Animal models of developmental motor disorders: parallels to human motor dysfunction in cerebral palsy

Abstract

Cerebral palsy (CP) is the most common motor disability in children. Much of the previous research on CP has focused on reducing the severity of brain injuries, whereas very few researchers have investigated the cause and amelioration of motor symptoms. This research focus has had an impact on the choice of animal models. Many of the commonly used animal models do not display a prominent CP-like motor phenotype. In general, rodent models show anatomically severe injuries in the central nervous system (CNS) in response to insults associated with CP, including hypoxia, ischemia, and neuroinflammation. Unfortunately, most rodent models do not display a prominent motor phenotype that includes the hallmarks of spasticity (muscle stiffness and hyperreflexia) and weakness. To study motor dysfunction related to developmental injuries, a larger animal model is needed, such as rabbit, pig, or nonhuman primate. In this work, we describe and compare various animal models of CP and their potential for translation to the human condition.

INTRODUCTION

Cerebral palsy (CP) is the most common form of motor disability of childhood, affecting more than 500,000 children and families in North America and approximately one in every 250 live births worldwide (Arneson et al. 2009; Bhasin et al. 2006; Johnson 2002; Oskoui et al. 2013; Paneth et al. 2006; Winter et al. 2002). Advances in perinatal care have improved survival rates for premature and very-low-birth weight infants who are at higher risk for developing CP. So, although the incidence of severe CP is declining among very-low-birth weight babies (Platt et al. 2007), better survival means the overall prevalence of mild CP and developmental coordination disorder is steady (McGowan and Vohr 2019; Oskoui et al. 2013). More than 85% of CP patients survive into adulthood (Blair and Stanley 2002), and life expectancy is related to the degree of impairments. Thus, improving outcomes is both a matter of improved quality of life and improving life expectancy for patients.

People with CP have a diversity of symptoms and severity; the common thread is impaired motor control arising during early development. CP is sometimes accompanied by other disorders, most often epilepsy (41% of individuals with CP also have epilepsy; see Christensen et al. 2014), deficits in vision (15%), and autism spectrum disorder (7%) (Rosenbaum et al. 2007). Developmental brain injuries and the motor dysfunction that accompanies them can also interfere with the ability of patients to interact with their environment, rest well, and acquire language during development. These impairments reduce both independence and quality of life. As children grow, secondary complications often appear, including abnormal postures and activity patterns that could contribute to altered bone structure, scoliosis, and hip subluxation (Chan and Miller 2014). Unfortunately, these conditions may also be painful; 75% of people with CP report experiencing pain (McDowell et al. 2017; Novak et al. 2012; Westbom et al. 2017), yet very little research is focused on the etiology of pain in CP. Associated conditions and secondary disorders are variable, but they often correlate with particular types of motor dysfunction and functional ability according to the Gross Motor Function Classification System (Condliffe 2016; Novak et al. 2012; Sewell et al. 2014).

Although there is evidence for a genetic contribution in some patients, neurological injuries during development, including hypoxia-ischemia, infection, and trauma, are thought to cause nearly all cases of CP. Thus, nearly all animal models of CP are based on various types of developmental injuries (summarized in Tables 1–3). Conditions associated with a diagnosis of CP include maternal infections and inflammatory response, prematurity, low birth weight, passage of meconium, neonatal seizures, and respiratory difficulties after birth (Nonaka et al. 1999). The most common perinatal brain injuries are hypoxic-ischemic encephalopathy and perinatal stroke (Nelson and Chang 2008), followed by traumatic injuries (Reddihough and Collins 2003). Because these conditions often include some amount of asphyxia, hypoxia is thought to contribute to CP (MacLennan 1999; Nelson 1988; Yudkin et al. 1995). Hypoxic ischemic injury as a direct cause for CP development is controversial because 1) it is hard to establish the exact correspondence between the injury and outcome, 2) delivery of oxygen after trauma did not impact outcome, and 3) a clear definition of hypoxic-ischemic encephalopathy is lacking (Fineschi et al. 2017). With this heterogeneity in causes, no single animal model is sufficient to examine all aspects of motor dysfunction in CP.

Table 1.

