Dementia associated with disorders of the basal ganglia

2.1 Input nuclei

The striatum, the main input nucleus of the basal ganglia, is primarily made up of ϒ-aminobutyric acid (GABA)-ergic medium-sized spiny neurons (MSNs), neurons with high abundance of dendritic spines comprising around 90% of all striatal neurons and a small proportion of cholinergic and GABAergic interneurons (Kawaguchi, 1993). These neurons relay the excitatory input they receive from cortical regions either monosynaptically to the output region of the basal ganglia, that is the SNr or GPi, or to the GPe which subsequently projects to the output structures. The classical view is that the striatal neurons directly innervating the output regions of the basal ganglia express dopamine receptor subtype 1 (D1) receptors and form the direct pathway. On the other hand, the latter neurons, which project to the globus pallidus, contain dopamine receptor subtype 2 (D2) receptors and make up the indirect pathway (Albin, Young, & Penney, 1995; DeLong, 1990). This expression pattern allows for the dopaminergic neurons in the SNc to modulate striatal neuron activity. Dopamine release increases the activity of the direct pathway through the activation of D1 receptors, while also decreasing the activity of the indirect pathway through D2 receptor activation through L-type calcium channels (Hernandez-Lopez, Bargas, Surmeier, Reyes, & Galarraga, 1997; Kravitz et al., 2010). Ultimately, despite the seemingly opposing actions of these pathways, the net outcome of dopamine release results in the activation of cortical motor regions.

The subthalamic nucleus, composed predominantly of glutamatergic neurons, mainly receives corticothalamic inputs, as well as projections from within the basal ganglia and specifically, from the GPe. This region projects to the globus pallidus or SNr, a projection known as the hyperdirect pathway, which has been suggested to be involved in the regulation and cessation of motor behavior and cognition (Nelson & Kreitzer, 2014; Wessel & Aron, 2013; Wessel, Conner, Aron, & Tandon, 2013; Wessel et al., 2016). Interestingly, the subthalamic nucleus has been highlighted as a target for deep brain stimulation in PD patients due to its ability to directly relay motor and premotor cortical information to the globus pallidus through this hyperdirect pathway (Dayal, Limousin, & Foltynie, 2017; Hamel et al., 2003).

2.2 Output nuclei

The two main output regions of the basal ganglia are the GPi and the SNr. They are primarily composed of GABAergic neurons and therefore form inhibitory projections with several thalamic and brainstem nuclei. The GPi and SNr are innervated by inhibitory projections coming from striatal neurons through the direct pathway, as well as excitatory projections from the subthalamic nucleus through the indirect pathway (Deschenes, Bourassa, Doan, & Parent, 1996). Some of the projections of the output nuclei are directed at the intralaminar nuclei of the thalamus which comprises both motor and limbic circuits, highlighting the potential involvement of the basal ganglia in cognitive functions as well as motor control (Fazl & Fleisher, 2018).

The tight organization of the basal ganglia demonstrates the high interconnectivity of their circuits and how impairment of one region may lead to abnormal functioning of another. In fact, exactly this interdependency of basal ganglia activity is what has been shown to be a driving factor for several neurodegenerative diseases, as discussed below.

3.1 A common pathway in the basal ganglia underlying psychiatric disorders and dementia

Dementia refers to a set of symptoms caused by brain disease or injury, resulting in persistent memory, personality, mood, and judgment impairment. Many of the neurodegenerative diseases, including AD, PD, HD, and DLB, are characterized by deterioration into some form of dementia sooner or later over the course of the disease. Cell death caused by abnormal protein accumulation and aggregation is a common feature of these diseases, which then leads to cognitive disturbances and decline into dementia.

The basal ganglia interact closely with the frontal cortex and have crucial roles in cognition and behaviors such as judgment and reasoning (Frank, Loughry, & O’Reilly, 2001) and damage to the basal ganglia can lead to some of the same cognitive impairments as damage to the frontal cortex. This is evidenced in individuals with frontotemporal dementia, particularly frontotemporal lobar degeneration, where atrophy in the frontal and temporal lobes of the brain leads to a profound decline in behavior and language ability (Rabinovici & Miller, 2010). Both frontotemporal dementia and frontotemporal lobar degeneration appear to have a shared pathological profile with certain psychiatric disorders, particularly bipolar disorder and major depressive disorder (Nascimento et al., 2019). This indicates that there may be a common underlying pathway driving cognitive disturbances in dementia and psychiatric disorders; the basal ganglia are highly likely to be involved as they have been shown to have abnormalities in both. For example, significant shape differences are observed in the striatum in bipolar disorder (Hwang et al., 2006) and frontotemporal dementia patients exhibit structural changes in the striatum and thalamus (Bede et al., 2018).

