A Critical Evaluation of Current Concepts in Cerebral Palsy

Abstract

Spastic cerebral palsy (CP), despite the name, is not consistently identifiable by specific brain lesions. CP animal models focus on risk factors for development of CP, yet few reproduce the diagnostic symptoms. Animal models of CP must advance beyond risk factors to etiologies, including both the brain and spinal cord.

Introduction

Cerebral palsy (CP) is the most common motor disability in childhood with a prevalence of 2.1 per 1,000 live births in developed countries, although a higher prevalence of CP is likely in lower income countries (152, 155). CP began as a description of physical symptoms and impairments by Dr. William John Little in 1862 (118, 119). Although Little originally termed this a “cerebro-spinal disorder,” the nomenclature quickly transitioned to the term cerebral palsy (118, 119). With the CP terminology, the brain and cerebral cortex became the focus of the etiology of the symptoms in both research and clinical interventions.

Since 1862, medical and scientific advances have been made in the prevention and medical management of CP. However, the physical impairments in individuals with CP remain incurable, and the prevalence has remained largely unchanged (155). Potential contributors to the lack of curative therapies for CP include the focus on the brain and cerebral cortex as the driver of the symptoms. Furthermore, CP is an inherently complex disorder due to the heterogeneity of etiologies and physical impairments.

The heterogeneity of CP and long-held clinical and scientific beliefs that CP is only a developmental brain injury has hampered animal models of this disorder. Specifically, the focus of animal models has been on inflicting a brain injury and subsequent description of the neuropathological changes in the brain. Yet, CP is a disorder of movement and postural control, symptoms absent in most animal models.

The following review will critically assess the current state of CP animal models in research, with a goal of promoting animal models that display the CP phenotype. We will briefly discuss the history of CP, symptoms and diagnosis of CP, risk factors for CP, the contribution of animal models to our understanding of CP pathophysiology, and gaps in our understanding of CP. The gaps in our understanding of the mechanistic underpinnings of CP are largely due to the sole, laser-like focus on a brain injury only. This limitation is further compounded by a lack of animal models with the physical symptoms of CP.

Is Cerebral Palsy an Accurate Term?

History of Cerebral Palsy

In 1862, Dr. William John Little, an orthopedic surgeon with a particular interest in tenotomy procedures, described children who had “congenital distortions” of their limbs (118, 119). Little described this disorder based on observations of muscle stiffness and limb deformity in children who had difficult births including prematurity, prolonged labor, breech presentations, and even nuchal umbilical cords (118, 119). In his original manuscript, Little referred to the symptoms as a “cerebro-spinal disorder” (118, 119) because of postmortem evidence of brain and spinal cord involvement. The abnormalities included venous and capillary congestion on the surface of the brain and spinal cord, and blood in the cerebrospinal fluid in the brain ventricles and within the subarachnoid space surrounding the brain and spinal cord (118, 119). Despite this early recognition of both brain and spinal involvement, CP quickly became referred to as “cerebral paresis” (49). This loss of the “spinal” component of this descriptive name may be due to the commonly held belief at that time that the spinal cord was more completely formed than the brain at birth (118, 119). Little hypothesized that this disorder was due to anoxia (asphyxia neonatorum) secondary to birth difficulties, recognizing a central nervous system disorder as the driver of the physical symptoms (118, 178). This was a novel hypothesis since the prevailing belief was that the limb deformities were from direct mechanical injury (118, 178). However, Little had sparse evidence to support this hypothesis, although he recognized the work of others who noted abnormal engorgement of the meninges and blood surrounding the brain and spinal cord in infants who died from suspected asphyxia neonatorum (118, 119). Over time, the spastic diplegic form of CP became known as Little’s Disease (49, 159). Although Little’s original concept of a “cerebro-spinal disorder” may be more accurate than the subsequent term of “cerebral paresis,” this long-held concept of an injury to the brain due to anoxia at birth has persisted despite evidence suggesting that CP may not be the result of an isolated injury at a single time or a single insult (147, 194).

