Inherency of Form and Function in Animal Development and Evolution

I discuss recent work on the origins of morphology and cell-type diversification in Metazoa – collectively the animals – and propose a scenario for how these two properties became integrated, with the help of a third set of processes, cellular pattern formation, into the developmental programs seen in present-day metazoans. Inherent propensities to generate familiar forms and cell types, in essence a parts kit for the animals, are exhibited by present-day organisms and were likely more prominent in primitive ones. The structural motifs of animal bodies and organs, e.g., multilayered, hollow, elongated and segmented tissues, internal and external appendages, branched tubes, and modular endoskeletons, can be accounted for by the properties of mesoscale masses of metazoan cells. These material properties, in turn, resulted from the recruitment of “generic” physical forces and mechanisms – adhesion, contraction, polarity, chemical oscillation, diffusion – by toolkit molecules that were partly conserved from unicellular holozoan antecedents and partly novel, distributed in the different metazoan phyla in a fashion correlated with morphological complexity. The specialized functions of the terminally differentiated cell types in animals, e.g., contraction, excitability, barrier function, detoxification, excretion, were already present in ancestral unicellular organisms. These functions were implemented in metazoan differentiation in some cases using the same transcription factors as in single-celled ancestors, although controlled by regulatory mechanisms that were hybrids between earlier-evolved processes and regulatory innovations, such as enhancers.

Cellular pattern formation, mediated by released morphogens interacting with biochemically responsive and excitable tissues, drew on inherent self-organizing processes in proto-metazoans to transform clusters of holozoan cells into animal embryos and organs.

Keywords: biogeneric, morphogenesis, pattern formation, cell differentiation, self-organization, phase transition, reaction-diffusion

Organization and functional integration are characteristics of multicellular organisms that distinguish them from mere associations of cells. Some organisms such as the social amoebae (e.g., Dictyostelium discoideum) (Bonner, 2009) and certain of the holozoans, a clade that includes both animal and nonanimal species (Sebé-Pedrós et al., 2017), exhibit both unicellular and multicellular life-cycle phases with a small range of differentiated cell types. But, it is in the animals or Metazoa, with up to hundreds of cell types constituting multifunctional, phenotypically plastic tissues, that organization and integration are most prominent. This review will focus on this group and the biological properties that distinguish it from its nearest relatives, the nonmetazoan holozoans.

There has been a wealth of approaches to the physiological relationships between, and evolutionary bases of, organization and integration in multicellular forms (see Arnellos et al., 2014 for a review). The main treatments have been within an evolutionary adaptationist perspective. These emphasize (1) the suppression of potential intercellular conflicts of genetically heterogenous cells leading to a “unicellular bottleneck” in development, (2) functional division of labor among cell types arrived at through cycles of natural selection for improved fitness, and (3) continuing evolution for integration among subsystems. Concepts that have been advanced to describe these processes include “alignment of fitness,” “export of fitness,” and “goal-directedness” (at the level of the organism as a whole), all of which, in most treatments, are proposed to be products of natural selection (Buss, 1987; Ruiz-Mirazo et al., 2000; Queller and Strassmann, 2009; Folse and Roughgarden, 2010). To this Arnellos et al. (2014) add the idea of “autonomy,” a concept related to function and goal-directedness, which has developed alongside the discourse on organization and integration (Ruiz-Mirazo et al., 2000; Ruiz-Mirazo and Moreno, 2004) and has been explored subsequently, in depth, in a monograph by Moreno and Mossio (2015). This property is seen as foundational to the possibility of Darwinian selection.

