Plasticity and Injury in the Developing Brain

Abstract

The child’s brain is more malleable or plastic than that of adults and this accounts for the ability of children to learn new skills quickly or recovery from brain injuries. Several mechanisms contribute to this ability including overproduction and deletion of neurons and synapses, and activity-dependent stabilization of synapses. The molecular mechanisms for activity dependent synaptic plasticity are being discovered and this is leading to a better understanding of the pathogenesis of several disorders including neurofibromatosis, tuberous sclerosis, Fragile X syndrome and Rett syndrome. Many of the same pathways involved in synaptic plasticity, such as glutamate-mediated excitation, can also mediate brain injury when the brain is exposed to stress or energy failure such as hypoxia-ischemia. Recent evidence indicates that cell death pathways activated by injury differ between males and females. This new information about the molecular pathways involved in brain plasticity and injury are leading to insights that will provide better therapies for pediatric neurological disorders.

Many disorders and injuries of the developing brain affect the basic mechanisms that allow the nervous system to be shaped by experience during childhood. These mechanisms provide the substrate for brain plasticity (kasosei in Japanese), which is much more active in children than in adults. Plasticity in the child’s brain is enhanced because the organization of networks of neuronal synapses as well as white matter pathways remain “under construction” well into adolescence and even later(1). Accordingly, the effects of intensive learning in school, exposure to a second language or practice in athletics has a much greater impact on children than adults. Several neurobiological mechanism contribute to brain plasticity, including an over-production of neurons in early development, apoptosis or programmed cell death of excessive neurons, overproduction and elimination of immature synapses in childhood, and continuous stabilization and strengthening of synaptic connections later in life(2). In this review we focus on some mechanisms for synaptic plasticity, and emerging evidence that these processes are disrupted in several pediatric neurological disorders.

Synaptic Plasticity

Synaptic plasticity is the most important mechanism that allows the developing brain to adapt to environmental influences and store information throughout life(3). This term includes changes that increase or decrease the strength or efficacy of synapses as well as the addition or pruning of synapses. Changes in the number of synapses are especially dynamic in the cerebral cortex in infancy and childhood in the human brain(4). Synapses are produced at a rapid rate in the postnatal period and reach a density that is twice the adult level by age two years, and then fall to the adult level by early adolescence. This process of synapse proliferation and pruning appears to be under the control of both intrinsic programs and environmental influences. The balance of activity between excitatory synapses that use glutamate as their neurotransmitter and inhibitory synapses that use γ-aminobutyric acid (GABA) as their neurotransmitter influence the stabilization of synapses and neuronal circuits (Figure 1) (2;5). Neurons that form synapses with the same neuron and which fire together repeatedly are more likely to form lasting circuits than those whose firing is not coordinated(6;7). Other neurotransmitters including acetylcholine and serotonin projections to the cerebral cortex influence the proliferation and pruning of synapses as well as the ability of neuronal circuits to rearrange in response to changes in sensory information. For example, the organization of cortical maps for somatosensory and auditory information in rodents is strongly influenced by release of acetylcholine from axons projected from the cholinergic nucleus basalis(8;9).

Figure 1.

Diagram of a neuronal dendrite and spine. Excitatory synapses that use glutamate form synapses on dendritic spins but inhibitory neurons form synapses on the body of the dendrite. Spines change shape in response to excitatory activity, mature spines are shorter than immature spines, which are receiving less excitatory input. PSD = postsynaptic density; BDNF= brain derived neurotrophic factor.

Changes in the strength of excitatory synapses are responsible for encoding memories in the brain as well as for other forms of plasticity of neural circuits(10). As shown in Figure 2, three major types of glutamate receptors respond to the neurotransmitter glutamate, including N-methyl-D-aspartate (NMDA) receptors, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors and metabotropic receptors linked to second messengers, such as the mGluR5 receptors(11;12). These glutamate receptors are anchored in the postsynaptic density (PSD) that is characteristic of excitatory but not inhibitory synapses(13). The PSD is a scaffold-like structure made up of hundreds of proteins including cytoskeletal elements and signaling molecules that change in response to development and synaptic activity(14). AMPA receptors linked to channels that carry sodium and/or calcium are responsible for most of the fast excitatory activity in the brain, and their number in the postsynaptic membrane determines the strength of the excitatory synaptic activity. NMDA channels also carry sodium and calcium, and they are opened by activation of specific receptors for glutamate and glycine. NMDA receptors are voltage-dependent and open only when there is enough AMPA receptor activity to depolarize the synaptic membrane(15).

