Nonhuman Primate Models of Neurodegenerative Disorders

Abstract

Alzheimer’s (AD), Huntington’s (HD), and Parkinson’s (PD) disease are age-related neurodegenerative disorders characterized by progressive neuronal cell death. Although each disease has particular pathologies and symptoms, accumulated evidence points to similar mechanisms of neurodegeneration, including inflammation, oxidative stress, and protein aggregation. A significant body of research is ongoing to understand how these pathways affect each other and what ultimately triggers the onset of the disease. Experiments in nonhuman primates (NHPs) account for only 5% of all research in animals. Yet the impact of NHP studies for clinical translation is much greater, especially for neurodegenerative disorders, as NHPs have a complex cognitive and motor functions and highly developed neuroanatomy. New NHP models are emerging to better understand pathology and improve the platform in which to test novel therapies. The goal of this report is to review NHP models of AD, HD, and PD in the context of the current understanding of these diseases and their contribution to the development of novel therapies.

Introduction

Neurodegenerative disorders are a group of age-related diseases characterized by progressive neuronal cell death. Alzheimer’s (AD), Huntington’s (HD), and Parkinson’s (PD) disease are examples of these disorders. Although each one of them has particular pathologies and symptoms, accumulated evidence points to similar mechanisms of neurodegeneration, including inflammation, oxidative stress, and protein aggregation (Granholm et al. 2008). In that regard, reports in all three disorders suggest that the accumulated misfolded proteins may have prion-like behavior. A significant body of research is ongoing to understand how these pathways affect each other and what ultimately triggers the onset of the disease to prevent, stop, or slow down disease progression. Mapping of the human genome was accomplished in 2003; improved genetic methods have accelerated the identification of an array of mutations linked to these disorders. New animal models are being developed to better understand pathology and improve the platform in which to test novel therapies.

Animal models of neurodegenerative disorders have been developed in multiple species in vertebrates and invertebrates aiming to understand different aspects of these diseases and identify new therapies (AD: Ji et al. 2016; Onos et al. 2016; HD: Brooks and Dunnett 2015; Howland and Munoz-Sanjuan 2014; Krench and Littleton 2017; PD: Chesselet and Richter 2011; McDowell and Chesselet 2012; Tenreiro et al. 2017). Experiments in nonhuman primates (NHPs) account for a small percentage of all research in animals. Yet the impact of NHP studies for clinical translation is much greater, especially for neurological disorders, as NHPs have a highly developed cerebral cortex and cognitive function, complex motor skills, and neuroanatomy. The mapping of the rhesus (Macaca mulatta) and common marmosets’ (Callithrix jacchus) genomes was accomplished in 2007 and 2014, respectively; they are used as a resource to identify similarities and divergences between species as well as to improve NHP models of disease. The goal of this report is to review NHP models of AD, HD, and PD in the context of the current understanding of these diseases and their contribution to the development of novel therapies.

AD

AD is the most common age-related neurodegenerative disorder and form of dementia. It is estimated that one in nine persons over 65 years old has AD and that over 5 million Americans are affected by AD. Five percent of all AD cases are inherited and are diagnosed between the age of 30 and 60 years old (early onset) (Younger/Early Onset Alzheimer’s & Dementia 2017. Available online).

AD patients present cognitive or intellectual symptoms (e.g., amnesia, aphasia, apraxia, and agnosia) as well as psychiatric ones (e.g., personality changes, depression, hallucinations, and delusions). In 2011 Alzheimer’s Association and the National Institute on Aging issue an updated criteria and guidelines for diagnosing AD and emphasized the urgency of defining preclinical AD stage. AD is usually diagnosed when patients are already in the middle stages of the disease and activities of daily living and sleep become difficult. During the more advanced, severe stages, AD patients may not be able to communicate, recognize him/herself and family members, may lose bladder and/or bowel control, and require help with all activities. Current AD treatments are symptomatic. For psychiatric symptoms, antipsychotic or antianxiety medications may be prescribed. For cognitive loss, enhancing cholinergic transmission by inhibition of the enzyme acetylcholinesterase is a usual treatment. N-methyl-D-aspartate receptor antagonists, which work by regulating the activity of glutamate, are used for improving memory and learning (Melnikova 2007).

Pathologically, AD is characterized by progressive neuronal loss in the hippocampus, entorhinal cortex, and basal forebrain cholinergic system combined with the presence of ß-amyloid plaques and tau-positive neurofibrillary tangles. Neuroinflammation is observed in areas of neurodegeneration. As AD progresses, it greatly compromises the cerebral cortex and eventually induces global brain deterioration (Braak et al. 1993; Schliebs 2005). The loss of cholinergic neurons is thought to be a leading factor in the decline of cognitive function; loss of serotoninergic and catecholaminergic neurotransmission are proposed as contributing factors (Palmer et al. 1987a; 1987b).