Hypoxia-ischemia-based models of developmental injury

Hypoxia-Ischemia	Model	Age	Phenotype	Extent of CNS Damage	Other Findings	Reference(s)
Rice-Vanucci
Chronic unilateral carotid occlusion followed by hypoxia	Mouse/rat	1–3 DPN	No gait abnormalities or changes in open field; impaired walking on rod, beam and inclined stair at 3 mo. Impaired ladder walking and decreased muscle size at 1 mo after longer hypoxia.	WMI: loss of oligodendrocytes (pre-OLs); Death of subplate neurons; cortex, thalamus and basal ganglia all affected by apoptosis and necrosis. Decreased striatal volume after longer hypoxia.	Less GluR1 and spine density in cortex, less dendritic branching in cortical neurons.	Back et al. 2002; Durán-Carabali et al. 2017; McQuillen et al. 2003; Ranasinghe et al. 2015; Sheldon et al. 1996; Stadlin et al. 2003
		5–10 DPN	Lower body weight; impaired performance in water maze, impaired righting, geotaxis and cliff aversion; lower respiratory rate.	WMI: loss of oligodendrocytes (pre-OLs); Cerebral infarct: subventricular layer and corpus callosum; hippocampal atrophy.	Elevated spinal and medullary 5HT; lower medullary glutamate and aspartate; apomorphine injection induced circling in adults.	Back et al. 2002; Bellot et al. 2014; Follett et al. 2000; McAuliffe et al. 2006; Rice et al. 1981; Sheldon et al. 1998; Skoff et al. 2001; Ten et al. 2003
Transient bilateral carotid ligation and subsequent hypoxia	Mouse/rat	5–7 DPN	Transient hindlimb paresis; Long-term motor deficits: impaired strength and rotarod performance.	WMI, PVL, activated glia.		Jelinski et al. 1999; Zaghloul et al. 2017
Chronic bilateral carotid occlusion with and without hypoxia	Rat	1–7 DPN	Lower body weight, impaired performance in open field and wire hanging; hyperactivity, altered gait, impaired righting, impaired geotaxis, and impaired motor coordinating ability in later life stages.	WMI in corpus callosum, subcortex, internal capsule; PVL, loss of pre-OLs; activated glia; elevated inflammatory mediators and increased brain water weight.		Cai and Rhodes 2001; Fan et al. 2005; Zhang et al. 2013
Chronic bilateral carotid occlusion	Dog	14 DPN	Lower body weight.	WMI; PVL.		Yoshioka et al. 1994
Transient bilateral carotid ligation followed by hypoxia	Pig	0–3 DPN	Significantly lower scores in blinded neurological exams that included reflex responses, motor activity and coordination.	PVL, activated glia in cerebral cortex, hippocampus and basal ganglia. Damage to hippocampus and basal ganglia more pronounced than cortex.		LeBlanc et al. 1991a, 1991b, 1991c
Instrumented Sheep
Bilateral carotid occlusion	Sheep	78–85% gestation	Tonic-clonic seizures; sequence of EEG changes predictive of injury.	WMI and cerebral cortex damage, hippocampus, striatum, and thalamus also injured.	FGF-2 increased after injuries; Disruption of BBB (peak permeability 4 h postinjury).	Chen et al. 2012; Petersson et al. 2002; Williams et al. 1990, 1992
Global HI
Occlusion of uterine arteries or hypoxia	Rat	86% gestation	Decrease in stride length and less movement in open field.	Activated glia, white matter injury, cell death in cortex.	KCC2 and GABAA receptor upregulation in cortex and hippocampus impaired by HI; decreased inhibitory synaptic currents in hippocampus.	Jantzie et al. 2013, 2014, 2015a, 2015b, 2018; Robinson 2005
	Rabbit	67–80% gestation	Variable. Hypertonia of limbs at rest; muscle shortening, hyper-reflexia, delayed or absent righting reflex, lower body weight, increased frequency of stillbirths.	Injury to basal ganglia and thalamus; WM loss in internal capsule and corona radiata; cortex (layers 4 and 6) and spinal cord (intermediate).	HI at E25 increases later WMI; elevated spinal 5HT, fewer lumbar motoneurons; motor deficits appear before myelination.	Buser et al. 2010; Derrick et al. 2004; Drobyshevsky and Quinlan 2017; Drobyshevsky et al. 2014a, 2015; Synowiec et al. 2019)
	Pig	1 DPN	Occasional seizures.	Neural death in basal ganglia.	Decreased EAAT2 and EAAT1.	Martin et al. 1997
	Sheep	80–88% gestation	Convulsions observed in some of the most severe animals.	Variable WMI; PVL; lesions in cortex and thalamus, reactive glia.	Cytokines and free radicals released.	Ikeda et al. 1998a, 1999; Marumo et al. 2001; Ohyu et al. 1999
	Nonhuman primate	96% gestation — term	Variable. Hyper/hypotonia; deficits in locomotion, suck, swallow; lack of righting, overactive startle response; seizures.	WMI, injury to basal ganglia, thalamus and brainstem nuclei, cortex involvement only in severe cases.		Myers 1975; Ranck and Windle 1959

CNS, central nervous system; DPN, days postnatal; EAAT1, excitatory amino acid transporter 1; EAAT2, excitatory amino acid transporter 2; GluR1, glutamate receptor 1; HI, hypoxia-ischemia; 5HT, serotonin; KCC2, postassium chloride cotransporter 2; OL, oligodendrocytes; WMI, white matter injury.

Table 3.