Figure 2.

Diagram of the three major types of excitatory neurotransmitter receptors for glutamate. NMDA receptors are activated when glutamate (glu) and glycine (gly) both occupy receptor sites and the membrane depolarizes, allowing magnesium (Mg⁺⁺) to leave the channel. Relief of the magnesium block allows calcium and sodium to pass through the channel. Most fast excitatory activity in the brain is mediated by AMPA receptor channels, which flux mostly sodium, but channels lacking the GluR2 subunit also pass calcium. In contrast to NMDA and AMPA receptors, metabotropic glutamate receptors are not linked to ion channels but to G proteins and second messenger systems such as phosphoinositide turnover that regulate intracellular calcium levels and protein translation.

Activation of NMDA receptors induces long term potentiation (LTP) which is associated with an increase in synaptic strength and is thought to be a physiologic correlate of memory formation in the hippocampus (Figure 3)(16). LTP induced by activation of NMDA receptors has been reported to be increased in the immature rodent brain compared to the adult brain, due in part to developmental changes in NMDA receptor subunits(16;17). The overall balance of excitation versus inhibition in the neonatal brain also appears to be shifted towards excitation in the neonatal period because receptors for GABA mediate excitatory activity at this stage due to developmental changes in chloride transporters(18). In contrast to activation of NMDA receptors, stimulation of mGluR5 receptors is associated with long term depression (LTD) of synaptic strength(19). AMPA receptors move back and forth between the postsynaptic membrane and the cytoplasm in a process called trafficking that is controlled by activity at NMDA and mGluR5 receptors(20;21). In LTP, high levels of NMDA receptor activity lead to insertion of more AMPA receptors into the postsynaptic membrane resulting in greater synaptic strength. Activation of NMDA receptors associated with physiologic induction of LTP also enhances production of brain derived neurotrophic factor (BDNF) by neurons (Figure 1)(22). BDNF binds to specific receptors on neurons and induces morphologic changes associated with LTP, including insertion of AMPA receptors into the postsynaptic membrane and changes in spine morphology(23). In contrast, activation of mGluR5 receptors leads to internalization of AMPA receptors, LTD and reduced synaptic strength(19).

Figure 3.

Long term potentiation (LTP) is a form of synaptic plasticity that increases the strength of synapses, and it is a physiological correlate of memory. NMDA receptor activation is necessary for LTP, and it stimulates insertion of more AMPA receptors into the synaptic membrane. Activation of NMDA receptors also stimulates the production of brain derived neurotrophic factor (BDNF) which also enhances insertion of NMDA receptors. Stimulation of mGluR5 receptors regulates protein translation and antagonizes LTP by stimulating trafficking of AMPA receptors away from the synapse into the cytoplasm. This increases long term depression (LTD).