Following this line of thought, Keijzer and Arnellos (2017) have questioned whether animal-type autonomy can itself be accounted for by selectionist and adaptationist accounts and suggest that it might not. What is required is a type of organization, characteristic of animal bodies and organs, that distinguishes metazoans from other multicellular organisms like plants and fungi, and also from the unicellular populations that were their direct ancestors, some of which have extant forms with multicellular stages. In his foreword to the Moreno and Mossio volume, Cliff Hooker problematizes the kind of organization that is found in its most intensive form in animals:

Organisation is not the same as order; pure crystals are highly ordered, so uniform they cannot show any organisation. Neither can gases, because they are too random to be organized. Organisation lies between the crystal and gas extremes but we don’t have a good theory that tells us exactly where and why. (Moreno and Mossio, 2015, p. vii)

I suggest that recent work on the mesoscale physics (i.e., processes at the 10−3–10−2 m scale) of cell collectives and on the products of the genetic toolkit that mediate these cell associations in animal tissues permit us to formulate a theory of the organizational properties of metazoan organisms. Specifically, I draw on earlier work on “dynamical patterning modules” (Newman and Bhat, 2008, 2009), developmental units in which the products of certain conserved genes mobilize physical forces and processes at the multicellular scale, and on “biogeneric materials” (Newman, 2016a). The latter are condensed forms of matter (in Hooker’s formulation, “lying between the crystal and gas extremes”) that, while arising from collections of living cells, exhibit predictable morphological motifs due to their having properties in common with nonliving materials such as liquids and elastic sheets. These concepts can help us understand how and why certain developmental forms and patterns of animal bodies and organs (multilayering, lumens, segmentation, appendages) have appeared recurrently throughout evolution.

In addition to their morphological characters, animal bodies exhibit coordinated and integrated functioning of distinct cell types. These alternative states of cellular activity, which depend on, but are not entirely reducible to differential gene expression, are partly conserved across the animal phyla, and the means of their progressive emergence from unicellular ancestors of the metazoans are only now becoming known (Brunet and King, 2017; Sebé-Pedrós et al., 2018). While the range and stereotypical character of the cell types of animals are not inherent to multicellular matter in the same way that the morphological motifs of these organisms are, their functions are intrinsic to the biology of their ancestral cells. The partitioning of these functions among up to several hundred different cell types depended on certain innovations in gene transcriptional regulation. The relations between the mechanisms of gene expression employed in the animals’ immediate unicellular ancestors and those regulatory mechanisms that distinguish them from other kinds of organisms are key to their ability to allocate inherent functions to specialized cells and will therefore be described in detail.

Beyond the inherency of tissue morphology and cell function described above, the development and evolution of complex bodies and organs involves a third category of processes that depends on, but is not a mere combination of, these two: the reliable spatial arrangement of different cell types within three-dimensional tissues (Salazar-Ciudad et al., 2003). These processes, collectively termed cellular pattern formation, are also inherent to animal systems, but in a way still different from morphogenesis and differentiation. They arise from the fact that biogeneric materials, having subunits that are living cells, are dynamically more complex than their nonliving counterparts. In multicellular systems, short- and long-range cell-cell communication via contact, chemical and mechanical signaling or electrical coupling, along with positive and negative feedback effects, produce lateral inhibitory effects, oscillations, synchrony, and repeated structures. Together, the animal-specific set of morphogenetic, differentiative, and patterning inherencies appear to be the organizational preconditions of subsequent metazoan evolution.

For each of the processes I discuss I will describe the main genes and molecular mechanisms involved in distinguishing animals in general from their inferred unicellular ancestors and some of the major organizational differences within Metazoa from one another. This has led to the identification of biological properties dependent on (1) novel genes or regulatory motifs coincident with emergence of Metazoa, (2) novel genes acquired after metazoan origins, and (3) ancestral genes repurposed to novel functions when they came to operate on the multicellular scale (Table 1).

If, as I suggest, inherency of form, function, and pattern make key features of animal body plans [and, as we propose elsewhere, those of plants (Benítez et al., 2018)] highly constrained and readily achieved, the role of natural selection for incrementally improved fitness becomes less decisive as the driving force of organismal diversification and complexification than generally believed. In the concluding section, I will discuss the relationship between these evolutionary perspectives.

