57: Repairing the Damaged Brain

• Damage to Axons Affects Neurons and Neighboring Cells

  • Axon Degeneration Is an Active Process

  • Axotomy Leads to Reactive Responses in Nearby Cells

• Central Axons Regenerate Poorly After Injury

• Therapeutic Interventions May Promote Regeneration of Injured Central Neurons

  • Environmental Factors Support the Regeneration of Injured Axons

  • Components of Myelin Inhibit Neurite Outgrowth

  • Injury-Induced Scarring Hinders Axonal Regeneration

  • An Intrinsic Growth Program Promotes Regeneration

  • Formation of New Connections by Intact Axons Can Lead to Functional Recovery

• Neurons in the Injured Brain Die but New Ones Can Be Born

• Therapeutic Interventions May Retain or Replace Injured Central Neurons

  • Transplantation of Neurons or Their Progenitors Can Replace Lost Neurons

  • Stimulation of Neurogenesis in Regions of Injury May Contribute to Restoring Function

  • Transplantation of Nonneuronal Cells or Their Progenitors Can Improve Neuronal Function

  • Restoration of Function Is the Aim of Regenerative Therapies

• An Overall View

For much of its history neurology has been a discipline of outstanding diagnostic rigor but little therapeutic efficacy. Simply put, neurologists have been renowned for their ability to localize lesions with great precision but until recently have had little to offer in terms of treatment. During the past decade this situation has begun to change.

Advances in our understanding of the structure, function, and chemistry of the brain's neurons, glial cells, and synapses have led to new ideas for treatment. Many of these are now in clinical trials and some are already available to patients. Developmental neuroscience is emerging as a major contributor to this sea change for three main reasons. First, efforts to preserve or replace neurons lost to damage or disease rely on recent advances in our understanding of the mechanisms that control the generation and death of nerve cells (see Chapters 52 and 53). Second, efforts to improve the regeneration of neural pathways following injury draw heavily on what we have learned about the growth of axons and the formation of synapses (see Chapters 54 and 55). Third, there is increasing evidence that some devastating brain disorders, such as autism and schizophrenia, are the result of disturbances in the formation of neural circuits in embryonic or early postnatal life. Accordingly, studies of normal development may provide an essential foundation for discovering precisely what has gone wrong in disease.

In this chapter we focus on the first two of these issues: How neuroscientists hope to augment the limited ability of neurons to recover normal function. We shall begin by describing how axons degenerate following the separation of the axon and its terminals from the cell body. The regeneration of severed axons is robust in the peripheral nervous system of mammals and in the central nervous system of lower vertebrates, but very poor in the central nervous system of mammals. Many investigators have sought the reasons for these differences in the hope that understanding them will lead to methods for augmenting recovery of the human brain and spinal cord following injury. Indeed, we shall see that several differences in regenerative capacity of mammalian neurons have been discovered, each of which has opened promising new approaches to therapy.

Next we shall consider an even more dire consequence of neural injury: the death of neurons. The inability of the adult brain to form new neurons has been a central dogma of neuroscience since the pioneering neuroanatomist Santiago Ramón y Cajal asserted that in the injured central nervous system, "Everything may die, nothing may be regenerated." This pessimistic view dominated neurology for most of the last century despite the fact that Ramón y Cajal added, "It is for the science of the future to change, if possible, this harsh decree." Remarkably, in the past few decades evidence has accumulated that neurogenesis does occur in certain regions of the adult mammalian brain. This discovery has helped accelerate the pace of research on ways to stimulate neurogenesis and to replace neurons following injury. Although this work is preliminary, and in some respects controversial, it now seems possible that in the future neuroscientists will be able to reverse Cajal's "harsh decree."

Because neurons have very long axons and cell bodies of modest size, most injuries to the central or peripheral nervous system involve damage to axons. Transection of the axon, either by cutting or by crushing, is called axotomy, and its consequences are numerous.

Axon Degeneration Is an Active Process

Axotomy divides the axon in two: a proximal segment that remains attached to the cell body and a distal segment that has lost this crucial attachment. Axotomy dooms the distal segment of the axon. Synaptic transmission soon fails at severed nerve terminals. After a delay, physical degeneration of the axon occurs and once it begins its progression is relatively rapid and inevitably proceeds to completion (Figure 57–1). Within the distal segment the neuronal membrane breaks down, the cytoskeleton is disassembled, and cytoskeletal components are degraded. This degenerative response is the first step in an elaborate constellation of changes that were initially described in 1850 by Augustus Waller, changes that are now called Wallerian degeneration.