Another psychiatric disorder that dementia may have overlapping mechanisms with is depression. Individuals with disorders of the basal ganglia are particularly susceptible to depression and patients with AD, PD, HD, DLB, frontotemporal lobar degeneration, and vascular dementia, to name a few, commonly experience depressive symptoms at some point during the disease (Baquero & Martin, 2015; Lanctot et al., 2017; Ring & Serra-Mestres, 2002). Up to 50% of PD (Cummings, 1992) and 30% of HD patients (Slaughter, Martens, & Slaughter, 2001) have depression, making it likely that basal ganglia damage in neurodegeneration may impact pathways underlying depression. Depression may also be a risk factor for dementia (Stern, Marder, Tang, & Mayeux, 1993) and neurodegenerative abnormalities in the basal ganglia may contribute to depression: in patients with major depressive disorder, length of illness, and number of depressive episodes is associated with decreased left putamen and left globus pallidus gray matter volume (Lacerda et al., 2003). This suggests that basal ganglia damage impacts higher level cognitive functions that are affected in both depression and dementia. Understanding this link will help elucidate if drugs used to treat depression can alleviate some of the symptoms of dementia and vice versa.

3.2 Basal ganglia and cognitive function in dementia

Individuals with dementia have a wide range of cognitive impairments, including memory loss as one of the first signs. This may be linked to the basal ganglia in some way due to their role in higher level cognitive functions. Procedural memory, or the ability to learn new skills, is particularly affected in PD and HD patients. In early PD, spatial learning and working memory, typically associated with nigrostriatal and dopaminergic pathways, are selectively impaired, indicating damage to networks connecting the prefrontal cortex and basal ganglia (Postle, Locascio, Corkin, & Growdon, 1997). In a study of PD patients with or without dementia, dementia was linked to an increased atrophy of regions of the basal ganglia including the caudate and the putamen (Melzer et al., 2012). Overall this suggests that damage to the basal ganglia may drive certain symptoms of dementia but how this occurs remains unclear.

In the healthy brain, the basal ganglia may regulate the ability to multitask and filter out distractions: individuals with better working memory have increased activation in the globus pallidus when faced with distracting stimuli, suggesting an improved efficiency at filtering out unwanted information (McNab & Klingberg, 2008). AD patients in particular have impaired working memory and attentional deficits (MacPherson, Della Sala, Logie, & Wilcock, 2007). It is therefore conceivable that basal ganglia pathways controlling executive function and memory are affected in neurodegenerative disorders.

In a drive to develop novel, effective, and targeted therapies for the dementias which currently have limited to no cure, it is crucial to understand how specific brain areas are impacted and how this may be shared between diseases. This review addresses shared basal ganglia dysfunction across neurodegenerative diseases and dementias and their implications.

5.1 Genes associated with Parkinson’s disease

PD itself is a multifactorial disease, arising from both genetic and environmental factors, which together contribute to disease progression through their effects on the accumulation and aggregation of misfolded proteins, clearance mechanisms, mitochondrial damage, and neuroinflammation. However, although more than 90% of cases are sporadic, a proportion of patients have a family history of the disease. To date, a number of genes have been identified as causative for PD, including mutations in the genes encoding for PARKIN, PINK1, SNCA (α-synuclein), DJ-1, and LRRK2. These proteins have been shown to play an important role in molecular pathways, which, when perturbed, can lead to neurodegeneration (Schapira, 2008). For instance, PINK1, DJ-1, and LRRK2 have been shown to encode proteins localized to mitochondria, suggesting their potential involvement in mitochondrial function and dynamics (Hao, Giasson, & Bonini, 2010; Pickrell & Youle, 2015; Truban, Hou, Caulfield, Fiesel, & Springer, 2017). We refer the reader to the review by Karimi-Moghadam, Charsouei, Bell, and Jabalameli (2018). Animal models of PD have contributed much to the current understanding of how mutations in these proteins can lead to mitochondrial impairment, oxidative stress, and disturbance in calcium homeostasis, which together can bring about disease pathology (Klein & Westenberger, 2012; Lill, 2016; Polymeropoulos et al., 1997).