The actual term “cerebral palsy” (CP), which continues to be used today, was coined in 1889 by Sir William Osler from a series of five lectures titled “The Cerebral Palsies of Children” (156, 178). In this series of lectures, Osler described numerous patients with symptoms of CP who had difficult deliveries (156, 178). Shortly thereafter, Dr. Sigmund Freud developed a classification system for CP and described other conditions associated with CP (65). His classification system was based on hypothesized timing of the injury occurrence, including antepartum, intrapartum, and postnatal (65, 178). However, Freud strayed from this research when he was unable to definitively describe unique neuropathological brain abnormalities to set CP apart from other conditions along with confusion of CP with other conditions resulting in similar impairments, including spinal disorders like poliomyelitis (178). The reasons for abandoning additional work in this area, in retrospect, were a herald for the possibility that CP is more than an injury to the brain.

Cerebral Palsy in Modern Medicine

Eventually, CP and its definition were revisited, although without a change in nomenclature. In 1959, the definition of CP was updated in a consensus report to a “persisting qualitative motor disorder due to non-progressive interference with development of the brain occurring before the growth of the central nervous system (CNS) is complete” (125). Although these authors recognized the incomplete development of the entire CNS at the time of the insult or injury, they unfortunately only included an insult to the brain as part of the definition of CP, excluding the role of the spinal cord. This definition of CP was further refined in 1964 in another consensus report by Bax et al. as primarily a “disorder of movement and posture due to a defect or lesion of the immature brain,” again without mentioning the spinal cord involvement (11). In 2004, the definition of CP was revisited yet again by a group of international experts. This definition, which continues to be used today, describes CP as a “group of permanent disorders of the development of movement and posture, causing activity limitation, that are attributed to non-progressive disturbances that occurred in the developing fetal or infant brain,” with further description of other potential associated conditions (169). Surprisingly, although the length of these descriptions of CP progressively increased with time, they exclude mention that CP symptoms also relate to spinal cord dysfunction (i.e., hyperexcitability/spasticity and delays in limb and trunk motor development). The involvement of abnormal spinal cord development is unmentioned in the current clinically accepted definition of CP, and the focus remains on the brain despite the severity or even presence of physical impairments being poorly predicted by brain imaging (10, 114, 203). However, despite the “progress” in the clinical consensus of CP, the definition of CP continues to focus on the brain without acknowledgment of developmental spinal cord dysfunction.

Is a Brain Injury Required for Diagnosis and Symptoms of Cerebral Palsy?

Diagnosis of Cerebral Palsy

CP is a clinical diagnosis, typically made before 2 years of age, with most being diagnosed in the first year of life (77, 90, 92). It is diagnosed based on both clinical symptoms and neurological signs, although not all signs or symptoms are present in an individual (Table 1). Typically, it is the delay in reaching normal developmental milestones that prompt a clinical evaluation for CP, especially in infants with risk factors for developing CP. Abnormal brain imaging is not a diagnostic requirement, nor is the presence of abnormal brain imaging diagnostic of CP if there are no motor symptoms.

Table 1.

Signs and symptoms of cerebral palsy

Neurological	Orthopedic	Cognitive	Vision/Hearing	Aerodigestive	Other
Ataxia	Femoral anteversion	Autism	Conductive hearing impairment	Aspiration pneumonia	Disorder sleep  initiation
Athetosis	Hip dislocation	Epilepsy	Cortical visual  impairment	Constipation	Disordered sleep  maintenance
Dystonia	Hip dysplasia	Intellectual impairment	Dyskinetic strabismus	Dysarthria	Failure to thrive/ poor growth
Gait disorder	Hip subluxation	Learning disabilities	High myopia	Dysphagia	Low bone mineral  density
Hyperreflexia	Joint contracture		Retinopathy of Prematurity	Gastresophageal reflux	Neurogenic bladder
Hypotonia	Patella alta		Sensorineural hearing  impairment	Obstructive sleep apnea	Muscle spasms
Impaired gross motor  coordination	Scoliosis			Respiratory disorders  (reduced forced  vital capacity,  forced expiratory  volume, peak  expiratory flow)	Pain
Impaired fine motor  coordination	Tibial torsion			Sialorrhea
Spasticity