The degeneration of transected axons was long thought to be a passive process, the consequence of separation from the neuronal cell body, within which most of the cell's proteins are synthesized. Lacking a source of new protein, the distal stump would, in essence, wither away. But the discovery and analysis in mice of a spontaneously occurring mutant called Wlds (Wallerian degeneration slow) challenged this view (Figure 57–2). In the Wlds mutant the distal stumps of peripheral nerves persist for several weeks after transection, about 10-fold longer than in normal mice. The Wlds mutation results in the fusion of two normal proteins into a novel protein. One of the contributing proteins is normally involved in biosynthesis of a metabolic cofactor, nicotinamide adenine dinucleotide (NADH), and the other is normally involved in ubiquitination, the process that covalently modifies proteins for degradation.

It remains unclear how the presence of the Wlds fusion protein slows degeneration so dramatically, but its discovery has been important in several ways. First, the very fact that degeneration can be slowed proves that it is not a passive consequence of separation from the cell body, but is rather an actively regulated response. Second, the nature of the Wlds protein has already given us clues about the nature of the active process that drives Wallerian degeneration.

Third, and perhaps most importantly, insight into how the Wlds fusion protein acts may be useful in devising treatments for neurological disorders in which axonal degeneration is prominent. A fatal disease of motor neurons, amyotrophic lateral sclerosis, falls into this category. Other possibilities include some forms of spinal muscular atrophy, Parkinson disease, and even Alzheimer disease. Axon degeneration that occurs in these diseases, as well as after metabolic, toxic, or inflammatory insults, resembles the degeneration that follows acute trauma and may be regulated in similar ways. Expression of the Wlds fusion protein significantly delays axonal loss and can even extend the life span of certain mouse models of motor neuron disease. Thus, while methods for saving transected distal axons are unlikely to be useful clinically for treating patients who have suffered traumatic injury, the same techniques could be useful in treating neurodegenerative diseases.

We return now to the injured axon itself. Even though the proximal portion of the axon remains attached to the cell body, it too suffers. And in some cases the neuron itself dies by apoptosis, probably because axotomy isolates the neuronal cell body from its supply of target-derived trophic factors. Even when this does not occur, the cell body often undergoes a series of cellular and biochemical changes called the chromatolytic reaction: The cell body swells, the nucleus moves to an eccentric position, and the rough endoplasmic reticulum becomes fragmented (Figure 57–1B). Chromatolysis is accompanied by other metabolic changes, including an increase in protein and RNA synthesis as well as a change in the pattern of genes that the neuron expresses. These changes are reversible if regeneration is successful.

Axotomy Leads to Reactive Responses in Nearby Cells

Axotomy sets in motion a cascade of responses in numerous types of neighboring cells. Among the most important responses are those of the glial cells that ensheath the distal nerve segment. The myelin sheath becomes fragmented and eventually removed.

This process is rapid in the peripheral nervous system, where the myelin-producing Schwann cells break the myelin into small fragments and engulf it. Schwann cells, which then divide, secrete factors that recruit macrophages from the blood stream. The macrophages in turn assist in the disposal of debris. Schwann cells also produce growth factors that promote axon regeneration, a point to which we will return later.

In contrast, in the central nervous system the myelin-forming oligodendrocytes have little or no ability to dispose of myelin, and the blood-brain barrier prevents the entry of macrophages, so removal of debris depends on a limited quantity of resident macrophages called microglia. These differences in cellular properties explain the observation that Wallerian degeneration proceeds to completion much more slowly in the central nervous system.

Axotomy also affects postsynaptic neurons. When axotomy disrupts the major inputs to a cell—as happens in denervated muscle, or to neurons in the lateral geniculate when the optic nerve is cut—the consequences are severe. Usually the target atrophies and sometimes dies. When targets are only partially denervated, their responses are more limited. In addition, axotomy affects presynaptic neurons. In many instances synaptic terminals withdraw from the cell body or dendrites of chromatolytic neurons and are replaced by the processes of glial cells—Schwann cells in the periphery and microglia or astrocytes in the central nervous system. This process, called synaptic stripping, depresses synaptic activity and can impair functional recovery.