Other models of developmental injury

Other Impairments	Model	Age	Phenotype	Extent of CNS Damage	Other Findings	Reference(s)
Cortical silencing
Chronic GABA agonist delivery to motor or sensorimotor cortex	Rat	7–28 DPN unilateral	No overt motor deficits	Fewer PV+ neurons in inhibited cortex; no loss in projecting neurons; compensation of contralateral cortex	Loss of PV+ spinal interneurons; no expansion in afferents; bilateral CST innervation	Clowry et al. 2004
	Cat	21–84 DPN unilateral, bilateral	Paw misplacement on horizontal ladder (end point control deficit); no impairment in limb trajectory; end point errors in reaching	Decreased area of medullary pyramid; cortical thinning and gliosis; increased neuron density in cortical layer 2 sub lamina; altered topography of corticospinal projections	Loss of spinal interneurons (ChAT+); increased vascularization of silenced cortex	Friel and Martin 2005; Friel et al. 2007, 2012; Li and Martin 2000; Martin and Ghez 1999; Martin et al. 2000, 2007
Injury
Unilateral damage to motor cortex	Rat	7 DPN	No behavioral deficits	Complete unilateral excision of motor cortex with no damage to underlying white matter	Preservation of more afferent synapses in ventral horn, loss of spinal interneurons (PV+)	Gibson et al. 2000
	Pig	“Young” 30–40 kg	Weakness in hindlimb and development of spasticity	Unilateral aspiration of cortical and subcortical structures near lateral sulcus, including white matter and deep nuclei	Increased latency of -motor evoked potentials in hindlimb of affected side; abnormal spreading of muscle activity bilaterally	Andreani and Guma 2008
Motor/sensorimotor restriction
MR, BoTox; SR, physical immobilization	Rat	2–28 DPN; SR 16 h/day	Severe impairments in rotarod, ladder and suspended bar tests; Slow to fast muscle fiber switch	Smaller size of spinal motoneurons	Smaller axon diameter in sciatic; increase sarcomere length and decreased density	Stigger et al. 2011a, 2011b
	Cat	MR from 21–91 DPN	Loss of supination component of grasping in restricted forepaw	Impaired development of CST projections in spinal cord		Martin et al. 2004
Toxins
Subcortical injection of glutamate agonist(s)	Mouse/rat	0–15 DPN	Variable; seizures, spasms, apnea, mortality	WMI peak susceptibility at 5–7 DPN; damage to cerebral cortex peaks later: lesions and necrosis of gray matter, loss of neurons in layers II–VI, striatum, hippocampus 5–11 DPN	AMPA- and NMDA-mediated loss of pre-OLs; peak susceptibility coincides with GluR4 and NMDAR expression in Pre-OLs	Follett et al. 2000; Gibson and Clowry 2003; Marret et al. 1995; Tahraoui et al. 2001
Subcortical injection of mitochondrial toxin	Rat	7 DPN	No change in gross motor behavior or use of forelimb	Loss of cortical axons; apoptosis in cortical layer V, fewer PV+ and CB+ neurons in motor cortex but no substantial loss of gray matter		Gibson and Clowry 2003
Carbon monoxide inhalation	Nonhuman primate	94–97% gestation	Moderate injury produced hypotonia; severe injury produced spasticity, hypertonia (extensor rigidity, clenching of fists), nystagmus, and death	Moderate injuries affected basal ganglia, some cortical necrosis; severe damage included increased intracranial pressure, hemorrhagic necrosis in basal ganglia, thalamus and cerebral cortex		Ginsberg and Myers 1974, 1976
Genetic models
Glycine receptor mutation	Mouse	Assessed at 28–63 DPN	Abnormal gait, hypertonia, exaggerated startle response, myoclonic jerks at 1 mo, more variable symptoms at 2 mo		Loss of large spinal motoneurons	Brandenburg et al. 2018

CB, calbindin; CNS, central nervous system; CST, corticospinal tracts; DPN, days postnatal; GluR4, glutamate receptor 4; OLs, oligodendrocytes; PV, parvalbumin.

ANIMAL MODELS OF CEREBRAL PALSY

Various animal models, including rodents, rabbits, cats, primates, and sheep, have been used to investigate CP (see Tables 1–3). Comparisons between animals and humans can be complicated by many factors, including differences in the pattern of development of the brain, spinal cord, and motor functions, as well as differences in specific neural connectivity and plasticity (Brandenburg et al. 2019; Clowry et al. 2014; Martin 2005). These topics will be addressed in the following sections of this review. In many animals (particularly rodents), there is a large disparity between the severity of injury to the central nervous system (CNS) and the motor phenotype; rodents can sustain quite severe brain injuries yet display mild or nonexistent motor deficits. Animal models that sustain severe brain injuries without concomitant severe motor dysfunction can still serve as useful models to study various aspects of brain injury and amelioration thereof. However, to examine the etiology of motor dysfunction, including spasticity and hyperreflexia, one must use an animal model in which this is a measurable outcome. In a surprising number of studies in which new animal models of developmental injury are introduced, neurological and behavioral phenotypes of the animal model are completely absent. A recent review by Rumajogee et al. 2016 summarized rodent models of CP; of the 13 models, only two included a description of motor function (see also Fan et al. 2005; Rice et al. 1981). The phenotypes in these rodents ranged from no deficits in behavior and reflexes (despite severe injuries) to prolongation of righting response and hyperactivity (Fan et al. 2005; Rice et al. 1981). Clearly, the focus of these studies was not on motor deficits, but the lack of information greatly complicates the choice of the most appropriate models for research.

Perhaps the most common animal models of CP are based on hypoxia-ischemia (HI), which is summarized in Table 1. Many rodent studies use variations of the Rice-Vannucci method of carotid artery occlusion (producing HI of the brain specifically), followed by global hypoxia (during which the animal is placed in a chamber with low oxygen levels) (Rice et al. 1981). A similar approach is used in instrumented sheep; during gestation, fetal sheep are partially extracted from the dam and surgically implanted with sensors and carotid ligation. After surgery the fetus is returned to the uterus to continue gestation for ∼1 wk, during which carotid occlusion is performed and fetal responses are monitored, resulting in severe CNS injuries and seizure-like activity (Chen et al. 2012; Petersson et al. 2002; Williams et al. 1990, 1992), although a description of motor deficits in sheep is completely absent in publications. Perhaps this is because termination of the experiments usually occurs upon birth, so deficits cannot be assessed. Targeting the injuries to the brain is useful for modeling injuries like neonatal stroke, but it is possible that global hypoxia ischemia or systemic inflammation may have significant effects outside of the brain that contribute to the etiopathology of CP, which is considered later in this review in sections on the involvement of the spinal cord and motor dysfunction in cerebral palsy. Animal models of global HI can be performed by either occlusion of circulation to the uterus during gestation or placing the pregnant dam in a low-oxygen environment. Reperfusion after occlusion of circulation adds another component to the injuries that are sustained in those models that would be absent when animals are simply placed in a hypoxic environment (Yang and Betz 1994; Pan et al. 2007).