The cause of AD is not clear. For most AD cases, a combination of genetic, environmental, and lifestyle factors, which varies between individuals, has been proposed. In that regard, having the apolipoprotein E ε4 gene on chromosome 19 increases risk for AD and is also associated with an earlier age of disease onset. Early onset has been linked to gene mutations on chromosomes 21, 14, or 1. Interestingly, each of these mutations affects the breakdown of abnormal amyloid precursor protein, generating amyloid plaques (Lane-Donovan and Herz 2017; Serretti et al. 2007). It has also been proposed that AD may share similarities with prion diseases based on evidence that ß-amyloid can be “seeded,” yet further confirmation is needed (Kovacs 2016).

Modeling AD in NHPs

NHPs’ cognitive abilities and brain complexity are well documented and have contributed to the understanding of the neuroanatomical basis of AD. This knowledge combined with the identification of AD risk factors has facilitated the development of NHP models of AD. Aging or targeted lesioning of rich cholinergic brain areas are typical methods to mimic AD in monkeys. New NHP models based on ß-amyloid and tau pathology are emerging.

Aged monkeys, like aged humans, develop behavioral and cellular abnormalities over time and some of them resemble AD (Albert 2002). Age-related deficits in memory and attention, deposits of amyloid plaques, and atrophy and/or loss of cholinergic and monoaminergic neurons are well documented in NHPs (Peters et al. 1996; Price et al. 1991). Upregulation of Aβ42 in the brain and bodily fluids is found in aged rhesus macaques (Zhao et al. 2017). Aged marmoset monkeys also present Aβ40 in vascular deposits, while Aβ42 is mostly found in plaques associated with swollen neurites but not abnormally phosphorylated tau (Geula et al. 2002). Although aging is not AD, aged models are excellent platforms to understand aging as a risk factor for AD, and, with certain limitations, as models of early AD.

The loss of cholinergic basal forebrain neurons and associated cognitive deficits can be mimicked in NHPs by stereotaxic injections of the cytotoxin ibotenic acid into the basal forebrain or by transection of the fornix (Kordower and Fiandaca 1990; Voytko et al. 1994). Lesions limited to the hippocampal region mainly impair monkeys’ performance in tasks of recognition memory (Zola et al. 2000) and learning two-choice discriminations. In comparison, the caudate nucleus is necessary for learning more difficult, gradually acquired discrimination tasks (Teng et al. 2000).

New models based on the intracerebral delivery of ß-amyloid plaques and neurofibrillary tangles are currently being pursued. It should be noted that evaluation of ß-amyloid transmission to NHPs was reported in 1993 (Baker et al. 1993). Three marmoset monkeys were intracerebrally injected with brain tissue from an AD patient. Analysis of the monkey brains 6–7 years later showed moderate numbers of amyloid plaques, dystrophic neurites, and cerebral amyloid angiopathy, but no neurofibrillary tangles. These pathologies were not found in age-matched controls (n = 6) or animals injected with normal brain tissues (n = 12). A few plaques were found in the brains of two of four marmosets that received brain tissue from three elderly patients with age-related pathology, of which two may have had possible prion disease. Results of a follow-up study in which marmoset monkeys were injected with AD brain homogenates, synthetic Aβ-peptides, or CSF found approximately 3.5 years later that β-amyloid seemed to be partially responsible for initiating or accelerating the process of cerebral amyloidosis (Ridley et al. 2006). In that regard, marmoset monkeys that were co-injected with β-amyloid fibrils and lipopolysaccharide to elicit inflammation presented plaques 5 months post surgery, suggesting that the inflammation accelerated β-amyloid deposition (Philippens et al. 2017).

Contributions of NHP Research to the Improvement of AD Treatments

Candidate AD therapeutic approaches can be grouped into neurorestorative and neuroprotective. Neurorestorative approaches using fetal tissue or stem cells to replace cholinergic neurons lost to the disease have been proposed (Sugaya 2003). Yet, the focus of the AD field has shifted towards the development of neuroprotective strategies due to the progressively widespread neurodegeneration and the complexity of the behaviors affected by AD.

Lifestyle modification has been proposed to decrease AD risk. Healthy diet and exercise are typical medical suggestions, in part because obesity is associated with type II diabetes, which in turn increases the risk of AD (Grizzanti et al. 2016). NHP studies in calorie restriction (CR) support this concept. Rhesus macaques following a CR diet have significantly better glucose regulation than age-matched controls, greater preservation of gray matter in frontal and parietal cortices, and better learning of a motor task (Kastman et al. 2012). Postmortem brain analysis of CR monkeys compared to age-matched controls showed that CR modulates inflammation and offset the burden of oxidatively damaged proteins (Sridharan et al. 2014; Willette et al. 2013). The consequences of a sedentary life are not limited to weight gain but also decreased cognitive ability. Studies in middle-age rhesus macaques trained to run daily in a treadmill for a period of 5 months have shown that improved fitness increases both the rate of learning and blood flow to the cerebral cortex, at least during the period of regular exercise (Rhyu et al. 2010).

