Is Being Plastic Fantastic? Mechanisms of Altered Plasticity after Developmental Traumatic Brain Injury

Abstract

Traumatic brain injury (TBI) is predominantly a clinical problem of young persons, resulting in chronic cognitive and behavioral deficits. Specifically, the physiological response to a diffuse biomechanical injury in a maturing brain can clearly alter normal neuroplasticity. To properly evaluate and investigate developmental TBI requires an understanding of normal principles of cerebral maturation, as well as a consideration of experience-dependent changes. Changes in neuroplasticity may occur through many age-specific processes, and our understanding of these responses at a basic neuroscience level is only beginning. In this article, we will particularly discuss mechanisms of TBI-induced altered developmental plasticity such as altered neurotransmission, distinct molecular responses, cell death, perturbations in neuronal connectivity, experience-dependent ‘good plasticity’ enhancements and chronic ‘bad plasticity’ sequelae. From this summary, we can conclude that ‘young is not always better’ and that the developing brain manifests several crucial vulnerabilities to TBI.

Introduction

Traumatic injury to the developing brain is distinct from adult traumatic brain injury (TBI) in many ways. One major difference is that the developing brain is more malleable to external stimuli, or more plastic, a characteristic that has often been touted as a significant advantage with regard to recovery of function. However, children who suffer TBI are well known to develop chronic cognitive and behavioral disturbances [Ewing-Cobbs et al., 1997, 1998; Fay et al., 1994; Jaffe et al., 1993; Levin et al., 1982, 2002]. Children also have a higher incidence of posttraumatic epilepsy [Annegers et al., 1998; Annegers and Coan, 2000] than adults. Infants and toddlers might arguably have the most plasticity and yet actually seem to have some of the worst developmental outcomes after significant TBI [Barlow et al., 2005; Ewing-Cobbs et al., 1997; Levin, 2003; Levin et al., 1992; Taylor and Alden, 1997]. Clearly, plasticity is not simply an all-or-none beneficial response that just happens to be more robust in development, it is a complex series of molecular, cellular and physiological events that must be carefully orchestrated for optimal developmental outcome. While the effects of ischemia, seizures and other brain injuries have been extensively studied during development, TBI, the number one cause of pediatric death and disability, has not. Here, we will describe the basic principles of developmental plasticity and review the current literature with regard to potential mechanisms underlying altered plasticity following TBI in the immature brain. Better delineating the nature and timing of these processes will provide a foundation for the therapeutic manipulation of neuroplasticity to enhance functional recovery in the developing brain.

Plasticity: Drastic or Fantastic?

Plasticity is defined by Merriam-Webster as ‘the capacity to vary in developmental pattern, phenotype or behavior according to varying environmental conditions’. In general, this ability has been viewed as an advantageous quality, one that permits the individual to change in response to external cues, and thus, make it better suited to survive and thrive in the current circumstances. However, to a large degree, the beneficial results of plasticity in the nervous system appear to be contingent upon the inciting stimuli. Physiological stimuli such as the presence of light-activating visual pathways would seem to be necessary for promoting normal development of the relevant neural circuitry. On the other hand, pathological or abnormal stimuli (or the absence of normal stimuli), such as persistent dark rearing, can result in aberrant neuroplasticity with subsequent abnormal developmental outcomes. These ‘abnormal’ outcomes may not be distinctly deleterious and, in fact, may confer some advantages to the individual/pathway/neuron in the ‘abnormal’ environment. However, upon exposure back into a normal environment, these changes are likely to be suboptimal.

Other examples of the ‘drastic versus fantastic’ nature of developmental plasticity are of tremendous potential clinical relevance. One type of beneficial plasticity is the response of the brain to enriching environmental conditions. In both animals and humans, stimulating environments trigger outgrowth of neural projections, resulting in increased dendritic arborization and cognitive enhancement [Greenough et al., 1973; Jacobs et al., 1993; Rosenzweig and Bennett, 1996]. The ability of the brain to benefit from environmental stimulation appears most robust during brain maturation. On the other hand, excessive stimulation of particular pathways, as may occur during seizures or other types of brain injury, can also promote outgrowth of abnormal neural connections, as well as development of circuitry that subsequently leads to epilepsy, and interferes with normal cognitive development.

Principles of Plasticity and Development after Injury

Type of Injury

Much of our current understanding about developmental plasticity and recovery from injury comes from studies of focal brain lesions induced at particular developmental stages. Focal sensorimotor cortex lesions sustained in development show better recovery than similar lesions in adulthood. This ‘younger is better’ paradigm was originally described in primates by Margaret Kennard [1942] and has been dubbed the ‘Kennard Principle’. This phenomenon has been ascribed to the enhanced plasticity and ability for reorganization present in the developing brain, particularly in the setting of a unilateral focal injury. Following surgical hemidecortication, anatomical recovery (brain weight, cortical thickness, brain region cross-sectional area) and behavioral outcome (Morris water maze, foot faults, beam walking, claw cutting) were best in neonatally lesioned rats, while rats lesioned as adults had the largest deficits [Kolb and Tomie, 1988]. Similar results were seen when kittens or adult cats underwent hemispherectomy – kittens performed close to controls, while adult-lesioned cats showed abnormal limb preference and reduced locomotion [Burgess and Villablanca, 1986].

Lesion location is also important in determining developmental vulnerability or resistance to functional impairment. Frontal lobectomy or lesions of the dorsolateral frontal cortex in monkeys resulted in less impairment on delayed response testing when the lesions were induced in infancy rather than as juveniles [Goldman, 1971]. In fact, monkeys that underwent dorsolateral frontal lesions in infancy performed similarly to nonlesioned controls at 1 year of age. However, when tested for visual pattern discrimination, monkeys with dorsolateral frontal cortex lesions from infancy actually showed worse 1-year outcomes than those lesioned as juveniles. On the other hand, orbitofrontal lesions showed no such sparing with age for multiple cognitive tasks (delayed response, visual pattern discrimination, delayed alternation, object reversal) [Goldman, 1971; Miller et al., 1973].

