3 DEVELOPMENTAL PLASTICITY

The human brain is an organ of staggering complexity, containing 86 to 100 billion neurons, an even larger number of glial cells, and something on the order of a quadrillion exquisitely precise synaptic connections. Proper functioning of the nervous system depends upon all of these connections forming correctly; but how does such a complex organ develop? It has evolved a strategy of high redundancy—that is, the developing brain produces far more nerve cells than it actually needs but kills many of them off, and those that survive go on to form an overabundance of synaptic connections before pruning back the misguided or otherwise exuberant ones. These processes are governed by genetics but are highly dependent upon environment and experience during the early postnatal period, factors that fine-tune the neuronal circuits as they are being laid down.¹

Growth Factors and Cellular Suicide

In the late 1940s, a young Rita Levi-Montalcini joined the laboratory of the renowned embryologist Victor Hamburger and set about understanding the relationship between the developing nervous system and the organs and other tissues it innervates. In his own experiments, Hamburger had removed the developing limbs from chick embryos and noticed that primary sensory neurons, which extend fibers to the muscles in the limbs, did not survive in the absence of their “target” tissues. He concluded that nerve cells depend largely on their final destination to mature into a given type.

Levi-Montalcini speculated instead that removing target tissues caused the nerve cells to undergo some kind of degenerative process. Working together, they repeated Hamburger’s experiments and confirmed his initial findings. Removing a limb bud caused the sensory neurons to die off and, conversely, grafting a supernumerary limb onto the embryo resulted in the survival of more cells. This led Levi-Montalcini to hypothesize that target tissues provide a feedback signal required for neuronal survival, and that the lack of that signal, or its low availability, causes cell death.²

Levi-Montalcini then continued trying to identify the signal and characterize its biological properties. Others had found that nerves rapidly grew into tumors that had been transplanted into chick embryos, leading Levi-Montalcini to the hypothesis that the transplanted tissue was secreting a diffusible factor that supported neuronal survival. Working with the biochemist Stanley Cohen, she added snake venom to sensory neurons growing in Petri dishes, and found that this actually produced more nerve fiber outgrowth than the tumors did.³

Cohen therefore suggested that they study mouse salivary glands, the mammalian equivalent of the snake venom gland. Fortuitously, they found the salivary glands to be a rich source of the feedback signal, so they succeeded in purifying the molecule and demonstrating that it was a small protein—which they called nerve growth factor (NGF). Cohen and Levi-Montalcini then went on to produce anti-NGF antibodies, and further demonstrated that the antibodies blocked the effect of the protein in newborn rodents and also on nerve cells growing in Petri dishes. These experiments showed conclusively that NGF is a diffusible protein that is secreted by certain tissues and promotes neuronal survival and differentiation.⁴

Levi-Montalcini’s work provided direct evidence that extensive cell death occurs during neural development, and neatly explained how the nervous supply exactly matches the size of organs and other target tissues. According to the neurotrophic hypothesis, nerve cells are initially overproduced but then compete for a limited supply of target-derived NGF; those that receive the signal survive and undergo maturation, whereas those that do not wither and die.

NGF was the first growth factor to be identified. Its discovery and characterization was an important milestone in our understanding of neural development, and Levi-Montalcini and Cohen shared the 1986 Nobel Prize in Physiology or Medicine for their work.

Since then, the development of molecular biological techniques has led to the identification of dozens of other so-called neurotrophic factors, each of which promotes the survival of a specific population or populations of cells in the developing nervous system. The membrane receptor proteins that mediate their effects have been identified, too, and we are also beginning to understand some details about how they work: binding causes the growth factor–receptor protein complex to be internalized by the cell and then transported back to the nucleus, whereupon it can switch genetic programs on or off.⁵

It soon became clear that extensive cell death is a normal feature of neural development in all organisms. This process is called programmed cell death. It regulates the size of neuronal populations, the proper spacing and positioning of cells, and the emergence of shape and form, among other functions, and is therefore vital to proper development of the brain.

Cell death is under genetic control, and requires “executioner” genes that encode enzymes called caspases. During development, the absence of neurotrophic signaling eventually switches these cell death genes on. Once the cellular suicide program has been activated, the caspase proteins begin to break the cell down from within: the cell’s DNA and scaffold proteins are cut into fragments, causing chromosome condensation, cell shrinkage, and membrane blebbing, all of which give the dying cell a characteristic appearance. Finally, immune cells called macrophages engulf and clear away the cellular debris.⁶

Synapse Formation

Immature neurons in the developing brain are highly promiscuous, forming many more synaptic connections than they need, before trimming back the exuberant, mismatched, and redundant ones.