Another useful way of modeling CP in animals is to make use of lipopolysaccharide (LPS) or other agents that produce an inflammatory response during development, as summarized in Table 2. Maternal infections are linked to a number of conditions diagnosed after birth, including CP, autism, and schizophrenia (Cordeiro et al. 2015). LPS induces an inflammatory response without subjecting animals to a live bacterial infection and is delivered either prenatally (via intrauterine injection) or postnatally (via IP injection). LPS starts a combination of cell-signaling cascades involved in inflammation (including interleukins, TNFα, and others) that may generate placental insufficiency and result in severe brain damage and, in many cases, death of the offspring (Carpentier et al. 2011; Cordeiro et al. 2015; Ginsberg et al. 2017). However, the motor phenotype produced by LPS alone appears milder (compare studies using nonhuman primates with HI vs. inflammation in Tables 1 and 2). Combining HI and inflammation (LPS) in the rat appears to produce more reliable motor deficits, including altered gait and coordination (summarized in Table 2) (Jantzie et al. 2014, 2018; Stigger et al. 2011a).

Table 2.

Inflammation based models of developmental injury

Inflammation	Model	Age	Phenotype	Extent of CNS Damage	Other Findings	Reference(s)
HI and LPS
IP injection in postnatal subjects / intrauterine injection in prenatal Global HI	Rat	∼80% gestation and 7 DPN	More movement in open field, altered gait, decreased paw pressure in stance, decreased coordination and consistency	Infarct of cerebral cortex, hippocampus, striatum and thalamus, activated glia, white matter injury	Pretreatment with LPS caused significantly larger HI-induced cerebral infarct than HI alone; CD14 and TLR4 expression changed	Eklind et al. 2005; Jantzie et al. 2014, 2015b, 2018; Maxwell et al. 2015; Stigger et al. 2011a; Yellowhair et al. 2018
	Sheep	53–84% gestation	No change in birth weight	White matter lesions; decreased bo. Oligodendrocytes	No exacerbation of injury from combination of inflammation and HI;	Nitsos et al. 2014
HI, LPS, and sensorimotor restriction (SR)
LPS: 80% gestation; HI: 0 DPN; SR: 16 h/day, 2–28 DPN	Rat	80% gestation: 28 DPN	Severe deficits in rotarod, ladder, and beam walk	Not assessed (previous studies on SR alone indicated disorganization of sensory and motor cortex)	soleus muscle (but not TA) smaller cross-sectional area with increased sarcomere length; slow to fast fiber transition in both muscles	Stigger et al. 2011a
LPS/live bacteria
IP injection in postnatal subjects/intrauterine injection prenatally	Rat	70% gestation to 6 DPN	Variable. Loss of body weight, increased frequency of stillbirths, impaired rotarod	Less damage in brains compared with other species; WMI, activated glia, damage to cortex, hippocampus, internal capsule	TNFα and IL-1β increased in brain; karyorrhexis of nuclei in white matter; neonatal challenge with LPS accelerated neuromotor development	Cai et al. 2000; Gilles et al. 1977; Poggi et al. 2005; Roberson et al. 2006; Stigger et al. 2011a
	Rabbit	70% gestation to 14 DPN	Lower birthweight, increased frequency of stillbirths, hypertonia, impaired posture, locomotion, feeding, and righting	WMI, PVL, subarachnoid hemorrhage, necrosis of WM and cortex, decreased density of Purkinje cells in cerebellum, inflammation, and microglial activation in PVWM	Decreased blood flow to cortex and WM; altered plasma and brain amino acids, decreased cortical 5HT (spinal not checked); altered tryptophan metabolism; elevated cytokines, iNOS	Ando et al. 1988; Debillon et al. 2000; Gilles et al. 1977; Saadani-Makki et al. 2008; Williams et al. 2017a; Yoon et al. 1997; Zhang et al. 2018a, 2018b
	Cat	2–20 DPN	Loss of body weight but no obvious neurological deficits	PVL, astrogliosis, necrosis, absence of cortical involvement		Gilles et al. 1976, 1977
	Dog	1–10 DPN	Hypoactivity, mortality	WMI in corpus collosum, forebrain; inflammation	Metabolic acidosis, hypoglycemia.	Young et al. 1983
	Sheep	65–83% gestation	No difference in birth weight	WMI, PVL, diffuse subcortical damage, 30% reduction of CST area, activated glia, decreased no. of oligodendrocytes	Elevated plasma IL6, IL8; inflammatory markers IBA-1, COX2 and GFAP; no acidosis.	Duncan et al. 2002; Gussenhoven et al. 2018; Paton et al. 2018; Yawno et al. 2013
	Nonhuman primate	74% gestation: 2 DPN	Low dose: no change in birth weight; higher dose: loss of body weight, weakness but no neurological deficit	Higher dose (prenatal): WMI, activated glia, necrosis; low dose (postnatal): increased cortical WM volume at 1 yr	Low dose: altered reaction to acoustic startle at 1 yr of age; emotionality.	Gilles et al. 1977; Willette et al. 2011

CNS, central nervous system; COX2, cyclooxygenase 2; CST, corticospinal tracts; DPN, days postnatal; 5HT, serotonin; IBA-1, ionized calcium-binding adapter molecule 1; IP, intraperitoneal; GFAP, glial fibrillary acidic protein; HI, hypoxia-ischemia; iNOS, inducible nitric oxide synthase; PVL, periventricular leukomalacia; PVWM, periventricular white matter; TLR4, Toll-like receptor 4; WM, white matter; WMI, white matter injury.