Although the mechanism of synaptic stripping remains unclear, two possibilities have been suggested. One is that postsynaptic injury causes axon terminals to lose their adhesiveness to synaptic sites so that they are subsequently wrapped by glia. The other is that glia initiate the process of synaptic stripping in response to factors released from the injured neuron or to changes in its cell surface. Whatever the trigger, the activation of microglia and astrocytes by axotomy clearly contributes to the stripping process. In addition, biochemically altered astrocytes, called reactive astrocytes, contribute to formation of a glial scar near sites of injury.

As a result of these transsynaptic effects, neuronal degeneration can propagate through a circuit in both anterograde and retrograde directions. For example, a denervated neuron that becomes severely atrophic can fail to activate its target, which in turn becomes atrophic. Likewise, when synaptic stripping prevents an afferent neuron from obtaining sufficient sustenance from its target cell, the afferent neuron's inputs are placed at risk. Such chain reactions help to explain how injury to one site in the central nervous system eventually affects regions far from the source of the injury.

Central and peripheral nerves differ substantially in their ability to regenerate after injury. Peripheral nerves can often be repaired following injury. Although the distal segments of axons degenerate, connective tissue elements surrounding the distal stump generally survive. Axonal sprouts grow from the proximal stump, enter the distal stump, and grow along the nerve toward its targets (Figure 57–3). The mechanisms that drive this process are related to those that guide embryonic axons. Chemotropic factors secreted by Schwann cells attract axons to the distal stump, adhesive molecules within the distal stump promote axon growth along cell membranes and extracellular matrices, and inhibitory molecules in the perineural sheath prevent regenerating axons from going astray.

Once regenerated peripheral axons reach their targets they are able to form new functional nerve endings. Motor axons form new neuromuscular junctions; autonomic axons successfully reinnervate glands, blood vessels, and viscera; and sensory axons reinnervate muscle spindles. Finally, those axons that lost their myelin sheaths are remyelinated, and chromatolytic cell bodies regain their original appearance. Thus in all three divisions of the peripheral nervous system—motor, sensory, and autonomic—the effects of axotomy are reversible. This is not to imply that peripheral regeneration is perfect. In the motor system recovery of strength may be substantial, but recovery of fine movements is usually impaired. Some motor axons form synapses on inappropriate muscle fibers, some peripheral axons never find their targets, and some neurons die. Nevertheless, the regenerative capacities in the peripheral nervous system are impressive.

In contrast, in the central nervous system regeneration after injury is poor (Figure 57–3). The proximal stumps of damaged axons can form short sprouts, but these soon stall and form swollen endings called "retraction bulbs" that fail to progress. Long-distance regeneration is rare. The longstanding failure of central regeneration has led to the pessimistic view that injuries to the brain and spinal cord are largely irreversible, and that therapy must be restricted to rehabilitative measures.

For some time neurobiologists have been seeking the reasons why regenerative capacity in the central and peripheral nervous systems differs so dramatically. The goal of this work is to identify the crucial barriers to regeneration so that they can be overcome. These studies have begun to bear fruit, and there is now cautious optimism that the injured human brain and spinal cord have a regenerative capacity that can eventually be exploited.

Before discussing these new developments it is helpful to consider the problem of neural regeneration in a broader biological context. Is it the ability of peripheral axons to regenerate that is unusual, or the inability of central axons to do so? It is in fact the latter. Obviously, central axons grow well during development. More surprisingly, axons in immature mammals can also regenerate following transection in the brain or spinal cord. Moreover, regeneration is robust in the central nervous systems of lower vertebrates such as fish and frogs, as exemplified by the studies of Roger Sperry on restoration of vision following damage to the optic nerve (see Chapter 54).

So why have mature mammals lost this seemingly important capacity for repair? The answer may lie in what the mammalian brain can do peerlessly, which is to remodel its basic wiring diagram in accordance with experience during critical periods in early postnatal life, so that each individual's brain is optimized to deal with the changes and challenges of internal and external worlds (see Chapter 56). But once remodeling has occurred, it must be stabilized. It is obviously useful to reassign cortical space to one eye if the other is blinded in childhood, but we would not want our cortical connections similarly rearranged in response to a brief period of unusual illumination or darkness. Maintaining constancy in the face of small perturbations in connectivity may therefore have the unavoidable consequence of limiting the ability of central connections to regenerate in response to injury. In this view our limited regenerative capacity is the Faustian biological bargain we have made for the possession of many precisely wired circuits that underlie our superior intellectual powers.