Symptoms of Cerebral Palsy

The primary symptoms of CP include spasticity (most common), dyskinesias (such as dystonia, athetosis, chorea), ataxia, hyper-reflexia, and in rare cases hypotonia (89, 152, 181). Spasticity is detected clinically as a velocity-dependent resistance of a muscle to stretch (112, 173). It is thought to be due to loss of descending inhibition of the motor neuron (102, 132). It is measured clinically by subjective assessment on physical exam and is typically scored in severity using the Modified Ashworth Scale and Modified Tardieu Scale (74, 135, 141). Dyskinesias are unwanted, often sustained, involuntary movements commonly due to an injury or pathology of deep brain structures (172). These can be evaluated clinically using the Barry Albright Dystonia Scale, Dyskinesia Impairment Scale, and Hypertonia Assessment Tool (1, 9, 98, 104, 139). Ataxia involves incoordination of movement, typically due cerebellar injury or maldevelopment (105). Hyper-reflexia is an increase in muscle stretch reflex, often elicited on physical exam via a tendon tap (i.e., patellar tendon, Achilles tendon, and biceps brachii tendons), with subjective grading of robustness of response (197). Hyper-reflexia, similar to spasticity, is thought to be due to a loss of descending inhibition of the motor neuron (197). Hypotonia, or decreased muscle tone, is most commonly associated with muscle or peripheral nerve disorders (117). However, it can occur in central nervous system, genetic, and metabolic disorders, although the mechanism underlying hypotonia is poorly understood (117). Hypotonia is graded clinically on presence and subjectively on severity, although no formal grading scales have been agreed upon.

These varied symptoms point to varied CP etiologies involving brain and spinal cord pathophysiology. Furthermore, the complexity and heterogeneity of CP is reflected by the secondary conditions associated with CP. These include epilepsy, musculoskeletal disorders (hip dysplasia, boney torsional deformities, scoliosis), autism, intellectual disability, chewing and swallowing disorders, respiratory dysfunction, sleep disorders, chronic pain, cortical visual impairment, hearing impairment, and bladder and bowel disorders (Table 1; FIGURE 1) (22, 44, 55, 111, 140, 148, 152, 154, 162, 169, 185). The most common form of CP is spastic CP (>80% of all diagnosed cases) (89), which almost certainly involves dysfunction of the spinal cord since spasticity is related to disinhibition of the motor neuron (102, 132). Spastic CP is diagnosed based on clinical symptoms, as noted above, with the presence of spasticity, which is also determined clinically. To advance our understanding of CP, it may be best to separate CP into those who predominantly display symptoms of spasticity and those who do not, since the spastic component is likely indicating significant involvement of the spinal cord. Furthermore, >80% of individuals with CP have spasticity (4). In this review, we will narrow our focus to spastic CP.

FIGURE 1.

Physical impairments and secondary disorders associated with spastic cerebral palsy in humans

In addition to spasticity, the essential diagnostic criterion for spastic CP includes impaired movement and/or coordination, which are determined clinically based on developmental norms for acquiring motor skills. In humans, impaired movement includes difficulty achieving motor milestones, crawling, standing, or walking, or difficulty with sitting or postural control. Note that although impaired movement is observed in monkey (red), cat (blue), rabbit (yellow), and mouse (green) models, spasticity is exhibited only in the mouse and rabbit. Neither cardinal symptoms nor signs of spastic CP are observed in any of the hamster, rat, or sheep models. Clearly, animal models that more closely resemble clinical phenotypes are needed for rational approaches to the understanding of the disease mechanisms and eventual translation to treatment.

What Is Known About the Etiology of Spastic Cerebral Palsy?

Symptoms of spastic CP give us insight into etiology. Spastic CP is classified topographically as unilateral (hemiplegia) or bilateral (diplegia or quadriplegia) (152, 177). Individuals with diplegic spastic CP tend to have greater involvement of the legs than the arms (i.e., disproportionately impaired spasticity and motor control of the legs compared with the arms), suggesting a rostral-caudal pattern of injury for individuals with this type of spastic CP (158). Some known risk factors for developing spastic CP include asphyxia, hemorrhagic or ischemic stroke (prenatal or neonatal), infection (prenatal or neonatal), brain malformation, trauma (inflicted or accidental), non-progressive genetic disorders/differences, and dysmaturation (Table 2) (5–7, 171). The mechanisms by which these risk factors cause or contribute to a neurodevelopmental injury to the brain is discussed in many reviews (54, 57, 99, 171, 182, 194–196, 205). Yet, there is scant information about the impact these risk factors have on spinal cord development. The spinal cord should not be ignored, since it is vulnerable to a developmental injury during this same period, and in particular the motor neurons (23, 189).

Table 2.