5.2 Loss of dopamine and physiological changes in the basal ganglia in Parkinson's disease

Loss of nigrostriatal dopaminergic neurons is a prominent feature of the disease. Interestingly, the dopaminergic neurons projecting to the striatum are more largely affected than those projecting to limbic regions, explaining the disturbances seen primarily in motor control (Galvan, Devergnas, & Wichmann, 2015; Kish, Shannak, & Hornykiewicz, 1988). This loss of dopaminergic neurons occurs in the early stages of the disease, with indications of this phenotype to precede the diagnosed onset of motor symptoms, suggesting their crucial role in modulating motor behavior (Iacono et al., 2015; Poewe et al., 2017). A common diagnostic tool for PD is the use of the radiolabeled ligand ¹²³I-ioflupane in single-photon emission tomography, which binds to dopamine transporters in the presynaptic space. The ¹²³I-ioflupane signal is reduced in the SNc and striatum of PD patients, which correlates well with the disease state, highlighting the importance of dopaminergic neuron death in driving pathology (Brooks, 2016; Kraemmer et al., 2014; Marshall et al., 2009).

The specific susceptibility of these neurons has been suggested to be due to a number of factors, particularly their high energetic demand, rendering them highly susceptible to mitochondrial dysfunction and oxidative stress, as well as their high dependence on L-type Ca²⁺ channels for their activity, making them vulnerable to altered Ca²⁺ influx (Bishop et al., 2010; Chan et al., 2007; Foehring, Zhang, Lee, & Callaway, 2009; Liss et al., 2005; Sulzer, 2007). The importance of these neurons is highlighted by the efficiency of L-DOPA, a dopamine precursor, at slowing, albeit not halting, disease progression (Barbeau, 1969). Moreover, the finding that nigrostriatal dopaminergic neurons are also selectively vulnerable to MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine), a toxin which leads to the development of a movement disorder very similar to PD, further highlights the pivotal role these neurons play in PD pathogenesis (Varastet, Riche, Maziere, & Hantraye, 1994).

SNc dopaminergic neurons harbor extensive axonal arbors that contact a large number of synapses, making their axons highly susceptible to alterations in axonal trafficking and abnormal mitochondrial activity (Arbuthnott & Wickens, 2007). Thus, due to the high metabolic demand of axonal terminals, somatodendritic regions of these neurons become less populated with mitochondria (Liang, Wang, Luby-Phelps, & German, 2007), which reduces their ATP-generating capacity and may subsequently result in an energy crisis (Nicholls, 2008). In support of this idea, human imaging and postmortem studies have demonstrated that some changes in the basal ganglia, such as the emergence of dysfunctional nigrostriatal axonal terminals (Kordower et al., 2013) and changes in dopamine turnover (Sossi et al., 2004), occur even before disease diagnosis. Recent positron emission tomography imaging studies have also demonstrated the loss of vesicular monoamine transporter 2 activity in the posterior putamen of patients with mild PD (Fazio et al., 2015; Hsiao et al., 2014). These findings suggest a reduced innervation of striatal neurons by SNc dopaminergic neurons and hence support the idea of an early loss of their axonal terminals. Furthermore, α-synuclein, the main component of the pathological Lewy bodies, has also been suggested to play a role in the early abnormal axonal pathology observed in PD patients through a yet unknown mechanism. This idea has arisen from the protein's increased propensity for fibrillation when mutated (Conway, Harper, & Lansbury, 1998; Narhi et al., 1999), its ability to regulate dopamine metabolism and transmission (Chen et al., 2008; Lam et al., 2011; Perez et al., 2002), and its toxic influence on mitochondrial function, as shown in animal models (Martin et al., 2006; Ordonez, Lee, & Feany, 2018). Given that α-synuclein is in fact a synaptic protein and is therefore able to exert its toxic effects at synaptic sites, it is reasonable to suggest its potential role in dopaminergic axonal dysfunction in the disease.