In the last decade, a number of NHP preclinical and clinical trials evaluating neuroprotective strategies have been performed with differing levels of positive behavioral effects (Akwa et al. 2005). Although several trophic factors are available today, nerve growth factor (NGF) remains the favorite for AD treatment. Several lines of research have confirmed NGF support of cholinergic neurons and more recently have linked it to decreased amyloid burden (Triaca et al. 2016). NGF cannot cross the blood brain barrier; thus, it requires chronic intracerebral targeted delivery, as intracerebroventricular administration induces adverse side effects (Winkler et al. 1997). Studies in rodents and NHP models of AD have shown that NGF delivery by direct protein or ex vivo gene therapy can protect cholinergic neurons from degeneration and sustain cholinergic function (Tuszynski et al. 1990). The safety, toxicity, and efficacy of autologous fibroblasts genetically modified to deliver NGF (Tuszynski et al. 1996) have been tested in NHPs before clinical trials. A phase I study has shown safety and some improvements (Tuszynski et al. 2005). Postmortem analysis of patients that received the therapy a decade earlier showed a positive response to NGF (Tuszynski et al. 2015). As an alternative to personalized cell preparation, adeno-associated viral vectors (AAV) (Mandel et al. 1999) and lentiviral vectors (Blesch et al. 2005) encoding for NGF have been tested in NHPs, with results similar to ex vivo treatments. The results of a phase I clinical trial, using bilateral stereotactic administration of AAV-NGF to the nucleus basalis of Meynert, demonstrated the feasibility of the approach (Rafii et al. 2014). AAV-NGF was well tolerated and able to produce long-term, biologically active NGF expression (Tuszynski et al. 2015). Based on these data, a multicenter, double-blind, sham surgery-controlled trial is currently ongoing (clinicaltrials.gov).

Several other therapies are currently being tested for AD treatment; of particular interest to the field are the ones targeting β-amyloid, such as vaccines, antibodies, and inhibitors or modulators of γ- and β-secretase and vaccines against tau protein. The role of NHP research in these studies has been mainly to assess tolerability and safety, and provide some evidence of efficacy in aged monkeys. Clinical trials using Aβ antibodies (solanezumab, gantenerumab, crenezumab) have not yet proven successful as disease-modifying therapies. New clinical trials targeting populations at risk of developing AD, compared to more advanced cases, may have a better chance to succeed (clinical trials.org) (Godyn et al. 2016; Marciani 2016).

HD

HD is an autosomal-dominant, progressive neurodegenerative disorder. It is estimated that 5 to 10 per 100,000 individuals have HD, with variations between populations (Harper 1992; Rawlins 2010); more than 30,000 Americans are affected by HD.

HD onset is usually observed in patients 30 to 50 years old; juvenal HD refers to cases with disease onset before 20 years of age (Lee et al. 2012). Patients with HD display dystonia and involuntary choreiform movements of the arms, legs, head, face, and upper body. They also develop dementia, with frontal-type cognitive impairments, and psychiatric manifestations, including alterations in mood, like depression, anxiety, anger and irritability, and obsessive-compulsive behaviors (Leroi and Michalon 1998; Mendez 1994; Rosenstein 1998). Although HD onset is defined by the development of the movement disorder, mutation carriers may present a combination of subtle clinical and neuroimaging parameters for several years before diagnoses. Treatments are mainly symptomatic, with limited efficacy, and can have severe side effects. The movement disorder can be treated with tetrabenazyne or antipsychotic drugs; the latter are also beneficial for severe anger or threatening behavior. Cognitive deficits are difficult to manage and are not responsive to cholinesterase inhibitors (Ha and Fung 2012).

Pathologically, HD is characterized by progressive loss of gamma aminobutyric acid (GABA) producing neurons in the striatum. Severe neuronal loss also occurs in deep layers of the cerebral cortex; neuroinflammatory markers are increased in areas of neurodegeneration. As HD progresses, neuronal loss and atrophy are observed in the globus pallidus, thalamus, subthalamic nucleus, substantia nigra, and cerebellum (Huntington 2003).

In 1993 the gene responsible for HD was identified in chromosome 4 (1993); a diagnostic test is currently available. The HD mutation consists of an expansion of a cysteine-adenosine-guanine (CAG) repeat encoding a polyglutamine tract in the N-terminus of the protein huntingtin. In normal conditions, huntingtin is found in the cytoplasm; it seems to have a role in normal cellular interactions, including transport vesicles, synaptic vesicles, microtubules, and mitochondria. N-terminal fragments of mutant huntingtin accumulate and form inclusions in the cell nucleus in the brains of patients with HD. Interestingly, huntingtin inclusions are rarely found in striatal neurons (compared to the cortex) and when present are observed in interneurons, which are typically spared in HD neurodegeneration (Saudou and Humbert 2016).

