1. Introduction
The emergence of a body axis is a central step in animal morphogenesis. In many cases, the animal’s body axis provides two separate guides for establishing the basic layout of the developing body plan: a symmetry axis and a polar direction. The alignment of the axis defines the symmetry axis of the animal’s body plan, while the polarity of the axis marks a breaking of symmetry, guiding the development of functional tissues asymmetrically along the axis. Here we take advantage of the relatively simple body plan of the freshwater animal
Hydra to separate these two features of the body axis and focus on the emergence of polarity along a developmental trajectory.
Hydra’s tubular body is characterized by a single oral–aboral body axis [
1]. The alignment of this axis defines the radial symmetry axis of the animal, whereas the polarity of the axis reflects the breaking of symmetry marked by the presence of a head at the oral end of the animal and a foot at the aboral end. The use of live imaging allows us to follow the establishment of the oral–aboral body axis at high spatial and temporal resolution throughout whole-body regeneration from small excised tissue strips, from the initial folding of the tissue towards the emergence of a fully developed mature animal.
The alignment of
Hydra’s body axis is supported by parallel arrays of supracellular muscle fibers called myonemes that are oriented parallel to the body axis in the ectoderm and in a perpendicular, circumferential orientation in the endoderm [
2,
3,
4]. This bilayered muscle-like organization provides structural support for the animal, upholding its tubular symmetry and allowing its behavioral movements. The polarity of the body axis is manifested at multiple levels, starting from the asymmetric distribution of various signaling molecules along the head-to-foot axis [
5,
6,
7], and eventually realized in the different morphological features and functional tissues in the animal [
1]. In particular, the head organizer localized at the tip of the mouth of the animal has a special role in keeping the integrity of the body plan and continuously defining the polarity of the body axis by constantly inducing position-dependent signals [
8]. During whole-body regeneration, after the removal of the existing organizer, a new head organizer has to form to establish and stabilize the emerging body axis polarity.
A tissue segment excised from a mature
Hydra has a strong memory of both the alignment and the polarity of the original body axis of the parent animal [
4,
9,
10,
11]. These inherited features provide important initial conditions that dominate the morphogenesis process and trigger the emergence of a regenerated body axis, stabilizing both its alignment and polarity. An excised tissue segment first folds into a hollow spheroid in order to regenerate [
4,
12]. We have previously shown that excised tissues inherit the parallel actin fiber arrays and that this order is partially maintained during the folding process, conferring an initial directionality in the folded spheroids [
4]. The alignment of the supracellular actin fibers in excised tissues confers a structural memory that persists during regeneration, and defines the alignment of the body axis of the regenerated animal. We have confirmed that this initial directionality and the structural inheritance it conveys are present down to the smallest tissue segments capable of regeneration.
The strong memory of body axis polarity is evident in bisected
Hydra that regenerate a head or foot according to their original polarity [
13]. Later work showed that the memory of polarity is retained even in small excised tissue segments [
9,
10,
11,
14]. Note that the supracellular actin fibers contain multiple actin filaments of mixed polarity. Thus, while these fibers can support the inherited alignment, they cannot provide a source for polarity memory since these contractile fibers are apolar. The Wnt signaling pathway is known to play a central role in the definition of body axis polarity in
Hydra [
5,
6,
7], and local activation of the Wnt pathway can induce head formation [
10,
15]. However, despite extensive research, it is still unclear what mediates the memory of polarity in regenerating tissues and specifies the location of the new head organizer. In particular, the initial response following bisection appears similar in head- and foot-facing wounds [
16,
17,
18], and the specification of diverse signaling trajectories characterizing head or foot formation is detected only 8–12 hours after bisection [
16,
18].
Our recent work provided evidence that the polarity of the emerging body axis can be identified in regenerating tissue segments relatively early after their folding into hollow spheroids by following the emerging defects in the supracellular order of the ectodermal actin fibers [
19]. The formation of these defects is an inevitable consequence of a topological constraint that is blind to the axis polarity. Nevertheless, the defect configuration faithfully marks the axis polarity long before the emergence of any morphological features [
19]. This result demonstrates that, indeed, at least for small tissue segments that fold smoothly into closed spheroids, some information for axis polarity exists at very early stages of the developmental trajectory, in agreement with the strong polarity memory realized long ago [
9], and, furthermore, that the polarity information is also manifested in structural elements of the supracellular actin fiber skeleton [
19].
The polarity of regenerating
Hydra tissues can also exhibit substantial plasticity and even be reversed under certain conditions, generated, e.g., by grafting different tissues or the application of exogenous Wnt [
10,
11,
20,
21]. Recent results highlight the importance of injury response in activating the Wnt signaling pathway associated with oral regeneration [
16,
17,
18,
22]. However, the injury outcome appears to be modulated by as yet unknown signals from the surrounding tissue. In particular, the appearance of a regenerating head at an injury site depends on the tissue context [
16,
17,
18]. Thus, the developmental trajectory seems to select the location of the emerging organizer among several alternative locations, heavily biased by memory from the parent animal [
11].