In seeking reasons for the poor regeneration of central axons, one critical question is whether the poor recovery reflects an inability of neurons themselves to grow or an inability of the environment to support axonal growth. This issue was addressed by Albert Aguayo and his colleagues in the early 1980s. They inserted segments of a central nerve trunk into a peripheral nerve graft, and segments of a peripheral nerve into the brain or spinal cord, to find out how the translocated axons would respond.

Aguayo found that the axons in the translocated segments promptly degenerated, leaving "distal stumps" containing glia, support cells, and extracellular matrix. The results were striking. Spinal axons that regenerated poorly following spinal cord injury grew several centi-meters when inserted into a peripheral nerve (Figure 57–4). Conversely, peripheral axons regenerated well through their own distal nerve trunk, but fared poorly when paired with a severed optic nerve (Figure 57–5).

Aguayo extended these studies to show that axons from multiple regions, including the retinal ganglion, olfactory bulb, brain stem, and mesencephalon, could each regenerate long distances if provided with a suitable environment. As we will see later, it turns out that regrowth of central axons is intrinsically limited. Nevertheless, these pioneering experiments focused attention on components of the central environment that inhibit regenerative ability and motivated an intensive search for the molecular culprits.

Environmental Factors Support the Regeneration of Injured Axons

In probing the differences between peripheral and central growth environments, initial searches were influenced by the results of experiments performed by Ramón y Cajal's student Jorge Tello-Muñoz nearly a century before Aguayo's studies. Tello transplanted segments of peripheral nerves into the brains of experimental animals and found that injured central axons grew toward the implants, whereas they barely grew when implants were not available.

This result implied that peripheral cells provide growth-promoting factors to the injured areas; factors normally absent from the brain. Ramón y Cajal reasoned that central nerve pathways lacked "substances able to sustain and invigorate the indolent and scanty growth" similar to those provided by peripheral pathways. Numerous studies over the succeeding century identified constituents of peripheral nerves that are potent promoters of neurite outgrowth. These include components of Schwann cell basal laminae such as laminin, and cell adhesion molecules of the immunoglobulin superfamily. In addition, cells in denervated distal nerve stumps begin to produce neurotrophins and other trophic molecules. Together these molecules nourish neurons and guide growing axons in the embryonic nervous system, so it makes sense that they also promote the regrowth of axons. By contrast, central neuronal tissue is a poor source of these molecules, containing little laminin and low levels of trophic molecules. Thus in the embryo both central and peripheral nervous systems provide environments that promote axon outgrowth. But only the peripheral environment retains this capacity in adulthood, or is able to regain it effectively following injury.

The practical implications of this view are that supplementing the central environment with growth-promoting molecules might improve regeneration. To this end investigators have infused neurotrophins into areas of injury or inserted conduits rich in extracellular matrix molecules such as laminin. In some attempts Schwann cells themselves, or cells engineered to secrete trophic factors, have been grafted into sites of injury. In many of these cases injured axons grow slightly more extensively than they do under control conditions. Yet regeneration remains limited, with axons generally failing to extend long distances. More important, functional recovery is minimal.

What accounts for such disappointingly limited regeneration? One limitation appears to be the existence of inhibitory signaling pathways that block the growth-promoting activity of cytokine factors. In the optic nerve, for example, the poor regeneration of the axons of retinal ganglion neurons is accounted for in part by the state of activation of a cytokine signaling pathway. The axonal growth-promoting effects of cytokines such as ciliary neurotrophic factors (CNTFs) involve activation of a receptor GP130, and GP130 signaling is counteracted by the activity of a suppressor of cytokine signaling called SOCS3. Thus deletion of the SOCS3 gene in mice augments the ability of CNTF to promote regeneration of retinal ganglion cell axons in the optic nerve (Figure 57–6).

Components of Myelin Inhibit Neurite Outgrowth

Fragments of central myelin are potent inhibitors of neurite outgrowth. Sprouting of spinal axon collaterals following injury is enhanced in rats treated to prevent myelin formation in the spinal cord (Figure 57–7). These findings implied that although both central and peripheral environments might contain a supply of growth-promoting elements, central nerves also contain inhibitory components. That myelin inhibits neurite growth may seem peculiar, but in fact myelination normally occurs postnatally, after axon extension is largely complete.

