Striatal neurotransmission
The major neuropathology of PD involves degeneration of dopaminergic neurons in the SNpc, which subsequently leads to loss of DA signal input in the striatum. DA is a key messenger in striatal neurotransmission that is primarily involved in the control of voluntary movement [
120,
121]. Thus, DA depletion and dysregulated striatal neurotransmission lead to the debilitating motor symptoms of PD patients. Increasing evidence indicates that LRRK2 may be involved in modulating the striatal neural network. LRRK2 is expressed in both striatal axons and dendrites [
103,
122] and its striatal expression level rises dramatically throughout the postnatal brain development period [
11,
60], pointing to its functional relevance in the central nervous system.
In contrast to transgenic mice which exhibit PD phenotypes, the first
LRRK2 KI mouse model generated harbouring the R1441C mutation within the Roc domain displays grossly normal DA neuronal morphology and projections even after aging [
93,
123]. However, upon stimulation with amphetamine (AMPH), a drug inducing synaptic DA release, R1441C KI mice lack the AMPH-induced increase in locomotor activity that is present in WT animals [
123,
124], suggesting a potential defect in the DA system. Similarly, we developed mice with homozygous LRRK2 R1441G KI mutation, which exhibit perturbed DA homeostasis at as early as 3 months of age, as evidenced by significantly reduced synaptosomal DA uptake compared to that of WT controls after reserpine treatment [
125]. Reserpine is a drug that irreversibly blocks vesicular monoamine transporter-2 (VMAT2). Given that the expression levels of both VMAT2 and dopamine transporter (DAT) in the striatum are similar between R1441G and WT animals, the lower DA uptake after reserpine treatment in the KI mice indicates an altered DAT function and increased susceptibility to DA depletion that may reflect the earliest presynaptic dysfunctions in PD mediated by pathogenic LRRK2. These R1441G KI mice also show greater impairment in locomotor activity induced by reserpine, with much slower motor recovery, evidencing the increased vulnerability to DA stress. These observations are important, because even without robust DA neuronal loss, the dysfunctional DA system is still a key feature of prodromal PD. While studies are available for both R1441C and R1441G KI models which have amino acid changes in the same R1441 residue (-C/G), those that explore common phenotypes in each mouse line are lacking. Moreover, the development of independent genetic lines with dissimilar genetic background in different laboratories renders the comparison of models rather difficult. Nevertheless, since both R1441C and R1441G KI mouse models show altered DA system, it is reasonable to expect that these two mutations—occurring at the same residue within the Roc domain—may share similar pathogenic mechanisms.
In accordance with mutations causing changes in the Roc-GTPase domain (R1441C/G) that mediate alterations in the striatal DA system, the G2019S variant, which causes an amino acid change in the kinase domain, also exhibits pathogenicity in DA synaptic function. The G2019S KI mice display an age-dependent reduction in basal DA levels, which is absent in WT mice [
126]. Challenging the G2019S KI mice with AMPH fails to induce DA release. In fact, the amount of DA release is reduced, which may be attributed to impaired DA packaging into vesicles or impaired DA exocytosis given the normal DA metabolite levels and reverse transport of cytosolic DA. Moreover, these G2019S KI mice display an increased homovanillic acid/DA turnover ratio (DA synthesis and metabolism ratio), consistent with an enhanced striatal DA turnover observed in the early stages of sporadic PD patients [
127]. The increased DA turnover is thought to be a compensatory mechanism against dopaminergic neurodegeneration, which may possibly explain the long prodromal phase before the onset of motor symptoms when the majority of DA neurons have died or are dysfunctional. In contrast to this study, another group reported no change in the basal striatal DA level in
LRRK2-G2019S KI mice [
128]. Interestingly, unlike our R1441G mice with no alterations in the levels of synaptic proteins [
125], these G2019S mice exhibit an age-dependent upregulation of DAT and downregulation of VMAT2, both of which are key regulators of DA homeostasis [
128]. This also suggests that different
LRRK2 mutations may exert different influences on DA synaptic proteins. The authors who produced the G2019S KI mice hypothesized that such alterations in synaptic protein levels contribute collectively to the perturbed neurotransmission
via upregulation of DAT, which results in increased oxidative stress generated by DA autooxidation and subsequent neuronal death [
129,
130]. Similar pathogenic events also occur in PD patients with reduced VMAT2 levels [
131].