In this section, I discuss the structural motifs that would inescapably characterize the metazoans once they emerged from an ancestral population of unicellular organisms. These characters were due to animal tissues being constituted as a new form of mesoscale matter by the innovation of molecular links between cells that also harnessed the cells’ motile activity. The physics of these living materials generated a specific array of morphological motifs with a constrained range of geometric and topological relationships among them. Later, with the appearance of additional novel genes (some of them phylogenetically unprecedented), new physical forces and constraints were mobilized in some of these multicellular entities, leading to successively more complex forms. The morphological properties described, which can be identified with phylotypic bodies and organs, were generally predictable properties of these materials and therefore inherent to animal tissues.

The animals or metazoans arose around 700 million years ago within an evolutionary lineage of eukaryotic cells – the holozoans – which were also ancestral to present-day unicellular choanoflagellates (reviewed in Newman, 2016b). The animals are multicellular, and in all metazoan phyla cell-cell attachment is mediated by members of the cadherin family of cell adhesion molecules (CAMs). Organisms of all these phyla express cadherins with a cytoplasmic domain that links to the actin cytoskeleton and permits cells to remain tightly bound to their neighbors, while they move past one another (reviewed in Newman, 2016a). With one exception [the ctenophores – “comb jellies” – a group that is an outlier of the metazoans in several respects (Lanna, 2015; King and Rokas, 2017; Whelan et al., 2017)], cell-cell attachment during development is mediated by the so-called “classical” cadherins (reviewed in Newman, 2016). These proteins are present in the morphologically simplest, and earliest diverging, animals, the sponges (Porifera) and the single extant species of Placozoa, Trichoplax adhaerens, the so-called basal metazoans. The cytoplasmic domain of the classical cadherins is not found in any nonmetazoan holozoans and has no counterpart in any other sequenced organisms. [The cadherins of ctenophores have a different cytoplasmic domain (Belahbib et al., 2018).] Such unprecedented coding sequences have been termed “de novo” or “orphan” genes (Bornberg-Bauer et al., 2015; Schmitz et al., 2018).

Materials with subunits that cohere but can nonetheless perambulate in a random fashion are, by definition, liquids. They exhibit surface tension and the propensity (like oil and water, for example) to phase-separate from other such materials with different cohesive properties. The liquid properties of metazoan cell clusters are based on biological functions rather than purely physical attributes. For example, in nonliving liquids, cohesion of molecular subunits is mediated by electrical forces, and the random movement of those subunits is due to Brownian motion driven by thermal energy (March and Tosi, 2002). In animal tissues the subunits are cells, whose cohesion is mediated by classical cadherins and whose random movement results from surface protrusive activity and amoeboid motion driven by cytoskeletal mechanics. Nevertheless, the physical laws that describe these two categories of liquid are sufficiently similar, so that many properties of the biological materials are predictably “generic” (Newman and Comper, 1990; Forgacs and Newman, 2005). Liquid tissues and other cell collectives with biologically based generic physical properties are thus termed biogeneric (Newman, 2016a).

The “liquid-tissue” state enabled by metazoan cadherins was a primitive defining condition of animal life. Liquid tissues have several emergent features that appeared independently of natural selection in the transition between ancestral colonial holozoans and ancestral metazoans. Liquid droplets, for thermodynamic reasons, assume the geometry with smallest surface-to-volume ratio, a sphere. Embryos and newly formed organ primordia are therefore spherical by default (Steinberg and Poole, 1982). In liquids that contain two different kinds of subunits (molecular species, in most physical examples), one of which has greater affinity for its own type than the other (due, in liquid-tissues, to different amounts or types of cadherins on cell surfaces), there is a separation of phases. The interface between the phases can be flat or curved, depending on the balance of self- and nonself-affinities, and in extreme cases, one liquid phase will completely engulf the other. Such configurations are common in cell mixing and sorting experiments (Forgacs and Newman, 2005). Gastrulation, the organization of a coherent mass of cells into closely apposed but nonmixing layers, which is the first step in body plan formation during development, is directly attributable to phase separation in some animal systems (Krieg et al., 2008). In other organisms, where the connection to phase separation is not as direct, more elaborate mechanisms of gastrulation may have built on morphological templates arising from this inherent tissue behavior early in their evolution (Newman, 2016a).

