of current tools to control mechanical conditions in embryos or their inability to reliably elicit significant responses.

Several recent studies have demonstrated that physical cues can pattern embryos at multiple levels, from cell behaviours, to tissue-level planar polarity, to cell identity. On the single-cell level, embryonic mesendodermal cells isolated from Xenopus can sense tension through hemi-desmosomes; this tension, when coupled to intracellular vimentin intermediate filaments through cadherin3, can direct cell protrusive activity away from the junction under tension (figure 4a; [31]). The ability to sense tension and direct protrusions in the opposite direction may underlie the collective ‘free-edge’ spreading (figure 4b; [31]) observed during mesendoderm mantle closure [34]. At the tissue level, early tension along the animal–vegetal axis of the Xenopus embryo generated during epiboly (figure 4c; [32]) appears to establish the anterior–posterior orientation of ciliated cells many hours later (figure 4c; [32]). The ability to store a record of tension in aligned microtubules (figure 4d; [32]) many hours after tension is removed hints at a durable memory of mechanical ‘history’ that has the potential to organize many polarized features in the later embryo. Such memory and the subsequent alignment of cilia-driven flow can be induced by externally generated tension (figure 4e; [32]). Similar mechanisms of memory, together with a dynamic readout of cell mechanical conditions may combine to maintain homeostatic control of cell number in the forming epidermis, for instance, by extruding cells when density is too high, and stimulating proliferation when density is too low (figure 4f) [33]. Findings from controlled experiments with cultured Madin Darby canine kidney (MDCK) cell sheets have been confirmed in the larval epidermis of growing zebrafish larva where over-dense conditions in the tail and fin can drive cell extrusion (figure 4g) [33]. The effects of gradients of mechanical cues can provide spatial and temporal cues to pattern cell behaviours and cell identities at a level equal to that provided by biochemical cues. The three cases discussed above demonstrate the methods needed to discern the difference between permissive and instructive cues. Like methods deployed to expose the role of biochemical cues, biomechanical methods require experimental control of mechanical stimuli and quantitative ‘read-outs’ of the cellular response.

Figure 4.

Mechanical cues can instruct cell behaviours and guide cell fate choices. (a,b) Internal collective movements of mesendodermal cell precursors in the Xenopus embryo migrate from the marginal zone (an annulus centered at the ‘equator’ of the early gastrula) to the future ventral regions of the embryo that will give rise to ventral organs such as the heart. Collective migration of these cells, which exhibit a mesenchymal phenotype is coordinated and highly directed. Both biochemical and biomechanical cues for this directed movement have been proposed. (a) In order to test the role of mechanical cues, single cells isolated from mesendoderm can be cultured on a two-dimensional ECM substrate and subjected to externally applied force. Tension from magnetic beads transmitted through cadherin and the intermediate filament keratin can redirect cell protrusions in Xenopus embryonic mesendoderm cells and (b) may guide in vivo collective migration during ‘free-edge’ spreading of embryonic mesendoderm in Xenopus. (Figure adapted from [31].) (c–e) Early patterns of strain generated by gastrulation movement of epiboly, the spreading of prospective epidermis to cover the entire surface of the embryo (c), were proposed to play a role in establishing the planar polarity of cilia-generated fluid flow over the tadpole skin. Directional fluid flow, from anterior-dorsal to posterior-ventral, is driven by the coordinated beating of multiciliated cells distributed over the epidermis. Early anisotropic tissue strain during gastrulation gives rise to polarized microtubules in the ectoderm that presage the planar polarity of multiciliated cells. (e) The patterning effect of strain on cilia and flow patterns can be recreated within naive tissues using micropipette aspiration to generate strain. (Figure adapted from [32].) (f,g) Extensive non-uniform growth of epithelial tissues such as the skin and in endothelial cells such as the kidney suggested mechanical cues could drive both addition and removal of cells to maintain homeostatic balance in cell density. (f) Experimental stretching of cultured epithelial cells revealed tissues could maintain homeostatic control over cell density by extruding cells when surfaces were under compression. (g) This same mechanism appears to control epidermal cell extrusion in the forming tail and fin of zebrafish larva. Density appears to be sensed through the tension-gated membrane channel Piezo. (Figure adapted from [33].) Permissions for re-use of figures provided courtesy of their respective authors.

