The When and Where of Floor Plate Induction

See allHide authors and affiliations

Science  27 Nov 1998:
Vol. 282, Issue 5394, pp. 1654-1657
DOI: 10.1126/science.282.5394.1654

The floor plate is a transient embryonic organizing center located at the ventral midline of the neural tube that profoundly influences the development of the vertebrate central nervous system. The specialized histological features of floor plate cells have long been recognized (1), but only comparatively recently have the remarkable patterning activities of this ventral midline neural cell group been revealed. Floor plate cells serve as a source of Sonic hedgehog, a cell surface and secreted protein that acts at distinct concentration thresholds to specify the identities of motor neurons and interneurons (2). In addition, floor plate cells secrete netrin-1, a chemotropic factor that directs the axonal trajectories of commissural interneurons and certain motor neurons (3). Appreciation of the specialized signaling properties of the floor plate has thus brought an enhanced interest in the origins of this neural organizing center.

Many studies have provided evidence that the differentiation of the floor plate requires inductive signals provided by axial mesodermal cells of the notochord that lie under the midline of the neural plate (4). Notochord signals can induce floor plate differentiation both in vitro and in vivo. Conversely, selective elimination of the notochord in vivo, without removal of floor plate precursors, results in the failure of floor plate differentiation (4). On the basis of these findings, a relatively simple view of floor plate differentiation initially emerged, emphasizing the notochord as a key cellular source of inductive signals. More recent data, however, suggest that there may be more to floor plate differentiation than a single inductive signal provided by the notochord. Indeed, one recent review has questioned the entire concept of induction of the floor plate (5).

Here we discuss recent advances in the understanding of the molecular steps of floor plate development, findings that have begun to shed additional light on the timing and position within the embryo at which floor plate differentiation is initiated. We argue that while these findings may indicate new complexities, they nevertheless do not erode the basic case for the operation of an inductive signal that directs floor plate differentiation. The issues at stake can be reduced to three basic questions: Does inductive signaling have a critical role in floor plate differentiation? What are the molecules that control floor plate differentiation? When and where is floor plate differentiation initiated?

Induction Between Linearly Related Cells

What is the contribution of inductive signaling to floor plate differentiation? It has been clear for some time that the notochord and floor plate do not fit easily into standard views of inducing and responsive cell groups. In large part this is because many of the key molecules that characterize floor plate cells are expressed at an earlier stage by the notochord (4). The discovery of the striking conservation of molecular properties by midline mesodermal and neural cells some time ago (6) raised the general issue of whether the notochord and floor plate derive from a common progenitor cell in the gastrula embryo and whether they acquire their characteristic properties simply as a consequence of their shared lineage, independent of any inductive signaling process. If this were the case, the observed dependence of floor plate differentiation on the notochord in vivo could be argued to reflect the incorporation of notochord-like cells into the ventral midline of the neural tube. It is primarily these issues that have resurfaced recently.

Fate mapping studies in vertebrate embryos have shown that precursor cells that give rise to both the floor plate and notochord can indeed be found in the node-organizer region of the embryo (79). Only those cells in the superficial layer of the node, however, are fated to generate both notochord and floor plate cells, and once cells ingress into the mesodermal layer of the node they contribute only to the notochord (8). These findings argue against a late contribution of prospective notochord cells to the floor plate. More generally, such fate mapping studies reveal little about the state of commitment of cells in and around the node. The common lineage of notochord and floor plate cells may be an indication merely of the fact that cells located at the midline of vertebrate embryos fail to disperse laterally (10). Thus, fate maps and lineage tracing alone do not provide evidence against inductive signaling.

A key discovery that both supported the involvement of inductive signaling and provided a better molecular understanding of floor plate induction was the identification of Sonic hedgehog (Shh), a member of the Hedgehog (Hh) gene family. Shh is expressed initially by cells in the node, later by axial mesodermal cells, and finally by floor plate cells themselves (see the figure, panels A to C). Shh can induce the ectopic differentiation of floor plate cells from neural precursors both in vivo and in neural plate tissue in vitro (4). Importantly, the concentration of Shh needed to induce floor plate differentiation is higher than that required for the generation of other ventral cell types (2, 11). These ectopic expression studies showed that Shh has all the properties expected of a floor plate-inducing factor. Moreover, Shh activity is required for floor plate differentiation in chick and mouse embryos: Inactivation of Shh signaling through the use of antibodies to Shh or by the targeted inactivation of the Shh gene leads to the failure of floor plate differentiation (12, 13). Significantly, the loss of Shh signaling prevents floor plate differentiation without obviously affecting the early development of the notochord, providing genetic evidence that the pathway of floor plate development differs from that of the notochord. Similarly, inactivation of the gene encoding the zinc finger transcription factor Gli2, a component of the downstream Shh-signaling pathway, blocks floor plate differentiation without perturbing the development or apparent signaling properties of the notochord (14, 15). These findings argue against the extreme model (see the figure, panel D) in which floor plate and notochord cells are equivalent, committed descendants of a common node progenitor. Instead they reveal a selective requirement for Shh activity in floor plate differentiation and strongly implicate intercellular signaling in this differentiation process.