Circumstances become more complicated if bilateral homotypic cortical areas are lesioned. Following bilateral motor cortex lesions, postnatal day 10 (P10)-lesioned rats showed better behavioral performance (Whishaw reaching task, beam walking) than adult or P1-lesioned animals, although anatomical measures (brain weight, cortical thickness) seemed to favor the adult-lesioned animals [Kolb et al., 2000]. Bilateral medial frontal lesions showed that the youngest rats (P3, P6) actually had worse anatomical and functional outcomes [Kolb et al., 1996]. Furthermore, bilateral medial frontal lesions in the perinatal period (P2) resulted in markedly reduced brain weight, dendritic arborization and spine density, leading to the conclusion that, in some settings, ‘earlier may be worse’ [Kolb and Cioe, 2000].

Of course, generally, TBI is a diffuse biomechanical injury that may have focal or multifocal components superimposed. Thus, while we can learn much from these age-at-lesion studies, it is essential to specifically model and study TBI in the developing brain [Prins and Hovda, 2003].

Recovery to Baseline

One important aspect of measuring physiological responses and functional outcomes during development is the fact that both physiology and function are often dynamically changing over time. This principle must be considered for both clinical management and research investigations involving the care and study of immature or pediatric populations. For example, in a more static (adult) model, a given function remains constant over time, as illustrated in figure 1a. If an injury occurs, there may be a permanent deficit ( fig. 1a, dotted line) or an acute reduction in function that slowly recovers over time. At some point, if and when the level of function returns to the premorbid level, recovery is said to be complete (fig. 1a, dashed line). However, this paradigm does not hold in a dynamic (developing) model. In a developing model, the level of function in normal subjects is changing over time. After developmental injury, while functional recovery may return to the premorbid baseline ( fig. 1b, dashed line), the baseline function of normal peers has already moved on (fig. 1b, solid line). Therefore, the duration of recovery, the rate of functional change across development and the presence/timing of critical windows all become important measures to determine whether the injured immature brain can actually fully recover to an age-matched baseline after injury. In the setting of pediatric brain injury, functional comparisons between injured and control subjects must be made at similar developmental ages.

Fig. 1.

Recovery of function after injury in different theoretical models. a Static (mature) model: baseline function does not change with time (solid line). Injury with no recovery shows downward step in function (dotted line). Injury with gradual return to premorbid baseline results in complete recovery (dashed line). b Dynamic (developmental) model: baseline function improves with time (solid line). Injury with gradual return to premorbid baseline results in incomplete recovery (dashed line). c Dynamic (interventional) model: baseline function shows no change with time (thin solid line), but following intervention, baseline function improves with time (thick solid line). Injury (followed by intervention) with return to premorbid baseline represents incomplete recovery (dashed line).

It is also possible that the immature brain may appear to have a complete functional recovery, but that the strength of this recovery is limited. One example of this is a study of unilateral excitotoxic injury in P7 rats. Early after injury, motor deficits are seen contralateral to the damage, but recovery occurs by 3–4 weeks. However, administration of an N-methyl-D-aspartate (NMDA) receptor antagonist after the time of full recovery is able to transiently reinstate the deficit [Felt et al., 2002].

One final related concept to be introduced is that of ‘growing into the lesion’. This idea purports that if a particular function is not normally well developed at the time of injury, then such a deficit may not necessarily be observed until a later developmental stage. In the study of dorsolateral versus orbital frontal lesions in monkeys described above [Goldman, 1971], the dorsolateral frontal lesioned animals, which appeared to show developmental sparing when tested 1 year after the lesion, actually demonstrated relative impairments on the delayed alternation task when tested after 2 years. This illustrates that developmental outcome should be interpreted not only based on age at injury, but also on age at testing.

Effects of Experience

An additional critical consideration when managing or investigating recovery from developmental brain injury is the presence of salient environmental stimuli. In a mature brain, functional enhancements may occur in response to appropriate environmental conditions, such as exploring an enriched environment or taking lessons in a foreign language. In these situations, the normally flat time course of function is now changed to a developmental one, with a change in function marked over time in the presence of the prescribed intervention (fig. 1c, thick solid line). The effects of injury on this now dynamic system may be mediated by a direct injury-induced deficit or by an injury-induced impaired responsiveness to the intervention. Thus, in situations where there is an environmentally induced increase in function, the appropriate comparison is not simply with uninjured controls, but also with uninjured controls exposed to the same environment or therapy ( fig. 1c). In this case, as described in the preceding discussion, recovery to premorbid baseline is an inadequate goal, as uninjured subjects experiencing the beneficial intervention will have gained enhanced functions over the same time period. In other words, when studying the recovery from pediatric TBI, one must control not only for age, but also for environmental conditions that can influence developmental trajectory (i.e. maternal behavioral factors, living conditions, social interactions).

Finally, it is important to realize that brain maturation occurs in situations where environmental conditions and training are superimposed upon normal developmental changes. One must carefully consider these complexities in the investigation and interpretation of injury recovery responses. Both normal maturational changes and response to environmental stimuli will affect functional outcomes measured during recovery from pediatric TBI.

Mechanisms of Altered Plasticity

The multiple variables operant during the recovery from developmental TBI are probably more easily addressed in basic science studies using well-characterized injury models and strict age- and environment-matched developmental controls than in clinical pediatric TBI studies, where the ability to control these variables is quite limited. Once an injury response or recovery mechanism is characterized in the preclinical setting, it would be possible to design well-controlled clinical studies to ensure that these findings are translatable back to pediatric brain-injured patients. In the following section, we will review the current status of basic science investigations into mechanisms of neuroplasticity following developmental TBI.