Mechanical patterning cues may operate like a classical morphogen gradient, inducing a range of cell identities across a broad field of responsive cells. These cues may be permissive or instructive. One case of mechanical induction might be found in early Drosophila where cells from the posterior end of the embryo can be mechanically induced to form mesoderm [35,36]. Another case can be made for the role of tissue compaction during patterning of the early tooth-field [37]. These cases suggest that physical mechanical cues can generate novel patterns of cell identities. Such a role has far-reaching implications during morphogenesis, organogenesis and later, as tissues maintain homeostasis. For instance, both cancer [38] and stem cells [39] react strongly to specific mechanical conditions. In the case of cancer, it is thought that the heterogeneous mechanical microenvironment drives phenotypic conversion of epithelial cells into invasive mesenchymal cells [40]; in the case of stem cells, the mechanical microenvironment of the niche may direct a cell to adopt a cell fate that best matches the mechanical conditions of that microenvironment [41].

In vivo studies in whole embryos or intact tissues, come with a caveat that mechanical processes may alter access or activity of preexisting growth factors or chemokines. Definitive tests to isolate mechanical properties such as stiffness from diffusion or ligand density can be more suited to in vitro analysis where the microenvironment can be carefully controlled. An elegant study of mammary epithelial branching was able to definitively rule out a role for biomechanical cues in favour of a role for diffusion of biochemical factors [42]. By using shaped wells for cultured epithelial cells the study could control shape (e.g. mechanics of constraints or geometric confinement) and proximity of wells (e.g. controlling gradient or absolute concentration of diffusible factors). Preliminary studies showed both cell- and tissue-boundary curvature correlated strongly with both proliferation and protrusive activity, consistent with a role for mechanical patterning in branching morphogenesis. An alternative biochemical explanation for the same observed behaviour is also suggested by the effects seen at high-curvature borders. Diffusive gradients are also increased at high-curvature boundaries, potentially reducing the concentration of an inhibitor of cell proliferation and protrusive activity. To test this possibility, cells were placed in high-curvature wells close enough to cells in nearby wells to increase the local concentration of any potential inhibitor. Remarkably, cells experiencing a higher density of nearby cells did not respond to the mechanical cue of high curvature. By carefully controlling the geometry of the wells and density of local cells, the study showed that curvature was a critical factor in altering the pattern of diffusion. Thus, while a mechanical role was suggested initially, such a role was isolated and ruled out by a ‘well-planned experiment’. Thus, while experiments in vivo may correlate highly with a plausible source of mechanical signalling it remains technically challenging to design experiments that would properly rule out a role for mechanics.

6. What to expect from complex mechanical signalling systems

Given that mechanics plays such a strong role in directing morphogenesis, there has been considerable support for a role of mechanics in guiding cell behaviours and patterning cell identities. As there are no gene loci for ‘stress’ or no tissue-specific knockout for ‘strain’, we must instead discuss how mechanical cues might be recognized and where productive investigation might be focused to improve our ability to isolate and test the role of mechanics in vivo. The following sections will focus on a conceptual framework for studying the potential role of mechanical cues by discussing mechanisms of mechanical signal propagation through an embryonic tissue.

For mechanical signals to have a role in patterning they must first be generated, then transmitted, and finally sensed and converted into a recognizable intracellular response. The sources of a mechanical signal could include sudden changes in tension or compression (e.g. delamination of a cell, wave of cell division, formation of new adherens junctions, changes in elastic modulus, etc.) or changes in shear strain (e.g. sensed by a tissue used as a traction substrate, onset of fluid flow, etc.). Such events generate mechanical energy in the form of molecular, cellular or tissue deformation. The net amount of energy will depend on the initial stresses produced and the strain and modulus of the transmitting structures.