The dopaminergic neurons in the SNc are autonomously active, generating broad action potentials (2–10 Hz) in the absence of synaptic input (Chan et al., 2007; Grace & Bunney, 1984; Mamelak, 2018). In the basal ganglia, this pacemaker activity is considered important for the maintenance of basal dopamine levels and is highly reliant on the activity of voltage-dependent L-type Ca²⁺ channels with the Cav1.3 subunit (Striessnig et al., 2006). This subunit allows for Ca²⁺ channels to be open during a hyperpolarized state, resulting in continuous oscillations of intracellular Ca²⁺ levels. However, this continuous Ca²⁺ influx can overload the Ca²⁺ buffering ability of SNc dopaminergic neurons and thus may be considered a key contributor to their susceptibility in PD (Surmeier, Guzman, Sanchez, & Schumacker, 2012). Ca²⁺ influx requires an equivalent amount of ATP molecules for its exclusion, making dopaminergic neurons at a constant high energetic demand. In fact, nigrostriatal dopaminergic neurons have also been shown to have a low Ca²⁺ buffering capacity resulting from their relatively low levels of Ca^2+-binding proteins such as calbindin. This allows excess intracellular Ca²⁺ to bind to intracellular proteins, which can trigger cell death pathways and further enter the mitochondria to then exert continuous mitochondrial oxidant stress (German, Manaye, Sonsalla, & Brooks, 1992; Surmeier et al., 2017). In comparison, the dopaminergic neurons in the ventral tegmental area, which are relatively resistant to pathology, express high levels of this protein, highlighting the importance of Ca²⁺ buffering capacity in cell susceptibility in PD.

It is important, however, to highlight that all the above mechanisms happen simultaneously, making it particularly challenging to target just one mechanism to improve neuronal viability. The fact that current treatments have been ineffective stresses the importance to understand what makes the basal ganglia susceptible in this disease and how they can be targeted for symptom alleviation.

6.1 Structural alterations of basal ganglia in Huntington's disease

Brain abnormalities in HD develop before evident symptoms or clinical diagnosis. There is 25% brain weight loss at advanced stages of the disease. The neuropathology consists of atrophy, neuronal loss, and astrogliosis of the basal ganglia and cortical structures. Striatal pathology starts caudally and then spreads rostrally in a dorsomedial to ventromedial direction (Fazio et al., 2015). Additionally, other regions of the brain are affected while the disease progresses, such as hippocampus, amygdala, and cerebellum (Rosas et al., 2003).

The loss of striatal and pallidal neurons in HD is selective; early in the disease there is loss of putaminal GABAergic cells projecting to the lateral globus pallidus, which affects the indirect pathway of the basal ganglia. The balance between indirect and direct pathways is consequently in favor of the direct pathway: there is unsuccessful inhibition and thus excitation of thalamic neurons is enhanced, resulting in random and inappropriate neuronal activity. In contrast, when the disease is advanced, cholinergic signals are lost and the caudate nucleus is atrophied. Detection of this state of the disease is visible through MRI or computerized tomography in the brain, where enlarged lateral ventricles of the caudate frontal horns are observed (Reiner et al., 2011).

In HD, there is also a significant shrinkage of the amygdala. This region plays a crucial role in the limbic circuit of the basal ganglia to trigger specific cognitive responses linked to motor activities. Indeed, declining emotion recognition in patients correlating with amygdala volume loss has been reported (Kipps, Duggins, McCusker, & Calder, 2007). On the other hand, the thalamus, subthalamus, and substantia nigra undergo cell loss, shrinkage, and astrogliosis as well but at lower levels. About 40% of GABAergic and dopaminergic neurons of the substantia nigra are lost, which could contribute to the akinesia symptoms seen in both HD and PD (Vonsattel, 2008).

7.1 Structural abnormalities of the basal ganglia in Alzheimer's disease

The prevalence of AD undoubtedly makes it a crucial disease to find a cure for; however, much of the research to date has focused primarily on the hippocampus and the neocortex. The basal ganglia have received little attention in this context, despite evidence of degeneration of structures in the deep gray matter which may contribute to AD symptoms (de Jong et al., 2008). Early evidence from Braak and Braak (1991b) indicated that, in cases of severe AD, there are large numbers of extracellular amyloid-β deposits in all thalamic nuclei and severe changes in the limbic nuclei due to the presence of neurofibrillary tangles, indicating an impairment of the limbic circuits (Braak & Braak, 1991a).