Modeling HD in NHPs

Modeling HD in NHPs has evolved over time. Neurotoxin models can be induced by direct stereotaxic injections into the striatum of an N-methyl-d-aspartate receptor agonist such as kainic, ibotenic, or quinolinic acid. These excitotoxins are reported to induce preferential loss of striatal GABAergic neurons, although some studies have found widespread neuronal loss in the lesion area (Beal et al. 1986; Burns et al. 1995; Coyle and Schwarcz 1976; Coyle et al. 1978; Davies and Roberts 1987; Ellison et al. 1987; Isacson et al. 1990; Schwarcz et al. 1983; Storey et al. 1994). In these models, systemic administration of apomorphine is administered to elicit dyskinesias; quinolinic acid has been shown to induce frontal cognitive deficits. An alternative modeling method uses chronic systemic administration of the mitochondrial complex II inhibitor 3-nitropropionic acid (3-NP) (Brouillet et al. 1995; Palfi et al. 1996, 2000). In baboon monkeys (Papio papio), 3-NP intoxication induces spontaneous and apomorphine-induced dyskinesia and a frontal-type cognitive decline (Palfi et al. 1996); in capuchin monkeys (Cebus apella), it also induces progressive dystonia (Palfi et al. 2000; Roitberg et al. 2002). The behavioral changes are associated with bilateral neuronal cell loss in the striatum, with a dorsoventral gradient, and relative sparing of the reduced nicotinamide-adenine dinucleotide phosphate positive interneurons and cholinergic neurons (Brouillet et al. 1995; Palfi et al. 1996; Roitberg et al. 2002).

With the identification of the mutated huntingtin gene emerged the possibility of generating genetic HD animal models. Due to the complexity of producing transgenic HD monkeys, investigators evaluated the effect of injecting lentiviral vectors encoding for the first 171 amino acids of the Huntingtin protein with 19 (wild-type: Htt171-19Q/19Q) or 82 (mutated: Htt171-82Q) polyglutamine repeats into the dorsolateral sensorimotor putamen of macaques (Palfi et al. 2007). In a first phase, cynomolgus macaques (Macaca fascicularis) were studied for 9 weeks. Apomorphine administration induced progressive chorea, dystonia, and ipsilateral turning behavior to the Htt171-19Q/82Q animals (n-4), while the Htt171-19Q/19Q (n = 2) or an additional buffer control (n = 1) showed no abnormal behavior. Postmortem, the Htt171-82Q subjects presented neuritic and nuclear huntingtin aggregates, astrocytosis, and neuronal loss. Then the investigators injected Htt171-82Q bilaterally into the dorsolateral putamen of three monkeys. Fifteen weeks after surgery, the animals progressively developed spontaneous dyskinesia of the legs, arms, and trunk and, in one case, tics that persisted for 30 weeks.

In 2008, the first transgenic monkey model of HD was reported (Yang et al. 2008). Mature rhesus oocytes were injected into the perivitelline space with high titer lentiviruses expressing exon 1 of the human HTT gene with 84 CAG repeats (HTT-84Q) and lentiviruses expressing the green fluorescent protein (GFP) gene, under the control of the human polyubiquitin-C promoter. Five live newborns were delivered at full term. All monkeys carried the transgenic mutant HTT and GFP genes, although the repeat length varied between subjects and ranged between 27 to 88 repeats. One animal survived for 1 month and two survived for less than a day. Based on preliminary observations, the severity, frequency, and onset of the involuntary movements seemed to depend on the length of the CAG repeats and the number of integration sites. The early death of the animals carrying a higher copy number of transgenes expressing a small amino-terminal HTT fragment suggests that the N-terminal mutant HTT fragments are pathogenic. Postmortem analysis of these three monkeys showed HTT aggregates or inclusions in the striatum and cortex, mostly located in the nuclei and neuropil; no clear signs of neurodegeneration were observed. Reports on the longitudinal evaluation of additional transgenic subjects identified progressive cognitive and motor impairment associated with a reduction in striatal volume observed with neuroimaging. Postmortem neuropathological analyses confirmed striatal neuronal loss (Chan et al. 2014, 2015). Germ line of the mutant huntingtin was confirmed in both embryonic stem cells generated from three male HD monkey founders and in second-generation offspring produced via artificial insemination (Moran et al. 2015).

Contributions of NHP Research to the Improvement of HD Treatments

Cell replacement strategies to alleviate GABAergic neuronal loss have been proposed using either fetal tissue sources or stem cell lines. In 3-NP-intoxicated monkeys, fetal striatal allografts induced recovery in a frontal-type cognitive task 2 to 5 months after surgery as well as reduced the occurrence of dystonia (Palfi et al. 1998). The clinical translation was first reported as favorable (Freeman et al. 2000), as the pathological brain evaluation of a patient that died 18 months after transplantation (due to cardiovascular disease) demonstrated survival of grafted cells, which presented typical morphology of the developing striatum and no huntingtin aggregates or neuritic dystrophies. Yet a follow-up report of the postmortem analysis of HD patients that received bilateral striatal fetal grafts 10 years earlier showed that the grafted neurons underwent HD-like neuronal degeneration (Cicchetti et al. 2009), including the presence of mutant huntingtin aggregates (Cicchetti et al. 2014).