Phase separation of cell populations in a liquid tissue mass illustrates the concept of dynamical patterning modules (DPMs) mentioned above (Newman and Bhat, 2008; Newman et al., 2009). In this example, cells with different homotypic and heterotypic adhesive strengths undergo spontaneous spatial redistribution to a predictable pattern by mobilizing, in multicellular clusters, a physical force (cell-cell adhesion) that is irrelevant to individual cells. The molecules that mobilize the physical effect are classical cadherins, which arose concomitantly with the metazoans. Other such examples of DPMs co-originating with Metazoa will be described in what follows. But in still other cases, proteins that evolved in unicellular ancestors took on novel DPM roles in the multicellular context, or novel gene products acquired later in animal evolution took on developmental functions by harnessing new physical processes in the form of DPMs.

In addition to having cytoskeletally linked classical cadherins, the metazoans are distinguished from all other life forms by the protein Wnt, which is similarly specified by an orphan gene unique to this group of organisms (reviewed in Newman and Bhat, 2009)1. Wnt induces reorganization of the cytoskeleton, acting via surface receptors (e.g., Frizzled, LRP5/6), β-catenin [itself unique to Metazoa among its close unicellular relatives, but with a homolog in the more distantly related D. discoideum (Miller et al., 2013; Gul et al., 2017)], and the PAR complex of adapters and protein kinases (Lang and Munro, 2017). The PAR complex predated the metazoans and serves to make the surfaces of cells nonuniform by mediating the insertion of different sets of integral membrane proteins on different portions (e.g., apical, basolateral) of the plasma membrane. In a manner that is physically analogous to polar molecules spontaneously organizing into micelles in water, liquid tissues containing apicobasally (A/B) polarized cells spontaneously form lumens and interior spaces (Forgacs and Newman, 2005).

To move beyond the simple forms of sponges and placozoans, two additional features were required, planar cell polarization (PCP) and the capacity to form a basal lamina. These effects, in turn, required several molecular innovations, using genes not present in the basal metazoans. One of these effects is a “noncanonical” Wnt-activated signaling pathway that reorganizes the cytoskeleton via the PAR complex but does so (in contrast to the canonical pathway described above) in a β-catenin-independent fashion. This pathway is facilitated by the inhibitory effect of the β-catenin antagonist Vang/Stbm, still another orphan gene, in this case unique to the eumetazoans, animals that develop from two or three distinct embryonic layers (diploblasts and triploblasts). The PCP pathway may have arisen in the common multicellular ancestor of the metazoans, but key components were lost in most sponge groups and placozoans, and it is incompletely present in ctenophores (Schenkelaars et al., 2016).

Unlike the canonical Wnt pathway, the PCP pathway causes cells to be polarized in their shapes or spatial orientation rather than over their surfaces. In a multicellular context, cells that elongate due to PCP align and intercalate, which causes the tissue mass to narrow in the direction of intercalation and elongate orthogonally to it. This tissue reorganization, termed convergent extension (reviewed in Forgacs and Newman, 2005), is the biogeneric counterpart of the deviation, in liquid crystals, from the spherical default shape of a liquid droplet due to alignment of polymers or anisotropic nanoparticles (reviewed in Newman, 2016).