Continuing research has consistently demonstrated that AD patients have shape changes in the basal ganglia, aided by recent advances in brain imaging techniques. Patients exhibit more rapid shape changes in the thalamus and basal ganglia compared with age-matched controls (Cho et al., 2014); the medial heads on the caudate nucleus and ventral lateral putamen are selectively degenerated (de Jong et al., 2011); and MRI indicates that volumes of caudate nucleus (Almeida et al., 2003), putamen, and thalamus (de Jong et al., 2011) are reduced in AD patients. Moreover, in all cases, the structural changes were correlated with cognitive decline. Thus, it is clear that, contrary to the standard declaration that the basal ganglia and associated subcortical structures are spared in AD, these regions are susceptible to structural changes during disease progression and may well underlie aspects of cognitive impairment.

The striatum, comprising of the nucleus accumbens, caudate nucleus, and putamen, appears to be particularly vulnerable in AD. Indeed, both diffuse plaques and neurofibrillary tangles are observed in striatal areas and amyloid-β deposits are found in the striatum in young, asymptomatic members of families with PSEN1 mutations (Klunk et al., 2007). Additionally, amyloid-β levels may increase in the basal ganglia over 10 years before expected symptom onset (Bateman et al., 2012), suggesting that it may be a brain region affected early in the disease and perhaps an underexplored therapeutic target.

As the basal ganglia are believed to drive higher level cognitive processes such as attention and memory, it is perhaps unsurprising that degeneration and shape changes would affect cognitive function in AD. Little, however, is known about this, as robust studies correlating alterations in subcortical volumes in AD with cognitive function are lacking.

7.2 Iron-associated basal ganglia dysfunction in Alzheimer's disease

Metabolic processing is known to dysregulate with age and is particularly affected in neurodegeneration. The basal ganglia (in particular the putamen, caudate nucleus, and globus pallidus) have a high metabolic rate and contain the highest levels of iron in the brain (Ramos et al., 2014), which may render them particularly sensitive to age-associated metabolic dysregulation. Even in the healthy aging brain, iron accumulates in cells leading to a region-specific increase in total iron levels (Connor, Menzies, St Martin, & Mufson, 1990). As aging represents the most significant risk factor for neurodegenerative diseases, iron may be involved as either a cause or a consequence. This is evidenced by studies showing that iron accumulation in the basal ganglia is a common feature of many neurodegenerative diseases and results in both cognitive and motor dysfunction. For example, iron has been found to co-localize with the insoluble protein aggregates that characterize neurodegenerative diseases such as AD, PD, HD, and DLB (Agrawal, Fox, Thyagarajan, & Fox, 2018; Joppe, Roser, Maass, & Lingor, 2019; van Bergen et al., 2016), suggesting a dysfunction in iron trafficking or metabolism as a cause or consequence of pathology. The heavy iron load in the basal ganglia may explain its increased susceptibility to neurodegeneration, so this represents an important brain region to further examine in this context.