We next consider how a mechanical signal might be transmitted. There are several possibilities but we shall first consider the direct transmission of mechanical signals by purely mechanical processes. The geophysics of earthquakes provides us with a good analogy where a sudden shift in rock, e.g. releasing or generating strain, at the epicenter can produce a signal that propagates through the surrounding rock. The source of the signal, the differential manner in which it spreads, and the impact at distant sites depends on the mechanical properties of the surrounding rock. Stiff rock transmits the signal with high fidelity while soft rock dissipates the signal. As the signal spreads from the epicentre it is sensed as an earthquake. Thus, mechanical events in the tissue may be transmitted directly as mechanical energy. Such transmitted mechanical energy should be detectable as a spreading wave of deformation. Direct mechanical transmission will be extremely fast, exceeding the speed of sound in water or approximately 1500 m s^–1; such a wave would pass over the entire embryo in less than 0.7 µs (1 mm diameter embryo) and evade detection by conventional high-speed cameras. Such high rates of transmission may be modulated somewhat by specialized connections between cells and the presence of rigid structures like bone. Most embryonic tissues are viscoelastic so that elastic propagating waves are very likely to be dissipated after travelling only a few cell diameters [43]; thus, mechanical cues that trigger responses at great distance or trigger responses after long time delays will need to be converted to more persistent biochemical signals. Thus, cases of mechanical signal transmission, or wave propagation, which can be detected in the laboratory will probably involve combinations of mechanical and biochemical signalling.

Mechanical energy may be sensed at any point during propagation and converted into a biochemical signal. Cells are uniquely specialized in converting mechanical to biochemical energy; biologists will recognize the conversion of a deformation to chemical signal in the principle of ‘conformational change’, a central paradigm of biochemical mechanisms that regulate the function of signalling molecules and molecular complexes in the cell. The challenge for those interested in the specialized conversion of cell- and tissue-scale mechanical energy into biochemical signals is how mechanical signals are filtered from the stochastic or biochemically induced dynamic changes in molecular complexes, e.g. how specific types of cell and tissue deformation are distinct from the conformational changes that underlie more generic biochemical signalling systems. These last considerations are ubiquitous throughout systems-based studies of cell signalling (see the ‘bow-tie’ model, [44]) and are not unique to the conversion of mechanical information to biochemical signalling.

To understand how cells sense mechanical signals, we must first consider the nature of the signal and how ‘mechanical information’ might be stored. As cells are largely viscoelastic they typically cannot sustain high levels of force, stress or strain for long before intracellular or extracellular structures remodel, dissipate energy to the surroundings and return the cell to a stress-free state. Embryonic tissues can clearly store and release elastic energy over the short term of seconds to minutes, but their long-term ability to store elastic energy is not well understood and may extend to hours or days depending on their composition and architecture For instance, tissues that incorporate oriented, elastic ECM such as collagen II fibrils may be able to store elastic energy. Such structures are not common in the early embryo but may dominate the mechanical structures and influence of mechanical cues during later development (e.g. during straightening of the Xenopus tailbud into the tadpole [45]) and organogenesis (e.g. during elaboration of complex branched epithelia [46]). This question is critical if we are to consider the role of mechanical gradients as dissipation is analogous to the role of a biochemical sink in the establishment of a growth factor gradient. To ‘lock-in’ or ‘fix’ a mechanical gradient requires that either the tissue stores the elastic energy or that the mechanical gradient is converted to more durable information before it has the chance to dissipate. The requirement that a mechanical gradient persists for a fixed time places an important constraint on any proposed mechanism of mechanical signalling just as one must consider the conversion of a chemical signal into a durable pattern of gene expression or cell behaviour. The similarities between the transmission of mechanical signals and biochemical ones are deep as the two processes can often be represented formally by the same mathematical equations [47]. Systems-based approaches to cell- and tissue-signalling have long considered these issues and we direct readers to recent discussions on the topic [48].

We next consider how cells or tissues might sense a mechanical signal. Cells may sense the passage of a mechanical cue or a mechanical gradient via intracellular or extracellular pathways. There are many diverse protein-based mechanosensors with more being identified and tested; to do proper service to the topic we refer interested readers to several of many excellent reviews on the topic [49–51]. We will briefly summarize several classes of these receptors and focus on how they may be integrated with spreading mechanical signals or standing mechanical gradients.