The evolutionary appearance of these “liquid-crystalline tissues” was accompanied by the capacity to form epithelia, planar tissues whose cells reside on a stiff extracellular layer, the basal lamina2. Direct cell-cell attachment in such tissues continues to be mediated by cadherins, as in the epithelioid mode of tissue organization that characterizes some nonplanar tissues in all metazoans. But the basal lamina constitutes a substratum upon which the liquid-tissue can adhere and spread, a physical process referred to as “wetting” (Jensen et al., 2015). Depending on the relative degree of affinity between the basal regions of the apicobasally polarized cells that produce the basal lamina and the substratum itself, and their lateral affinity to one another, the epithelium will be columnar, cuboidal, or squamous.

Basal laminae are absent in the placozoan and most sponge species but are present in all eumetazoans (reviewed in Newman, 2016). The basal lamina is predominantly composed of the extracellular matrix (ECM) protein type IV collagen, which is produced by all metazoans, but only in the more morphologically complex, eumetazoans and the homoscleromorph sponges are its molecular subunits chemically crosslinked into a basal lamina (Fidler et al., 2017). This crosslinking is catalyzed by the enzyme peroxidasin, which is specified by an orphan gene unique to this group of organisms (Fidler et al., 2014)3. The capacity to form epithelia, in coordination with ability of the main body mass and subsidiary buds to undergo convergent extension, permitted diploblastic, and eventually triploblastic, eumetazoans to generate elongated bodies, appendages, tentacles, and tissue ridges, folds, and clefts. These morphological motifs were not only enabled by these new tissue organizational properties, but they were also made readily achievable.

With the evolutionary emergence of a third tissue layer, which came to be sandwiched between the two epithelial germ layers in one or more diploblastic ancestors, the triploblasts, the most ecologically successful of the metazoan groups, came into existence (reviewed in Newman, 2016). The middle layer – mesoderm – enabled generation a new array of morphological motifs, which were correspondingly inherent to these new organisms. These intrinsic organizational propensities led to triploblasts having more complex body plans than diploblasts, and, in contrast to the latter, geometrically complex organs containing multiple cell types. An important reason for this was that the surface and enteric epithelia in triploblasts could continue to serve (as in diploblasts) as a boundary between the organism and its environment, while the mesodermal layer, having no analogous protective role, could disaggregate into a new physical state consisting of cells dispersed in a matrix, a mesenchyme. Mesenchymal tissues, on their own, or in interaction with an adjacent epithelium, could form structures that were impossible in simpler animals, but predictable, based on the additional physical effects to which they were susceptible (reviewed in Newman, 2016b).

The morphologically simplest triploblasts develop with a single body cavity, the digestive tube. Such organisms, including flatworms like planaria, have ovaries and testes, and clusters of neurons, termed ganglia. Most triploblasts, however, such as arthropods, mollusks, and chordates, have a second body cavity, the coelom, within which digestive tube is suspended. In these coelomate triploblast lineages, organ complexity took a dramatic leap. This is because instead of their being just two epithelial-mesenchymal interfaces, as in acoelomates, there were now four. The interaction of the epithelium of the body surface (the ectoderm) with its underlying mesenchyme produced, in various species, bristles, hairs, feathers, teeth, and limbs, while the interaction of body lining epithelium (the endoderm) with its overlying mesenchyme produces intrinsic and extrinsic elaborations of the digestive tube, respectively, villi and crypts, and the liver and pancreas. The blood vessels and heart, lungs, and urogenital organs, as well as various glands are generated by invagination and evagination, folding and branching, and thickening and thinning of composite epithelial-mesenchymal layers in other regions of the developing embryo (reviewed in Newman, 2016a).

Similar to the rise of the animals as a whole, and in their turn, the eumetazoan subgroup, the appearance of the triploblasts was accompanied and facilitated by novel gene products. These included structural ECM molecules such as fibronectin and tenascin (Hynes, 2012), glycans bearing unusual sugar moieties, and enzymes such as Has2 [which may have had a viral origin (Blackburn et al., 2018)] that generate ECM polysaccharides such as hyaluronan, which induce cell motility (Spicer and Tien, 2004). These molecules led to the ingression, typically accompanied by disaggregation of an epithelial layer (epithelial-mesenchymal transformation; EMT), in one or more diploblastic ancestors. Because of the heterogeneity of the mesoderm-inducing components, it is possible that triploblasts might not be monophyletic (Burton, 2008).

