Regulation of gastrulation movements by emergent cell and tissue interactions

Abstract

It is during gastrulation that the primordial germ layers are specified, embryonic axes become morphologically manifest, and the embryonic body plan begins to take shape. As morphogenetic movements push and pull nascent tissues into position within the gastrula, new interactions are established between neighboring cells and tissues. These interactions represent an emergent property within gastrulating embryos, and serve to regulate and promote ensuing morphogenesis that establishes the next set of cell/tissue contacts, and so on. Several recent studies demonstrate the critical roles of such interactions during gastrulation, including those between germ layers, along embryonic axes, and at tissue boundaries. Emergent tissue interactions result from - and result in - morphogen signaling, cell contacts, and mechanical forces within the gastrula. Together, these comprise a dynamic and complex regulatory cascade that drives gastrulation morphogenesis.

Graphical abstract

Introduction

Lewis Wolpert is credited with saying that “it is not birth, marriage, or death, but gastrulation that is the most important time of our lives”. Indeed, he has a point: a great many critical events comprise this relatively short phase of embryonic development. At its conception, the embryo consists only of a mound of pluripotent cells, but by gastrulation’s end, an animal form has begun to take shape. It is during gastrulation that the anteroposterior (AP) and dorsoventral (DV) embryonic axes are established, the primordial germ layers are specified, internalized, and subsequently shaped into a rudimentary body plan with organ anlagen. Newly formed tissues thin and expand by epiboly movements, and lengthen along the nascent AP axis during gastrulation concomitant with mediolateral (ML) narrowing in a highly conserved process termed convergence and extension (C&E) [1]. This taking of shape, or morphogenesis, is the essence of gastrulation, and is accomplished through an intricate series of individual cell and collective tissue behaviors that are precisely coordinated in both space and time with embryonic axis formation (reviewed in [2]).

It has long been understood that signaling molecules, termed morphogens, diffuse throughout the developing embryo to instruct the fate of cells in a concentration-dependent manner. Localized sources of such morphogens and their antagonists establish gradients across an entire embryo or tissue, and thus affect the fate and behavior of cells at a distance [3,4]. Many embryos possess small regions termed “organizers” that are the source of many morphogens simultaneously, and which function during gastrulation to orchestrate cell fates and morphogenetic behaviors throughout the entire embryo [5,6]. Such global signaling patterns AP and DV body axes, specifies germ layers, patterns tissue sub-types, and regulates gastrulation movements [7]. As this inductive cascade unfolds, new cellular interactions are established: between cells of adjacent germ layers, between neighbors with different positional values along an axis, and at tissue boundaries within the nascent germ layers (see graphical abstract). These cellular interactions in turn inform subsequent morphogenetic movements, and therefore comprise an emergent aspect of gastrulation. Each is regulated by a distinct set of molecules, and each produces a distinct cell behavior that contributes to proper shaping of the embryo. In this review, we discuss recent advances in our understanding of dynamic cellular interactions that drive morphogenesis during gastrulation and their underlying cellular and molecular mechanisms.

Interactions between cells of adjacent germ layers

At the onset of gastrulation, mesoderm and endoderm germ layers internalize through the blastopore (or its equivalent) and the mesoderm comes to lie between the deep endoderm and superficial ectoderm. New interactions are therefore established between cells within adjacent germ layers, and these vertical interactions are critical to gastrulation morphogenesis. Although cells of different germ layers can and do interact, they must also form stable boundaries between them that keep the layers segregated. In Xenopus embryos, separation of ectoderm and mesoderm is regulated by Eph-ephrin signaling [8]. Eph receptors and their ephrin ligands often exhibit complementary expression patterns between adjacent tissues and facilitate repulsive interactions between them, thereby forming a tissue boundary (reviewed in [9]). During Xenopus gastrulation, particular combinations of Ephs and ephrins have been shown to regulate inter-germ layer interactions and segregation [8] (Fig. 1A’). Frizzled7 was also identified as critical for Xenopus germ layer segregation during gastrulation by triggering downstream non-canonical PKC-dependent signaling [10]. An overexpression screen for molecules that disrupt boundaries between germ layers also identified the cell adhesion molecule EpCAM as a potential regulator of germ layer segregation [11].

Figure 1. Emergent cell and tissue interactions during gastrulation.