Prevalence of cerebral palsy by live birth based on risk factors and percentage of individuals with cerebral palsy diagnosis who had risk factors

Risk Factor	Prevalence of CP		Percentage (%) of Individuals with CP
Birth prematurity	<28-wk gestation	82 per 1,000 live  births (155)	73% (90)
	28- to 31-wk gestation	43 per 1,000 live  births (155)
	32- to 36-wk gestation	7 per 1,000 live  births (155)
In utero insults (i.e.,  uterine ischemia, intrauterine  infections, uterine  inflammation)	Odds ratio of developing CP if maternal  infection during pregnancy 1.55,  with greater risk if infection  at 21–40 wk of gestation (153)		12% of those with spastic  CP (79)
Brain malformations			8.6–11.1% (70,  107, 108)
Stroke	1 per 4,000 live births (122)		1% (from postnatal  stroke) (72)
Hypoxic-ischemic  encephalopathy	1.5 per 1,000 live births (110)		14.5% of term  infants (76)
Asphyxia			2–10% (53,  165)
Twin gestation	Odds ratio of CP diagnosis if  twin gestation 6.62 (153); 7–12  per 1,000 live births  (20, 113, 160, 209)

The second column shows the prevalence of cerebral palsy per live births according to risk factors (first column). The last column displays percentage of individuals living with cerebral palsy who had a particular risk factor exposure. Note that some individuals with cerebral palsy may have been exposed to more than one risk factor.

Therapeutic interventions for spasticity in CP may also provide insight into etiology. Interventions to reduce spasticity symptoms generally target motor neuron hyperexcitability, although the targeting depends on the intervention (i.e., ranging from a specific muscle to the entire spinal cord). These spasticity-reducing interventions act on the motor neuron either through direct modulation of neurotransmitter input or indirectly through various mechanisms (FIGURE 2). For example, baclofen is a GABA_B receptor agonist commonly used for spasticity treatment in individuals with spastic CP (25, 134). GABA is one of two inhibitory (hyperpolarizing) neurotransmitters in the central nervous system, with the metabotropic GABA_B receptors located in both the brain and spinal cord (136). Baclofen is often administered intrathecally, targeting the spinal cord (18). An intervention that indirectly mitigates motor neuron hyperexcitability is a selective dorsal rhizotomy (FIGURE 2) (27, 133). This is a procedure that removes some afferent input to the motor neuron, some of which occurs directly via reduction of Ia afferent inputs through selectively sectioning portions (rootlets) of the lumbar sensory nerves (usually L₂–S₁ with generally 25–50% of rootlets sectioned), and some of which may occur indirectly. The end result is a reduction or elimination of spasticity in the legs via interruption of the sensory excitatory input to the motor neuron (133). Both of these interventions target leg spasticity, which typically involves all muscles of the lower limb at varying degrees. Symptoms of spastic CP always involve the legs, with variable involvement of the trunk and arms, supporting a differential rostro-caudal burden of disease, suggestive of spinal cord involvement.

FIGURE 2.

Medical and surgical interventions in the treatment of spastic cerebral palsy

All of the current clinical treatments for spastic CP involve both the brain and the spinal cord. Indeed, most treatments exclusively affect the spinal cord, motor neuron and/or motor unit alone. Treatment with baclofen, a presynaptic GABA agonist, may be orally (systemic) or intrathecally (targeting spinal cord neurons) administered. Gabapentin is a mild GABAmimetic and binds the α2δ subunit of voltage-gated Ca²⁺ channels. Clonazapam and diazepam are similar GABA receptor agonists. Tizanidine and clonidine primarily reduce interneuron activity within the spinal cord via agony of α2 adrenergic receptors. Selective dorsal rhizotomy selectively removes Ia afferent inputs to motor neurons, reducing their reflex excitability. Dantrolene works at the level of skeletal muscle, inhibiting the release of Ca²⁺ from the sarcoplasmic reticulum.

It is likely that spasticity is mediated by spinal cord circuits, since spasticity results from disinhibition of the motor neuron (102, 132). This is in direct conflict with the long-held conceptual framework that spastic CP is exclusively a developmental brain disorder. We will critically evaluate the premise of spastic CP as a brain disorder, assess preclinical animal models of this disorder, and propose a shift in the paradigm to consideration of spastic CP as a central nervous system disorder affecting both the brain and the spinal cord.

How Is Spastic Cerebral Palsy Modeled?