Abnormal Plasticity in Pediatric Brain Disorders

Several acquired and genetic pediatric disorders disrupt brain function primarily by targeting synaptic mechanisms involved in neuronal plasticity of the developing brain(1;2). The common theme for these disorders is that they disrupt specific steps in the pathways that lead to changes in numbers of synapses or in the strength of synapses based on synaptic activity(24). For example, neurofibromatosis type 1 (NF-1) is caused by a mutation in the tumor suppressor gene for the protein neurofibromin, which is a GTPase activating protein (GAP)(25). Mutations in neurofibromin result in over-activity of Ras, a small GTPase switch that controls signaling from growth factor receptors in the neuronal membrane to intracellular signaling pathways in the nucleus(26). Neurofibromin controls the mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI-3) pathways that are involved in cell growth as well as learning and memory. Work in a mouse model of NF indicates that learning problems are associated with excessive inhibition by GABA neurons in the hippocampus, and learning can be improved by a drug that antagonizes GABA(26). The cholesterol lowering drug lovastatin has also been shown to improve learning in the NF-1 animal model by reducing Ras activity(27). Mice that lack neurofibromin in the majority of cortical neurons and astrocytes fail to form cortical barrels in the somatosensory cortex(28). Excessive activity of the similar small GTPase Ras-like protein Rheb is also involved in the pathogenesis of tuberous sclerosis (TS)(29). TS is caused by mutations in tuberin (TSC2) or hamartin (TSC1), which binds to TSC2, and normally they function together to inhibit Rheb. Excessive activity of Rheb caused by mutations in TSC1 or TSC2 results in excessive activity of a protein kinase known as murine target of rapamycin (mTOR) which regulates protein synthesis and cell growth(30). Mice with haploinsufficiency in TSC1 showed social and cognitive deficits without cerebral pathology or seizures, suggesting that cognitive problems in the disorder result primarily from disrupted neuronal function rather than effects of tubers or other lesions(31).

Neuronal Plasticity in Fragile X Syndrome

Fragile X syndrome is well known to pediatric neurologists as the most common inherited form of mental retardation with characteristic dysmorphic features and neurobehavioral abnormalities including social avoidance, anxiety and autistic-like behaviors(32). Neuropathological studies of human brain as well as studies in a mouse model of fragile-X syndrome showed that neuronal architecture is abnormal in fragile X syndrome with long, thin and tortuous dendritic spines which appear immature(33). Fragile X syndrome is caused by a loss of function mutation in the fragile-X mental retardation protein (FMRP), an mRNA binding protein that regulates translation(34). In 2002 Huber et al reported that LTD is increased in the hippocampus from mice without FMRP(35). LTD is normally triggered by protein synthesis in response to with activation of metabotropic glutamate receptors, and these results suggest that FMRP normally antagonizes protein synthesis induced by these receptors (Figure 4)(36). Additional experiments in cell culture showed that internalization of AMPA receptors is increased in neurons lacking FMRP compared to controls, which is consistent with the finding that LTD is enhanced(36). Experiments in slices of visual cortex from FMRP deficient mice also showed that LTP is very impaired, and it could be restored by a general antagonist of metabotropic glutamate receptors(37). These data suggest that the deficit in LTP is mediated by excessive activity of metabotropic receptors. They are consistent with the observation that a metabotropic antagonist can also reverse behavioral and anatomic abnormalities in fruit flies in which a homologue of Fmr1 has knocked out(38). These results are also consistent with another experiment in which mice were bred to produce both knockout of Fmr1 and a 50% reduction in the expression of mGlur5 receptors(39). The reduction in mGluR5 receptors in this model corrected the defect in ocular dominance plasticity caused by Fmr1 knockout. A study of plasticity in the neocortex of Fmr1 knockout mice in the early postnatal period showed that spike timing-dependent plasticity (STDP), which depends on NMDA receptors, was also absent while LTD was robust(40). Another study in young adult Fmr1 knockout mice showed that impaired induction of LTP induced by theta burst afferent activity in the hippocampus could be restored by infusion of BDNF(41). Another recent study in the Drosophila model of fragile X syndrome found that dFMRP is positively regulated by sensory input during late brain development and is required to limit axon growth and activity-dependent pruning of axons branches(42). These studies indicate that there is morphological, biochemical and electrophysiological evidence of impaired synaptic plasticity in fragile X syndrome, and suggest that pharmacological intervention might be possible in the future.

Figure 4.

The gene for Fragile X mental retardation protein (FMRP) is mutated in Fragile X syndrome. The normal function of FMRP is to antagonize the effects of stimulation of mGluR5 receptors which causes internalization of AMPA receptors and long term depression (LTD). A mutation in the gene for FMRP therefore leaves mGluR5 action unopposed, causing AMPA receptors to become internalized and depressing synaptic function.

