Answer
A morphogen is a protein or signaling molecule that drives the differentiation of primoridal tissue into its final form – it is the source, genesis, of the shape, morphê.
In the spinal cord, multiple morphogenetic gradients conspire to create a broad variety of cell types.
Ontogeny Recapitulates Phylogeny
In the history of life of Earth, organisms have increased in complexity, beginning with single-celled organisms, which bound together to create multi-cellular organisms, then organisms with body plans, and then finally the vertebrates and invertebrates. If we were to trace our ancestry as humans, we would pick up the trail at the bony fish, pass through the earliest beasts of the land, neither mammal nor reptile, to our furry, milk-making ancestors, and then finally to humans.
Whenever a new human being is created from the fusion of a sperm and egg, this evolutionary history repeats itself, sped up billions of times and acting on the miniature scale of a single fetus. Though it is not a perfect reconstruction, the similarities are striking enough that they gave rise to the polysyllabic slogan "Ontogeny recapitulates phylogeny" – the process of an organism coming into being repeats the process of evolution.
Due to the complexity of this task, and the centrality of this literal self-replication to our understanding of biological principles, the process of building an organism out of one cell and a bunch of raw material has become its own branch of biology: developmental biology.
The development of the nervous system has drawn particular attention not only because the intricacy and complexity of neural tissue necessitate that nature use her wiliest tricks and deepest principles, but also because the creation of the nervous system touches directly on issues of profound philosophical import: where does our perception come from? Why can it be trusted? How are experience and sensory data balanced with hard-wired, genetically-encoded principles?
Developmental biology has a reputation of blandness, driven by its detail-orientation and its plethora of acronyms. In a sense, it is to biology what biology is to the other sciences: it smacks of "stamp-collecting".
This reputation is unfair. The study of developmental biology is driven by these feelings of wonder and passion: wonder at the astounding observation that intricate structures arise from an undifferentiated clump of matter, and a passionate, pre-scientific commitment to either nature or nurture as the determinant of human destiny.
So with these grand feelings in mind, let us dive from the clouds of abstraction into the weeds of biological detail.
Establishing Structure: Reaction-Diffusion and Pizza
The term morphogen dates to 1952, when it was invented by Alan Turing, father of the computer.
In his essay "The Chemical Basis of Morphogenesis", Turing sets out to explain the curious fact that clumps of cells that start out spherically symmetrical can break that symmetry and become simply bilaterally symmetric – the same on the left side as on the right.
Consider: if I cut a cheese pizza in half, the two halves look the same, no matter where I chose to make the cut. If I try the same thing with a T-bone steak, the two halves won't match. The cheese pizza has circular symmetry, while the steak has no symmetry. If I take a single slice of pizza, I can't just cut willy-nilly like I could with the whole pizza, but if I cut through the middle of the slice, I get two equal-sized halves. Pizza slices are bilaterally symmetric, like squares, equilateral triangles, and vertebrates. I used to take advantage of this symmetry to make "pizza-sandwiches" in high school.
Now, say I want to make a pizza that is half pepperoni and half cheese. This is a simple task if you can stand above the pizza – you simply pick a dividing line and then sprinkle the pepperoni only on one side. Note that a half pepperoni and half cheese pizza is no longer circularly symmetric!
The problem biology faces is similar: our circularly-symmetric object, the zygote, has to set a dividing line between, e.g., the front of the organism and the back, which gets rid of, or breaks, the circular symmetry. In keeping with "ontogeny recapitulates phylogeny", this symmetry-breaking is first used to define where to put the anal and oral termini of the gut.
The trouble is, biology has no way of "standing above the pizza". Instead, the pizza slices must decide, on their own, whether they are on the pepperoni half or the cheese half, and produce the relevant topping. This is obviously a tall order!
In fact, using simple differential equations, it's impossible to solve! Turing solved it by considering a more complicated class of differential equations: the "Reaction-Diffusion" equations. The importance of his solution is that it arises naturally from the interactions of relatively simple components.