To study the mechanisms underlying known, suspected, and yet to be identified etiologies of spastic CP, animal models are essential. Animal models of spastic CP must display motor symptoms. Yet, most animal models of CP have focused on the risk factors for developing spastic CP rather than the mechanisms causing the spasticity and motor symptoms. Furthermore, these insults or risk factors impact not only the brain but also spinal cord development and specifically motor neuron abnormalities. The pathognomonic symptoms of spastic CP imply that motor neuron abnormalities are the mechanistic driver of the spasticity and motor impairment. Animal models of spastic CP must move beyond the prevailing research paradigm of modeling risk factors. Progressing beyond descriptions of risk factors for spastic CP to defining the mechanisms causing the symptoms facilitates development of targeted curative interventions, as opposed to merely modulating the symptoms. To move the research paradigm forward, it is imperative to understand the state of animal models for spastic CP.

Animal Models of Spastic Cerebral Palsy Risk Factors

Traditionally, animal models of spastic CP have focused on a brain injury as the driver of symptoms. Yet, few of the commonly accepted animal models actually display the diagnostic symptoms of spasticity and motor impairment (FIGURE 1). Most animal models of spastic CP are derived based on known risk factors for developing this permanent disorder.

The most common risk factor for development of spastic CP is birth prematurity (4, 155, 158, 181). Birth prematurity has been modeled via various methods for inducing preterm birth in mice, rats, rabbits, and sheep (15, 26, 37, 47, 48, 52, 69, 71, 109, 130, 151, 166, 168, 210). Due to the preterm birth or the research protocols, these animals do not survive, thus do not develop CP symptoms. Furthermore, a notable issue with animal models using birth prematurity is that most species are born at different stages of developmental maturity (34, 51, 204). These differences in developmental stages may not be reflective of the developmental timing of preterm birth in humans.

In utero insults are known risk factors but are far less common causes of spastic CP than birth prematurity (176, 207). In utero insults that have been modeled as risk factors for spastic CP include reduced uterine blood flow (uterine ischemia) and intrauterine infections or inflammation (51, 73, 147). Uterine ischemia has been modeled in the rabbit, sheep, and monkey, with physical symptoms observed only in monkeys and rabbits (40–42, 50, 91, 130, 142, 166, 168). In monkeys, feeding difficulties, weakness, hypertonia, and abnormal movement lead to them perishing due to the feeding difficulty (142). Rabbits demonstrate physical impairments including difficulty with head control, locomotion, swimming, and feeding (40–42, 45, 46). The husbandry for these animals is arduous, since the feeding impairment necessitates orogastric delivery of rabbit milk for survival (40–42). Uterine inflammation via introduction of an endotoxin or live bacteria has been extensively studied in mice, rats, rabbits, and sheep, with many of these animals displaying brain abnormalities. However, none survived a sufficient length of time to display motor symptoms (14, 15, 26, 28, 37, 38, 43, 48, 52, 58, 71, 120, 151, 170, 210).

Brain malformations, although rare, are also a risk factor for developing spastic CP (158, 184). Brain malformations have been modeled through administration of a glutamate analog, which results in neuronal brain migration abnormalities in hamsters (78, 129). Despite the presences of these neuronal migration abnormalities, these animals do not appear to develop motor symptoms (78, 129).

Stroke is a well-known risk factor for development of unilateral spastic CP (89, 181). Unilateral spastic CP is a type of CP that is increasing in frequency (89, 181). Stroke has been induced in animals via unilateral carotid artery occlusion, although the animals do not develop motor symptoms (17, 115).

Hypoxic-ischemic encephalopathy is a risk factor for severe forms of CP, including spastic quadriplegic or dyskinetic CP (81). Spastic quadriplegic CP is a form of CP in which an individual has difficulty with movement of arms and legs, and typically also has difficulty with postural or trunk control. Dyskinetic CP may also be associated with impaired movement of all four limbs and the trunk, typically due to other movement difficulties, including dystonia, chorea, and athetosis. These are the most common types of CP to occur in infants born at term gestation (19, 80, 89, 181). Hypoxic-ischemic encephalopathy is modeled in a variety of animal species via bilateral carotid artery occlusion (29, 137, 150, 193) or unilateral carotid artery occlusion/ligation and exposure to hypoxia (21, 95, 97, 145, 146, 167, 183, 188, 191). Despite the extreme reduction in perfusion to the brain in these animal models, motor symptoms were infrequently seen (137, 150). In cats with bilateral carotid artery ligation, delays in visual and auditory orientation accompany delays in standing, walking, and running (137, 150). In a monkey, hypoxic-ischemic encephalopathy was modeled via placental abruption and delayed uterine membrane opening, with the animal exhibiting lethargy, decreased activity, easy startle, and difficulty with maintaining upright posture (164).