Synaptic Abnormalities in Rett Syndrome

Like Fragile X syndrome, Rett syndrome (RTT) is an X-linked developmental disorder of cognition and behavior that has a major impact on the development and plasticity of synapses(43). Most males with RTT do not survive, but girls develop characteristic stereotyped hand-wringing movements, severe cognitive impairment, acquired microcephaly, seizures, and disorders of breathing and autonomic dysfunction after a period of relative normality in the first year of life. Early clinical descriptions as well as neuropathology and imaging studies suggested a disorder of neuronal development, and studies of neurotransmitter receptors in postmortem brain found abnormalities in the expression of glutamate and GABA in cerebral cortex(44;45). Microarray studies of postmortem brain also found abnormalities in expression of genes associated with developing synapses(46). Mutations in the transcriptional repressor MeCP2 are responsible for most cases of RTT, and this protein is expressed primarily in neurons(47). The timing of its expression during development is delayed until just before the formation of synapses(48-50). Biopsies of nasal epithelium from girls with RS found that neurogeneis and early development of neurons are normal but establishment of mature synapses is blocked(51). Mecp2 deficient mice have morphologic abnormalities in cerebral cortex with a reduced number of thin dendritic spines and immature postsynaptic densities at excitatory synapses(52). These mice have impaired learning and cognition along with deficits in hippocampal plasticity including both LTP and LTD(53;54). Altered basal inhibitory rhythms and enhanced hyperexcitability have also been recorded in the hippocampus of Mecp2 deficient mice(55). This is consistent with the finding that neurons from Mecp2 mice are more sensitive to excitotoxic cell death and hypoxia than controls(56). The maturation and gene expression in hippocampal neurons has also been found to be abnormal(57;58). Phosphorylation of Mecp2 has been shown to regulate activity dependent transcription of the brain derived neurotrophic factor (BDNF) as well as enhancing the growth of dendrites and dendritic spines(59). In contrast, over-expression of Mecp2 in neurons in vitro has been shown to increase axonal length and dendritic complexity(60). The level of BDNF expression has been shown to affect the progression of neurologic impairment in Mecp2 mutant mice, with loss of BDNF worsening impairment and BDNF overexpression prolonging life(59;61). Administration of an ampakine drug that activates AMPA glutamate receptors has been shown to increase expression of BDNF and improve breathing abnormalities in Mecp2 null mice(62). RTT is an important example of the broad impact that disruption in synaptic developmental and activity dependent neuronal plasticity have on the developing brain.

Synaptic Plasticity and Vulnerability to Hypoxia-Ischemia

Some of the same mechanisms responsible for synaptic plasticity can also become mechanisms for injury if the developing brain is subjected to stresses such as hypoxia-ischemia, infection or certain metabolic disorders(1;11). For example the voltage dependent NMDA glutamate receptors can be opened in response to hypoxia-ischemia due to a combination of membrane depolarization from energy deficiency and accumulation of glutamate in the synaptic cleft due to inadequate removal by energy dependent transporters (Figure 5)(63). In this situation, the enhanced function of immature NMDA receptors that enables heightened plasticity in the immature brain can become a liability, making developing excitatory neuronal circuits vulnerable to injury. Enhanced function of NMDA receptors contributes to neuronal injury that occurs in response to the asphyxia in term infants who can develop preferential injury to circuits in the cortex and/or basal ganglia in response to severe asphyxia(15). Nearly complete asphyxia from cord compression often results in preferential injury to the peri-rolandic cortex, putamen and thalamus, which are connected by circuits that use glutamate as their neurotransmitter(64). In contrast, less severe but more prolonged asphyxia is more likely to produce multi-cystic encephalomalacia involving the cerebral cortex but sparing the basal ganglia. These special patterns of injury are probably related to the selective vulnerability of developing neuronal circuits that reflect their normal adaptive role in brain plasticity(63;65).

Figure 5.

Glutamate excitotoxicity results from over-activity of glutamate ionotropic receptors. Injury to neurons is caused by over-stimulation of NMDA receptors which flood the cytoplasm with calcium. AMPA receptor stimulation is also necessary to depolarize the neuronal membrane, allowing NMDA channels to open. Damage to pre-oligodendrocytes in the premature brain is caused predominantly by activation of AMPA receptors which lack GluR2 subunits and are permeable to calcium at that stage in development.