Models of CP that are not based on HI or inflammation are summarized in Table 3. Most of these models include application of toxins or surgical injury to the brain, with some notable exceptions. Sensorimotor restriction in rats during postnatal development (in which hindlimbs are immobilized for 16 h/day) produces the most severe deficits with the most clear parallels to clinical CP observed in rodent models (Stigger et al. 2011a). Another noninjury model is a genetic mouse model with a mutation in glycine receptors [Spa (Glrb^−/−) mouse] that results in hyperekplexia and spasticity. Although this mouse shows spasticity that parallels some aspects of clinical CP, the mutation itself is not linked to any known cases of CP (Brandenburg et al. 2018).

Although mice are a natural choice for exploring links between genetics and CP, they are a poor model of motor deficits. In contrast to rodents, rabbits (members of the Leporidae family) after either HI injury or LPS (Tables 1 and 2) show prominent motor deficits, including muscle stiffness, deficits in locomotion and feeding, and hyperreflexia (Derrick et al. 2004; Saadani-Makki et al. 2008; Synowiec et al. 2019). Only the most severely injured rabbit kits require hand feeding and show early mortality. Thus, the postnatal maturation of rabbits after developmental injury is an extremely useful model for studying motor deficits over time. Both children with CP and HI rabbits have hypertonia and hyperreflexia; the H reflex is increased in size with a decrease in its rate-dependent depression (Achache et al. 2010; Futagi and Abe 1985; Synowiec et al. 2019; Tekgül et al. 2013). Another animal model that appears to be a good model of human CP is the pig. After HI or cortical injury, their phenotype includes lower scores on neurological exams, seizures, and spasticity (see Tables 1 and 3) (LeBlanc et al. 1991a, 1991c; Martin et al. 1997). Larger animal models (namely sheep) have been indispensable for finding biomarkers of fetal distress during various insults, including hypoxia and inflammation (Ikeda et al. 1998b, 1999; Marumo et al. 2001; Nitsos et al. 2014; Ohyu et al. 1999; Petersson et al. 2002; Williams et al. 1992). Nonhuman primates have been studied postnatally after developmental insults (Gilles et al. 1977; Ginsberg and Myers 1974, 1976; Myers 1975; Ranck and Windle 1959; Willette et al. 2011) and show motor dysfunction, including hypertonia (clenching of fists and extensor rigidity), seizures, and exaggerated startle response (Ginsberg and Myers 1974, 1976; Myers 1975; Ranck and Windle 1959). Although primates are the best animal model of motor deficits related to CP, their use poses other problems, including ethics and cost. Rabbits seem best suited to bridge the gap between smaller and larger animals. Rabbits have the advantages of larger animals, including parallel motor deficits to humans, wide availability, simple housing needs, a short reproductive cycle, large litters, and a similar development pattern to humans (Derrick et al. 2004; van der Merwe et al. 2019; Weiss and Disterhoft 2015). Thus, rabbit models of CP are emerging as the most practical for studying motor deficits.

MOTOR DYSFUNCTION IN CEREBRAL PALSY

Given that most motor deficits in CP develop after birth, it is critical to track the deviation from motor milestone in children who are at risk of CP along with the animal models that mimic them. A common motor deficit in CP is spasticity, which has been characterized by hypertonia, hyperreflexia, clonus, spasms, co-contraction, and improper control of voluntary muscle activity (Carr et al. 1992; Dietz 1999; O’Sullivan et al. 1998; Poon and Hui-Chan 2009). Increased resistance to imposed stretch, as an indicator of spasticity, progressively increases in children with CP ≤4–6 yr of age, with ∼50% of children exhibiting marked tone in ankle plantarflexors (Hägglund and Wagner 2008). Contractures also start to develop near 2 yr of age (Willerslev-Olsen et al. 2018), whereby joint range of motion is reduced and joint position is altered. Children with CP have fixed, flexed postures of upper and lower extremities that often require surgical correction (Horstmann et al. 2009; Ozkan and Tuncer 2012; Shamsoddini et al. 2014). It is unclear the extent to which muscle stiffness and contractures are a result of excessive, involuntary but neutrally driven muscle activity or the result of increased musculotendinous stiffness from mechanical changes in muscles (shortened muscles, changes in the structure of muscle fibers, connective tissue, and the extracellular matrix induced by muscle atrophy) (Lieber and Fridén 2019; Mathewson and Lieber 2015; Willerslev-Olsen et al. 2013).

Perhaps there is both a feedforward and feedback relationship between development of the musculature, joint stiffness, and neural drive. Impaired muscle growth is present in humans, the rabbit HI model of CP, and two mutant mouse models (the spa mouse model of spasticity and the Celsr3/Foxg1 mouse model of cortical impairment) (Barber and Boyd 2016; Brandenburg et al. 2018; Han et al. 2013; Synowiec et al. 2019; Verschuren et al. 2018), contributing to altered loading and adaptation of the extracellular matrix (reviewed in Gough and Shortland 2012; Lieber and Fridén 2019). Muscle atrophy appears to play a large role in the development of contractures, since it precedes the development of passive joint stiffness and excessive stretch reflexes in the medial gastrocnemius (Willerslev-Olsen et al. 2018). In mice, rabbits, and humans, muscle atrophy appears along with losses in numbers of motoneurons (Brandenburg et al. 2018; Drobyshevsky and Quinlan 2017; Han et al. 2013; Marciniak et al. 2015). However, at least one model of rodent hypoxia involves motoneuron loss without loss in muscle mass (Watzlawik et al. 2015), suggesting muscle loss could be secondary. Because muscle atrophy is a large problem causing motor dysfunction in patients, finding an animal model that displays this phenotype is important for gaining insights into the mechanism underlying contractures in children with CP. A recent study using the rabbit HI model of CP found muscle shortening, atrophy, and increased sarcomere length in the tibialis anterior of rabbits after prenatal HI (Synowiec et al. 2019). The flexed postures found in these rabbits could be the result of shortened musculotendinous tissue in response to muscle atrophy. A mechanical contribution is also suggested by the fact that stillborn kits also have flexed postures (Derrick et al. 2004).