Due to the aggressive and progressive nature of HD, effective neuroprotective strategies are urgently needed. The oral drug riluzole interferes with glutamate activity, and it has been shown to be neuroprotective in 3-NP-intoxicated monkeys (Palfi et al. 1997). In HD patients, riluzole preserved brain glucose metabolism and increased production of neurotrophins (Cicchetti et al. 2009). Gene therapy for the delivery of ciliary neurotrophic factor has also been tested in monkeys treated with 3-NP with positive effects (Mittoux et al. 2000). A new generation of gene therapy strategies aiming to deliver short hairpin RNA (shRNA) to silence the expression of mutated huntingtin is currently being investigated in NHPs (Dufour et al. 2014; McBride and Clark 2016; McBride et al. 2008, 2011; Monteys et al. 2017).

PD

PD is the second-most common age-related neurodegenerative disorder after AD, affecting 1% of the population over 65 years old. PD is diagnosed by the presence of typical motor symptoms, including tremor, rigidity, bradykinesia (slowness of movement), and postural instability. Over time, PD progression may severely affect walking, talking, or completing simple tasks. PD patients also present nonmotor symptoms such as depression, orthostatic hypotension, fatigue, urinary problems or constipation, and sleep disruptions (Titova et al. 2016). Most importantly, the onset of nonmotor symptoms may precede motor dysfunction for many years, and their identification is proposed for diagnosis of prordromal PD. In that regard, in 2015 the International Parkinson and Movement Disorder Society presented a report with guidelines for refined PD diagnosis and produced criteria for prodromal PD (Berg et al. 2015).

Dopamine (DA) replacement (usually combined with carbidopa) is the mainstay therapy for PD motor symptoms. Dopamine agonists and anticholinergics can be used alone or in combination with levodopa. Monomine oxidase B inhibitors, such as rasagiline and selegiline, are many times preferred for early PD or as a complement to DA replacement. PD nonmotor symptoms are not alleviated by DA replacement therapy; their identification is critical to provide symptomatic relief and/or include lifestyle modifications as part of the treatment plan. Disruption of the basal ganglia neural circuitry by targeted ablations or, more recently, deep brain stimulation, are possible surgical options. Compared to AD and HD, advancements in PD treatments have greatly improved patients’ quality of life. Yet as PD progresses, drug efficacy declines and side effects, such as dyskinesias, increase.

The pathologic hallmarks of the disease are the loss of dopaminergic neurons in the substantia nigra pars compacta and the presence of intracytoplasmic inclusions called Lewy bodies (LBs) and Lewy neurites (LNs), which are found in neurons of the central (CNS) and peripheral (PNS) nervous system (Braak et al. 1995; Del Tredici and Braak 2016). LBs and LNs are composed of over 70 different proteins (Wakabayashi et al. 2007); α-synuclein (α-syn) is their main component. PD motor symptoms are related to the loss of dopamine in the striatum, the main area of projection of nigral dopaminergic neurons. PD nonmotor symptoms are associated with neuropathology in other areas of the CNS and PNS (e.g., cardiac dysautonomia symptoms are related to sympathetic ganglia neurodegeneration).

The cause of PD is still unknown. Identified risk factors include aging, exposure to environmental toxins, and genetics. In 1997, the first gene mutation linked to PD was identified: a single point mutation of alanine to threonine in the 53^rd (A53T) amino acid residue in the α-syn protein sequence of patients with familial early-onset PD (Polymeropoulos et al. 1997). Following this discovery, additional mutations in the α-syn gene (SNCA) have been identified in familial PD cases, such as A30P, G51D, and triplication (Kruger et al. 1998; Lesage et al. 2013; Singleton et al. 2003). Several other gene mutations including in LRRK2, PINK1, and DJ-1 in addition to SNCA, have been linked to PD (Puschmann 2013). Five percent of all PD cases are estimated to be familial; it should be noted that the LRRK2 mutation G2019S has been observed in sporadic cases, and its incidence varies between populations (e.g., 40% of PD cases in the African Arab population) (Lesage et al. 2006).

Although the etiology of PD is unclear, inflammation, mitochondrial dysfunction, oxidative stress, and protein aggregation have been identified as mechanisms contributing to PD neurodegeneration (Blesa et al. 2015; Hirsch et al. 2012; Mullin and Schapira 2015). The identification of α-syn as a component of LBs and LNs has led to hypotheses on PD neurodegeneration centered on potential α-syn toxicity and its potential spread from the PNS to and through the CNS in a prion-like fashion (Braak et al. 2004; Chu and Kordower 2015). An alternative and maybe complementary theory is that α-syn simultaneously damages CNS and PNS neurons, but differing regional plasticity affects which dysfunction and symptoms appear first (Engelender and Isacson 2016).