Mesenchymes and their mature counterparts, connective tissues, lack direct integration of cell-cell attachment with cytoskeletally driven motility and are therefore not liquid tissues by the definition above. They are nonetheless biogeneric materials, variously viscous, viscoelastic, or solid, depending on the composition of their ECMs or (as described below) the state of mechanical stress of the contained cells. As with liquid- and liquid-crystalline tissues, mesenchymal tissues have characteristic inherent forms. One prominent example is cell condensation, wherein groups of cells embedded in ECM are drawn closer to each other (a structural motif unavailable in epithelioid tissues), forming transient clusters with epithelial-type contacts. Condensations serve as the primordia of hairs, teeth, and feathers, as well as the bones of vertebrate fins and limbs (reviewed in Forgacs and Newman, 2005).

Cnidarians, which are diploblasts, can form solid-phase exoskeletons consisting of calcium-based mineral or the polysaccharide chitin. In later diverging triploblastic forms such as bivalve mollusks and arthropods, exoskeletons continue to be produced by epithelia. But the presence of mesoderm enables some triploblasts, such as echinoderms and chordates to also produce endoskeletons (reviewed in Brown, 2011). These are typically based on patterned arrays of mesenchymal condensations (Hall and Miyake, 2000) and become solid by the production of collagenous ECMs, containing a high density of proteoglycans in cartilage, or hydroxyapatite mineral in bone and the dentin of teeth. (Tooth enamel is a mineralized exoskeletal tissue.)4

Mesenchymal tissues can also undergo a different rigidifying transformation termed “fluid-to-solid jamming” (Mongera et al., 2018). This is a phase transition between the liquid-tissue state and a solid state [analogous to glass transitions in supercooled fluids (Sadati et al., 2014)], in which persistent cytoskeletally mediated stresses at a supracellular scale solidify the tissue. Jamming is employed as a mechanism (alternative to convergent extension) of elongation of the primary body axis in zebrafish embryos. It is impaired by knockdown of a cadherin, suggesting that it depends on a shift from mesenchymal ECM-based tissue cohesion to an epithelioid mode of cohesion, like mesenchymal condensation. The nominally “solid” phase after jamming therefore exhibits some liquid-tissue properties. Specifically, cell-scale stresses fluctuate rapidly, enabling cell rearrangements, thus effectively melting the tissue at the growing end and permitting longer-term supracellular-scale stresses to guide morphogenetic flows in fluid-like tissue regions.

Metazoan tissues thus have inherent morphogenetic propensities due to their liquid, liquid-crystalline, planar, and solidifying nature. These comprise mutlilayering and lumen formation (which is seen in all groups, even the most basally diverging ones), body elongation and appendage formation (which are seen in all eumetazoans), and organ formation (found in all triploblasts). The bodies of triploblasts can also become segmented by the deployment of oscillatory processes intrinsic to cells and tissues (see below, “Pattern Formation: Inherency in the Integration of Form and Function”). The structural motifs of organs, tissue multilayers, folds, ridges and projections, straight and branched tubes, segments and lobes resemble those that produce the phyletic body plans and employ the same conserved set of developmental toolkit genes. Since the relevant biogeneric properties, the genes that mobilize them, and the inherent forms they facilitate, arose very early in the history of the metazoans, gradualist natural selection did not bear the exclusive burden of producing the major structural features of animal bodies and organs.

Morphological motifs are not cell types, however, and it is through the functional specializations of their cells that the physiology of animal life is realized. I describe next how ancestral and derived inherency rather than classical gradualist trial-and-error scenarios plausibly accounted as well for the evolution of cell differentiation in metazoans.