Synapse formation (or synaptogenesis) is best understood at the neuromuscular junction, where the motor neuron nerve terminal comes into contact with skeletal muscle tissue. Cajal had recognized early on that these synapses are far more accessible and easier to study than those in the brain, which are much smaller and more densely packed. “Since the full grown forest turns out to be impenetrable and indefinable,” he wrote in his autobiography, Recollections of My Life, “why not revert to the study of the young wood, in the nursery stage, as we might say?”

At the neuromuscular junction, the motor neuron releases the neurotransmitter acetylcholine, which binds to receptors on the muscle fibers, causing them to contract. Initially, however, neither the nerve terminal nor the muscle is ready to perform this signaling process. The end of the developing nerve fiber takes the form of a growth cone—a dynamic structure covered with finger-like projections called filopodia—which detects chemical cues in the local environment to guide the growing tip of the nerve fiber to its proper destination, laying down new material as it proceeds. Likewise, the immature muscle mass has not yet split into individual muscle cells, and its acetylcholine receptor molecules are evenly distributed beneath the membrane.

Immature neurons in the developing brain are highly promiscuous, forming many more synaptic connections than they need, before trimming back the exuberant, mismatched, and redundant ones.

Synapse formation and maturation are highly dependent upon interactions between the immature nerve and muscle. As the growth cone extends its tip to the muscle fiber, it releases a burst of acetylcholine, and this causes a redistribution of the acetylcholine receptors, which first cluster and then become immobilized at specific locations in the membrane. Innervation of the muscle by the nerve increases the conductance of the receptors already present in the muscle, and also elicits the synthesis of new receptor molecules, which are inserted into the muscle membrane.

Consequently, the muscle mass eventually splits into individual muscle fibers, each with a specialized receptor zone called an endplate. When the process is complete, there will be approximately 20,000 acetylcholine receptors per square micrometer of endplate, several thousand times the density of other regions of the muscle membrane.

At the earliest stages of development, neuronal growth cones split and send branches of the immature nerve fiber to more than one muscle fiber. But as development proceeds and the neurons mature, the number of synaptic connections gradually decreases. Spontaneous electrical activity stabilizes some of the connections, and experience strengthens them further. This process is also at least partly dependent upon the availability of growth factors in the muscle cells. Thus, nerve fiber branches that do not receive an adequate supply of growth factors are retracted, and synapses that are not strengthened by activity and experience are stripped away, such that every mature motor neuron innervates just one fiber.⁷

Synapses in the brain and spinal cord differ from the neuromuscular junction in several important ways. While the neuromuscular junction connects nerve to muscle, brain synapses connect neuronal elements to each other—a nerve terminal to a nerve cell body, an axon, or a dendrite. And whereas a mature motor neuron connects to a single muscle fiber, neurons in the brain are estimated to form an average of about 10,000 synaptic connections with other cells. Given their small size, complexity, and inaccessibility, we know far less about how brain synapses form. They are, however, believed to assemble in basically the same way as those at the neuromuscular junction.

In all organisms, synapse formation begins during embryonic development and continues during the early postnatal period. In humans, functional synapses have been observed at 23 weeks of gestation. The few postmortem studies that have been carried out so far suggest that synapses form at different rates within different brain regions, but that typically the number of synapses in most or all regions peaks during the first year of life. In the visual cortex, for example, the formation and stabilization of synapses is highly dependent upon visual experience (see below), and the number of synaptic connections reaches its highest density at between 2.5 and 8 months of age. By contrast, some regions of the developing frontal cortex continue to create new synapses well into the third year of life.⁸

Synaptic Pruning

Unwanted neuronal connections are eliminated from the developing nervous system by a process called synaptic pruning. Until relatively recently, it was widely believed that synaptic pruning in the cerebral cortex occurs mostly at puberty and is completed during early adolescence. In the past few years, however, it has become clear that synaptic pruning in the prefrontal cortex continues well into the third decade of life before the total number of synapses in the brain stabilizes to adult levels.⁹

Thus, while the human brain reaches its full size by about 16 years of age, the prefrontal cortex does not reach full maturity until this pruning is complete, and these gradual brain changes are associated with changes in behavior. The frontal cortex is associated with complex functions such as decision-making and evaluation of rewards and, because it takes so long to reach full maturity, adolescents tend to place great emphasis on gaining approval from their peers, and often engage in risky behavior to do so. As synaptic pruning refines the prefrontal circuitry during the second and third decades of life, the executive functions improve, and adults behave more responsibly.¹⁰

Synapse formation and pruning occur extensively in the embryonic brain, and are vital for its proper development. Yet, neither process is restricted to development: the adult brain continues to create new synapses, and to eliminate unwanted ones, throughout life, and we now know that both these processes play important roles in learning, memory, and other aspects of normal brain function (see chapter 4).