Changes in Neurotransmission

A major potential mechanism of alterations in developmental neuroplasticity following TBI involves changes in neurotransmission (table 1). Clinically, dysfunctional neurotransmission has been implicated in many sequelae of pediatric TBI, including memory impairment and attention problems [Blitzer and Lombroso, 2003; Griffin et al., 2003; Johnston, 2004; Konrad et al., 2003]. In both immature and adult TBI models, abnormalities have been reported in multiple neurotransmitter systems, including glutamatergic [Biegon et al., 2004; Giza et al., 2006; Kumar et al., 2002; Miller et al., 1990; Osteen et al., 2004; Sihver et al., 2001], γ-amino-butyric acid (GABA)ergic [Lowenstein et al., 1992; Sihver et al., 2001], cholinergic [Dixon et al., 1996, 1997, 1999; Gorman et al., 1996; Shao et al., 1999; Sihver et al., 2001] and aminergic [Walter et al., 2004] neurotransmission.

Table 1.

Neurotransmitter changes

Neurotransmitter	System component	Change after developmental TBI	Change after adult TBI
Glutamate	NMDA receptor binding		↓ H 2 h, 12 h, 7 days
			↓ C 2 h, 12 h, 7 days1,9,13
	NMDA receptor expression	↓ H 2 days, 4 days	↓ H 6 h, 12 h3
		↓ C 1 day2	↓ C 1 day, 2 days4
	NMDA-mediated calcium flux		↓ C 1 day, 2days4
	NMDA receptor electrophysiology	↓ H LTPi 1 day, 2 days5	↓ H LTPi 4 h, 2 days6
			↓ H LTPm 8 weeks7
	glutamate antagonist effect on neuron survival	↓ H	↑ H
		↓ C8	↑ C (many)
	other glutamate receptor-related genes
	(GluR4, GluR5, KA1, GRIP)	↓ H <1 day16
GABA	GABA receptor binding		↓ H 2 h
			↓ C 2 h, 12 h9
Acetylcholine	mACh receptor binding		↓ H 2 h, 12 h
			↓ C 12 h9
	mACh receptor expression		↓ H 4 weeks10
	vesicular ACh transporter	↑ H <1 day16	↑ H 4 weeks10
	choline acetyltransferase		↓ H 1 h
			↓ FC 1 h, 1 day
			↑ PC 1 h, 1 day, 5 days11
	cholinergic fiber regrowth	↓ H 2 weeks12	↓ H 2 weeks14
Dopamine	aromatic amino acid decarboxylase activity	↑ C 2 days15
	catechol-O-methyltransferase and D1 receptor genes	↑ C <1 day16

H = Hippocampus; C = cortex; LTPi = long-term potentiation induction; LTPm = long-term potentiation maintenance; mACh = muscarinic acetylcholine; FC = frontal cortex; PC = parietal cortex.

¹Miller et al., 1990.

²Giza et al., 2006.

³Kumar et al., 2002.

⁴Osteen et al., 2004.

⁵D’Ambrosio et al., 1998.

⁶Sick et al., 1998.

⁷Sanders et al., 2000.

⁸Pohl et al., 1999.

⁹Sihver et al., 2001.

¹⁰Shao et al., 1999.

¹¹Gorman et al., 1996.

¹²Prins et al., 2003.

¹³Biegon et al., 2004.

¹⁴Phillips et al., 1994.

¹⁵Walter et al., 2005.

¹⁶Prins and Giza, 2006.

Indiscriminate activation of glutamate receptors is a fundamental component of biomechanical neural injury [Faden et al., 1989; Katayama et al., 1990; Kawamata et al., 1992]. This initial flood of glutamate is followed by an apparent downregulation of NMDA receptors [Biegon et al., 2004; Giza et al., 2006; Kumar et al., 2002; Miller et al., 1990; Osteen et al., 2004; Sihver et al., 2001]. The time frame of these reductions ranges from a few hours to several days after injury depending upon injury model, age at injury and methodology used to measure receptors (mRNA, protein, receptor binding).

In uninjured immature animals, the NR2A subunit of the NMDA receptor shows dynamic developmental regulation, increasing dramatically in many brain regions during maturation – somatosensory cortex [Flint et al., 1997], visual cortex [Roberts and Ramoa, 1999], cerebellum [Takahashi et al., 1996] and hippocampus [Tovar and Westbrook, 1999] – and also increasing in response to specific environmental conditions, i.e. nurturing maternal behaviors [Liu et al., 2000], switching from dark to light rearing [Quinlan et al., 1999a, 1999b] and enriched environment housing [Tang et al., 2001]. It appears that upregulation of NR2A is a fundamental molecular process that underlies maturation of these excitatory circuits. Interestingly, fluid percussion injury causes a reduction in cortical NR2A in the first 1–2 days after injury in adults [Osteen et al., 2004] and P19 pups [Giza et al., 2006; Santa Maria et al., 2005]. Furthermore, pups also show a longer-lasting age-specific reduction in the NR2A subunit in the hippocampus until postinjury day 4 [Giza et al., 2006; Santa Maria et al., 2005]. Electrophysiological studies demonstrate impaired induction of long-term potentiation (LTP) with preserved long-term depression in the developing hippocampus after TBI [D’Ambrosio et al., 1998]. Impairment of LTP induction without affecting long-term depression has been shown to result from NR2A-specific inhibition of the NMDA receptor [Liu et al., 2004]. This raises the intriguing idea that impaired glutamatergic neurotransmission underlies deficits in cognition and plasticity after TBI and that perhaps the developing brain is uniquely vulnerable to the consequences of diminished excitatory activity.