In addition to the role of LRRK2 in presynaptic neuronal termini,
LRRK2 mutation also affects postsynaptic function in spiny projection neurons (SPN) that populate the striatum [
118]. The classical simplified model of neuronal circuitry governing movement control divides SPN into two populations based on their innervation source and projection targets: direct-pathway SPN (dSPN) "promoting movement” and indirect-pathway SPN (iSPN) “suppressing movement” [
118,
132]. DA signalling has opposing effects on the activities of the two SPN populations: it increases the dSPN activity while decreasing the activities of iSPN. The balance between the two pathways receiving dopaminergic input controls motor coordination, and the net output is to promote movement. In PD, SPN loses their dopamine input from the SNpc, which causes an imbalance between the two pathways and eventually results in the characteristic motor symptoms [
133]. Such perturbation in synaptic transmission in SPN has been observed in both R1441C and G2019S KI mice, an impairment mediated by abnormal reorganization of synaptic receptor proteins in dSPN [
134]. In PD patients, disrupted signalling in SPN has been associated with hypokinetic symptoms [
135], suggesting a role of LRRK2 in modulating SPN neurotransmission. Interestingly, LRRK2-mediated defects in SPN are stronger in R1441C than in G2019S mice, indicating increased pathogenicity conferred by the GTPase-mutant R1441C [
134]. For instance, a recent study on both R1441C and G2019S KI mice has shown decreased excitability of iSPN that is associated with impaired motor learning only in R1441C mice but not in G2019S mice [
136]. Similarly, R1441C KI mice show an aberrant increase in synaptic PKA activity in dSPN, which is not altered in G2019S mice [
124]. It is known that the altered PKA activity results in abnormal synaptogenesis in the developing SPN, so the aberrant PKA activity in R1441C mice suggests a pathogenic role of mutant
LRRK2 in early-onset dysfunction of SPN. These may help to explain the higher PD penetrance rate in R1441C/G/H patients compared with that of G2019S carriers [
73].
Dopamine D2 receptors (D2R) are postsynaptic proteins highly expressed at the postsynaptic termini of iSPN and regulate DA levels in the synaptic cleft. Both R1441C and G2019S KI mice failed to respond to quinpirole (a selective D2 receptor agonist that reduces locomotor activity), implying that the two mutant variants of LRRK2 desensitize D2R [
123,
137,
138], possibly through a common pathway that is yet to be discovered. With the desensitization of D2R, the use of antagonist and agonist radioligands of dopamine D2R for functional molecular positron emission tomography (PET) imaging in human patients [
139] may help reveal early D2R dysfunction before the occurrence of significant neurodegeneration prior to PD diagnosis.
Altogether, considerable evidence suggests that the majority of LRRK2 variants alter not only presynaptic but also postsynaptic striatal neuronal activities, compromising the functions of striatal SPNs and subsequent neural outputs that control movement. Interestingly, while R1441C/G and G2019S LRRK2 mutations confer similar presynaptic alterations in response to AMPH challenge, the reports on postsynaptic phenotypes show greater pathogenicity conferred by R1441C over G2019S. Although there is no overt neurodegeneration in SNpc of LRRK2 KI mice, perturbations in nigro-striatal neurotransmission pathways observed in KI models may reflect early synaptic dysfunction leading to PD in humans, suggesting potential therapeutic benefits of targeting pre- and post-synaptic functions prior to the disease onset.
Locomotor symptoms
Mice are advantageous over other animals in modelling PD, as in mouse models degeneration in the nigrostriatal system correlates with motor impairments that can be easily assessed by various behavioural tests [
140]. However, there is no PD mouse model that displays both genuine parkinsonian traits and neuropathological changes thus far. PD mouse models induced by neurotoxins such as 1-methyl-4-phenyl-1,2,3,6,-tetrahydropyridine (MPTP), rotenone and paraquat show motor dysfunction as a result of neurodegeneration induced by specific toxic insults to the SNpc. Nevertheless, these toxin-based models of PD do not fully recapitulate neuropathology of PD, as they usually lack LB pathology [
141] and only display an acute neuronal loss in contrast to the progressive neurodegeneration over a long period of time in PD patients.
LRRK2 transgenic mouse models overexpressing pathogenic variants such as R1441C or G2019S exhibit DA neuronal loss, which leads to levodopa-responsive locomotor dysfunction [
36,
110‐
112], indicating a dose effect of
LRRK2 pathogenic gene on the DA system.
In fact, behavioural studies in
LRRK2 KI mouse models have revealed a modest locomotor impairment that may correlate with subtle motor changes occurring in the prodromal phase of PD patients. For instance, we previously investigated the combined effects of gene mutation (
LRRK2-R1441G), environment (rotenone, a pesticide and mitochondrial toxin) and aging, by oral gavage of low-dose rotenone in the R1441G KI mice for approximately half of their lifespan, and observed an increased locomotor deficit that was absent in age-matched WT mice [
142]. This may serve as a novel paradigm in PD mouse model research to demonstrate an interplay of mutant
LRRK2 with other causative factors that can result in motor impairments resembling PD symptoms.