Asphyxia is another risk factor of CP that is typically associated with the development of spastic and dyskinetic quadriplegic CP (19, 80). Despite the severity of the motor symptoms in humans with CP, animal models using asphyxia have had minimal, temporary motor symptoms that improve with time (88, 115, 143, 144, 189, 192).

Thus the glaring flaw in most animal models based on recapitulating risk factors for spastic CP is the absence of symptoms that characterize spastic CP.

Animal Models of Spastic Cerebral Palsy Symptoms

As noted previously, impairments in movement or posture are essential for diagnosis of CP, with spasticity being the most common symptom (89, 152, 169, 181). Accordingly, these symptoms should also be critical for modeling this condition successfully in animals.

The spa mouse, originally described in 1961, has symptoms inducible with handling or startle consisting of tip-toe walking, arched back, and difficulty righting (32, 33). In addition, these mice also are smaller than their littermates at maturity and display muscle rigidity, myoclonic jerks, exaggerated startle response, and spasticity (12, 23, 84, 138), a phenotype strikingly similar to symptoms in humans with spastic CP (FIGURE 1). Although not exhaustively chronicled, the phenotypical abnormalities in the spa mice predominantly affect the trunk and limb muscles, particularly the hind limb (35, 212). The spa mouse is not a genetic model of spastic CP but rather a phenotypic model of spastic CP via motor neuron disinhibition. Moreover, although not commonly recognized as an animal model for spastic CP, these mice have been used to model symptoms and interventions in spastic CP (35, 205, 212). In 1994, the genetic basis for these motor symptoms was found to be due to a homozygous insertion of LINE-1 in the beta subunit of the glycine receptor gene, resulting in a splicing error of this subunit (103). We have recently shown that this abnormal neurotransmitter signaling in spa mice results in excessive motor neuron loss, with these mice having fewer motor neurons, which are smaller in size (23). Specifically, spa mice had over 50% reduction in the tibialis anterior motor neurons compared with wild-type mice, with motor neuron somal size reduced. This motor neuron loss disproportionately affected the larger motor neurons compared with wild-type mice (23). Corroborating evidence of reduced numbers of motor neurons with suggestion of increased motor unit size in adult humans with spastic CP has been reported in one small study (128). However, correlation of these findings to impact on physical function is not clear due to the clinical measures used and the large variability in physical abilities of the participants with CP (128).

This early developmental difference in the CNS neurotransmission may underlie the spastic CP phenotype, which notably manifests on clinical exam as muscles with velocity-dependent resistance to stretch, hyperreflexia, motor incoordination, and gait disorder. The embryonic and perinatal periods are a critical time for spinal cord development, specifically developmental motor neuron loss. By the beginning of the third trimester, the motor neuron pool reaches its maximum number of motor neurons (82, 161, 180). Thereafter, the number of motor neurons is reduced into the neonatal period, a highly regulated process dependent on many factors, including motor neuron excitability (101, 116, 127).

Excessive motor neuron excitability, also known as hyperexcitability, is manifest clinically as spasticity due to the motor neuron activity stimulating excessive muscle activation, such as with stretch and reflex testing, a predominant symptom of spastic CP. Motor neuron hyperexcitability is multifactorial (59) and can be seen with changes in motor neuron size (i.e., smaller motor neurons) (85–87), motor neuron morphology (i.e., dendritic branching) (189), and imbalances in neurotransmitter influences (39, 64, 67, 96, 100). Imbalances in neurotransmission include excessive glutamate neurotransmission (depolarizing) and/or loss of GABA or glycine neurotransmission (both hyperpolarizing), both of which cause hyperexcitability. These neurotransmitter perturbations can occur at differing time points in development and impact both the brain and the spinal cord. For example, CNS excitotoxicity via glutamate agonists and alteration of glutamate receptor expression may contribute to susceptibility to in utero brain injury during fetal development (39, 64, 100). CNS excitotoxicity could also occur postnatally via disruption of descending (i.e., brain or corticospinal tract injury) or local (interneuronal) glycine and GABA pathways, resulting in the disinhibition of motor neurons (67, 96). Regardless of their source, reduced inhibitory inputs contribute to motor neuron hyperexcitability, likely due to a net gain in depolarizing inputs and the continued presence of persistent inward calcium and sodium currents (60, 61, 83).