Examples of cell and tissue interactions in Xenopus (A–A”), Zebrafish (B,B’), and Drosophila (C–C”) gastrulae. A) A’ depicts vertical interactions that establish a boundary between nascent mesoderm (red) and ectoderm (blue). A” depicts accumulation of Myosin at the boundary between axial (purple) and paraxial (red) mesoderm and at AP cell junctions. Dorsal is to the right. B) B’ depicts migration of endoderm cells (yellow) along the mesoderm layer (red). Gray triangles represent Wnt and BMP signaling gradients with important roles in gastrula patterning and morphogenesis. Dorsal is to the right. C) C’ depicts Myosin accumulation at boundaries between AP body parasegments (shown in pink, purple, and green). Anterior is to the left.

In spite of boundaries between them, or perhaps facilitated by them, communication between germ layers contributes substantially to gastrulation morphogenesis. Xenopus neuroectoderm, for example, undergoes C&E when explanted [12,13], but this is accomplished via a different suite of cell behaviors depending on whether it is isolated from the underlying mesoderm [14]. A recent study also implicated mesoderm internalization in establishing cell polarity and asymmetric stabilization of planar cell polarity (PCP) signaling components (discussed further below) within the overlying epidermis [15]*. Evidence suggests that this is likely a response to mechanical strain resulting from mesoderm internalization, however, rather than chemical signaling. Indeed, application of external strain can polarize cells in Xenopus gastrulae [15,16], and mechanical forces resulting from mesoderm and/or endoderm internalization have also been suggested as a contributing factor during germ band extension in Drosophila [17,18]. Similarly, friction between anterior axial mesoderm and the overlying neuroectoderm in zebrafish gastrulae is required for proper morphogenesis of the neural plate [19]**. Finally, the enveloping layer (EVL), the outermost epithelial cell layer, is required for doming and spreading of the blastoderm underlying epiboly during zebrafish gastrulation [20]*. As above, this process reportedly does not depend on a signaling molecule, but rather a reduction of surface tension within the EVL [20]. Additional examples of tissue mechanics regulating morphogenesis are well documented and are further reviewed in [21].

Other cases of inter-germ layer communication involve very specific chemical signals. Internalized endoderm cells, which spread via random walk beneath mesoderm in zebrafish gastrulae [22], express the chemokine receptor Cxcr4, while cells in the overlying mesoderm layer express its ligand Cxcl12 (Fig. 1B’) [23,24]. This signal from the mesoderm coordinates endoderm migration with that of mesoderm through regulation of Integrin-dependent cell adhesion [24] and/or instructive chemotactic signaling [23]. Another receptor-ligand pair, the complement protein C3a and its receptor C3aR, is critical for radial intercalation during Xenopus gastrulation [25]*. Unlike ML intercalation underlying C&E (see below), radial intercalation describes the insertion of cells into a layer above or below their own, which promotes thinning and spreading of the tissue underlying epiboly. In Xenopus gastrulae, cells of the deep layer express C3aR, and the complementary expression of its ligand in the superficial level was proposed to drive intercalation of deep cells into this adjacent cell layer via chemotaxis [25]. In mouse gastrulae, radial intercalation of cells from the mesoderm into the extraembryonic visceral endoderm layer is also critical to formation of the definitive endoderm. Fgf8 and Sox17 expression are required for this process [26], but whether chemotaxis is involved is unknown. Together, these examples highlight the importance of vertical cell and tissue interactions in driving gastrulation morphogenesis.

Interactions between neighboring cells along the AP axis

The dorsal gastrula, or Spemann-Mangold, organizer is induced before gastrulation, but it is during gastrulation that its inductive activities establish AP and DV embryonic axes that in turn provide spatial coordinates for gastrulation morphogenesis. Extension of the AP body axis is the most dramatic of the gastrulation movements, and it requires AP axis patterning. Indeed, Xenopus dorsal mesoderm explants must contain tissues with different AP positional values to elongate [27], and normal AP axis patterning is likewise required for Drosophila germ band extension [28,29]. In both systems, interactions between neighboring cells at different AP positions produce polarized cell intercalations that drive AP neighbors away from one another (Fig. 1A”, C’), leading to C&E of the embryonic body. Cell intercalation can be the result of at least two distinct cell interactions: protrusion-based cell shuffling or asymmetric junction remodeling [30].