Despite the absence of severe motor dysfunction regardless of aging [
143], there is a subtle but significant reduction in locomotion in the aged R1441C KI mice in more complex tasks such as the Vertical Pole test and the Ladder-and Beam Walk test. These motor tests reflect perturbation of the nigrostriatal system [
144] by measuring fine changes in motor activities of mice. Interestingly, one study reported a hyperkinetic phenotype in young
LRRK2-G2019S KI mice, which is correlated with the reported increase in DA release upon stimulation [
145]. Such hyperkinetic behaviour, however, decreases with age, mirroring the subsequent reduction in the extracellular DA level [
126,
137]. The authors postulated that the hyperactivity in early age may reflect a compensatory mechanism prior to nigrostriatal system impairments [
145]. Nevertheless, how the different
LRRK2 mutations contribute to early changes in DA system and subsequent motor changes in both mouse models and PD patients is still unclear and requires further research.
Thus far, several lines of evidence from KI mouse models have implicated a potential role of pathogenic LRRK2 in producing subtle but progressive motor deficits with age and signs of olfactory dysfunction. Future studies investigating the LRRK2-relevant behavioural changes in KI mice may aim to use a panel of common tests on different aspects of PD motor and non-motor symptoms: bar and drag test, stepping test or pole test for measuring akinesia/bradykinesia; rotarod test for measuring balance, strength and coordination; tail-suspension test or forced swim test for assessing depression and behavioural despair [
140].
Mitochondrial dysfunction
Mitochondria play a critical role in neuronal energy homeostasis. Dysregulated mitochondrial homeostasis leads to disrupted bioenergetics in nigrostriatal DA neurons that are particularly vulnerable to dysfunction and degeneration [
139,
150]. Mitochondrial dysfunction is one of the key pathogenic features of both familial and idiopathic PD [
151,
152], and is associated with ATP deficiency and nigrostriatal DA neurodegeneration. Post-mortem PD brains show reduced mitochondrial Complex I activity [
153,
154]. Moreover, pharmacological inhibition of mitochondrial Complex I by neurotoxins such as MPTP leads to selective degeneration of DA neurons in mice [
155]. Genetic studies also revealed that mutations of various mitochondrial genes such
PRKN,
PINK1 and
DJ-1 are causative factors for PD [
156], further suggesting mitochondrial dysfunction as one of the pathogenic events in PD.
LRRK2 under physiological conditions is mainly localized in cytoplasm, but is also found in nucleus and mitochondria. Its pathogenic mutations have been increasingly linked with mitochondrial dysfunction in PD [
151,
157]. In heterozygous and homozygous
LRRK2-G2019S KI models, mitochondrial morphological abnormalities in aged striatum occur in a gene dose- and age-dependent manner [
126]. The abnormal shape resembles beads-on-the-string morphology, which possibly denotes defects in fission and fusion [
158], suggesting that the LRRK2 variant may impair mitochondrial fission—a key process for maintaining healthy mitochondrial network. Interestingly, alterations in mitochondrial shape are not limited to striatum but also occur in other brain regions such as cortex and to a lesser extent the hippocampus, indicating that the LRRK2 variant-driven changes in mitochondrial morphology may be tissue-specific. Similarly, we recently reported abnormal mitochondrial morphology denoting impaired fission and disrupted clearance of damaged mitochondria in the striatum of aged
LRRK2-R1441G KI mice [
159]. Most importantly, our previous work also revealed that the R1441G KI mice receiving low doses of rotenone for over half of their lifespan to mimic the effect of environmental influence over time in PD, show Complex I deficiency [
125]. Such defects were absent in the WT controls receiving the same treatment. Therefore, the study results suggest combined effects of LRRK2 variant and other risk factors—aging and environment toxins—that result in mitochondrial dysfunction, leading to PD.
Collectively, LRRK2 KI mice, despite having different mutations, display pathologies on mitochondrial function and morphology in vivo, suggesting the importance of physiological LRRK2 in maintaining mitochondria homeostasis and DA neuronal survival. Therefore, targeting pathogenic LRRK2 to mitigate mitochondrial dysfunction appears to be a reasonable option for delaying neurodegeneration and thereby modifying the course of the disease. This also highlights LRRK2 KI mice as a suitable LRRK2-PD model system for studying mitochondrial dysfunction in PD. Further work is required to pinpoint the exact mechanism of LRRK2-mediated defects, including whether or not the kinase hyperactivity is involved.
