Introduction
Autosomal-dominant missense mutations in the leucine-rich repeat kinase 2 (
LRRK2) gene (PARK8, OMIM 609007) cause familial late-onset Parkinson’s disease (PD) [
59,
95].
LRRK2 mutations also occur in 1–2% of sporadic cases [
28,
70] and recent genome wide association studies (GWAS) showed that common variations in the
LRRK2 locus increase the risk of disease, pointing to a crucial role of LRRK2 in the pathogenesis of PD. LRRK2-associated PD is clinically and pathologically indistinguishable from the idiopathic form [
18], although some differences in motor and non-motor features have been reported [
46]. The majority of LRRK2 autoptic cases report progressive degeneration of dopamine (DA) neurons located in the substantia nigra pars compacta (SNpc), and α-synuclein (α-syn)/ubiquitin-positive intraneuronal cytoplasmic inclusions in surviving neurons [
31,
33], although a pleomorphic pathology associated with other neurodegenerative diseases has also been observed [
95]. Despite the undisputed genetic link between LRRK2 mutations and PD, the pathogenic mechanisms through which LRRK2 mutations affect PD onset and progression remain debated [
17,
49]. LRRK2 is a complex multi-domain protein belonging to the ROCO family, characterized by the presence of a GTPase and a serine-threonine kinase domain surrounded by a number of protein-protein interaction domains [
16,
51]. The most common LRRK2 pathogenic mutations are represented by Gly2019Ser (G2019S) in the kinase domain, followed by the hotspot mutation Arg1441Cys/Gly/His/Ser (R1441C/G/H/S) in the GTPase domain [
19,
73]. The G2019S mutation results in a two to threefold increase in LRRK2 kinase activity, which appears to be crucial for LRRK2-induced neurodegeneration in vitro [
26,
90,
91]. More recently, the cellular activity of LRRK2, probed with anti-autophosphorylation antibodies against Serine 1292 [
67,
72] and by measuring the phosphorylation of a subset of Rab GTPase which are
bona fide LRRK2 cellular substrates [
76], revealed a homogeneous increase of LRRK2 kinase activity in the presence of pathogenic mutations, which is not limited to the G2019S mutant as it occurs in vitro.
Various LRRK2 rodent models have been generated in the attempt to replicate the dysfunction and/or degeneration of the nigro-striatal dopaminergic pathway in vivo. Unfortunately, these models provided conflicting data. Mice overexpressing human G2019S or R1441C/G mutations through BAC technology did not show overt dopaminergic neurodegeneration [
39,
40,
53] but reduced striatal DA content or basal extracellular levels in vivo when compared to non-transgenic wild-type controls [
4,
53]. Consistently, the K
+-evoked DA release was reduced in striatal slices from BAC hG2019S mice [
40]. In mice where hG29019S [
13,
64] or hR1441C [
88] overexpression in SNc was achieved through the CMV/PDGF promoter, an 18–50% reduction in the number of nigral DA cells was observed at old ages (16–21 months). In these mice, no changes of in vivo DA content was observed [
64], although the K
+-evoked DA release from striatal slices was reduced [
88]. Conditional expression of hG2019S [
41] or hR1441C [
83] in SNc also did not cause nigral DA neuron loss; only a mild reduction in the density of TH terminals was observed in 16-month-old mice [
41]. In these mice, hG2019S overexpression caused a reduction of DA content and release from striatal slices [
41]. Lack of nigro-striatal degeneration [
37,
74,
93] or changes in DA content [
37,
93] were also confirmed in transgenic rats overexpressing hG2019S or hR1441C mutations. In vitro, a reduction of the K
+-evoked DA release in BAC overexpressing rats was found [
74]. Finally, no overt neurodegeneration [
29,
82,
92] or changes in striatal DA content [
29,
82] were observed in G2019S or R1441C knock-in (KI) mice, although in vivo microdialysis revealed a 60% reduction in both spontaneous and amphetamine-induced DA release in 12-month-old G2019S KI mice [
92].