Searches for the inhibitory components of central myelin turned up an embarrassment of riches. Several classes of molecules found at higher levels in central myelin compared to peripheral myelin are able to inhibit neurite outgrowth when presented to cultured neurons. The first to be discovered was identified when an antibody generated against myelin proteins proved to be capable of partially neutralizing myelin's ability to inhibit neurite outgrowth. Use of this antibody to isolate the corresponding antigen yielded the protein now called Nogo. Two other proteins, myelin-associated glycoprotein (MAG) and oligodendrocyte-myelin glycoprotein (OMgp), initially isolated as major components of myelin, have been found to inhibit the growth of some neuronal types.

Intriguingly, Nogo, MAG, and OMgp each bind to common membrane receptors, NogoR and PirB (Figure 57–8). In mutant mice lacking PirB, regeneration of severed corticospinal axons is enhanced; the extent of axonal regeneration in mice lacking NogoR or its three ligands is uncertain. Identifying physiologically relevant constraints on regeneration may be complicated by the presence of still undiscovered inhibitors. But if many inhibitory components trigger the same intracellular signaling pathway, then interference with that pathway might neutralize the impact of many inhibitors in one fell swoop.

Injury-Induced Scarring Hinders Axonal Regeneration

Myelin debris is not the only source of growth-inhibiting material in the injured brain or spinal cord. As noted earlier, astrocytes become activated and proliferate following injury, acquiring features of reactive astrocytes that generate scar tissue at sites of injury. Scarring is an adaptive response that helps to limit the size of the injury, to reestablish the blood-brain barrier, and to reduce inflammation.

But the scar itself hinders regeneration in two ways: through mechanical interference with axon growth and through growth-inhibiting effects of proteins produced by cells within the scar. Chief among these inhibitors are a class of chondroitin sulfate proteoglycans (CSPG) that are produced in abundance by reactive astrocytes and directly inhibit axon extension by interaction with tyrosine phosphatase receptors on axons (Figure 57–8). Attention has therefore focused on ways of dissolving the glial scar by infusion of an enzyme called chondroitinase, which breaks down the sugar chains on CSPG. This treatment promotes axon regeneration and functional recovery in animals. Drugs that reduce inflammation and decrease scarring, notably prednisolone, are also beneficial if administered shortly after injury, before the scar forms.

An Intrinsic Growth Program Promotes Regeneration

So far we have emphasized differences between the local environments of peripheral and central axons. However, environmental differences cannot completely account for the poor regeneration of central axons. Even though they can regenerate in peripheral nerves, central axons grow much less well than peripheral axons when navigating the same path. Thus adult central axons may be less capable than peripheral axons of regeneration.

In support of this idea, experiments in tissue culture have shown that the growth potential of central neurons decreases with age, whereas mature peripheral neurons extend axons robustly in a favorable environment. One potential explanation for this difference is variation in the expression of proteins thought to be critical for optimal axon elongation, such as the 43 kDa growth-associated protein or GAP-43. This protein is expressed at high levels in embryonic central and peripheral neurons. In peripheral neurons the level remains high in maturity and increases even more following axotomy, whereas in central neurons its expression decreases as development proceeds.

Is this reduced ability of central axons to regenerate irreversible? Hope for reversibility is provided by studies involving "conditioning lesions." Recall that primary sensory neurons in dorsal root ganglia have a bifurcated axon, with a peripheral branch that extends to skin, muscle, or other targets, and a central branch that enters the spinal cord. The peripheral branch regenerates well following injury, whereas the central branch regenerates poorly. However, the central branch will regenerate successfully if the peripheral branch is damaged several days before the central branch is damaged (Figure 57–9). Somehow prior injury or conditioning lesion activates an axonal growth program.

One component of the growth program responsible for regeneration of the central branch appears to be cyclic adenosine monophosphate (cAMP). This second-messenger molecule activates enzymes that in turn promote neurite outgrowth. Levels of cAMP are high when neurons initially form circuits; they decline postnatally in central but not peripheral neurons. In some instances increased supplies of cAMP or proteins normally activated by cAMP can promote regeneration of central axons following injury. Accordingly, drugs that increase cAMP levels, or that activate targets of cAMP, are being actively considered as therapeutic agents to be administered following spinal cord injury. In addition, expression of GAP-43 can promote the regeneration of sensory axons past the site of a spinal cord lesion.