Intracellular mechanosensors have been discovered through single-cell studies and fall into several classes based on their localization within the cell and their proximity to load-bearing structures. Strain within load-bearing sites such as focal adhesions to ECM, or adherens junctions in epithelia, or desmosomes in the basolateral cortex of cells have each been shown to contain scaffold proteins that undergo conformational changes to expose otherwise hidden or cryptic-active sites. Cell strain may also be detected by protein-sensors in less critical sites, such as within the cytoskeleton or plasma membrane. Such protein-based mechanosensors may change conformation in response to strain in a way that hides active sites or exposes new ones, these active sites can serve an enzymatic function, e.g. provide a transient ‘on-off’ switch, or might attract other proteins that stabilize the site and create a long-lasting toggle ‘on’ or ‘off’ switch. The differences between these two sensory schemes could function to activate transient cell behaviours, or to drive the cell to change phenotype or identity.

Extracellular mechanosensors are similar to intracellular systems and rely on mechanical cues to alter cryptic-active sites to open or close under varying degrees of tension or compression. The composition and presentation of the heterogeneous extracellular network of proteins can be quickly modified to change cell behaviour as well as identity [52]. Fibronectin is a classic example of this type of mechanism, requiring cell-generated tension to polymerize [53]. Increased or decreased sites for cell attachment can alter cell behaviours as well as phenotype. Tension or slack within ECM fibrils may also open sites for remodelling, for instance, opening cryptic polymerization sites to reinforce otherwise weak ECM or opening proteolytic sites that to prune overly stiff networks [54]. In addition, many growth factors are bound within the ECM [55], and their release may depend on mechanical cues.

7. Effectors of mechanical signalling: mechanical, chemical, genetic, etc.

Once a mechanical signal has been converted into a biochemical form, there are many conventional pathways that can control the cellular response. In both plants and animals, mechanical signalling pathways appear to integrate with a wide range of conventional signalling pathways (see one of many recent reviews including [49,56–58]). Numerous cases of mechanical signalling have been identified in cultured cells, and we can break these cases down into three general categories. First, response to a mechanical stimulation might take the form of an induced mechanical process, for instance, triggering another round of mechanical signalling or may simply actuate a movement or deformation locally. Second, the response might alternatively take the form of a biochemical signal and induce a specific cell behaviour such as protrusion or assembly of a polarized structure such as a cilium. Third, the response might drive a transcriptional response. In fact, all three responses may be simultaneously triggered. We leave the reader to consider the many diverse potential responses; however, we would note that mechanical, gene regulatory and biochemical signalling pathways will have many similarities, exhibiting feedback, accommodation, time delays, saturation, de-sensitization, etc. For instance, just as one might consider the minimum number of ligands or the duration of ligand presentation before triggering a receptor we would want to know the threshold for a mechanical signalling system, or whether the cells are measuring force or strain, or how cells measure mechanical gradients.

Lastly, the combination of mechanical and biochemical signalling in the embryo presents unique challenges not found in cell culture. The most critical is that cell movements often bring tissues into contact with new sources of chemical signalling (e.g. splitting of the neuroepithelial eye-field by hedgehog secreted by migrating mesenchymal cells [59]). Known as secondary induction, this process is uniquely dependent on the success of earlier mechanical processes, yet does not involve a classic transformation of mechanical energy into biochemical energy that typifies our earlier discussions. Cycles of movement and induction are common during development and provide a permanent record of past mechanical processes. By making use of mechanical processes, this cycle may provide robust milestones to the genetic and environmental programmes of development.