In contrast to brain injuries that occur in term infants, periventricular leukomalaciawhite (PVL) is a prominent feature of injuries that occur prior to 32 weeks gestation(66). This pattern of selective vulnerability has been linked to the sensitivity of late oligodendrocyte progenitor cells to damage from glutamate mediated excitotoxicity and oxidative stress(67). The lower incidence of PVL seen in infants older than 32 week gestation is correlated with a large decline in populations of oligodendrocyte precursors and onset of myelination at this time. The vulnerability of the late oligodendrocyte progenitor cells to injury depends on their expression of AMPA receptors that lack GluR2 subunits which makes them able to flux high amounts of calcium that is toxic to cells(68;69). The inadequacy of enzymes that detoxify nitric oxide and other oxygen free radicals in the oligodendrocyte progenitors also contributes to their vulnerability in the preterm brain(70). Studies in human postmortem brain from babies of different gestational ages indicates that the period of vulnerability of the white matter in the fetal and preterm brain coincides with expression of GluR2 lacking AMPA receptors(71). Oligodendrocytes in the term and neonatal brain express AMPA receptors with GluR2 subunits, making them less vulnerable, while neurons in the cerebral cortex are expressing receptors without GluR2 receptors, making them more vulnerable(71). These data indicate that the age dependent selective vulnerability of white matter and neuronal structures in the developing brain is related to developmental programs for expression of glutamate receptors. Although little is known at this point about the normal function of glutamate receptors in oligodendrocyte development, it is possible that they mediate communication between activity in axons and oligodendrocytes.

Excitatory Receptors and the Cascade of Injury

Excessive activation of glutamate receptors in neurons and oligodendroglia initiates a cascade of events that result in injury to neurons or pre-oligodendrocytes as shown in Figure 6. Calcium that enters cells though NMDA and AMPA glutamate receptors as well as additional calcium entering through voltage sensitive calcium channels can flood the cytoplasm and mitochondria(72). This enhances the production of the toxic free radical gas nitric oxide produced by activation of nitric oxide synthase (nNOS) and oxygen free radicals(73;73-76). Nitric oxide alone or combined with superoxide ions to form peroxynitrite is toxic for mitochondria, and mitochondria increase production of their own reactive oxygen species (ROS) in the face of hypoxia(77). Calcium flooding and enhanced production of ROS in mitochondria combined with impaired adenosine triphosphase (ATP) production secondary to hypoxia increases levels of lactic acid and can cause cerebral edema. Very severe hypoxia-ischemia and mitochondrial injury can lead to critical energy failure with implosion of cell membranes and the pathological process of necrosis (Figure 6). However, less severe insults are more likely to lead either survival and recovery or apoptosis or programmed cell death, which is triggered by events within the nucleus that cause chromatin condensation and DNA fragmentation(78). Apoptosis is especially prominent in the developing brain compared with the adult, probably because it is naturally activated in the developing brain to eliminate excess cells that will not be needed in the mature brain(78;79). After a hypoxic-ischemic injury to the brain in neonatal rodents, apoptosis is observed for a week or more after the insult, suggesting that it continues to be triggered long after the insult.

Figure 6.

Cascade of events that follows a hypoxic-ischemic insult and over-stimulation of glutamate receptors in the brain. This cascade evolves over days to weeks and involves signaling to the mitochondria and nucleus. Very severe injury to the mitochondria leads to complete energy failure and destruction of cell membranes associated with necrosis. Milder insults activate apoptosis, which is much more prominent in infants than in adults. In addition to an extrinsic FAS death pathway (not shown), caspase-dependent and non-caspase dependent pathways activate apoptosis in the nucleus. Work in rodents indicates that these pathways are sexually dimorphic with males preferring the PARP-AIF non-caspase pathway more than females. This pathway is also more easily activated by glutamate receptor stimulation. In contrast, the cytochrome C- caspase 3 pathway is more active in females. ROS = reactive oxygen species; AIF = apoptosis inducing factor; CYTO-C = cytochrome C; PARP = Poly (ADP-ribose) polymerase.