GABAergic neurotransmission impairments, reduced inhibition and hyperexcitability have clearly been demonstrated after TBI in adults, particularly in hippocampal circuitry [Golarai et al., 2001; Toth et al., 1997]. Post-TBI reductions in GABA receptor binding have also been reported in mature animals [Sihver et al., 2001]. Reduction or genetic blockade of GABAergic activity impairs plasticity during critical periods of brain maturation [Hensch et al., 1998]. In the immature brain, there is evidence of hippocampal hyperexcitability early after fluid percussion injury [D’Ambrosio et al., 1999]. However, the specific contribution of GABAergic dysfunction to this hyperexcitability should be interpreted cautiously, as other mechanisms (such as glial dysfunction) have also been implicated.

Cholinergic pathways are clearly important for cognition and normal development [Berger-Sweeney, 2003; Ricceri, 2003]. An impairment of this neurotransmitter system has been reported in adult models of TBI, including early reductions in choline acetyltransferase, the biosynthetic enzyme for acetylcholine [Gorman et al., 1996]. Following adult TBI, decreased muscarinic acetylcholine receptor binding and protein levels are seen [Shao et al., 1999; Sihver et al., 2001], and evoked release of acetylcholine is diminished [Dixon et al., 1997]. Perhaps in response to this injury-induced cholinergic dysfunction, vesicular acetylcholine transporter levels actually increase chronically after adult TBI [Dixon et al., 1997; Shao et al., 1999].

While there have been no direct measures of post-TBI cholinergic neurotransmission in the immature animal, one study examined alterations in the regrowth of cholinergic projections following developmental TBI and found evidence of impaired neuroplasticity in the hippocampal dentate gyrus [Prins et al., 2003]. Despite ample evidence of altered cholinergic neurotransmission after TBI in mature animals [Dixon et al., 1999; Shao et al., 1999] and clear developmentally mediated changes in cholinergic receptors [Adams et al., 2002; Miyoshi et al., 1987], further work is necessary to more precisely determine how this neurotransmitter system is involved in plasticity following TBI to the immature brain.

Chronic alteration in dopaminergic systems has been implicated in some of the cognitive and attention problems seen in patients after pediatric TBI. Recently, Walter et al. [2004] used positron emission tomography to estimate the activity of aromatic amino acid decarboxylase (the final enzyme in dopamine biosynthesis) after fluid percussion injury in newborn and juvenile piglets. In this study, severe TBI triggered a significant upregulation of amino acid decarboxylase, but only in neonates and not in the juvenile piglets [Walter et al., 2004]. This is the first experimental evidence of TBI-induced age-specific effects on dopaminergic neurotransmission. The behavioral and developmental consequences of these changes remain to be seen.

Molecular Alterations: Gene and Protein Expression

Complex pathophysiological events, such as the response to TBI and the normal path of cerebral maturation, are particularly amenable to investigations using molecular screening techniques to delineate major pathways involved in these multifaceted processes. Using standard microarray techniques, we investigated the postinjury gene expression profile in the preweanling male rat brain (P19). Specifically, pups were killed, fresh brains were harvested and brain regions dissected immediately on ice within 24 h of both mild and severe lateral fluid percussion injury (FPI) and compared with sham animals (n = 4 per group). FPI was directed via a 3-mm-diameter craniotomy located 3 mm posterior and 6 mm lateral to the bregma, as described previously [Giza et al., 2002; Prins et al., 1996]. Mild injury was defined as loss of toe pinch response <45 s and severe injury as loss of toe pinch >120 s. This loss of toe pinch represents a simple postinjury physiological measure which correlates with atmospheric pressure, duration of apnea and expression of c-fos expression in our prior work [Giza et al., 2002]. In no case was the dura breached, and similarly injured animals at this age do not go on to develop a histological lesion [Gurkoff et al., 2006]. RNA was isolated from the ipsilateral cortex and hippocampus using guanidine thiocyanate extraction [Chomczynski and Sacchi, 1987]. Affymetrix rat gene RG U43A arrays were used. Analysis of expression data was performed using Genesifter software (www.Genesifter.net), setting the significance at changes greater than 2.0 fold up- or downregulated compared with sham. This analysis was chosen to minimize the amount of false positives, while realizing that differentially expressed genes with higher individual variability or a lower magnitude of change might not be detected as significantly different.

For mild injury, 60 genes were differentially expressed in injured versus sham animals. Forty genes were altered in the cortex (30 up and 10 down) and 32 were significantly changed in the hippocampus (26 up and 6 down). Following severe injury, 96 genes showed significant differences from sham. In the cortex, 58 genes were changed (48 up and 10 down), while in the hippocampus, 60 genes demonstrated differences (44 up and 16 down).

Categorizing the differentially expressed genes by primary function after mild or severe injury demonstrates global patterns of response that allow some general interpretations to be made. As might be expected, genes associated with stress/reactive oxygen species, inflammation or cytoskeletal proteins comprised a larger proportion of the significant changes marked by upregulation ( fig. 2 ). Conversely, among all downward gene expression changes, a larger proportion was made up of genes related to neurotransmission/plasticity and growth factors/hormones. These patterns remained consistent between mild and severe injuries.

Fig. 2.

Gene expression changes by functional category after mild and severe developmental brain injury. a Proportion of gene changes ascribed to specific functional categories following mild injury. The greatest proportion of downregulated genes can be classified as genes related to neurotransmission/plasticity or growth factors/hormones. b Proportion of gene changes ascribed to specific functional categories following severe injury. Again, genes associated with neurotransmission/plasticity and growth factors/hormones represent a much larger percentage of downregulated than upregulated genes. ROS = Reactive oxygen species.

Of the genes designated as neurotransmitter/ion channel/plasticity genes, 4/7 were downregulated by mild injury and 9/13 by severe injury. Genes coding for growth factors/hormones were also predominantly reduced, 3/3 by mild and 3/4 by severe injury (table 2).

Table 2.