Advancing our insight into the emergence and stabilization of body axis polarity requires an experimental strategy that can expose regenerating tissue segments to an array of initial and boundary conditions and examine their effects on the developmental outcome. To this end, we adapt a methodology of creating
frustrating conditions for the regenerated tissue. We have recently shown that grafting two tissue rings with opposite orientations into a single tissue segment exposes the plasticity and reorganization capabilities of the regenerating axis polarity with sensitivity, also, to the original position of the tissue along the body axis of the parent animals [
11]. Here, we further develop this methodology by following the regeneration of rectangular tissue strips. Choosing this initial geometry is motivated by our previous observations that the initial folding of tissue strips creates another type of frustrating initial condition for the inherited polarity: excised tissue strips fold in a purse string-like manner, bringing together their originally head- and foot-facing sides [
4]. Any inherited gradients associated with positional information denoting tissue localization along the body axis of the parent animal would become highly distorted by this folding process (unlike bisected
Hydra or excised tissue rings where the initial tissue polarity is maintained), yet excised strips regenerate almost exclusively into mature animals with proper morphology [
4].
How does a tissue spheroid reorganize following the folding process to regenerate properly along a well-defined body axis? To gain insight into this process, we utilize live microscopy combined with specific markings on the excised tissue strip and follow the tissue deformations and cytoskeletal organization during regeneration. We verified that indeed there is a contact region between the two opposing sides of the tissue strip. Despite the convoluted initial folding step, we find that the regeneration process proceeds along a well-defined trajectory with the tissue deforming in a continuous manner and reorganizing in a way that largely maintains the original tissue polarity. The new head always emerges from a region that originated from the head-facing side of the strip, a short distance away from the initial contact site between the two opposing edges of the strip. Likewise, the foot of the regenerated animal forms from a region originating from the foot-facing side of the excised strip. The location of the mouth of the regenerated animal at the tip of its head coincides with the position of an aster-like defect in the organization of the supracellular actin fibers, which emerges early on in the regenerating tissue spheroid. Conversely, two horseshoe-like defects in the actin fiber organization emerge (and subsequently merge) in the region that becomes the foot of the regenerated animal. This stereotypical and highly reproducible folding and regeneration process, culminating with the formation of a highly ordered mature Hydra, make regenerating tissue strips an excellent model system for further inquiry into the emergence and stabilization of body axis polarity in animal morphogenesis.
4. Discussion
Our observations show that, despite the apparent frustration created by the adhesion between the head-facing and foot-facing sides of an excised tissue strip, the folding process and subsequent regeneration follow a well-defined stereotypical developmental trajectory, eventually resulting in an ordered mature body plan. Importantly, the new head emerges within regenerating strips from a similar location (relative to the original position in the parent animal) in a highly reproducible fashion. Given the complexity of the folding process and the tissue rearrangements during regeneration, as well as the inherent flexibility of Hydra tissues that are capable of forming a new organizer at any site within the excised tissue, it seems unlikely that this reproducibility can only reflect pre-patterned biochemical signal gradients instilled in the tissue or signals emerging from injury sites (which in this case span a considerable portion of the tissue). Rather, the highly reproducible regeneration trajectory suggests that the integration of different mechanisms, involving both mechanical and biochemical processes and supported by their mutual feedback, constrains the tissue and attracts the tissue’s dynamics towards a very well defined trajectory.
An important feature to emphasize here is that despite the inherent animal-to-animal variability and further variations in the initial conditions introduced by the rather crude excision step, the strip regeneration process is highly stereotypical, not only in terms of the final outcome but also in the trajectory taken. We show that the initial tissue folding (
Figure 1), the dynamics of the tissue (
Figure 2), and the cytoskeletal reorganization during regeneration (
Figure 3 and
Figure 4) are all remarkably reproducible, exhibiting essentially the same patterns in all samples. In particular, the location of the regenerated head coincides with the site of an aster-like defect in the organization of the supracellular actin fiber that is already apparent ~8 h after excision. The canalized trajectory taken by the system is particularly striking given the frustrating initial folding which does not preserve the original oral–aboral axis of the parent animal, as clearly illustrated by our observations that the most oral and aboral edges of the tissue adhere to each other (
Figure 1G). Nevertheless, the morphogenesis process is highly constrained toward an ordered body plan.
Superficially, the strong canalization of
Hydra regeneration seems to contradict its flexibility to regenerate a mature animal with a highly ordered body plan from a wide variety of initial conditions and under constraining boundary conditions. However, the form of strong canalization presented by
Hydra regeneration appears to be different from that envisioned by Waddington long ago and that has been serving since then as a guiding principle in developmental systems [
25]. This classical picture of canalization envisions a static landscape with imprinted trajectories that guide the dynamics of a developing system in a program-like manner. We suggest that
Hydra provide an example of dynamic canalization which is of a completely different nature from this classical canalization. Instead of imprinted pre-patterned trajectories, the dynamics itself can stabilize its own attractors for proper development; the dynamical rules are instilled rather than the trajectories themselves. Support for the potential for whole-body regeneration under a wide range of configurations and highly variable internal and external conditions is actually provided by the ability of the tissue to translate weak inherited cues to converge the dynamic trajectory into a stable attractor. In other words, among the multiple different trajectories possible in regeneration, those leading to an ordered body plan are selected by weak inherited cues which provide the system with enough information to converge to a stable attractor of the dynamics. From the mechanical side, the tissue-folding process driven by internal generated forces, the tissue reorganization, and the actin-fiber organization leading to an emerging aster-like defect must all be coordinated with the underlying biochemical signals to support a stable, ordered body plan. The convergence of the regeneration process thus suggests that the mechanical processes and the bio-signaling events associated with axis formation are strongly intertwined and that the robust development arises through constraints embedded in these dynamics and is stabilized by mutual feedback mechanisms [
26,
27]. Further work is needed to establish this picture of dynamic canalization in
Hydra and extend this concept to other organisms.