Cell shuffling was first described as a mechanism for C&E in the dorsal mesoderm of Xenopus gastrulae [31], and has since been observed in numerous tissues and species, including the paraxial mesoderm and neural plate of mice [32,33], axial mesoderm of zebrafish [34] and ascidians [35], and the dorsal hypodermis of C. elegans [36]. During cell shuffling, cells exhibit ML intercalation behavior (MIB), which entails elongation of the cell body perpendicular to the axis of extension and formation of bipolar protrusions at their medial and lateral ends [37,38]. These protrusions are thought to gain traction on adjacent cells, thereby pulling them in between one another and separating AP neighbors [1]. In Xenopus dorsal mesoderm, this process involves contractile actomyosin networks [39] that are anchored at Cadherin-based adhesions between neighboring cells [40]. Although cellular protrusions required for this mode of intercalation are generally associated with mesenchymal or mesenchymal-like cells [33,37,41–43], they have also been observed during epithelial C&E [32,36,44]. Furthermore, the polarity of these protrusions is regulated by PCP signaling in vertebrate embryos. The PCP pathway was first discovered in Drosophila for its role in polarizing cells within the plane of an epithelial sheet [45,46], and is thought to act as a “molecular compass” that translates global embryonic polarity cues into cellular polarity through cell-cell interactions (reviewed in [47–49]). In vertebrate gastrulae, PCP signaling polarizes cells with respect to the AP axis in part through the asymmetric localization of its core components to anterior and posterior cell membranes [50–54], which act upstream of small GTPases to regulate cell shape, actin-based protrusions, and myosin contractility [64,65,68,69]. PCP loss of function impairs ML cell elongation and causes normally bipolar protrusive activity to become randomized during C&E [56], thereby reducing biased intercalations and tissue extension [32,50]. Indeed, loss of PCP function disrupts C&E of the embryonic body in zebrafish [44,55], Xenopus [56–58], mouse [59–64], chick [65,66], and ascidian gastrulae [67]. The PCP pathway in both flies and vertebrates shares components with Wg/Wnt signaling, leading to the hypothesis that Wnt ligands may represent the global signals that instruct PCP signaling. Some (so called non-canonical) Wnts are indeed required for C&E in some vertebrate species, including Wnt5 and Wnt11 in zebrafish [42,43], Xenopus [70,71], and chick [72]. wnt5 is expressed in a P to A gradient in zebrafish, and wnt5 mutants cannot be rescued by global expression of wnt5 RNA [43], suggesting a possible instructive role for this gradient in establishment of planar polarity. Indeed, a recent report describes how exogenous Wnt5 and Wnt11 ligands repolarize PCP components Prickle3 and Vangl2 away from the Wnt source in Xenopus gastrulae [52]**, but whether they play this role in vivo is unknown. Furthermore, studies in Xenopus explants point to graded Activin/TGFβ/Nodal-like signaling as sufficient to confer AP tissue polarity that drives MIB behavior upstream of or in parallel to PCP signaling [27]. Underscoring the essential role of such global axis patterning by TGFβ signaling in C&E, zebrafish gastrulae in which excess BMP activity disrupts DV axis patterning also exhibit defective C&E cell movements [73]. The ventral to dorsal BMP gradient limits C&E to dorsolateral regions of the gastrula by negative regulation of PCP gene expression [58,73] and disruption of a Cadherin-based cell adhesion gradient along the DV axis [74].

The second type of cell interaction underlying C&E - asymmetric junction remodeling - also drives biased cell intercalations with respect to the AP axis, but is distinct from cell shuffling in several important ways. Asymmetric junction remodeling is typified by Drosophila germ band extension, during which cell interfaces preferentially shrink along the DV axis and expand along the AP axis (either by T1 transitions or rosette resolution), thus separating AP cell neighbors [29,75,76]. Unlike the biased protrusive activity that drives cell shuffling, the driving force of this flavor of intercalation is thought to be the preferential accumulation of non-muscle Myosin along shrinking cell interfaces [76]. This mode of intercalation is most often observed within epithelial tissues with well-defined apical junctions, like the neural plate of chick and mouse embryos [32,65], but was also described within the dorsal mesoderm of Xenopus [77] (Fig. 1A”). In either case, asymmetric localization of Myosin is essential for biased intercalations that drive C&E and is polarized with respect to the nascent AP axis. In vertebrate examples of asymmetric junction remodeling, this polarity is also regulated by PCP signaling. A recent study of explanted Xenopus dorsal mesoderm describes PCP-dependent localization of Septins to cell interfaces between AP neighbors (Fig. 1A”), which restrict Myosin contractility to these junctions to ensure properly biased cell intercalations [77]. During Drosophila germ band extension, however, it is not PCP signaling but combinations of Toll-like cell surface receptors that encode the positional identity of cells and polarize Myosin contractility with respect to the AP axis [78]**. This positional code is then translated into cell intercalation behavior by preferential accumulation of Myosin at DV interfaces between cells with different Toll-like receptor profiles [78,79] (Fig. 1C,C’). Modeling further posits that this type of combinatorial code is sufficient to inform AP positional identity, and requires a minimum of three cell-surface molecules to be mixed and matched along the body axis [79]**. Myosin-dependent tension at DV interfaces stabilizes and recruits additional Myosin, creating a robust feed-forward loop that ensures proper directionality of neighbor exchanges [80]. Regardless of the mode of cell intercalation, these examples demonstrate that morphogenetic machinery within cells must be polarized with respect to the nascent embryonic axes to effectively promote axis extension.