Gene expression changes after mild and severe developmental TBI

Category	Gene	Cortex		Hippocampus
		mild	severe	mild	severe
Cell cycle	cell division cycle 2 homolog	2.19	2.25
	cyclin D3	2.14
	DNA polymerase α-subunit IV (primase)				2.05
	DNA topoisomerase 2α		2.06
	growth arrest and DNA damage inducible 45α	3.14	3.08		−3.3
	high-mobility group box 2	2.01	2.02	2.46	3.04
	nerve growth factor-inducible antiproliferative putative secreted
	protein PC3		−2.13
	mini chromosome maintenance deficient 6		2.42		2.4
Cytokine	chemokine ligand 3, macrophage inflammatory protein 1α		12.05
	chemokine receptor (LCR1)			−3.06	−2.16
	granulin, epithelin 1 and 2				2.09
	Gro		7.48
	interleukin-1β		5.58	22.03	3.1
Cytoskeletal	α-tubulin			2.26–2.36	2.3–3.44
	α-actinin-2-associated LIM protein, PDZ and Lim domain 1			2.32
	annexin I	2.68	2.31	3.78	3.0–5.31
	annexin II				2.55
	annexin IV, ZAP36			2	2.36
	MAP 1b		1.91
	nexilin, myosin heavy chain – nonmuscle		2.24
	PDZ and LIM domain 1, CLP36				2.01
	SH3 domain-binding protein (CR16)				−2.21
	transgelin		3.45	3.41	6.01
	tropomyosin 1α				−2.19
	vimentin	2.14	2.6	2.46	2.82
Glia	fatty acid-binding protein 7		2.28		2.14
	glial fibrillary acid protein	3.2	2.8		3.61
	S100	2.44	2.13		2.04–2.08
Growth/hormone	bone morphogenetic protein 3	−2.56
	cortistatin	−2.37	−2.06
	preoptic regulatory factor 1		−1.83
	transthyretin	−22.56 to −37.05	−5.65 to −13.5	−2.83	9.71
Inflammation	complement component 1s, serine protease			2.09	3.08
	complement component C3			4.47
	complement protein C1q β-chain	2.01	2.3	2.13	3.08
	contrapsin-like protease inhibitor-related protein (CPi-26)			2.52	5.51
	Fcγ receptor	2.79	2.42–2.61	2.65–3.25	2.04
	immunoglobulin-binding protein 1	−2.49
	interferon regulatory factor 1				2.4
	monocyte differentiation antigen CD14		2.09	2.47	2.85
Ion channel	calcium-activated K+ channel, Kcnma1	−1.95
	RCK2 K+ channel				−2.09
	voltage-gated K+ channel, Kcna2			−2.01	−2.17
	voltage-gated K+ channel, Kcng1	2.02
	voltage-gated Na+ channel I, Scn1a	−2.21
	voltage-dependent anion channel, porin 31HL, RVDAC1	−2.16
Metabolism	3-ketoacyl-CoA thiolase				2.18
	3-methylcholanthrene-inducible UDP-glucuronosyltransferase				2.01
	aminoimidazole ribonucleotide carboxylase		−2.98
	arachidonate 12-lipoxygenase		2.13
	arginase II		2.51
	cytochrome b5		3.29
	cytochrome P450				2
	cytochrome P450 4A3			−2.19
	cytochrome P450, subfamily 51				−2.08
	diabetes-inducible cytochrome P450		−2.26
	dihydropyrimidinase	2.81
	farnesyl diphosphate synthase		2.97
	isopentenyl diphosphate-dimethylallyl diphosphate isomerase		−2.04
	lipocalin 2	3.66	2.81–5.33	3.42–12.18	2.95
	mitochondrial genome: cytochrome oxidase subunits I, II, III, ATPase6				−3.28
	nonspecific lipid transfer protein		2.6		2.06
	pyruvate dehydrogenase phosphatase isoenzyme 1	−2.23
	retinol-binding protein 1	2.76	2
	ribonuclease, RNase A family 4	2.38	2.14
	suppressor of potassium transport defect 3 (SKD3 protein)	−2.18
Neurotransmitter	catechol-O-methyltransferase	2.52	2.56
	D1 dopamine receptor		2.39
	excitatory amino acid transporter 2, glutamate transporter 1			2.38
	GluR4-c				−2.0 to −2.25
	GluR5-2				−3.27
	kainate receptor subunit (KA1)				−2.06
	Na/H-coupled glutamine transporter, SP120				−2.07
	vesicular acetylcholine transporter				2
ROS	catalase		2.26
	γ-glutamylcysteine synthetase		2.32
	glutathione peroxidase	2.31	2.64
	heme oxygenase	3.07–3.16	2.11–5	2.07	2.55–2.79
Signal transduction	activin A receptor type II-like 1, serine-threonine kinase			−2.2
	CDP-diacylglycerol synthase		2.19
	Edg-1 orphan receptor	2.11
	lectin galactose binding, IgE-binding protein	5.33	2.18–3.02	2.75–2.93	4.65–5.28
	lectin galactose binding, soluble 1				2.86
	Mob-1			9.8	9.4
	platelet-derived growth factor receptor α			−2.38
	PLC δ4	2
	pyrimidinergic receptor P2Y	2.18
	Ras homolog gene family member V		2.6
	Ras-related protein, arachidonic acid epoxygenase				−2.03
	rhodopsin kinase		1.8
	transmembrane 4 superfamily member 3	2.11
Stress	glutathione S-transferase		2.42
	heat shock protein 27.1	8.79	2.13–3.93	2.09–3.21	5.13
	heat shock protein 70	2.12	2.11–3.72		6.73
	heat shock protein 70.2		3.38		8.64
	interferon-induced mRNA			2.41	2.32
	metallothionein 1a			2.88	2.09
	metallothionein 2 and 1	2.71	1.87–2.33	3.6	2.76
Synaptic plasticity	Arc		−2.03
	activin receptor interacting protein, S-SCAM				−2.74
	AMPA receptor interacting protein GRIP				−2.97
	Munc13-3 protein				2.06
Transcription factor	activating transcription factor 3, leucine zipper protein		2
	ASM15	2.94	2.31
	C-myc			3.07	3.93
	Cbp/p300 interacting transactivator 2, MRG1		2.64
	CCAAT/enhancer binding protein β, CELF	2.47	1.84–9.3	2.51	2.84
	DNA topoisomerase 2α				2.45
	early growth response 1, Krox-24, zif268, NGFIa		−2.09
	early growth response 2, Krox-20	2.14	−2.84
	immediate-early serum-responsive JE gene	3.68–4.12	3.37–5.16	44.85	4.19–39.25
	neuro-D4 protein		2.31
	NGFIb		−2.45
	nucleoporin p58				2.13
	pleiomorphic adenoma gene-like 1				2.47