In a previous longitudinal study, we reported that G2019S KI mice had enhanced motor behavior compared to both WT mice and mice carrying the D1994S kinase-dead mutation [
43]. In this follow-up study, we sought to investigate the mechanisms underlying such phenotype, and in particular, whether G2019S LRRK2 is associated with dysregulation of nigro-striatal DA transmission. Indeed, in vivo [
93] and in vitro [
54] evidence that the G2019S mutation can be associated with increased DA release has been presented. Here, different aspects of striatal DA transmission were evaluated, namely the integrity of the nigro-striatal DA pathway, in vivo and in vitro striatal DA release, expression and function of proteins involved in synaptic load (DA transporter, DAT) or vesicle storage (vesicular monoamine transporter type 2; VMAT2) of DA, and, finally, the levels of endogenous α-syn, and its Serine129 phosphorylated (pSer129 α-syn) or 3,4-dihydroxyphenylacetaldehyde (DOPAL)-bound forms, which are considered markers of synaptic damage.
Discussion
We previously reported that G2019S KI mice have a hyperkinetic phenotype relying on elevated kinase activity [
43]. In this follow-up, we show that 12-month-old G2019S KI and WT mice bear similar numbers of nigral DA neurons and striatal DA terminals, in keeping with previous studies in the same genotype [
29,
92] or in BAC G2019S overexpressing mice [
40,
53] and rats [
37,
74,
93], as well as similar extracellular DA levels and depolarization-evoked striatal DA release, in line with that found in 22-month old R1441G KI mice [
82]. Since the LRRK2 kinase inhibitor Nov-LRRK2-11 [
43] also failed to affect striatal DA release in vivo in any genotypes, we conclude that the nigro-striatal DA system is morphologically intact and the exocytotic properties of DA neurons are functionally preserved in G2019S LRRK2 carriers.
The lack of changes of striatal DA release is at striking variance with the 60% reduction in basal striatal extracellular DA levels reported, in the absence of motor phenotype change, in 12-month-old G2019S KI mice [
92]. We cannot easily explain this difference since both strains of G2019S KI mice are backcrossed on C57BL and bear similar kinase-enhancing mutations on exon 41 [
29,
92]. Indeed, kinase activity appears to be elevated in both strains, as evaluated by in vitro kinase assays on synthetic substrates [
29,
92]. We confirmed this finding in vivo, showing that, in good agreement with previous work on brain lysates of BAC G2019S mice [
72], pSer1292 levels were ~8-fold higher in the striatum of G2019S KI mice compared to WT controls. pSer1292 appears a more reliable marker of kinase activity with respect to ATP γ-phosphate incorporation measured in in vitro assays. Indeed, Ser1292 LRRK2 is an autophosphorylation site, and pSer1292 levels correlate with in vivo kinase activity [
72]. pSer1292 levels are a more reliable readout of in vivo LRRK2 kinase activity even compared to pSer935 levels, since LRRK2 is phosphorylated at Ser935 by other kinases [
14,
20,
67]. The discrepancies between these two strains of G2019S KI mice might be explained by quantitative differences in kinase activity along with inter-individual genomic variability, motor tests used, or environmental conditions.
Despite the lack of changes of DA release, the levels and functions of proteins involved in DA synaptic load (DAT) and vesicular storage (VMAT2) were altered in 12-month-old G2019S KI mice. Strikingly, these changes were age-dependent, since they were not observed in 3-month-old animals, indicating these changes are elements of an orchestrated, progressive response relying on the interaction between a genetic factor (the G2019S mutation) and aging, i.e. the main risk factor in PD.
DAT was upregulated in G2019S KI mice, which might represent a vulnerability factors for DA neurons [
56]. Indeed, DAT overexpression has been associated with an increase of oxidative stress and neuronal degeneration [
50] likely because cytosolic DA accumulation causes the buildup of reactive oxygen species and quinones, generated by DA autoxidation [
25,
77,
79]. Moreover, cytosolic DA is metabolized by monoamine oxidase A to DOPAL, which causes synaptic dysfunction and terminal loss acting via different mechanisms, including cross-linking with α-syn [
9]. Finally, environmental toxins causing PD, such as the toxic metabolite of MPTP, MPP
+, are taken up by DA neurons through DAT. In fact, the greater susceptibility of BAC hG2019S overexpressing mice to the parkinsonian toxin MPTP can be explained by DAT upregulation [
32].