We can envision complex programmes of development operating like computer codes running in parallel on multiple processors. In order to keep such programs running smoothly, computer scientists have devised a range of cross-processor communications for these programs, such as ‘block control’ and pauses for ‘synchronization’. These adaptations allow some parts of computer programs to disperse tasks to different processors where they run independently. However, as the program advances to completion it occasionally holds up all processors to ensure information needed by dispersed tasks can be shared before individual processors return to their tasks. Synchronization of dispersed tasks is important within single cells, for instance, the spindle checkpoint during mitosis that indicates all chromosomes are properly aligned and attached to a spindle. We consider programmes that direct morphogenesis, organogenesis and generic self-assembly would similarly split tasks among different tissues and cells. Each task would be started as cells become committed but would require feedback from neighbouring cells before tasks are complete and cells become specified. We might consider that subroutines of self-assembly are largely independent but, like parallel-processing computer programs, will depend on cross-communication to complete the task of embryogenesis.

8. Developmental defects, birth defects and mechanical failure

We conclude by speculating on the role of mechanics in producing developmental defects. To understand the origins of birth defects as well as the mechanisms of evolution, we need to understand how networks of transcriptional regulation, signalling and mechanical processes are integrated to create large-scale structures. Defects may arise from faults in any of these systems; to begin the process of identifying the contributions of mechanical processes to birth defects, we define a range of ‘mechanical’ defects and turn to discuss their contribution to structural and functional birth defects.

We propose several general classes of developmental defects can result from defects in mechanical processes. In one case, the tissue or organ may not match the normal shape, size or composition at the equivalent stage. Alternatively, based on the normal tables of development, a process may lag behind or advance ahead of the normal rate of development for the embryo or adjacent tissues. Lastly, a defect might manifest as an overt mechanical failure in a morphogenetic event where a tissue ruptures, two tissues slip when they should stick or as tissues buckle or fold in the wrong direction. Conversely, morphogenesis might rely on tissue buckling, rupture or slippage, and the absence of these events may lead to a defect. To be recognized, a developmental defect would have to persist and generate a phenotype observable at a specific stage when embryos are screened. However, lower costs and advanced technologies are providing new imaging tools that allow long-term recording of developing embryos. When long-term recordings are evaluated with newly developed image analysis tools, detailed movements and patterns of gene expression are revealed so that even subtle defects, transient disruptions or changes in developmental rates can be detected. Findings from these new methods are forcing us to consider the consequences of a developmental defect, which can include either stabilization of the new phenotype, robust recovery and restoration of a normal phenotype, or an avalanching defect, where the developmental defect in one tissue spreads to other tissues leading to a strongly penetrant new phenotype. New phenotypes may also reflect compensation where the impact of the defect is mitigated, where subsequent deviant shapes are transformed to more normal ones, or events in the embryo re-synchronized. A truly robust programme of development may include repair of a mechanical failure. Given the few studies of variation in normal development, it is entirely possible that a ‘defect’ observed after a molecular or mechanical manipulation may actually fall within the tolerances of normal developmental processes.

Compensation to minimize the effect of a developmental defect may lie within the robust nature of mechanical structures. For instance, engineering principles of safe design allow some structures to automatically redistribute forces if one subunit is lost. Energy stored in one part of the structure can be transferred to other parts without the exchange of mass or energy-bearing compounds. Mechanical engineers design many structures to prevent avalanching defects, for instance, channelling destructive energies of an automobile crash away from the driver and passengers. Engineering such programmes of robustness adds cost at all levels from design, testing and manufacturing. Future efforts to understand the robust programmes of development will require deeper understanding of the connection between the design phases (e.g. evolution) and manufacturing phases (e.g. gene regulatory and biochemical signalling networks) and their operation during morphogenesis (e.g. mechanics of development).

9. Conclusion

Developmental biologists have returned to questions about the mechanical design of embryos posed in the 1800s and are now integrating the principles of mechanical design with principles of genetics and cell biology. The new synthesis of biomechanics, biophysics, cell biology and genetics is critical to establish theoretical and experimental frameworks needed to carry out the goals outlined by early proponents of the field such as Morgan and His. In advancing the field of developmental biology and the science of self-assembly, we make available new tools for clinicians, bioengineers and basic biomedical researchers.

Competing interests

I declare I have no competing interests.

Funding

This work was supported in part by grants to L.A.D. from the NIH (R01 HD044750 and R56 HL134195) and the NSF (CMMI-1100515).

Disclaimer

Any opinions, findings and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the NIH or NSF.