Regionally, the overall numbers of genes up- or down-regulated were similar in both cortex and hippocampus, although there were some interesting differences based on gene function. Genes involving cell cycle, metabolism and reactive oxygen species were more often altered in the cortex, usually being upregulated. The gene categories showing injury effects preferentially in the hippocampus included cytoskeletal, inflammation and neurotransmission/plasticity. Most cytoskeletal and inflammation genes were increased, but those involving neurotransmission/plasticity were generally reduced.

Impairment of glutamatergic systems has long been implicated in deficits of learning and memory. Interestingly, several genes coding for proteins associated with glutamate receptors and signaling were downregulated, including GluR4, GluR5, KA1 and GRIP (glutamate receptor interacting protein) ( table 2 ). Following mild injury, the glutamate transporter EAAT2 (excitatory amino acid transporter 2) was upregulated (table 2). Upregulation of the glutamate transporter and downregulation of glutamate receptor subunits may both serve as means of functionally downregulating the glutamatergic system following the indiscriminate excitatory amino acid release triggered by biomechanical injury. The glutamatergic activity-dependent gene Arc also showed significant cortical downregulation following severe injury.

Among other neurotransmitter systems, the cholinergic marker VAChT (vesicular acetylcholine transporter) shows an upregulation in the hippocampus following severe injury, consistent with results seen in adults described above [Dixon et al., 1999; Shao et al., 1999]. Dopaminergic-related genes (catechol-O-methyltransferase and the D₁ dopamine receptor) also showed upregulation after severe injury, albeit only in the cortex (table 2). Upregulation of dopamine biosynthetic enzyme activity has also been reported after fluid percussion injury in neonatal pigs [Walter et al., 2004], as mentioned earlier.

Gene expression profiling can provide a broad picture of a complex response, but ultimately, proteins are the effectors of this response at the cellular level. Thus, proteomic approaches are also valuable in breaking down the response to injury, particularly in the developing brain. The advantages of a proteomic approach are that it will identify changes that occur independent of transcriptional alterations, including increased/decreased protein degradation and, most importantly, posttranslational modification. For these reasons, gene array and proteomic results do not necessarily match; nonetheless, these two approaches can be strongly complementary. A proteomic approach was utilized in the preweanling rat (P17) looking at proteins from the dorsal hippocampus 24 h after controlled cortical impact (CCI) injury. Significant changes relevant to postinjury plasticity were detected and included reductions in several proteins associated with the cytoskeleton (actin β-chain −37%, tubulin α₁α₂-chain −45%, tubulin β-chain −53%), protein processing (glucose regulated protein 78 −51%) and signal transduction (phosphatidylinositol transfer protein α isoform −27%). Increases in free radical/cell stress proteins were also detected, including a 62% upregulation of superoxide dismutase [Jenkins et al., 2002].

In addition to these broad gene and protein profiling approaches, several investigators have targeted studies of specific plasticity-related genes following developmental TBI. Using a subset of genes specifically induced by depolarization (but not by trophic factors), Giza et al. [2002] reported age- and severity-specific transcriptional changes in the hippocampus. Nerve growth factor-induced protein B, a transcription factor, is upregulated in a severity-dependent fashion in the immature brain, but is increased independent of severity in adults. Synaptotagmin IV, a presynaptic calcium-binding protein that in distinction to other members of the synaptotagmin family inhibits synaptic activity, is increased only in developing animals after severe injury and not at all in adults [Giza et al., 2002].

Brain-derived neurotrophic factor (BDNF) is an activity-dependent molecule closely associated with experience-dependent plasticity which undergoes distinct developmental regulation. After mild lateral FPI in P19 rats, relative reductions in BDNF mRNA were noted between the ipsilateral and contralateral cortex and hippocampus, and protein levels of BDNF were significantly reduced (−10%) in the ipsilateral occipital cortex at postinjury day 7 [Griesbach et al., 2002]. Studies in adults have shown that exercise-induced upregulation of BDNF and subsequent cognitive enhancement is impaired early after injury [Griesbach et al., 2004], suggesting that this reduction may also have relevance to plasticity and cognition in the developing animals. Most importantly, the immature animals demonstrated a pattern of impressive contralateral BDNF upregulation in both cortex and hippocampus up to 2 weeks after injury, perhaps revealing a distinct compensatory mechanism for plasticity in the developing brain [Griesbach et al., 2002].