Although it is not clear whether feedforward changes in the periphery stem from motoneuron loss in the CNS, feedback from the periphery certainly alters central drive to motoneurons. Increases in musculotendinous stiffness result in less sensory activation during tasks like walking so that activation of spinal networks must rely more on a weakened and disordered descending input (Condliffe et al. 2019; Frisk et al. 2017). The extent to which hypertonia and hyperreflexia are driven by decreased function of corticospinal tracts (CST), changes in spinal motoneurons, or alterations in the muscle due to injury has never been directly tested. Injuries occurring during critical developmental time periods for the brain, brainstem, spinal cord, and musculature could disrupt the formation of specific circuits or disrupt the excitatory-inhibitory balance within circuits. Strengthening weakened pathways and promoting circuit development may be very difficult to overcome at later time points, although intensive treadmill training in children 5–14 yr of age with CP can produce plastic changes in the corticospinal tract that accompany improvements in gait (Hurd et al. 2017; Willerslev-Olsen et al. 2015). Larger rewiring may be possible if intensive training is initiated before 2 yr of age, when myelination of the leg CST is still immature (Hurd et al. 2017).

INVOLVEMENT OF THE SPINAL CORD

Damage to the motor cortex, basal ganglia, thalamus, cerebellum, and CST is known to cause motor problems in CP, but the contribution of spinal circuits to dysfunction has been woefully understudied. Spinal motoneurons are the final common pathway in the CNS, and thus not only their activity but their proper development may be disrupted by several converging pathways. Two lines of evidence suggest that spinal circuits may play a larger role in CP than has previously been thought; 1) animal models show that maturation of spinal circuits is altered by injury to descending tracts or silencing of the motor cortex (Jiang et al. 2016, 2018; Smith et al. 2017), and 2) in both humans and animal models, global injuries that are sufficient to cause brain damage also cause cell death in the spinal cord (Clowry 2007; Drobyshevsky and Quinlan 2017; Natsume et al. 1995; Sladky and Rorke 1986). Postmortem tissue from asphyxiated infants and rabbit kits affected by HI showed cell death in the spinal cord and decreased numbers of lumbar motoneurons (Drobyshevsky and Quinlan 2017; Sladky and Rorke 1986) along with injuries to deep nuclear structures of the basal ganglia, thalamus, and brain stem (Natsume et al. 1995). In short, perinatal HI causes injury throughout the CNS, including the spinal cord. Whereas cortical and subcortical injuries have been studied a great deal, how injuries to the spinal cord contribute to motor dysfunction has largely been overlooked.

Cortico-, rubro-, vestibulo-, and reticulospinal projections to the spinal cord are mostly glutamatergic (Du Beau et al. 2012) but dampen spinal excitability by activating spinal inhibitory interneurons (Jankowska et al. 1976). Thus, damaging these tracts can reduce inhibitory tone in the spinal cord and contribute to hypertonia, although this has never been directly tested (Deon and Gaebler-Spira 2010; Sanger 2003; Volpe et al. 2017). When the late-developing corticospinal tract is disrupted in its growth, there is an aberrant strengthening of the earlier-arriving pathways that have likely established synaptic contacts in the spinal cord before injuries occur. For example, reticulospinal projections from the brainstem are the first to arrive in the spinal cord, followed by serotonergic fibers from the raphe nucleus (9 wk postconception in humans) and by rubrospinal tracts (Kudo et al. 1993; Perreault and Glover 2013; Sundström et al. 1993). Corticospinal tracts are the last to arrive, at ∼24 wk postconception in humans, although their refinement and maturation continues for years (Clowry 2007; Eyre et al. 2000). When the late-arriving CST is impaired by injury, the relative importance of other inputs in driving spinal activity is increased, including inputs from other (earlier arriving) descending tracts (Cooper et al. 2017) and afferent fibers (Jiang et al. 2016; Tan et al. 2012). Other altered pathways in human participants with CP include more excitable cortico-reticulospinal pathways (Smith and Gorassini 2018). Thus, it appears that perinatal injuries that cause CP reduce the influence of the late-arriving CST in the spinal cord while increasing the relative importance of the early-arriving reticulospinal and rubrospinal inputs (Williams et al. 2017b).

Perturbations in the normal development of descending tracts from the brain likely cause downstream disturbances in the development of spinal neurons. These disturbances include the maintenance of excessive synaptic drive from sensory afferents, which may cause hyperreflexia through overstimulation of motoneurons. This could occur through two different processes; 1) since exuberant afferent synaptic contacts are pruned upon arrival of the CST in the spinal cord, injuries that impair CST growth result in the maintenance of these synapses, and 2) after lesions of the CST, afferent sprouting has been demonstrated (Tan et al. 2012) due to competition between spinal afferents and CST connections (Jiang et al. 2016). In addition, fewer spinal motoneurons are present in people with CP and animal models of HI injury and genetic perturbation of cortical projecting neurons (Drobyshevsky and Quinlan 2017; Han et al. 2013; Marciniak et al. 2015). Motoneuron loss does not appear to involve cell death directly but rather altered developmental processes that could include the pruning stage in late gestation during which motoneurons are normally reduced in numbers.