Possible mechanisms of floor plate development in the chick embryo.

Panels A and B show transverse sections through neural tube-stage chick embryos. Shh gene expression is detected by in situ hybridization histochemistry. (A) Restriction of Shh expression to the notochord at early stages of caudal neural tube development. (B) Expression of Shh by both the notochord and floor plate at late stages of neural tube development. (C) The position of notochord (N) and floor plate (FP) cells. (D) A model of floor plate development in which floor plate and notochord cells derive from a common progenitor cell in the node, independent of inductive signaling. (E) A model in which the Hedgehog-mediated induction (red arrows) of floor plate cells begin in the node and continues after neural tube closure. (F) A model in which the induction of floor plate differentiation occurs primarily after neural tube closure. Floor plate cells may also serve as a source of Hedgehog signals that induce additional neural tube cells to acquire a floor plate fate.

Timing of Floor Plate Induction

Studies of Shh signaling in mouse and chick embryos have not, however, resolved the major question of the time and place at which floor plate differentiation begins. Because notochord and floor plate share many molecular properties, notably expression of the winged helix transcription factor HNF3β and Shh (4), individual precursors of these two cell groups cannot be distinguished within the node. Nevertheless, the expression of such genes by node cells leaves open the possibility that the induction of floor plate differentiation begins within the node itself (16). For example, it is possible that a subset of cells within the node that expresses Shh induces adjacent cells to embark upon a program of floor plate differentiation (see the figure, panel E). Alternatively, there may not yet be a distinction between notochord and floor plate precursors within the node, in which case Shh signaling could act in a stochastic manner to direct a subset of equivalent node cells to a floor plate fate. The possibility that floor plate differentiation begins in the node is, of course, still compatible with a requirement for Shh-mediated inductive signaling.

Several lines of evidence, however, argue against the idea that the specification of floor plate cells occurs exclusively within the node, at least in avian embryos. A major fraction of cells destined to populate the floor plate reside in a region anterior to the node in gastrula embryos (7, 17). These more anterior cells do not yet express definitive floor plate markers (16), nor do they acquire floor plate properties when grown in isolation (4). The differentiation of these anterior cells into floor plate thus appears to take place outside the node and at a later developmental stage (7). The importance of a later contribution of signals from the axial mesoderm is also consistent with the finding that at most axial levels, prospective floor plate cells do not express the full complement of floor plate properties, including Shh and netrin-1, at the time that they first occupy the ventral midline of the caudal neural tube, even though such markers are expressed by the notochord at this stage (see the figure, panel A). When taken together with the absence of a floor plate after selective notochord removal, these results support the idea that the progression of floor plate differentiation requires a later or sustained period of signaling from the axial mesoderm as it extends under the caudal neural tube (see the figure, panel F). Moreover, cells in more lateral regions of the neural plate and neural tube can acquire floor plate properties even at much later stages of development if they migrate medially and populate the ventral midline of the neural tube (18). Finally, floor plate cells themselves can induce the differentiation of more lateral neural tube cells to acquire a floor plate fate (19) (see the figure, panel F), a process of homeogenetic (like-begets-like) induction that may underlie the marked increase in the number of floor plate cells that occurs after neural tube closure (19) and may ultimately reinduce floor plate cells after notochord removal. Thus, a substantial proportion of floor plate cells appear to be specified relatively late and to derive from progenitor cells that reside within the neural epithelium itself rather than within the node.

Insights from Zebrafish Mutants

Although the analysis of floor plate differentiation in avian and mammalian embryos presents a reasonably coherent picture of this inductive process, recent genetic analyses of zebrafish development suggest that in this organism, the pathway of floor plate differentiation may be more complex (5). Null mutations in the zebrafish Shh gene, otherwise called sonic you, eliminate a group of lateral floor plate cells but leave intact a more medial strip of floor plate cells (20). At first glance, these results might be construed as evidence against an essential role for Hedgehog signaling in floor plate induction. However, zebrafish embryos, in contrast to their amniote counterparts, express two other hedgehog genes, echidna hedgehog (ehh) and tiggywinkle hedgehog (twhh), in midline mesodermal and neural cells, respectively. The twhh gene, like Shh, is initially expressed by cells in the embryonic shield (the zebrafish counterpart of the node), and later ehh is expressed by axial mesodermal cells (21, 22). It remains possible, therefore, that multiple hedgehog genes cooperate in the induction of floor plate cells in zebrafish whereas Shh is solely responsible in avian and mammalian embryos.