Defects in autophagy and endo-lysosomal pathways
An orchestrated network of endo-lysosomal and autophagic pathways is crucial for maintaining protein homeostasis and turnover in damaged organelles in the brain. A defective network in humans may trigger pathologies similar to PD. Evidence suggests that dysfunction of these pathways is associated with LRRK2 mutations, such as abnormally high lysosomal pH in
LRRK2-G2019S KI mice-derived cortical neurons, impaired clearance of autophagosomes and defective autophagosome-lysosome fusion in R1441C-overexpressing neurons [
36,
160,
161]. Furthermore, we have shown that aged mutant
LRRK2-R1441G KI mice display accumulation of LAMP2A and HSPA8/HSC70 proteins in the striatum [
162], suggesting defective chaperone-mediated autophagy, a specific route for clearance of various proteins such as α-synuclein
via LAMP2A-mediated lysosomal degradation [
162,
163]. We have also demonstrated impaired clearance of defective mitochondria in
LRRK2-R1441G KI mouse embryonic fibroblasts (MEFs), associated with abnormal clustering of mitochondria and impaired DRP1-ERK signalling under mitochondrial stress [
159]. Similarly,
LRRK2-G2019S KI mice show accumulation of LAMP2A and downregulation of a number of key proteins in the autophagy-lysosome pathway, including LAMP2, mTOR, TFEB and GBA1 [
164], further confirming the involvement of LRRK2 in autophagic pathways.
Although the detailed pathogenic mechanisms of
LRRK2 variants in PD are still unclear, hyperactive kinase activity appears to be an important disease marker of both LRRK2 and non-LRRK2 PD.
LRRK2-G2019S KI mice with hyperactive LRRK2 show impaired mitophagy, which is rescued with therapeutic LRRK2 kinase inhibition in MEFs derived from the same mouse as a proof of concept [
160,
165]. Another study demonstrated that the hyper-kinase activity in G2019S KI mice impairs neuronal autophagy by specifically modulating axonal transport of the autophagosomes in a kinase-dependent manner [
166]. Such impairment is associated with an increased level of motor adaptor protein JNK interacting protein 4 recruited on the autophagosomes, and hyperactive variant of LRRK2 was proposed to abnormally activate the motor protein kinesin, which delays autophagosome transport, eventually resulting in an abnormal accumulation of protein aggregates. This is also reflected by the accumulation of α-synuclein deposits in the Lewy neurites of neuronal axons where autophagosomal transport actively occurs [
167]. This demonstrates the suitability of mutant
LRRK2 KI mouse model in recapitulating the axonal pathology preceding axonal degeneration in PD [
168]. Overall, different
LRRK2 KI mouse studies have provided robust evidence of impaired autophagy-lysosomal system, reflecting similar defects in PD.
Synucleinopathy
α-Synucleinopathies are a collection of diseases characterized by abnormal accumulation of α-synuclein aggregates in neurons, nerve fibres or glial cells, including PD, dementia with Lewy bodies, multiple system atrophy and rare conditions including inherited metabolic disorders, e.g., Gaucher's disease [
169]. In PD, the collapse of intracellular transport and degradation system for damaged organelles, together with the spontaneous aggregation of certain proteins, leads to accumulation of different pathogenic protein entities formed primarily from α-synuclein and tau [
170,
171]. Autopsy examinations have revealed that the majority of
LRRK2-PD patients present with LB-like α-synuclein pathology. Immunohistochemical analysis of brain samples of PD patients revealed that a significant portion of α-synuclein-positive LBs contained LRRK2 [
172], further supporting the association between LRRK2 and α-synuclein. Approximately 24% of LBs in idiopathic patients were LRRK2-positive, whereas in confirmed
LRRK2-PD patients with G2019S mutation, the proportion of LRRK2-reactive LBs increased to 50% [
173], implicating a pathogenic role of LRRK2 in synucleinopathies. α-Synuclein is known to interact with some of the endocytic kinase substrates of LRRK2, such as Rab3a, Rab8 and Rab35 [
174,
175], suggesting that LRRK2 and α-synuclein share similar pathogenic pathways
via pathogenic hyper-phosphorylation of these Rabs, which may lead to mishandling of α-synuclein.