Alterations in spinal reflexes have been observed in human participants with CP, generally demonstrating a decrease in inhibitory reflexes and an increase in excitatory reflexes to produce a more excitable spinal cord compared with age-matched controls. The lower suppression of H reflexes by other afferents or from repeated activation of the H reflex at high frequency suggests that presynaptic inhibition of Ia afferents is reduced in CP (Achache et al. 2010; Synowiec et al. 2019). The excitability of reciprocal inhibitory pathways is unchanged in the resting muscle (Achache et al. 2010; Leonard et al. 2006), but during voluntary contractions it is enhanced in ankle flexors (Berbrayer and Ashby 1990; Brouwer and Smits 1996) and reduced in ankle plantar flexors (Leonard et al. 2006). In the upper limb during development, there is a lack of pruning of heteronymous reflexes in the triceps brachii in response to taps applied to the biceps brachii tendon (O’Sullivan et al. 1998), similar to that seen for the persistence of radiating brainstem reflexes (reviewed in (Smith and Gorassini 2018). The inhibitory component of the cutaneomuscular reflex occurring at a short (spinal) latency is also reduced (Evans et al. 1991; Gibbs et al. 1999), especially in participants with poor walking abilities (Condliffe et al. 2016). In terms of excitatory spinal reflexes, pathways mediating group II and propriospinal pathways (Achache et al. 2010) and the spinal excitatory component of the cutaneomuscular reflex (Gibbs et al. 1999) are enhanced in leg muscles, with mixed effects of the latter in the upper limb (Evans et al. 1991). Thus, there are several changes in the excitability of spinal reflex pathways in CP that may be related to ongoing motor function. Animal models that display a prominent motor phenotype with hyperexcitable and radiating reflexes should be utilized to investigate these issues.

CRITICAL DEVELOPMENTAL PERIODS

Certain areas of the CNS are more susceptible to damage at certain times in development in both humans and animal models. In humans, insults during a critical period between 26 and 34 wk of gestation most often give rise to CP (Back et al. 2001; Volpe et al. 2017). In full-term infants, the cerebral cortex and underlying subcortical and periventricular white matter are most vulnerable, perhaps due to the vascular supply (Redline and O’Riordan 2000; Redline and Ravishankar 2018; Stavsky et al. 2017) or fragility of oligodendrocytes during myelination (Graham et al. 2016). Critical periods seem to correspond to periods of myelination, development of the cortical subplate, and cerebral white matter and invasion of the spinal cord by corticospinal tracts. The cortical plate is critical for proper cortical development and shows susceptibility to damage from insults that also cause CP in infants and in animal models (Hadders-Algra 2018; Hoerder-Suabedissen and Molnár 2015; McClendon et al. 2017; Pogledic et al. 2014; Sheikh et al. 2019). In humans, most major white matter tracts are not significantly myelinated until early childhood (Dietrich et al. 1988; Nakagawa et al. 1998), and continuing myelination contributes to the enhancement in total white matter volume through the second decade of life (Giedd et al. 1999). Interestingly, motor deficits arise before myelination of the CNS in some infants and in the rabbit hypoxia-ischemia (HI) model, and in the rabbit motor deficits are associated with loss of conductivity in unmyelinated fibers and fewer axons at the level of the internal capsule (Drobyshevsky et al. 2014b). In children, global MRI scores appear to be more predictive of a CP outcome than white matter scores (George et al. 2018), and one meta-analysis found that very early predictions of CP (>5 mo corrected age) were most accurate when based on neurological exams, including Prechtl Qualitative Assessment of General Movements (Novak et al. 2017).

In humans and most mammals, including nonhuman primates, the brain develops mainly in utero, with the areas of active proliferation of oligodendrocytes (OLs) in the deep periventricular white matter appearing most vulnerable to injuries. Like humans, rodents, rabbits, and sheep show greater damage to white matter when hypoxic-ischemic insults are timed to coincide with the developmental peak in pre-OLs (Back et al. 2006; Buser et al. 2010). In rabbits, the appearance of pre-OLs at ∼80% gestation coincided with a developmental window of enhanced white but not gray matter susceptibility to hypoxia (Buser et al. 2010). This critical period is present earlier in sheep; the fetus shows more white matter injury at midgestation than when injuries occurred in a more mature fetus (Penning et al. 1994). It is not clear whether the severity of motor deficits increased in rabbits and sheep when injuries were timed to coincide with vulnerability of OLs, only that white matter injuries were maximized. In contrast to larger mammals, rodents show postnatal brain development based on several markers of brain development, including myelin formation (Dobbing 1974; Dobbing and Sands 1973, 1979; Dobbing et al. 1971; Smart et al. 1973). Thus rodent models of CP use postnatal injuries or cortical silencing techniques most often at postnatal day 7, a time that is thought to be roughly equivalent to the third-trimester in humans (Back et al. 2002; Bellot et al. 2014; Edwards et al. 2017; Follett et al. 2000; McAuliffe et al. 2006; Rice et al. 1981; Sheldon et al. 1998; Skoff et al. 2001; Ten et al. 2003). Interestingly, whereas brain development is comparable between rodents at a 1-wk postnatal and a third-trimester human, spinal development and motor abilities are very different. Mice are capable of locomotion and weight-bearing at just over 1 wk of age (Fox 1965) and rats at just over 2 wk (Gramsbergen 1998), so based on development of coordinated motor skills, rodents at 1–2 wk postnatal are equivalent to 1 or 2 yr of age in humans. Rodents achieve motor milestones rather quickly after birth, considering that their CNS development is delayed in comparison with other mammals. Growth of the CST to spinal targets occurs prenatally in humans, primates, and cats, whereas it occurs postnatally in rodents (Hsu et al. 2005, 2006; Martin 2005). Development of the rabbit CST is not well studied; myelination of the CST begins at P5, whereas functional maturation is not complete until 2 mo of age (Franson and Hildebrand 1975). In rodents, the CST is fully mature at ∼30 days of age (Hsu et al. 2006; Reh and Kalil 1982), in nonhuman primates full maturity of the CST takes 36 mo (Olivier et al. 1997), and in humans it occurs sometime between 10 and 20 yr of age (Welniarz et al. 2017; Yeo et al. 2014). Rodent motoneurons form synaptic contacts with developing muscles later in gestation than humans (55–80% gestation in rodents vs. ∼25% gestation in humans) (Diamond and Miledi 1962; Ezure and Tanaka 1996; Fidziańska 1980; Juntunen and Teräväinen 1972; Perreault and Glover 2013). Polyneuronal innervation of muscle fibers is lost relatively late in both rodents and rabbits compared with humans (perinatally in humans compared with the 2nd postnatal wk in rabbits and rodents) (Gramsbergen et al. 1997; Soha et al. 1987), and motoneurons display mature firing patterns later in development (postnatal days 0–5 in rodents vs. ∼40% gestation in humans) (Tadros et al. 2015; Vinay et al. 2000). Thus rodents have the most condensed developmental period for CST and spinal motoneuron growth compared with humans, primates, rabbits, and cats, with spinal invasion of the CST and motoneuron maturation starting later than other species and being completed earlier in postnatal development. Timing injuries in animal models to coincide with the development of descending projections and sensory circuits in the spinal cord may be beneficial for studying motor deficits.