Quite paradoxically, the increase of DAT activity was associated with a blunted neurochemical and behavioral response to GBR-12783. This is consistent with microdialysis works reporting a 35–50% reduction of nomifensine-induced DA release in the striatum of KI mice constitutively expressing R1441G LRRK2 [
39] or temporally expressing G2019S LRRK2 [
93]. Previous studies in cells have shown that DAT expression levels inversely correlate with the potency of DAT blockers [
12], a phenomenon also observed for the serotonin transporter [
65]. Membrane DAT is in equilibrium between oligomeric and monomeric forms, and it has been hypothesized that higher DAT expression leads to higher DAT oligomerization, and DAT oligomers have lower affinity for DAT blockers with respect to monomers [
38].
In line with that found in G2019S overexpressing mice [
41], G2019S KI mice showed reduced striatal VMAT2 levels. This reduction was robust and consistent with the two different antibodies, one of which validated in VMAT2
+/- mice [
15]. Reduction of VMAT2 is observed also in PD patients [
55] and is pathogenic in PD. In fact, filling synaptic vesicles via VMAT2 is a way to keep cytosolic DA levels in a nontoxic range; accordingly, VMAT2 deletion induces neurodegeneration [
80] whereas VMAT2 overexpression protects DA neurons [
10,
42]. However, despite VMAT2 reduction further enhanced the already higher DAT/VMAT2 ratio in nigro-striatal DA neurons, thus increasing their vulnerability [
56], G2019S KI mice did not show overt neurodegeneration (up to 19-months at least) or even significantly enhanced levels of DOPAL-bound α-syn, a marker of DA cytotoxicity. This questions the physiological meaning of the 50% reduction of VMAT2 observed in the striatal homogenate of G2019S KI mice. Indeed, contrary to that expected from the Western blot data, an increase in tetrabenazine-sensitive vesicular DA uptake was measured in G2019S KI mice in vitro. Although we cannot rule out the possibility that such increase is compensatory in nature, the possibility that this discrepancy relies on technical reasons should be considered. In fact, the reduction of VMAT2 levels measured in striatal homogenate might not faithfully reflect a reduction of active VMAT2 expressed on mature, release-prone synaptic vesicles. In fact, VMAT2 levels measured by Western blot encompass also VMAT2 contained in immature secretory vesicles trafficking from the soma to presynaptic vesicle membrane, or recycling from the plasma membrane [
30]. Interestingly, Sonsalla and collaborators [
30] proved a disparity between tetrabenazine binding measured in striatal homogenate and striatal synaptic vesicles at 24 h after MPTP, showing that, under certain conditions, tetrabenazine binding measured in striatal homogenate may not be representative of vesicular VMAT2. On the other hand, the major limitation of a whole-brain preparation of synaptic vesicles is heterogeneity. VMAT2 is present not only in striatal dopaminergic terminals but also in noradrenergic, serotoninergic and histaminergic terminals in striatal and extrastriatal areas. Since there is no possibility to dissect out the contribution of the different populations of VMAT2-positive synaptic vesicles in this whole-brain preparation, we cannot prove that the observed increase of vesicle uptake is really due to VMAT2 expressed on striatal vesicles, or is the net result of all changes of VMAT2 activity in different nerve terminals and brain areas.
Nonetheless, in favor of the hypothesis that vesicular DA uptake might be increased rather than reduced in G2019S KI mice, G2019S KI mice were relatively more resistant than WT controls to the hypolocomotive action of 1 mg/Kg reserpine in vivo, which is opposite from that expected from DA depleted vesicles. We can speculate that the greater resistance to reserpine might be due to a greater competition for VMAT2 of reserpine and cytosolic DA (the increase in DAT activity and the reduced DA turnover might overwhelm the buffering capacity of VMAT2, thus causing an increase in cytosolic DA). Alternatively, we might speculate that synaptic vesicles in G2019S KI mice are more enriched in DA, although only a trend to an increase in extracellular DA levels or in the K+-induced DA release was observed in the G2019S LRRK2 carriers.