Other zebrafish mutants, notably no tail (ntl) and cyclops (cyc), also have a profound influence on the differentiation of the midline. Their phenotypes raise the issue of whether floor plate development can proceed in the absence of inductive signaling from the axial mesoderm (5). The ntl gene encodes a T-box protein closely related to the mammalian brachyury (T) protein, and like mouse brachyury mutants, ntl mutants also exhibit defective notochord development (23, 24). Nevertheless, the floor plate is present and even overrepresented (24). One suggested explanation for the persistence of floor plate differentiation in the absence of the notochord in ntl embryos is that axial mesodermal cells are still present at the midline, despite the absence of overt structural features of notochord differentiation. Consistent with this idea, cells underlying the midline of the neural plate still express Hedgehog genes (24). Alternatively, ntl function may normally be required to promote the formation of axial mesoderm, with the consequence that in ntl mutants, unspecified progenitor cells within the node are still capable of responding to local Hh signaling but generate only floor plate cells. Thus, the phenotype of the ntl mutation does not argue against a requirement for Hedgehog-mediated inductive signaling in floor plate generation, but instead would seem to indicate that a differentiated notochord is not a required source of such signals, at least in zebrafish.

Cyc mutant embryos exhibit a midline phenotype that in some ways is complementary to that of ntl embryos, in maintaining a notochord but lacking a floor plate, at least at early developmental stages (25). Initially, mosaic analyses suggested that the loss of floor plate cells in cyc mutants resulted from a perturbation in the ability of neural cells to respond to axial mesodermal signals (25). The cyc gene has, however, recently been shown to encode a nodal-related TGFβ superfamily member that is expressed by cells in the embryonic shield but not later by neural tube cells (26, 27). Moreover, the rescue of floor plate differentiation appears to require cyc expression in embryonic shield cells rather than in neural cells (27). How then may cyc act? The finding that cyc embryos possess a notochord and express Hh genes yet lack a floor plate could reflect the existence of a Hedgehog-independent but cyc-dependent pathway of floor plate differentiation in zebrafish (27). Alternative possibilities are that cyc function is required to control the proliferation of axial mesodermal cells or to maintain their inductive signaling properties. In this context, the high Shh concentration threshold requirement for floor plate differentiation (2, 11) could mean that a partial attenuation of Hh signaling from the axial mesoderm in cyc embryos would lead to the loss of floor plate cells but the preservation of other ventral cell types: the cellular phenotype of cyc mutants. A third possibility is that cyc signaling controls the expression of a mesodermal signal that acts in parallel with Hh proteins, perhaps by sensitizing neural cells to Hh signaling. The possibility that cyc regulates organizer development is supported by the recent finding of a second nodal-related gene, squint (sqt), that exhibits partially redundant functions with cyc in the formation of the embryonic shield (28).

An additional gene required for floorplate formation is one-eyed pinhead (oep), which encodes an epidermal growth factor (EGF)-related protein (2931). The failure of floor plate differentiation in oep mutants may reflect a cell-autonomous role in floor plate precursors themselves (30) or a function in the control of axial mesoderm differentiation (29), or both. Thus, despite marked defects in floor plate differentiation in many zebrafish mutations, the cellular and molecular analysis of these mutant phenotypes leaves open the possibility that these genes are involved primarily in the regulation of axial mesoderm differentiation, affecting floor plate differentiation only secondarily. Consequently, floor plate differentiation in zebrafish may operate under guidelines more closely related to those in avian and mammalian embryos than has recently been envisaged (5, 20).

Does this mean that the complete picture of floor plate differentiation is now apparent? Almost certainly not. Many aspects of the early cellular interactions that control the decision of axial midline cells to embark upon distinct pathways of notochord and floor plate differentiation need to be defined more clearly. In addition, there may be factors expressed by axial mesodermal cells that regulate the perception of Hh signals by neural cells and, if so, their characterization could help to clarify many of the unresolved issues discussed above. Zebrafish genetic screens have identified many additional mutations that perturb midline mesodermal and neural differentiation (32), and it seems likely that the molecular analysis of some of these mutants will reveal novel components in the pathway of floor plate differentiation. The need to define the relative contributions of Hh-dependent and -independent signaling to floor plate differentiation should maintain this intriguing cell group at the center of developmental studies for some considerable time.

References and Notes

View Abstract

Navigate This Article