Modeling PD in NHPs

NHP models of PD are based on identified risk factors including aging, neurotoxins, and genetics. Progress in the understanding of the mechanisms of PD pathology and the complexity of PD is shaping the methods of model development as well as the evaluation of the animals.

Aged NHPs have proven to be useful platforms to understand the role of aging in PD. Like humans, monkeys develop age-related dysfunction of the nigrostriatal system and motor impairments, such as slight tremor, stooped posture, and gait and balance disturbances (Emborg et al. 1998). Changes in α-syn expression, including translocation into the neuronal bodies, have been observed with age (Chu and Kordower 2007).

Neurotoxin NHP models of PD are induced by the administration of 6-hydroxydopamine (6-OHDA) and 1-methyl-4-phenyl-1,2,3,6 tetrahydropyridine (MPTP). 6-OHDA is a catecholaminergic neurotoxin (Senoh et al. 1959a, 1959b) with a molecular structure similar to dopamine, which facilitates its specific uptake by catecholamine transporters. Once inside neurons, 6-OHDA impairs mitochondrial function through direct interaction with complex I; it also rapidly undergoes autoxidation, generating reactive oxygen species and eliciting inflammation, which further promotes neurodegeneration. 6-OHDA does not cross the blood-brain barrier; thus, to exert CNS toxic effects, it must be injected directly into the brain, usually unilaterally into the nigrostriatal system (Rodriguez-Pallares et al. 2007). Systemic administration of 6-OHDA induces catecholaminergic neurodegeneration in the PNS and can be used to replicate cardiac dysautonomia (Joers et al. 2012, 2014).

MPTP is a highly lipophilic mitochondrial complex I inhibitor that easily crosses the blood-brain barrier. The enzyme MAO-B, which is found in astrocytes in the brain and in multiple peripheral organs, is responsible for converting MPTP into its toxic metabolite 1-methyl-4-phenylpyridinium. Similar to 6-OHDA, 1-methyl-4-phenylpyridinium is preferentially uptaken by the dopamine transporter system; it induces neurodegeneration by impairing mitochondrial function (Dauer and Przedborski 2003; Mizuno et al. 1987; Nicklas et al. 1985; Rappold and Tieu 2010). Nigral α-syn expression has been found to increase after MPTP dosing and, in some cases, it may accumulate (Halliday et al. 2009; Kowall et al. 2000; McCormack et al. 2008).

MPTP administration to monkeys can be done sc, im, iv, or via the carotid artery, using different dosing regimens (Emborg 2007). Similar to PD patients, monkeys exposed to MPTP present tremor, bradykinesia, hypokinesia, rigidity, and disturbance in fine motor skills as well as posture, balance, and gait abnormalities (Burns et al. 1983; Langston et al. 1984). Systemic dosing of MPTP induces a bilateral PD motor syndrome, while intracarotid artery injection induces hemiparkinsonism. MPTP can affect other neuronal groups beyond dopaminergic nigral neurons, which contribute to the onset of motor and nonmotor symptoms, including prefrontal cognitive deficits (Schneider 1990). Cardiac catecholaminergic loss, mimicking postganglionic autonomic dysfunction typical of PD, has been described, although the effects seem to be temporary (Goldstein et al. 2003). Catecholaminergic neurons in the locus coeruleus and pedunculopontine nucleus are also affected (Herrero et al. 1993a, 1993b; Masilamoni et al. 2011, 2017; Pifl et al. 1991). Similar to PD, MPTP can induce neuronal loss in caudal intralaminar thalamic nuclei, which may precede the development of motor symptoms in PD and may account for some of the cognitive deficits in attentional set-shifting (Villalba et al. 2014). Cholinergic neuronal loss in the peduculopontine nucleus of aged MPTP-treated monkeys was reported to contribute to abnormal gait and posture (Karachi et al. 2010).

Typical anti-parkinsonian medications are effective against MPTP-induced PD motor symptoms, and prolonged daily use of L-DOPA therapy induces dyskinesias, as observed in humans.

Intracerebral injection of viral vectors encoding for mutated α-syn or administration of LB extracts have been tested for modeling genetic-like PD in NHPs (Vermilyea and Emborg 2015). Nigral overexpression of human α-syn wild type and A53T using AAV vectors induced PD-like motor symptoms, significant nigral dopaminergic cell loss, and α-syn aggregates in common marmoset monkeys (Eslamboli et al. 2007; Kirik et al. 2003). AAV and lentiviral vectors encoding for A53T α-syn were tested in cynomolgus (Koprich et al. 2016) and rhesus macaques (Yang et al. 2015), inducing nigral cell loss and α-syn accumulation and aggregation; behavioral changes were not reported. AAV-induced overexpression of Parkin and A53T α-syn decreased striatal dopaminergic markers and α-syn accumulation and phosphorylation in cynomolgus, motor symptoms were not observed. Intracerebral inoculation of α-syn fibrils has not yet been used in monkeys, although several reports in rodents are available (e.g., Luk et al. 2012; Paumier et al. 2015). Cadaveric LB extracts have been injected into the striatum or nigra of cynomolgus macaques with or without previous MPTP challenge (Recasens et al. 2014), inducing some decreases in striatal and nigral dopaminergic markers and increases in α-syn expression, yet PD motor symptoms were not detected.