Formation of New Connections by Intact Axons Can Lead to Functional Recovery

So far we have discussed interventions designed to enhance the limited regenerative capacity of injured central axons. An alternative strategy focuses on the significant, although incomplete, functional recovery that can occur following injury even without appreciable regeneration of cut axons. If the basis for this limited recovery of function can be understood, it may be possible to enhance regeneration and function.

A rearrangement of existing connections in response to injury may contribute to recovery of function. We have learned that axotomy leads to changes in both the inputs to and the targets of the injured neuron. Although many of these changes are detrimental to function, some are beneficial. In particular, the central nervous system can, following injury, spontaneously undergo adaptive reorganization that helps it regain function. For example, after transection of the descending corticospinal pathway, which occurs with many traumatic injuries of the spinal cord, the cortex can no longer transmit commands to motor neurons below the site of the lesion. Over several weeks, however, intact corticospinal axons rostral to the lesion begin to sprout new terminal branches and form synapses on spinal interneurons whose axons extend around the lesion, thereby forming an intraspinal detour that contributes to limited recovery of function (Figure 57–10).

Similar instances of functional reorganization have been demonstrated in the motor cortex and brain stem. These compensatory responses attest to the latent plasticity of the nervous system. The ability of the nervous system to rewire itself is most vigorous during the critical periods of early postnatal life (see Chapter 56) but can also be reawakened by traumatic events in adulthood.

How can the nervous system's rewiring ability be improved? It is possible that some of the beneficial effects of grafts in experimental animals reflect reorganization of intact axons rather than regeneration of transected axons. As the nervous system's plasticity becomes better understood, therapeutic strategies that promote specific changes in circuitry may become possible. Perhaps most promising is an approach in which cellular or molecular interventions that promote growth are combined with behavioral therapies that result in circuit rewiring.

The failure to grow a new axon is by no means the worst fate that can befall an injured neuron. For many neurons axotomy leads to the death of the cell. Efforts to improve recovery following injury therefore need to consider neuronal survival as well as axonal regrowth. Since neuronal death is a frequent consequence of severe neural insults, such as stroke and neurodegenerative disease, improved ways of retaining or replacing neurons would have broad utility.

The loss of cells following injury is not unique to the nervous system, although in other tissues new cells are often effective at repairing damage. This regenerative capacity is most dramatic in the hematopoietic system, where a few stem cells can repopulate the entire adaptive immune system. In contrast, it has long been believed that the generation of neurons is complete by birth. Because of this, approaches to regeneration have often focused on finding ways to spare neurons that would otherwise die.

This traditional view is changing, prompted by Joseph Altman's discovery in the 1960s that neurogenesis continues into adulthood in some parts of the mammalian brain (Figure 57–11). This finding challenged fundamental tenets of existing dogma, and the idea that new neurons could form in the hippocampus and olfactory bulb of postnatal rodents was met with skepticism for three decades.

More recently, however, the application of better cell labeling technologies has amply confirmed Altman's results and extended them to nonhuman primates and, even in a limited way, to humans via material obtained at autopsy. In the dentate gyrus of the hippocampus, for example, precursor cells divide throughout life. Some die soon after they are born and others become glial cells, but a substantial minority differentiate into granule cells that are indistinguishable from those born at embryonic stages (Figure 57–12). New neurons are also added to the adult olfactory bulb. They are generated in a subventricular zone far from the bulb itself, then migrate to their destination (Figure 57–13). In both cases the new neurons extend axons and dendrites, form synapses, and become integrated into functional circuits. Thus neurons born at embryonic stages are gradually replaced by later-born neurons, so that the total number of neurons in these regions of the brain is maintained.

The function of neurons born in mature animals is not completely understood, but the cells appear able to recapitulate many of the properties of neurons that arise in the embryo. When the generation of new neurons in the adult is prevented, certain behaviors mediated by the olfactory bulb and hippocampus are degraded. Conversely, some behavioral alterations are accompanied by alterations in the tempo of adult neurogenesis. Adult neurogenesis can be decreased in animal models of depression and chronic stress, whereas enrichment of the habitat of an animal or an increase in the physical activity of otherwise sedentary rodents can increase the generation of new neurons.