Each cell is presumed to both secrete diffusible chemicals and also to use those chemicals to regulate internal reactions, some of which produce other diffusible chemicals. These are the chemicals he calls "morphogens", and the model is known as a "Reaction-Diffusion Model".
The surprising result is that there need not be any primordial asymmetry in the model in order to generate asymmetric results. By analyzing the form of the differential equations that describe the dynamics of reaction-diffusion models on a hollow sphere, Turing discovered that tiny deviations, like those caused by random jostling and even thermal noise, are amplified†. The eventual result is the development of an axis defined by the location of these deviations.
With an axis like this in hand, it is straightforward to construct models that use simpler mechanisms to generate further specification. In a zygote, the initial process sets up a division into "front half" and "back half", and then further refinement separates "brain" from "spinal cord" and "front of brain" from "back of brain", and so on.
Let's turn now to a concrete example of such a model, the differentiation of the spinal cord.
Morphogens and the Spinal Cord
One primordial axis runs from the head of the animal to the tail: it is the anterior-posterior axis. Along this axis, a special structure arises, known as the notochord. The notochord is a flexible but stiff semi-compressible organ, much like the tongue. In the early chordates, this structure served as a simple spinal column.
In vertebrates, the notochord exists solely to organize development. It releases a morphogen called "noggin", which causes the cells located close to the notochord to become neurons. These cells fold into a tube, which becomes the central nervous system. The spinal cord remains a tube, while the brain continues to fold and twist to create the structure we are familiar with.
This undifferentiated tube needs to eventually take on the dorsal-ventral pattern that characterizes the functional anatomy of the spinal cord – motor neurons on the bottom, sensory neurons on the top.
The notochord is also critical for this process. The main morphogen secreted by the notochord for specifying cell fates in the spinal cors is known as "Sonic Hedgehog" (seriously).
This rather bizarre name deserves a bit of comment. The hedgehog family of genes was named for the effect of mutations to this gene family in flies. Rather than growing sparsely and generally, the bristles on hedgehog mutant flies grew densely and on the spine, giving the pupae the appearance of a hedgehog. As various members of the hedgehog family were discovered, they were named after species of hedgehog: Indian hedgehog, Pygmy hedgehog, and so on.
Due to the relative difficulty of genetic work in vertebrates compared to invertebrates, the vertebrate version of hedgehog was one of the last hedgehog genes to be discovered, and the community had already used up the various species of hedgehogs on the invertebrate versions. Partly as a joke, the name Sonic Hedgehog was suggested, and it stuck. Turns out, if you want the good names, you've gotta go fast.
But I digress. Secretion of Sonic Hedgehog, or Shh, causes the cells of the spinal cord that are close to the notochord to change their gene expression patterns, giving rise to the motor neurons of the ventral spinal cord.
On the other side of the spinal cord, the dorsal side, other cells release a different set of morphogens, primarily of the Bone Morphogenetic Protein, or BMP, family. These drive the cells of the dorsal spinal cord to become spinal interneurons, which mediate messages to the motor neurons from sensory neurons and from neurons projecting from the cortex. BMPs and other morphogens also induce the neural precursor cells that didn't make it into the tube to become the sensory neurons of the dorsal root ganglia, among other things.
Of course, "Bone Morphogenetic Protein" seems like a silly name for a protein that regulates the development of spinal cord neurons. In fact, the BMPs have many roles in development, not just differentiating neuron clases in the spinal cord – they tell cells whether to become neurons or skin cells, for example, and they are, in fact, responsible for the morphogenesis of bones.
This is one of the startling facts about developmental biology: despite the fact that a bewildering array of structures must arise to create an entire organism, the set of distinct organizing molecules for specifying that panoply can fit comfortably on an index card. Discovering the secret to devising such an economical and flexible instruction set for a self-organizing complex system is one of the goals of developmental biology.
Footnotes
† The critical role for random fluctuations and uncertainty in changing the theoretical behavior of the differential equations in this model presages the eventual development of chaos theory.