NEUROMODULATION

In addition to the well-documented effects of neuromodulators on synaptic strength and excitability, monoamines, including serotonin (5HT) and dopamine, have roles in neural development as well. Children with CP who had higher endogenous levels of dopamine neurotransmission had better outcomes after therapy (Diaz Heijtz et al. 2018), whereas people with CP who were given levodopa showed overall improvements in motor functioning (Brunstrom et al. 2000; Rosenthal et al. 1972). A host of studies on brain injuries in adults also suggest that dopamine could be therapeutic (Bradley and Damiano 2019). In contrast, selective serotonin reuptake inhibitors (SSRIs) taken prenatally are associated with adverse developmental outcomes in babies, including lower motor scores (Johnson et al. 2016; Salisbury et al. 2016), although separating the effects of SSRIs and maternal depression remains controversial (Malm et al. 2016). If redundancy is an indicator of importance, 5HT is critical for development. Three different sources of 5HT ensure fetal exposure during development; maternal 5HT reaches the fetus, the developing fetal brainstem begins producing 5HT early in development, and the placenta itself provides a third source of 5HT (Bonnin and Levitt 2011). Neuromodulators affect maturation of synapses, in particular slowing maturation of inhibitory inputs (Branchereau et al. 2002). Previous studies have shown that during prenatal development, rather than being refined to specific synapses, 5HT is released into the cerebrospinal fluid (CSF) via volume transmission (Bunin and Wightman 1999). The concentration of 5HT in the CSF peaks during perinatal development; in rabbits this peak is reached around birth, whereas in humans it appears at ∼3 yr of age (Bunin and Wightman 1999; Drobyshevsky et al. 2015; Hentall et al. 2006; Lacković et al. 1988; Lewinsohn et al. 1980). The timing of this peak seems to correlate with the appearance of spasticity around birth in rabbits and the peak of spasticity at 4 yr of age in humans (Drobyshevsky et al. 2014a; Hägglund and Wagner 2008). Thus it is remarkable that 5HT is increased beyond the normal peak after HI injury (Bellot et al. 2014; Drobyshevsky et al. 2015). Paradoxically, after maternal infections, the metabolism of tryptophan is altered to favor the production of kynurenine instead of serotonin in both the human and rabbit placenta and the brain of rabbits (Manuelpillai et al. 2005; Williams et al. 2017a). After inflammation-induced injury, lower levels of serotonin have been found in the rabbit brain, particularly in the somatosensory cortex (levels in the spinal cord were not tested) (Williams et al. 2017a). The changes in 5HT could have many diverse effects in the brain and spinal cord, including altering persistent inward currents, which promote repetitive firing in motoneurons (and muscle stiffness) (D’Amico et al. 2013a, 2013b; Harvey et al. 2006a, 2006b; Hounsgaard et al. 1988; Hsiao et al. 1997, 1998; Li et al. 2007), altering activity of chloride transporters, and affecting dendrite outgrowth (Ben-Ari et al. 2012; Bos et al. 2013; Chang and Balice-Gordon 2000; Harvey et al. 2006b; Hsiao et al. 1997, 1998; Rörig and Sutor 1996; Wirth et al. 2017; Xu et al. 2004). Serotonergic antagonists are sufficient to reduce joint stiffness in rabbits affected by HI with flexed postures (Drobyshevsky et al. 2015). Future treatment strategies utilizing serotonergic and dopaminergic approaches should be more fully explored.

CONCLUSIONS

Animal models of disease and injury have advanced our ability to treat and cure a variety of conditions. Unfortunately, there has been very slow progress in finding better treatments for CP. More studies are needed on all aspects of this condition, using carefully chosen model systems: genetic risk factors, strategies to reduce brain injuries, and amelioration of motor symptoms.

GRANTS

This study received funding support from NINDS NS104436.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

K.A.Q. analyzed data; K.A.Q. interpreted results of experiments; C.F.C., M.G., and K.A.Q. drafted manuscript; C.F.C., M.G., and K.A.Q. edited and revised manuscript; C.F.C., M.G., and K.A.Q. approved final version of manuscript.