Molecular markers of cellular stress and injury have also been examined following developmental TBI. While these changes may be more relevant to neuronal survival and cell death (see next section), cells that are sublethally damaged may survive in a dysfunctional state. With either cell loss or localized cellular dysfunction (or both), the remaining neuronal circuits may thus be altered and potentially interfere with normal cerebral maturation. Glutathione peroxidase, a critical enzyme involved in superoxide metabolism, is upregulated in the adult mouse cortex and hippocampus following CCI. However, in the P21 mouse, this increase is not seen, indicating a potential age-specific vulnerability to oxidative stress in the immature brain [Fan et al., 2003]. Other cellular stress molecules implicated in an age-specific pathophysiological response to developmental TBI include isoforms of heme oxygenase, which show a >2-fold induction in the P19 cortex and hippocampus following both mild and severe lateral FPI (table 2). The developmental implications of heme regulation after TBI have been recently reviewed [Chang et al., 2005]. Metallothionein I/II are also important molecules in the response to oxidative stress and show developmental regulation with lower basal levels in the immature brain. However, deafferentation injury (occipital cortex ablation) induces a far higher and earlier induction of metallothionein I/II in the lateral geniculate in 10-day-old mice compared with adults [Natale et al., 2004]. Metallothionein I/II gene expression is also significantly upregulated in the cortex and hippocampus of P19 rats following TBI, as described above (table 2).

Cell Death: Necrosis and Apoptosis

Histological damage and neuronal loss are not direct measures of plasticity, but can certainly have effects upon developmental plasticity by disconnecting circuits, causing local calcium accumulation in dying cells, triggering inflammation, altering the extracellular milieu and reducing trophic support. Postinjury cell death is distinctly different in the immature rat brain, which is particularly sensitive to excitotoxicity in the neonatal period (1st postnatal week) [Ikonomidou et al., 1989, 1996], but seemingly less so in the 2nd to 3rd postnatal weeks, when compared with the adult brain. In fact, both weight drop and lateral FPIs seem to result in little neuronal loss in the preweanling (P15–19) rat [Adelson et al., 2001; Bittigau et al., 1999; Gurkoff et al., 2006; Prins et al., 1996]. This absence of overt cell death suggests that the molecular and physiological perturbations seen following developmental TBI are due in greater part to neuronal dysfunction rather than destruction. Tissue destruction is also dependent upon injury type, as CCI injury delivered to P21 mice [Tong et al., 2002] or P17 rats [Card et al., 2005] clearly does induce histological cell death.

However, in the neonatal period, the injury response is distinct. There is a peak in vulnerability to an excitotoxic lesion at P6–7 (hypobaric/ischemic injury or direct NMDA injection) [Ikonomidou et al., 1989]. This susceptibility to NMDA-mediated excitotoxicity is also seen after a weight drop/contusion TBI is administered to P7 rats. In this model, NMDA antagonists (but not α-amino-3-hydroxy-5-methylisoxazole-4-propionate antagonists) provide neuroprotection from necrotic neuronal death [Ikonomidou et al., 1996]. However, treatment with NMDA antagonists also appears to trigger a substantial wave of delayed apoptotic cell death in widespread brain regions 24–48 h after treatment [Bittigau et al., 1999; Pohl et al., 1999]. Furthermore, this delayed apoptotic effect has been seen in the absence of injury after other neonatal treatments that impair neural activity, including common anticonvulsants and general anesthetics [Bittigau et al., 2002; Jevtovic-Todorovic et al., 2003]. There is some evidence that this drug-induced enhancement of developmental apoptosis has long-term effects on plasticity, as cognitive impairments in spatial learning have been reported in neonatally treated rats, with anesthetics [Jevtovic-Todorovic et al., 2003] or MK801 [Adelson et al., 2002].

In addition to acute and subacute injury-induced cell death, there may also be a chronic evolving lesion after TBI. This has been well described in the adult brain after different models of injury, and ongoing neurodegeneration may occur as long as 1 year after the insult [Conti et al., 1998; Fox et al., 1998]. In this issue, using a developmental TBI model (P21 CCI in mice), Pullela et al. [2006] describe cortical lesion expansion and progressive hippocampal neuron loss as late as 4.5 months after the original injury.

Perturbation of Neural Connectivity and Function

Changes in neuronal circuitry are a critical part of normal development, and perturbations of this connectivity have been implicated in pediatric TBI-induced learning disabilities and psychiatric disorders [Kumar and Cook, 2002; Levin, 2003]. Lateral FPI in immature animals has been shown to result in alterations in anatomical connections that are not associated with functional improvement. Ip et al. [2002] reported increased dendritic branching in the parietal cortex of sham animals housed in enriched environment (EE), and these animals have been shown to be cognitively superior in spatial learning tasks [Fineman et al., 2000; Giza et al., 2005]. While injured animals reared in EE showed no dendritic expansion, injured animals maintained in standard housing conditions did show significant injury-induced dendritic expansion [Ip et al., 2002]; however, in this case, subsequent studies revealed that these injury-induced anatomical increases did not manifest as any functional advantage on cognitive tests [Fineman et al., 2000; Giza et al., 2005].

Using a clever system of pseudorabies virus (PRV) injection into the entorhinal cortex and retrograde transsynaptic transport, it has been possible to begin looking at functional connectivity after developmental TBI [Card et al., 2005]. The strength and pattern of the PRV immunohistochemical signal differed according to the number of synapses crossed, allowing for qualitative assessment of patterns of neural connections. Twenty-eight days after CCI on P17, the overall distribution of the PRV signal did not change, suggesting that the primary pathways remained intact. However, a significant increase in the magnitude of the PRV signal was detected in the neocortex and overall, leading the authors to conclude that this methodology would prove useful in further experiments to compare and contrast neural network development in the immature brain following injury.

Electrophysiological studies have also demonstrated impairment of neural function after TBI. Twenty-four to 48 h following lateral FPI in P24–31 rat pups, the threshold for generation of a CA1 population spike was increased and no induction of LTP was possible [D’Ambrosio et al., 1998]. These results clearly show impaired responsiveness of this excitatory circuit after developmental TBI. Abnormalities in the generation and maintenance of LTP have also been reported following lateral FPI in adult animals [Sanders et al., 2000; Sick et al., 1998].