Where do neurons generated in the adult brain come from? The principle that embryonic neurons and glia arise from multipotential progenitors also applies to neurons born in adults. Stem cells are the source of neurons in the adult as well as the embryo. The discovery and characterization of neurogenesis from adult stem cells has influenced research on recovery from injury in two important ways.

First, the findings that endogenously generated neurons can differentiate and extend processes through the thicket of adult neuropil and can be integrated into functional circuits led researchers to speculate that the same could be true for transplanted neurons or precursors. In the past decade the idea of replacing lost neurons has progressed from science fiction to a tantalizingly testable hypothesis. Second, since neural precursors can be induced to divide and differentiate, strategies designed to augment this innate ability are now being considered, with the goal of producing neurons in large enough numbers to replace those lost to injury or neurodegenerative disease. At present such strategies are not part of clinical practice, but the intensity of the research devoted to these goals is reason for optimism.

Transplantation of Neurons or Their Progenitors Can Replace Lost Neurons

For many years neurologists have transplanted developing neurons into experimental animals to see if the new neurons could reverse the effects of injury or disease. These attempts have had promising results, most notably in the treatment of loss of dopaminergic cells in Parkinson disease, but their application to human patients has been fraught with difficulty.

One problem is the difficulty obtaining and growing developing neurons in sufficient numbers. Modifying neurons by introducing new genes so as to improve their chances of functioning in a new environment has also been challenging. In many cases the grafted neurons are already too mature to differentiate properly or to integrate effectively into functional circuits.

With the discovery that neural precursors transplanted into the adult brain differentiate into neurons, these obstacles may soon be overcome. Several classes of precursors have been transplanted successfully, including neural stem cells and committed precursors. In many cases these cells differentiate into neurons that are more characteristic of the transplant site than of their site of origin. This result supports the idea that the local environment plays a major role in determining neuronal cell type, and suggests that precursors need not be derived separately for each neuronal type. Conversely, the plasticity of such precursors is not unlimited, so differentiation along specific pathways needs to be established in culture before engrafting the cells.

To date this has been achieved most successfully with embryonic stem cells (ES cells). These cells are derived from early blastocyst stage embryos and can give rise to all cells of the body. In principle, the ability to direct the differentiation of ES cells along specific pathways in culture permits the generation of large numbers of cells for transplantation. Furthermore, methods for generating specific classes of neural precursors and neurons from ES cells have been devised. For example, it is possible to generate neurons that possess many or all of the properties of the spinal motor neurons that are lost in amyotrophic lateral sclerosis or to generate the dopaminergic neurons lost from the striatum in Parkinson disease, and so to engraft such neurons into the spinal cord or brain (Figure 57–14).

This technology has been enhanced recently by the molecular reprogramming of skin fibroblast cells to create induced pluripotent stem (iPS) cells (Figure 57–15). These iPS cells have a distinct advantage over ES cells; their production does not require use of embryos, effectively bypassing a minefield of practical, political, and ethical concerns that have hindered research using human ES cells. Another advantage of iPS cells is that they can be generated from an individual patient's own skin cells, neatly avoiding issues of immunological incompatibility. Many hurdles need to be overcome before iPS cells can be used clinically in regenerative medicine. Nevertheless, these cells are already being used in chemical screens to identify compounds that counteract the cellular defects that underlie human neurodegenerative disease.

Stimulation of Neurogenesis in Regions of Injury May Contribute to Restoring Function

What if, following injury in adults, one could stimulate endogenous neuronal precursors to produce neurons capable of replacing those that have been lost? Two recent findings suggest that this idea is not so far-fetched.

First, neural precursors capable of forming neurons in culture have been isolated from many parts of the adult nervous system, including the cerebral cortex and spinal cord, even though neurogenesis in adults is ordinarily confined to the olfactory bulb and hippocampus. This diversion of cell fate leads to the idea that in the adult, neurogenesis occurs in specialized niches that contain local permissive factors.

Second, the generation of new neurons can be stimulated by traumatic or ischemic injury (akin to stroke), even in areas such as the cerebral cortex or spinal cord in which neurogenesis normally fails to occur. However, the fact that recovery after stroke and injury is poor demonstrates that spontaneous compensatory neurogenesis is insufficient for tissue repair. Injury-induced neurogenesis has been enhanced in experimental animals by administration of growth factors that promote neuronal production from progenitors grown in culture. If such interventions could be adapted to humans, the range of neurons subject to replacement would be greatly increased.