The first report of transgenic PD rhesus macaques used lentiviral vectors encoding for human A53T SNCA that were injected into fertilized oocytes (Niu et al. 2015). Compared to a stillborn control, an A53T animal had elevated levels of α-syn expression in the nigra, but no change in the S129 phosphorylation of α-syn a marker of possible pathology. Brain MRIs of 1.5- (n = 2) and 2.5- (n = 2) year-old A53T NHPs did not show evidence of neurodegeneration in the MRI, although the animals exhibited some difficulty in fine motor skill tests. Lastly, a 2.5-year-old A53T monkey presented cognitive deficits and an increase in the amount of time spent walking in circles compared to an age-matched control, which the investigators interpreted as increased anxiety.

Contributions of NHP Research to PD

NHP PD models have been essential for the development of efficacious therapeutic strategies (Capitanio and Emborg 2008). They provide a platform to identify the neuronal circuitry affected by the disease, generate rational surgical approaches, like deep brain stimulation, and evaluate the efficacy and safety of novel DA replacement therapies. Their role continues to be critical to ensure the safety and efficacy of inovative compounds aiming to decrease the adverse effects associated with long-term use.

As an alternative to dopamine replacement by drugs, nigral fetal grafts were tested in monkeys and translated to human trials with controversial clinical results (Fitzpatrick et al. 2009). In addition to lack of efficacy, some patients developed graft-induced dyskinesias. These uncontrolled abnormal movements were not related to L-DOPA medication and did not decrease when L-DOPA dosing was reduced or stopped. Another potential complication was the postmortem finding of LB-like pathology in the cells grafted a decade earlier (Kordower et al. 2008; Li et al. 2008). Although the impact of the aggregates on the functionality of the grafts is unknown, most important was the fact that PD could be transferred to the grafted cells. Several other cell sources have been tested in monkeys, including retinal pigmented and carotid ganglion cells, which have been proposed as an alternative source for cell replacement. Although many of these novel cells have been translated into the clinic and positive results were found in early phase trials, phase II studies have not demonstrated efficacy. More recently, stem cell lines conditioned or not to develop a dopaminergic phenotype have been developed, and reports of transplants in monkeys using these cells have been published (Vermilyea and Emborg 2017).

Gene therapy has been proposed as an alternative method to induce functional restoration by delivering enzymes responsible for dopamine production (to increase dopamine availability) or GABA synthesis (to affect neurotransmitter balance in the basal ganglia neural network). Based on rodent and monkey preclinical data (Hadaczek et al. 2010), a phase I open label clinical trial using AAV2 encoding for aminoacid decarboxylase (AADC), the enzyme necessary for the decarboxylation of L-DOPA into dopamine, was performed. AAV2-AADC was infused into the putamen; the results of the trial met criteria for safety (Christine et al. 2009). Currently, a new AAV2-AADC phase I trial is ongoing, in which the viral vector suspension is infused into the putamen using MRI-guided convection and the target population is PD patients with fluctuating responses to L-DOPA (clinicaltrials.gov).

The first PD gene therapy trial proposed to use intracerebral injections of AAV2 encoding for glutamic acid decarboxylase (GAD, the rate-limiting enzyme for the synthesis of GABA) into the subthalamic nucleus to inhibit overactive glutamatergic neurons. As rodent and NHP data (Emborg et al. 2007) suggested feasibility, a phase I clinical trial was performed that confirmed the safety of the approach (Kaplitt et al. 2007) and suggested efficacy (Feigin et al. 2007). A follow-up, double-blind, sham-surgery-controlled randomized phase II trial in which AAV2-GAD was bilaterally delivered into the subthalamic nucleus of patients with advanced PD demonstrated efficacy (LeWitt et al. 2011).

Due to the progressive nature of PD, neuroprotective strategies are urgently needed. Several compounds that can be administered systemically such as MAOB inhibitors, dopaminergic agonists, immunophilines, gangliosides, and nicotinic agonists have being evaluated, but none has shown clear clinical efficacy (Stocchi 2015). Glial derived neurotrophic factor (GDNF) is a potent dopaminotrophic factor that requires targeted delivery to be effective without inducing side effects (Lapchak et al. 1997). Intracerebral infusions have presented controversial results; the method of delivery and protein distribution have been identified as issues that need to be addressed (Sherer et al. 2006). Gene therapy is envisioned as a more efficient way to chronically deliver trophic factors (e.g., Kordower et al. 2000). AAV2 encoding for GDNF has been shown to be effective in PD monkeys (Kells et al. 2010). A phase 1, open-label, dose escalation, safety and tolerability study of AAV2-GDNF for patients with advanced PD is currently ongoing. AAV2 encoding for neurturin (a trophic factor related to GDNF) has shown a similar preclinical success than GDNF (Herzog et al. 2008), suggesting that it could be used as an alternative treatment. AAV2-neurturin clinical trials, first targeting the putamen and then the putamen and nigra, showed encouraging results in phase I but failed in phase II studies (Kordower 2016).