Animal models with developmental neurotransmitter imbalances exist. Spa, spasmodic, and oscillator mice all have glycine receptor mutations, which reduce glycinergic neurotransmission (23, 84, 138, 212). Various glycine receptor subunits are expressed at high levels in discrete brain stem nuclei and more diffusely in both the gray and white matter of the spinal cord, with lower level regions within the forebrain and cerebellum (66, 123, 126, 131, 211). Within the brain stem, motor and auditory neurons have abundant glycine receptors, linked to audition and motor control (123). Within the spinal cord, motor neurons and interneurons involved with motor control (30, 36, 199, 200) and dorsal horn and sensory interneurons (including those associated with nocioception) (75, 94, 124, 190) express glycine receptors (123). Outside the central nervous system, glycine receptors are expressed on spermatozoa, with chloride channel activity associated with the acrosome reaction and fertilization process (24, 174). Indeed, in cases of glycine receptor mutation, fertility is markedly reduced, including in spa mice (13, 175). Homozygous knockout mutant mice display a phenotype similar to cerebral palsy and survive into maturity (23, 84, 138, 212). Gephyrin mice have both glycine and GABA receptor mutations via a deletion of exon 1 of the gephyrin gene and deletion of the upstream sequences responsible for initiating transcription and translation of gephyrin (56). Gephyrin mice have early symptoms suggestive of a severe motor impairment (8, 56, 62, 63, 93, 201). However, they do not survive past P1 due to inability to suck, progressive hyperextended posturing, and progressive apnea (8, 56, 62, 63, 93, 201).

These animal models demonstrate that abnormalities in glycine and GABA synapses impact motor control and physical function. Notably, these animal models have germline mutations, meaning that glycine receptor mutations are present in all cell types, not just motor neurons. In this respect, they are imperfect models; however, since glycine is so highly expressed in the spinal cord and the brain stem, the behavioral abnormalities inherent in mice that have GABA receptor mutations are largely avoided (2, 3, 31, 208). Glycine and GABA are particularly critical neurotransmitters since they are excitatory (depolarizing) in embryonic and early postnatal development, with a transition to their inhibitory (hyperpolarizing) influence later in life (16, 68, 106). Regulation of chloride transport is responsible for this transition from depolarizing to hyperpolarizing actions. Driving this transition is the developmental downregulation of the sodium-potassium-chloride cotransporter (NKCC1) and simultaneous upregulation of the neuron-specific potassium-chloride cotransporter (KCC2) (93, 96, 121, 187), resulting in changes in the intracellular chloride concentration (FIGURE 3). For all motor neurons, extracellular chloride concentration is ~110 mM (16, 157, 179). However intracellular chloride concentration differs based on maturity, with immature neurons being ~30 mM vs. ~5 mM in mature neurons (179, 198). This change in intracellular chloride concentration results in changes in the equilibrium potential of chloride. Specifically, in the embryonic and early postnatal period, the equilibrium potential for chloride is ~20 mV, with maturity being ~75 mV (16, 157, 202). However, the resting membrane potential of the cell is ~65 mV, with the action potential threshold of ~45 mV in development and maturity. Thus, during early developmental periods, activation of GABA and glycine receptors results in chloride efflux, depolarizing the motor neuron (FIGURE 3) (16, 206). As postnatal maturity progresses, the equilibrium potential for chloride becomes more negative than the resting membrane potential, due to upregulation of the KCC2 channel, resulting in chloride contributing to hyperpolarization (inhibition) of the motor neuron when GABA and glycine receptors are activated (FIGURE 3) (16).

FIGURE 3.