It is therefore plausible that VMAT2 uptake elevation compensates for the loss of VMAT2 protein and protects from cytosolic DA toxicity, even in the presence of upregulated DAT. Whether this adaptive change will be effective throughout the life-span of G2019S KI mice is unknown, since we have investigated G2019S KI mice up to 19 months. However, it is also possible that other compensatory mechanisms will come into play to preserve DA homeostasis and DA neuron integrity.
In this respect, one important finding of the present study is that pSer129 α-syn levels are elevated in the striatum of G2019S KI mice. Since this was not paralleled by an elevation of total α-syn levels, we concluded that G2019S LRRK2 facilitates this posttranslational modification of α-syn. This is in line with a recent study showing that the formation of pSer129 α-syn-positive inclusions in nigral DA neurons in response to intranigral α-syn fibrils injection is accelerated in BAC hG2019S rats [
86]. pSer129 α-syn [
24] is the predominant form of syn in Lewy bodies [
1], and for this reason it has been hypothesized to favor α-syn aggregation, thus contributing to PD [
57]. However, the role of pSer129 α-syn phosphorylation in α-syn toxicity in vivo is still under debate [
57,
81]. In fact, from the published literature it appears that depending on which kinase is involved in Ser129 α-syn phosphorylation, either neurotoxicity (G protein receptor kinases) [
11,
71] or neuroprotection (Polo-like kinase 2) [
58] can ensue. Moreover, LRRK2, and more intensely G2019S LRRK2, can directly Ser129-phosphorylate α-syn in vitro [
63]. Pinning down the pathway underlying Ser129 α-syn phosphorylation might help understand whether this modification is protective or pathogenic for DA neurons.
Conclusion
We previously reported that G2019S KI mice have enhanced motor performance starting at 6 months of age [
43]. We now show that this behavior is not sustained by enhanced DA release, suggesting that other mechanisms, such amplified postsynaptic D1 receptor signalling [
54,
61] or glutamate release [
3], might contribute. Of note, this study reveals for the first time that G2019S KI mice progressively develop (between 3 and 12 months) dysfunctions of plasma membrane and vesicular DA transporters, along with an overload of pSer129 α-syn inclusions in striatum. These adaptive changes were not associated with overt nigro-striatal DA degeneration or changes of striatal DA release, indicating DA homeostasis is preserved, at least up to 19-months. Nonetheless, they might represent vulnerability factors to DA neurons. A more stringent analysis of the time-course of these changes might help elucidate how this response of DA terminals is orchestrated, and how these factors relate to each other and, ultimately, to G2019S LRRK2. In fact, there is no evidence that G2019S LRRK2 can directly affect DAT or VMAT2 trafficking, although this possibility is worth investigating due to the role of LRRK2 in endosome and autophagosome pathways [
68]. Nonetheless, G2019S LRRK2 could do so indirectly, through pSer129 α-syn. Indeed, α-syn stimulates DAT activity [
23,
36] and G2019S LRRK2 has been shown to increase this property by phosphorylating α-syn at Ser129 [
27]. Preliminary evidence that α-syn controls VMAT2 activity has also been collected in cells where α-syn knockdown causes an increase and A53T α-syn overexpression a reduction [
22,
44]: nonetheless, the role of the pSer129 remains unknown. Whatever the mechanistic interactions between these players are, the gradual development of this response offers a wide time-window for a pharmacological intervention (e.g. with LRRK2 inhibitors) that could establish the role of the kinase vs non kinase activities of LRRK2.
In conclusion, G2019S KI mice might represent a presymptomatic model of PD, a valuable tool to verify a “multi-hit” hypothesis of PD [
78], where genetic variables (G2019S LRRK2), an established risk factor (aging), and internal (e.g. DA, α-syn) or environmental (e.g. MPTP) factors interact to shape the emergence of the parkinsonian phenotype.