New therapies aiming to address synucleinopathy are now proposed, including silencing the α-syn gene or increasing α-syn clearance (Wong and Krainc 2017). Interestingly, in vervet monkeys (Chlorocebus aetiops), nigral injection of AAV-shRNA to knockdown α-syn induced a region-specific decrease in TH-positive nigral cell number and striatal innervation compared to animals that received scrambled shRNA; no behavioral changes were reported (Collier et al. 2016). Administration of kinase inhibitors are currently in the process of being evaluated (clinical trials.org). Without regards of the therapy, patients’ PD stage at the time of the trial will be an issue for neuroprotective success. By the time of diagnoses, PD patients have already lost 50% of nigral dopaminergic neurons, suggesting that the chances for inducing neuroprotection are limited. A recent example is the result of the evaluation of the peroxisome proliferator receptor gamma agonist termed pioglitazone, which in PD animal models including NHPs showed positive effects (Swanson et al. 2011), and although its clinical translation showed some effects, it failed to achieve significance (Investigators 2015).

Conclusions

The contribution of NHP research to the AD, HD, and PD fields has yielded a number of advancements toward the understanding of neural networks affected by each disorder, and basically provided a canvas (and still does) to assess the safety and feasibility of novel therapies. It is clear that much work is still needed, especially toward the development of neuroprotective strategies. In that regard, the identification of biomarkers of early disease onset will play a critical role for successful clinical translation. In the case of HD, mutation carriers can be targeted, yet the age of disease onset is not easily predictable. For AD and PD, the identification of vulnerable populations will be more difficult, as in most cases disease onset depends on a combination of factors. Can NHP studies help untangle the clues?

Whether β-amyloid, huntingtin, and α-syn have the potential to act as prions is still debatable (Collinge 2016; Stopschinski and Diamond 2017), yet they seem to have a role in neurodegeneration, similar to inflammation and oxidative stress (Gasiorowski et al. 2017; Kannarkat et al. 2013; Manoharan et al. 2016; Silajdzic et al. 2013). A new generation of NHP models is emerging. Although most of these models need validation and characterization, they represent an opportunity to assess the efficacy of novel therapies, which cannot be tested in toxin models. Furthermore, transgenic models will facilitate the search of biomarkers of disease onset and progression, as the animals can be studied from birth. New genomic editing technologies (e.g., Gaj et al. 2016) that allow for creating point mutations, instead of overexpression of a mutated protein, hold promise for the creation of true-to-life models.

Common marmosets have emerged as an alternate NHP resource (Cyranoski 2014) as they present certain advantages for the generation of genetic models compared to more traditional NHP species, like rhesus macaques. Marmosets regularly give birth to twins or triplets compared to rhesus singletons, and they have a shorter lifespan (~16 vs. ~35 years), which facilitates the study of age-related disorders (Abbott et al. 2003; Schultz-Darken et al. 2016). Yet, when planning a research project, investigators should also consider the species brain volume, behavior, and genetic makeup, among other things, and how these issues relate to the hypothesis to be analyzed. For example, a mutation in the α-syn gene that leads to the replacement of the amino acid alanine by threonine at the 53^rd residue (A53T) has been linked to early-onset familial PD. In normal conditions marmosets and most other vertebrates, with the exception of humans, apes, and macaques (Hamilton 2004), naturally have a threonine at the 53^rd residue of the α-syn sequence. Thus, studies on the impact of α-syn pathology will benefit from using rhesus as a model organism.

As mentioned in the introduction, NHP studies represent a small but important percentage of the overall research enterprise. Clinical trials solely based on rodent studies have been shown to have lower rates of translational success compared to the ones following a stepwise approach that included NHP investigations (Zeiss 2017; Zeiss et al. 2017). NHP’s role is most relevant for longitudinal studies, evaluating first-in-class and invasive therapies, beyond exploratory studies but to assess specific questions of efficacy and safety (Zeiss 2017). This approach requires a thorough understanding of the disease as well as the model to be used. Although NHP models of neurodegenerative disorders continue to be improved, they are not perfect, and many need to be validated. Neurodegenerative disorders are complex and their presentation varies between individuals; therefore, one model will not necessarily provide all needed answers. The best a scientist can do is to match the NHP model and the experimental design to the scientific question at hand and apply rigorous and unbiased scientific methods for data collection and analysis. Ultimately, all results, positive and negative, should be shared to maximize the usefulness of the study and to contribute to finding a cure for neurodegenerative disorders.