Interactions at tissue boundaries

During embryonic axis patterning, smooth and continuous morphogen gradients are translated into discrete tissue types. This means that cells just below a signaling threshold adopt one identity, but their immediate neighbors just above the threshold can adopt a distinct identity, resulting in a boundary between them. Unlike boundaries between germ layers, boundaries between neighboring and distinct tissues within the same germ layer are often oriented within the plane of the AP axis (rather than vertically, see graphical abstract). This orientation therefore puts them in the position (literally and figuratively) to regulate cell behaviors underlying embryonic AP axis extension. The process of Drosophila germ band extension described above is thought to require more than simple patterning along the AP axis, but interactions at tissue boundaries between AP body segments (Fig. 1C,C’). The expression domains of Toll-like cell surface receptors that comprise the AP positional code in germ band cells described above [78] likely coincide with parasegment boundaries at which Myosin accumulation and cell intercalations are enhanced compared with non-boundary cell interfaces [79], demonstrating an outsized role for these boundaries in C&E. Indeed, Drosophila embryos with mutations in pair-rule genes, which establish boundaries between AP body segments, exhibit reduced axis extension [29].

The above example illustrates how boundaries perpendicular to the AP axis can promote C&E. The notochord boundary between axial and paraxial mesoderm of vertebrate embryos also contributes to C&E, but instead runs parallel to the AP axis. Similar to germ layer boundaries, the notochord boundary is established by modulation of inter-tissue adhesion mediated by Eph-ephrin signaling [81] (Fig. 1A”). Complementary expression of Eph receptors and ephrin ligands in the axial and paraxial mesoderm is associated with Myosin accumulation at this boundary, which reduces Cadherin clustering to suppress cell-cell adhesion across the boundary [81]. Unlike DV boundaries in the germ band, however, this Myosin accumulation promotes ML cell polarity and intercalation toward (i.e. perpendicular to) the boundary, thereby extending - rather than shortening - it [38,82]. The influence of this boundary can even reach several cell diameters to promote MIB at a distance [38]. The exact mechanism by which the notochord boundary promotes perpendicular intercalation is unknown, but tissue boundaries in other developmental contexts may provide clues. Boundaries between Drosophila imaginal disc compartments and at the leading cell edge during dorsal closure also accumulate Myosin [83,84], which creates increased tension that promotes cell intercalation toward the boundary [85–88]. Once again, these examples imply an important role for mechanical forces in addition to chemical signaling during morphogenesis. Interestingly, notochord boundary-associated cell polarity is independent of PCP signaling, whose role in ML cell polarization was discussed above. In both ascidian and zebrafish embryos, intact notochord boundaries and PCP signaling are both required for - and cooperate to promote – C&E [89] (Williams & LSK, unpublished data). These studies underscore the critical role of tissue boundaries in polarizing cell behaviors underlying axis extension.

Conclusions

Gastrulation entails a sequence of dynamic and precisely coordinated patterning, cell fate specification, and morphogenetic processes. Early signaling events define embryonic polarity and germ layers, and provide instructions for these germ layers to fold and reshape into a nascent body plan. As these inductive and morphogenetic processes progress, new interactions are established that facilitate additional short and long-range intercellular signaling, and so on. Importantly, nascent tissue interactions are not simply the result, but also drivers, of gastrulation movements. An appreciation of cell-cell communication beyond morphogen gradients is critical to our understanding of morphogenesis during gastrulation and beyond, and the spatiotemporal coordination of these numerous complex signaling events with one another and with embryonic patterning will be critical areas for future study.

Highlights.

• New cell and tissue interactions represent an emergent property of gastrulation.

• Interactions arise between germ layers, along body axes, and at tissue boundaries.

• Interactions entail cell signaling, cell adhesion, and/or mechanical forces.

• Emergent interactions drive gastrulation morphogenesis.