Sensory Experience and Critical Periods

Much of our understanding of how sensory experience shapes developing neural circuits comes from another classic set of experiments, performed by the physiologists David Hubel and Torsten Wiesel in the 1960s. Using microelectrodes to examine the properties of cells in the primary visual cortex of cats, they had identified neurons that responded highly selectively to visual stimuli consisting of dark bars moving in specific directions.¹¹ They went on to show that these orientation-selective cells are arranged in alternating columns which respond preferentially to visual inputs from one eye or the other.¹² These ocular dominance columns give the primary visual cortex its characteristic striped appearance, and one of its other names, the striate cortex.

Thus, inputs from the left and right eyes converge in the primary visual cortex, and compete for space there, and in another set of experiments Hubel and Wiesel showed how this competition is driven by visual experience. They reared newborn kittens with one eyelid sutured shut, and found that this had a dramatic effect on development of the visual cortex. As a result, the ocular dominance columns that would normally receive inputs from the closed eye failed to develop, whereas those receiving inputs from the open eye grew to be far larger than they should. Importantly, though, the experiments also showed that the effects were reversible, but only if the eye was reopened before the kittens reached a certain age.^13,14

This marked another significant advance in our understanding of neural development. It showed that proper development of the visual cortex is highly dependent upon visual stimulation, and established the critical period—a narrow developmental time window during which the nervous system is especially sensitive to particular environmental stimuli—as a key concept not only in developmental neuroscience but also in psychology.

This work—for which Hubel and Wiesel were subsequently awarded the Nobel Prize—also led to an effective treatment for amblyopia (lazy eye), an eye condition that affects about 4% of children. Amblyopia occurs because of improper eye development, and results in reduced vision, misaligned eyes, and poor depth perception. It can be treated by patching up the other eye, which forces the child to use the lazy eye and thus drives the development of that part of the visual pathway. The best outcome is achieved if the treatment is started before 8 years of age.

Subsequent research has shown that the other sensory systems are similarly dependent upon experience for their development. It has also revealed that the timing of the critical period for plasticity in the visual cortex is controlled by the maturation of inhibitory interneurons. Interneurons typically have short fibers that are restricted to single regions of the brain; they synthesize and release the neurotransmitter gamma-aminobutyric acid (GABA), which inhibits neuronal activity. They play important roles in integrating information and regulating the activity of neuronal networks.

The brain contains various types of interneurons, but many of these have not yet been properly characterized, and we probably still do not fully appreciate their diverse forms and functions. But one type in particular—the large basket cells—are evidently responsible for plasticity in the developing visual system.

Large basket cells are present in the primary visual cortex, but they mature slowly. When newborn mice first open their eyes, a protein called Otx2 is transported along the optic nerve from the retina to the visual cortex, where it accumulates inside the large basket cells. At this stage, the large basket cells are still immature, forming numerous weak inhibitory connections with their neighboring neurons. When the concentration of Otx2 reaches a certain level, the molecules enter the nucleus, where they activate a genetic program that promotes maturation of the large basket cells.¹⁵

As this program unfolds, the large basket cells begin to refine their connections. Certain synapses are stabilized and strengthened while others are eliminated by pruning. Meanwhile, the maturing network of large basket cells is gradually ensheathed by a net of extracellular matrix proteins, which strengthens the new synaptic connections further. Thus, sensory experience refines the microscopic structure of the visual cortex by driving maturation of the large basket cells, which puts the brakes on plasticity by consolidating the emerging circuitry at a time when its representation of the world is most accurate.¹⁶

Sensory experience refines the microscopic structure of the visual cortex by driving maturation of the large basket cells, which puts the brakes on plasticity by consolidating the emerging circuitry at a time when its representation of the world is most accurate.

In keeping with this idea, deleting one of the genes needed for GABA synthesis, or administering a drug that blocks or reduces GABA-mediated inhibition, prevents the experience-dependent plasticity of ocular dominance columns in mice. Similarly, an infusion of brain-derived neurotrophic factor (BDNF), a growth factor needed for the survival and maturation of large basket cells, accelerates closure of the critical period. Conversely, when an enzyme that breaks down the extracellular net is injected into the mouse brain, it reopens the critical period; and the transplantation of immature interneurons into the brains of newborn mice induces a second period of plasticity corresponding with maturation of the transplanted cells.¹⁷

Thus, “critical periods” are not as critical as we once thought they were. The surprise discovery that the timing, control, and closure of critical periods is dependent upon the maturation of long-range inhibitory circuits immediately suggested ways in which they might be “reopened” in later life. Indeed, clinical trials are now under way to test whether drugs that block GABA-mediated inhibition might benefit adults with amblyopia by restoring plasticity in the visual cortex.¹⁸