Certain disorders are characterized by iron dysregulation (Dusek, Jankovic, & Le, 2012) but the impact of this in AD is not fully understood. An early study of the basal ganglia in AD revealed that patients who have basal ganglia mineralization have a higher rate of psychotic and motor disturbances, such as delusions, depression, parkinsonism and myoclonus, compared with age- and gender-matched AD patients without mineralization in the basal ganglia (Forstl, Burns, Cairns, Luthert, & Levy, 1992). Autopsy studies have shown that the highest concentrations of iron in the brain exist in the basal ganglia including the globus pallidus, putamen, and substantia nigra (Conway et al., 1998). Amyloid-β deposition and aggregation has been shown to be accelerated by metal ions such as iron and zinc (Lovell, Robertson, Teesdale, Campbell, & Markesbery, 1998) and iron has been demonstrated in vitro to increase the aggregation and cytotoxicity of amyloid-β peptide (Galante, Cavallo, Perico, & D'Arrigo, 2018). Iron also appears to be involved with plaques and tangles, the two hallmark pathologies of AD. Amyloid-β plaques co-localize with iron (van Bergen et al., 2016), suggesting that iron may either exacerbate amyloid-β aggregation or that the plaques may trap iron in some way. In mouse models, iron has been demonstrated to correlate with amyloid-β plaque morphology, with evidence for the formation of an iron–amyloid complex (Telling et al., 2017). Iron has also been found to bind to tau protein, inducing phosphorylation and aggregation (Liu, Fan, Yang, Wang, & Guo, 2018), thereby promoting the accumulation of hyperphosphorylated tau which leads to neurofibrillary tangles. Neuropathological studies appear to support this idea as iron levels in eight regions of the AD brain, including the globus pallidus and caudate nucleus, are raised compared to healthy controls, implicating the basal ganglia as part of the pathological profile of AD (Tao, Wang, Rogers, & Wang, 2014). It appears iron metabolism is impaired in some way in AD and other neurodegenerative diseases but how this relates to the basal ganglia or precisely to disease progression requires further studies. An improved understanding of this would aid the development of treatment strategies, such as developing iron chelators to sequester excess iron in the brain.

8.1 Pathology

Lewy bodies have been linked to cell death and the extent of Lewy body pathology correlates with clinical pathology in both DLB (Tiraboschi et al., 2015) and PD with dementia (Hurtig et al., 2000). Lewy body pathology is found widespread in postmortem brains from both groups of patients, especially affecting the frontal cortex, midbrain and brainstem nuclei, dorsal efferent nucleus of the vagus nerve, basal forebrain nuclei, and limbic regions (Jellinger & Korczyn, 2018; Neef & Walling, 2006). This widespread pathology makes the spreading of the disorder more difficult to define than in PD, as it does not appear to extend in a rostro-caudal manner (Walker et al., 2015), although systems to stage the disorder have been proposed (McKeith et al., 2005).

Gray matter atrophy occurs in frontotemporal, occipital and parietal cortical areas and subcortical atrophy of amygdala, caudate, putamen, and pallidum (Almeida et al., 2003; Beyer, Domingo-Sabat, & Ariza, 2009; Gazzina et al., 2016; Sanchez-Castaneda et al., 2009; Watson, O'Brien, Barber, & Blamire, 2012). Like PD, dopaminergic and cholinergic neurons appear especially vulnerable in the disease and in some cases have been found to be more degenerated than in PD (Colloby, McParland, O'Brien, & Attems, 2012; Muenter et al., 1998; Outeiro et al., 2019; Thomas et al., 2017). Dopamine synthesis, transporter binding, and storage are all decreased in patients with DLB (Gomperts et al., 2016; Marquie et al., 2014; Nikolaus, Antke, & Muller, 2009). Cholinergic deficits in the cortex are observed due to the loss of neurons in the basal forebrain, leading to cholinergic denervation and signaling deficits (Grothe et al., 2014; Mazere, Lamare, Allard, Fernandez, & Mayo, 2017).

Cell death due to an impairment of axonal transport, autophagy–lysosomal, ubiquitin–proteasome, and mitochondrial pathways have all been strongly implicated in the disease (Manecka, Vanderperre, Fon, & Durcan, 2017; Power, Barnes, & Chegini, 2017; Sato et al., 2018; Volpicelli-Daley, 2017). Importantly, whether this is directly due to Lewy bodies is unclear, as they may play a role in protecting the cell through the uptake of misfolded and nonfunctional proteins (Beyer et al., 2009), as there are cases of individuals with a significant Lewy body pathology at autopsy that do not demonstrate symptoms of dementia (Parkkinen, Pirttila, & Alafuzoff, 2008). Oligomers of α-synuclein may also mediate the cellular toxicity in the disease (Outeiro et al., 2019). An additional factor that may play a role in cell death is the co-occurrence of AD pathology in patients with DLB, which is suggested to be an increasingly important part of the disorder rather than a comorbidity (Deramecourt et al., 2006; Jellinger & Korczyn, 2018). This may lead to synergistic effects and increased cell death. Indeed, AD and DLB co-pathologies have been shown to lead to a greater temporal lobe atrophy (van der Zande et al., 2018) and greater baseline amyloid-β imaging predicts more severe neurodegeneration in DLB patients (Sarro et al., 2016).