Maturation of inhibitory neurotransmission is dependent on chloride channel expression

During development, the opening of ionotropic chloride channels due to GABAergic or glycinergic receptor activation (glycine receptor pentamers with gephyrin clustering shown in orange) leads to chloride efflux (blue arrows) from the neuron, depolarizing the membrane (black bilayer). This is due to high intracellular chloride concentrations (~30 mM) engendered by the abundant expression of NKCC1 chloride transporter (teal) and low expression of the KCC2 chloride transporter (green). The relative expression of these transporters reverses during later postnatal neuronal maturation, such that KCC2 becomes much more abundant and NKCC1 expression is negligible. As a result, internal chloride concentrations in mature neurons is much lower (~5 mM), such that GABAergic or glycinergic receptor activation leads to influx of chloride through ionotropic channels and the hyperpolarization of the membrane. Note that potassium (and sodium in NKCC1) is also co-transported through these channels (yellow arrows).

Animal models with KCC2 mutations have been developed, with rodents homozygous for the KCC2 mutation displaying abnormal in utero and neonatal motor impairments and the inability to breathe (93, 186). Similar to gephyrin mice, mice with KCC2 mutations do not survive (8, 56, 62, 63, 93, 186, 201). Due to such a severe motor phenotype and lethality, the KCC2 and gephyrin mice may be suitable for modeling CP via studying interventions to reduce the severity of the phenotype, as long as this promotes survival. Importantly, these types of studies may provide insight into potential treatments for individuals with spastic CP.

How Do We Target the Gaps in Cerebral Palsy Research?

As discussed, critical gaps in the concept of CP and in animal modeling of CP persist. To truly reflect CP, the model must have an abnormality of movement or posture, such as a change in locomotion, difficulty feeding, or impairment in righting (152, 169). Many of the current animal models are focused on the risk factors for the developmental brain injury due to the long-held dogma about the etiology of CP. Animal models have largely ignored the impaired posture and movement that is essential to the diagnosis of CP or the mechanisms driving the impaired movement and posture.

Animal models of CP must overcome the gaps in CP research, including the long-held misconceptions regarding an isolated brain injury as a driver of the symptoms. Animals for modeling CP must survive without spontaneous neurological recovery to facilitate study of the CNS mechanistic drivers of the phenotype over a lifespan. Animal models that focus on inflicting a developmental brain injury without regard for a postural or movement abnormality are not models of CP, since many infants at risk of CP also have brain abnormalities, although most do not develop CP (90, 203). White matter injury of the brain is a common feature of animal models, although it is neither predictive of CP development nor prognostic of movement disorder severity in CP (114). Brain lesions are neither sensitive nor specific to CP, since not all individuals with CP have abnormal brain imaging (10, 203). Furthermore, the vast majority of our clinical interventions to therapeutically modulate spastic CP symptoms act on the spinal cord, motor neuron, and/or motor unit (133, 163) (FIGURE 2). Thus animal models with a developmental difference of movement regardless of whether they display gross anatomical evidence of a brain injury should be embraced as models of CP and closely evaluated with regard to mechanistic causes of the movement disorder (23, 205, 212).

Recently, the NIH published a strategic plan for CP research (149). One recognized area of focus was the need for multiple animal models to address the complexities of the biology of CP (149). Also noted was that the focus is not only on the brain but also should include the impact of any potential developmental disruption, including genetic/genomic differences, on the neurodevelopment of the spinal cord and musculoskeletal system (149). However, these complex developmental disruptions cannot be studied if appropriate and varied animal models are not recognized, encouraged, available, and used. To study the basic science of CP, a variety of animal models with differing neurodevelopmental insults are needed. It is critical that all models display a phenotype of CP to identify whether the neurodevelopmental insult results in motor impairments. After all, the motor impairment is both essential for diagnosing CP (169) and a critical marker for tracking response to therapeutic interventions.

Conclusions

CP, although being defined as a developmental brain injury, is not clearly or consistently traceable to specific brain lesions. The single commonality of CP is the presence of a developmental disturbance resulting in motor or postural disabilities. Although many animal models of CP have focused on inflicting a brain injury, few have consistently reproduced the motor symptoms of CP, an essential component of the diagnosis. Animal models of CP must display disordered movement or posture. In addition, animal models of CP should be heterogeneous in etiology to include developmental disruptions that affect the entire CNS. Since scientific advancements are identifying new etiologies of CP, including genetic disorders and neurotransmitter imbalances, animal models of CP also need to advance. Animal models of CP also need to go beyond risk factors that are known or presumed to continue to advance the science of not only the etiology of CP but also CP prevention and cure. Critically, the field needs to embrace the original conception of CP as a “cerebro-spinal” disorder, both in our clinical work and in animal models.