Transplantation of Nonneuronal Cells or Their Progenitors Can Improve Neuronal Function

Cells other than neurons are lost after brain injury. Among the most profound losses are those of oligodendrocytes, the cells that form the myelin sheath around central axons. The stripping of myelin continues long after traumatic injury and contributes to progressive loss of function of axons that may not have been injured directly.

Although the adult brain and spinal cord are capable of generating new oligodendrocytes and replacing lost myelin, this cellular production line is insufficient to restore function in many cases. Since several common neurological diseases, most notably multiple sclerosis, are accompanied by a profound state of demyelination, there is strong interest in providing the nervous system with additional oligodendrocyte precursors in order to augment remyelination.

Neural stem cells, multipotential progenitors, ES cells, and iPS cells can give rise not only to neurons but also to nonneural cells, including oligodendrocytes and their direct precursors. Indeed, at present human ES cells are being channeled into oligodendrocyte progenitor cells and implanted into injured spinal cords of experimental animals. Transplanted cells that differentiate into oligodendrocytes enhance remyelination and substantially improve the locomotor ability of experimental animals (Figure 57–16).

Restoration of Function Is the Aim of Regenerative Therapies

We need to bear in mind that efforts to replace central neurons or to enhance the regeneration of their axons would be of little use if these axons were unable to form functional synapses with their target cells. The same fundamental questions asked about axon regeneration in adults therefore apply to synaptogenesis: Can it happen, and if not, why not?

It has been difficult to address these questions because axonal regeneration following experimentally induced injury is usually so poor that the axons never reach appropriate target fields. However, several of the studies discussed earlier in this chapter offer hope that synapse formation is possible within the dense adult neuropil. In fact, axon branches that regenerate following injury can form synapses on nearby targets. For example, Aguayo and his colleagues found that retinal axons were able to regrow into the superior colliculus when they were channeled through a peripheral nerve that had been grafted into the optic nerve (Figure 57–17A). Remarkably, some collicular neurons fired action potentials when the eye was illuminated, showing that functional synaptic connections had been reestablished (Figure 57–17B).

Likewise, neurons that arise endogenously or are implanted by investigators can form and receive synapses, raising the possibility that behaviors can be restored. Thus there is reason to believe that if injured axons can be induced to regenerate, or new neurons supplied to replace lost ones, they will wire up in ways that help restore lost functions and behaviors.

Axons can regenerate and form new synapses following injury, but regeneration is far more widespread and effective in peripheral axons than in central axons. Recent studies have identified several key factors that limit regeneration of central axons. These include insufficient supplies of growth-promoting factors, pathways laden with growth-inhibitory factors, impenetrable scars, and an intrinsic reluctance of adult central axons to grow.

This is a discouraging array of obstacles, but by understanding them we can hope to manipulate them. If this can be done, it should be possible to enhance regeneration following injury and thus provide restoration of function to many patients for whom there is currently little hope.

Given the complexity of the problem, it is perhaps overly optimistic to expect that any single intervention will suffice. Instead, combined approaches may be needed. For example, the enzyme treatment that breaks down chondroitin proteoglycans in the glial scar is far more effective when combined with administration of neurotrophic factors. Likewise, implantation of cellular bridges that promote regeneration may need to be combined with administration of drugs to neutralize inhibitory factors that would otherwise halt growth as axons leave the bridge and enter the neuropil to form synapses.

Another consequence of axonal injury, and of many neurodegenerative diseases, is the death of neurons. This is particularly serious because in most parts of the brain and spinal cord the neurons we are born with are the only ones we will ever have. Here two new developments provide hope where there had been little: the discovery that new neurons are formed and integrated into functional circuits in a few parts of the brain, and the suite of technological advances that has allowed large numbers of neural precursors to be generated for implantation.

Finally, we should not avoid the relationship between the failure of regeneration following injury and the stabilization of connections that occurs at the end of critical periods. Myelination, which occurs largely at the end of a critical period, may have the secondary effect of preventing further, large-scale rearrangement of synaptic connections. Likewise, astrocytes may not only nurture synapses but also contribute proteoglycans that limit the ability of axons to reach out to new targets. Thus, caution will be needed to ensure that treatments aimed at fostering recovery following injury do not end up promoting formation of maladaptive circuits.