LRRK2 KI models have been widely used to study the potential interaction between LRRK2 and α-synuclein in the development of synucleinopathy in PD. Varying severities of α-synuclein pathology have been reported in
LRRK2 KI mice, possibly due to the intrinsic variability of protein aggregation. For example,
LRRK2-R1441C KI mice do not show increased aggregation of proteins including α-synuclein, tau and ubiquitin [
123]. Similarly, no α-synuclein or tau pathology is found in the striatum of
LRRK2-R1441G KI mice [
125]. However, we have recently shown that the striatum of
LRRK2-R1441G KI mice displays more severe age-dependent accumulation of α-synuclein oligomers than that in age-matched WT littermates [
162]. These oligomers represent the toxic species that induce neuronal death [
176].
In contrast to R1441C/G KI mice [
123], the mice with G2019S mutation show more varied pathologies. One study showed that the
LRRK2-G2019S mice have increased tau phosphorylation correlated with positive puncta of phosphorylated tau, but without α-synuclein pathology [
126], while others have reported age-dependent elevations of phosphorylated α-synuclein and α-synuclein inclusions in the striatum of G2019S KI mice [
128]. A potential modulatory role of LRRK2 in fibril-induced α-synuclein aggregation has been shown by comparing primary cortical neuronal cultures derived from G2019S KI mice with those from
Lrrk2 KO mice [
177]. The G2019S KI neurons show increased vulnerability to α-synuclein accumulation triggered by fibrils, whereas the
LRRK2 KO neurons showed robust resistance to aggregation [
177].
While the molecular interaction between LRRK2 and α-synuclein is still unresolved, studies have investigated whether the pathogenicity is mediated
via the hyper-kinase activity conferred by the pathogenic
LRRK2 mutations. Interestingly, primary neurons derived from G2019S KI mice have impaired α-synuclein metabolism, which is reversed by pharmacological inhibition of LRRK2 kinase activity [
160]. Similarly, multiple human induced pluripotent stem cell-derived neural lines derived from fibroblasts of G2019S PD patients show increased accumulation of α-synuclein, which can be reduced with LRRK2 inhibitors [
178]. The authors postulated that the increased accumulation of α-synuclein could be attributed to mutant LRRK2 that negatively affects the autophagy-lysosomal system. Administration of a LRRK2 kinase inhibitor into α-synuclein transgenic mice also significantly decreases the α-synuclein accumulation in several regions of the brain, including neocortex and striatum [
175], further confirming the role of LRRK2 hyperactivity in synucleinopathy.
Although a growing body of evidence suggests that the α-synuclein accumulation may be mediated by the hyper-kinase activity of mutant LRRK2, contradictory results have been reported. Hippocampal neurons derived from a transgenic BAC mouse expressing the G2019S variant protein show a mild but reversible elevation in α-synuclein pathology [
179]. The same neuronal cultures treated with α-synuclein pre-formed fibrils which resulted in the development of α-synuclein pathology showed no response to LRRK2 kinase inhibitor treatment. A more recent study conducted by the same group showed that neurons derived from
LRRK2-G2019S transgenic mice exhibit a robust tau pathology that is also insensitive to the treatment with a LRRK2 kinase inhibitor [
180], adding to the complexity of the interaction between LRRK2 and pathological proteins of PD. It should be noted, however, that α-synuclein, being a dynamic protein in nature, may develop into pathological forms
via different mechanisms, which may or may not interact with LRRK2 for further pathogenicity.
In mouse models,
LRRK2 mutations alone have produced mild-to-no signs of severe synucleinopathy. The pathogenicity of LRRK2 variants may be enhanced in the presence of other factors such as aging and PD-causing α-synuclein mutations. For instance, one study performed a stereotactic injection of AAV2/9 vectors expressing a human A53T α-synuclein (a pathogenic PD-causing variant which is aggregation-prone) into the striatum of both young and aged
LRRK2-G2019S KI mice and WT mice [
181,
182]. This resulted in greater nigral degeneration and pSer129 α-synuclein aggregates in the aged
LRRK2-G2019S KI mice compared to their age-matched WT littermate controls [
183]. These demonstrated an interplay between pathogenic LRRK2, mutant α-synuclein and aging in accelerating development of α-synuclein pathology implicated in PD. Similarly, A53T α-synuclein transgenic mice overexpressing
LRRK2 display degeneration of DA neurons and aggregation of α-synuclein [
96], which are absent in transgenic mice overexpressing LRRK2 alone. These findings clearly show that cross-breeding of
LRRK2 KI mice with mice carrying other genetic risks facilitates a deeper elucidation of the interplay of LRRK2 with other PD factors that may speed up the development of α-synuclein pathology. Therefore,
LRRK2 KI mice represent a good
in vivo model to observe different stages of synucleinopathy under the co-influence of LRRK2 and secondary factors such as aging and genetics.