Sonic hedgehog, the morphogen

A gradient of Shh specifies progenitor domains for motor neurons (MN) and four classes of interneurons (VO–V3)

The process of induction, in which signals from one tissue change the differentiation of another, is a key concept in development. Although observations of inductive processes date from the early embryological literature, it wasn't until the 1990s that the molecular underpinnings began to be defined. A few key papers along the road to understanding how the early central nervous system (CNS) is patterned along the dorso-ventral (DV) axis provide an excellent illustration of how the study of development was ushered into the age of molecular biology.

Patterning of the CNS along the DV axis — for example, the specification of motor neurons in the ventral half of the spinal cord and sensory neurons in the dorsal half — begins just as the neural plate is folding along its midline to form the neural tube. A series of grafting and ablation experiments by van Straaten and Drukker showed that the notochord — a rod-like structure of mesodermal tissue that is just ventral to the neural tube — could induce the development of a specialized structure along the ventral midline of the neural tube called the floor plate. The floor plate was known to have effects on the guidance of some spinal-cord axons, and, along with the notochord, it was implicated as a signalling centre.

Very early on in the 1990s, Jessell and colleagues set out to test whether signals from the notochord or floor plate might be responsible for controlling the DV pattern of neuronal differentiation. Through grafting experiments in the chick embryo, they showed that both the notochord and the floor plate could induce ectopic patterns of ventral motor neuron layers. Furthermore, removal of the notochord or floor plate resulted in the loss of these ventral motor neurons. So, both notochord and floor-plate cells could release signals that are required for the initial DV patterning of the CNS. Induction by notochord and floor-plate explants resulted in distinct subsets of neuron types at defined distances from the ectopic tissue source. The authors proposed a model in which a diffusible signal produced in the notochord or floor plate results in a gradient of the signal across the DV axis, and this gradient is responsible for the pattern of neurons induced. The search began for the molecular identity of the diffusing signal.

The next breakthrough came from the cloning of vertebrate homologues of the Drosophila segment polarity gene hedgehog (hh). Two groups simultaneously cloned hh homologues — McMahon and colleagues in the mouse, and Ingham and colleagues in zebrafish. In both cases, the expression pattern of one homologue in particular, sonic hedgehog (Shh), implied that it might be involved in ventral CNS patterning. Shh was expressed in both the notochord and the floor plate at the appropriate times in development that were expected for the inductive signal. Furthermore, ectopic expression of Shh could induce floor-plate-specific genes. Shh was predicted to be a secreted protein, so it was an obvious candidate for the diffusible signal that was predicted by the earlier study of Jessell and colleagues.

Subsequent work has shown that differentiation of the floor plate might be more complex, perhaps beginning during gastrulation in the node, but it has also confirmed that Shh is indeed the inductive signal for ventral patterning of the CNS. Furthermore, we now know that a gradient of Shh released by the floor plate is important for this patterning. Distinct neuronal subsets respond to different concentrations of Shh along this DV concentration gradient. Indeed, Shh has come to be known as a classic example of a morphogen — a signal that regulates the spatial pattern of cell differentiation in a concentration-dependent manner — largely due to these groundbreaking experiments on the patterning of the CNS.