Experience-Dependent Enhancement: Good Plasticity?

Normal cerebral maturation in animals and humans can be substantially modified by an individual’s environment. In fact, the effects of EE during development have been well characterized as a form of ‘good plasticity’ that results in lasting anatomical – rats [Greenough et al., 1973; Rosenzweig and Bennett, 1996] or humans [Jacobs et al., 1993] – and behavioral enhancements [Leggio et al., 2005; Noble et al., 2005; Venable et al., 1988].

Using the lateral fluid percussion model of TBI in P17–20 rat pups, Fineman et al. [2000] reported that injured pups showed no overt histopathology or impairment in spatial learning 1 month after the injury, when housed in standard conditions. However, they discovered that injured pups reared in an enriched environment for 17 days failed to develop the increases in cortical thickness and improved spatial learning capabilities seen in their uninjured controls [Fineman et al., 2000]. Thus, while there was no clear cortical atrophy, there was a failure of experience-dependent plasticity. Enriched environment rearing also resulted in an increase in cortical pyramidal neuron dendritic arborization, but this was inhibited in animals that underwent FPI prior to EE housing [Ip et al., 2002]. Since this injury model does not produce measurable neuronal loss [Gurkoff et al., 2006] and since the animals demonstrate normal open field exploratory behavior [Fineman et al., 2000], the impairment of plasticity seen is presumed to be due to diminished neural function or responsiveness. This period of unresponsiveness to EE has been shown to partially recover by 2 weeks after injury. When P19 pups were subjected to moderate-severe lateral FPI and allowed to recover for 2 weeks prior to differential housing (standard vs. EE), both sham and injured animals showed shorter latencies in the Morris water maze, indicating an augmentation of spatial learning. However, when tested for memory in the probe trial, uninjured EE animals still outperformed the FPI EE animals, which showed lasting deficits in spatial memory [Giza et al., 2005]. Thus, the 2-week postinjury delay allowed for enrichment-induced benefits in task acquisition, but augmentation of memory functions was still impaired.

By examining molecular signals for changes that correlate with diminished ‘enrichability’, it is interesting to remember that the injury-induced reductions in the NR2A subunit of the NMDA receptor occur within the first postinjury week. Preliminary work looking at whether EE housing rescues the levels of NR2A after injury reveals that these reductions occur independent of housing conditions [Santa Maria et al., 2005].

Chronic Sequelae: Bad Plasticity?

While the increased neuroplasticity of the developing brain has generally been thought of as beneficial, it is also known that abnormal neural connectivity can result in worsened function and this can be thought of as ‘bad plasticity’. One means by which functional deficits may occur is in conjunction with progressive lesions over time, as discussed above. If the injury process evolves over a long time period, then neural connections are being disturbed at the very time when circuit maturation should be occurring. In the model of developmental (P21) mouse CCI, no significant differences in latencies to find the hidden Morris water maze platform were detected when the animals were tested 5 weeks after injury, but in adulthood (over 4 months after injury), the injured animals showed clear acquisition deficits [Pullela et al., 2006]. This delayed impairment is likely due to ongoing injury progression and abnormal reorganization of damaged networks.

A second mechanism by which ‘bad plasticity’ can manifest negative cognitive consequences is when the initial injury triggers a trophic response but the regrowth of damaged fibers is perturbed. This would occur in the absence of a progressive lesion, so ongoing deafferentation could not be the underlying mechanism. Following an entorhinal cortical lesion in juvenile (P28) rats, the cholinergic axons to the hippocampus regrow in a characteristic manner, and this regrowth is correlated with changes in hippocampal synaptophysin staining. No distinct spatial memory deficits are seen. Animals subjected to fluid percussion injury 1 day prior to entorhinal cortex lesion showed altered patterns of cholinergic fiber regrowth and concomitant impairments of spatial memory [Prins et al., 2003].

A final mechanism of ‘bad plasticity’ involves self-propagating neural activation that can result in reinforcement of excitatory circuits, ultimately leading to spontaneous seizures. Chemically induced status epilepticus using lithium pilocarpine clearly has age-dependent patterns of acute cell death, mossy fiber sprouting and subsequent epileptogenesis [Sankar et al., 1998]. It is generally accepted that the pediatric population is more vulnerable to both acute posttraumatic seizures and to the development of late posttraumatic epilepsy [Annegers et al., 1998; Annegers and Coan, 2000]. Recently, chronic recordings after P32–35 lateral FPI in rats have shown a high incidence of posttraumatic epilepsy as the animals mature [D’Ambrosio et al., 2004]. Furthermore, the electrical pattern of the neocortical seizures changes with postinjury time and with cerebral maturation, starting as focal seizures, but over time evolving into spreading seizures and eventually into limbic seizures involving hippocampal structures bilaterally [D’Ambrosio et al., 2005]. These studies provide strong evidence for ongoing circuit rearrangement and abnormal plasticity after developmental TBI and result in a clinically relevant and potentially disabling condition.

Summary

TBI is predominantly a condition of the young brain, and this younger age group has often been credited with enhanced plasticity that results in improved recovery. However, plasticity is a complex and balanced process that must be precisely executed for optimal maturational outcome. Injuries to the immature brain must be evaluated carefully, and recovery to preinjury baseline is clearly an inadequate endpoint in cases where ongoing development and/or ongoing experience-dependent enrichment are occurring. The pathophysiology of acute TBI results in a multitude of processes that can perturb normal developmental plasticity, including faulty neurotransmission, alterations in molecular signals, necrotic and apoptotic cell death, changes in neural connectivity and function, inhibition of experience-dependent ‘good’ plasticity and activation of self-propagating ‘bad’ plasticity. Realizing that the dogma ‘younger is better’ may be incorrect is the first step to carefully investigating these complex injury effects and devising mechanistic interventions to promote recovery from these often devastating injuries.