Apoptosis can be triggered either by signals from the mitochondria to the nucleus or by signaling from the extrinsic cell surface Fas death receptor to the nucleus, and both these pathways are enhanced by oxidative stress(80;81). As shown in Figure 6, energy failure in mitochondria can trigger two types of cell death signals to the nucleus, one mediated by activation of caspase 3 and a second non-caspase pathway triggered by apoptosis inducing factor (AIF). AIF is a flavoprotein that is released from stressed mitochondria and travels into the nucleus where it activates apoptosis(82). The major signal for release of AIF from mitochondria is a reduction in levels of the high energy substrate nicotinamide adenine dinucleotide (NAD⁺) in mitochondria in response to activation of the DNA repair enzyme (ADP-ribose) polymerase (PARP) in the nucleus(83). PARP is activated during hypoxia-ischemia by oxidative damage to DNA. The second major apoptotic cell death pathway depends on activation of caspase 3, and is triggered by movement of cytochrome C from mitochondria into the nucleus(84). Recent data indicate that these two cell death pathways are sexually dimorphic, with the caspase-dependent pathway more prominent in females and the PARP-AIF pathway more prominent in males(85;86). The non-caspase PARP-AIF pathway appears to be more easily triggered by activation of excitatory glutamate receptors and activation of nNOS than the caspase-dependent pathway. Neurons from males have been shown to preferentially release more AIF from mitochondria into the nucleus in response to glutamate and NO^. than females, and females preferentially release more cytochrome C(87). In neonatal mice, genetic knockout of the Parp gene is protective in males but not females while caspase 3 inhibiting drugs protect females but not males(88). Inhibitors of PARP and nNOS also protect adult male rodents from injury, but they paradoxically increase injury in females(89;90). This information indicates the cell death pathways in the brain may differ according to genetic sex and this could be relevant to the excess of males with cerebral palsy and other forms of brain injury.

Neuroprotective Therapies for Brain Injury

The perspective that many cell signaling pathways that are involved in brain plasticity are also involved in the pathogenesis of brain injury in the developing brain is an important one for thinking about how to design neuroprotective therapies (Table 1)(91). The corollary of this concept is that some therapies that protect the brain could also impair plasticity or kill neurons if applied in excessive amounts. For example it is clear that NMDA glutamate channel blockers and drugs that activate GABA receptors can cause apoptosis in the developing brain(92). There is clearly a benefit to risk assessment with all agents that are likely to have protective effects, and careful study in animal models as well as follow-up in human trials is warranted. The antibiotic minocycline, which is protective in some models of brain injury, can also enhance injury in neonatal hypoxia-ischemia(93). The observation that signaling pathways involved in cell death are sexually dimorphic is also important both for pre-clinical studies and clinical trials. Recent data indicate that drugs that block glutamate receptors, nNOS, caspase 3 and PARP are likely to act differently in males and females(89;94;95). The cytokine erythropoietin, which has neurotrophic and neuroprotective effects and is protective against hypoxia-ischemia in neonatal mice, also appears to be more effect in females than in males(96;97). Although at least one study found that hypothermia was more protective in female than in male 7 day old rats, the two reported successful clinical trials of hypothermia for term infants with asphyxia did not show any differences according to sex(98;99).

Table 1.

Potential Interventions for Neuroprotection Against Brain Injury

Hypothermia

NMDA, AMPA glutamate receptor antagonists

Nitric Oxide Synthase Inhibitors

Caspase Inhibitors

Acetylcysteine

PARP Inhibitors

Erythropoietin

Growth Factors (NGF, BDNF)

Mitochondrial ATP-sensitive K+ Activators

Conclusion

A major difference between the nervous system in infants and children compared to adults is the capacity for greater plasticity in the developing brain. The molecular signaling pathways involved in brain plasticity are being discovered at an increasing rate, and it is clear that they are disrupted in some common pediatric disorders. These discoveries may lead to better treatments for currently untreatable disorders such as fragile X syndrome, neurofibromatosis and tuberous sclerosis. Brain damage in response to hypoxia-ischemia and other insults often involves overstimulation of these same plasticity mechanisms. Recent evidence indicates that cell death pathways are strongly influenced by genetic sex. A better understanding of molecular pathways involved in plasticity and injury will lead to progress in pediatric neurology.