Leucine‐rich repeat kinase 2 (LRRK2) inhibitors differentially modulate glutamate release and Serine935 LRRK2 phosphorylation in striatal and cerebrocortical synaptosomes

Abstract Mutations in leucine‐rich repeat kinase 2 (LRRK2) gene have been pathogenically linked to Parkinson's disease, and pharmacological inhibition of LRRK2 is being pursued to tackle nigro‐striatal dopaminergic neurodegeneration. However, LRRK2 kinase inhibitors may have manifold actions, affecting not only pathological mechanisms in dopaminergic neurons but also physiological functions in nondopaminergic neurons. Therefore, we investigated whether LRRK2 kinase inhibitors differentially modulate dopamine and glutamate release from the mouse striatum and cerebral cortex. Spontaneous and KCl‐evoked [3H]‐dopamine and glutamate release from superfused synaptosomes obtained from wild‐type and LRRK2 knock‐out, kinase‐dead or G2019S knock‐in mice was measured. Two structurally unrelated inhibitors, LRRK2‐IN‐1 and GSK2578215A, were tested. LRRK2, phosphoSerine1292 and phosphoSerine935 LRRK2 levels were measured in all genotypes, and target engagement was evaluated by monitoring phosphoSerine935 LRRK2. LRRK2‐IN‐1 inhibited striatal glutamate but not dopamine release; GSK2578215A inhibited striatal dopamine and cortical glutamate but enhanced striatal glutamate release. LRRK2‐IN‐1 reduced striatal and cortical phosphoSerine935 levels whereas GSK2578215A inhibited only the former. Neither LRRK2 inhibitor affected neurotransmitter release in LRRK2 knock‐out and kinase‐dead mice; however, they facilitated dopamine without affecting striatal glutamate in G2019S knock‐in mice. GSK2578215A inhibited cortical glutamate release in G2019S knock‐in mice. We conclude that LRRK2‐IN‐1 and GSK2578215A modulate exocytosis by blocking LRRK2 kinase activity, although their effects vary depending on the nerve terminal examined. The G2019S mutation unravels a dopamine‐promoting action of LRRK2 inhibitors while blunting their effects on glutamate release, which highlights their positive potential for the treatment of PD, especially of LRRK2 mutation carriers.


| INTRODUC TI ON
Leucine-rich repeat kinase 2 (LRRK2) is a 2527 amino acid, multifunctional protein endowed with a kinase domain and a Ras of complex (ROC) domain with GTPase activity, surrounded by protein-protein interaction domains. 1,2 Mutations in the LRRK2 gene are associated with familial late-onset 3,4 and sporadic 5 Parkinson's disease (PD). To further emphasize the role of LRRK2 in idiopathic PD, genome-wide studies have revealed LRRK2 to be associated with an increased risk of PD. 6 At present, at least six mutations, that is, the G2019S and I2020T in the kinase domain, the R1441C/G/H and in the ROC domain, and the Y1669C in the CoR domain, have been consistently shown to be pathogenic. 7 Limited evidence for the p.N1437H mutation in the ROC domain has also been presented. 8 In particular, expression of the G2019S mutation is associated with enhanced kinase activity and neurodegeneration in vitro. 9,10 This has boosted the development of LRRK2 kinase inhibitors as novel disease modifying agents, able to attenuate nigrostriatal dopamine (DA) neuron loss in PD. [11][12][13] Nonetheless, the possibility that LRRK2 inhibitors interfere with cell homeostatic functions, in the same or different neuronal populations or tissues, exists, 14 which raises safety issues about this class of compounds. Thus, comparing the effects of LRRK2 inhibitors on different neuronal populations, in both wild-type (WT) and LRRK2 mutant expressing animals, is mandatory.
Among the various cellular functions modulated by LRRK2, exocytosis appears attractive because LRRK2 can regulate neurotransmitter release via multiple routes, 15 for example, by modulating vesicle mobility and trafficking, [16][17][18] SNARE protein assembly, 18,19 and presynaptic Ca ++ entry. 20 Given the pathogenic role of LRRK2 in PD, a wealth of studies focused on in vivo and in vitro DA release using LRRK2 knock-out (KO) mice, [21][22][23] G2019S 24,25 or R1441C 26 knock-in (KI) mice, hG2019S or hR1441G overexpressing mice [27][28][29][30][31] or rats. 32,33 Fewer studies attempted to address the role of LRRK2 in the release of other neurotransmitters, focusing specifically on in vitro glutamate (GLU) release in the cortex 16,27,34 and hippocampus. 35 None of these studies, however, performed a simultaneous analysis of DA and GLU release within a specific or different brain areas, to investigate whether LRRK2 control of neurotransmitter release is similar across different subpopulations of nerve terminals.
Moreover, only a few studies employed more than one LRRK2 kinase inhibitor, leaving to speculation whether these molecules, in addition to sharing class-specific properties have peculiar effects. In fact, it has been previously shown that pharmacological blockade of kinase activity results in rapid dephosphorylation of LRRK2 at Ser935, an index of kinase activity inhibition and disturbance of LRRK2 binding to 14-3-3, 36 followed by delayed LRRK2 degradation through the ubiquitin-proteasome pathway. 37 LRRK2 inhibitors might have a different ability to influence such mechanisms, as shown in primary astrocytes where only GSK2578215A 38 among a panel of 6 different LRRK2 inhibitors, was able to induce protein destabilization. 37 This would suggest that LRRK2 inhibitors might have not only a different potency but also a different mode of interaction with LRRK2 kinase pocket. In fact, while the ability of LRRK2-IN-1 (IN-1) to inhibit LRRK2 was dramatically reduced (by 190-folds) in A2016T mutants, 39 that of GSK2578215A was minimally affected (7-folds). 38 For these reasons, in this study we investigated whether two structurally unrelated LRRK2 kinase inhibitors, such as IN-1 and GSK2578215A, differentially affect the spontaneous and KClevoked [ 3 H]-DA and GLU release in superfused synaptosomes from the mouse striatum and cerebral cortex. Synaptosomes represent a basic preparation of nerve endings, suitable for studying exocytosis since they preserve the release machinery (ATP-and Ca ++dependent release), express membrane and vesicular transporters, and expose autoreceptors. In this preparation, the KCl-evoked neurotransmitter efflux relies on exocytotic Ca ++ -dependent and, partly, Na + -dependent mechanisms, whether spontaneous efflux is essentially non exocytotic. 40 Moreover, the superfusion conditions adopted in this study ensure a rapid removal of the neurotransmitter from the medium, thus minimizing neurotransmitter uptake and autoreceptor activation, 41,42 which might confound the effect of the depolarizing stimulus and LRRK2 inhibitors on exocytosis. The effects of IN-1 and GSK2578215A were first investigated in synaptosomes from WT mice, then in synaptosomes from mice with constitutive deletion of LRRK2 (KO mice) or knock-in for the LRRK2 D1994S kinase-dead mutation (KD mice) to confirm their pharmacological specificity. Since LRRK2 inhibitors are expected to be used in G2019S carriers first, their effects were also investigated in synaptosomes from mice expressing the LRRK2 kinase-enhancing G2019S mutation (G2019S KI mice). 21,24,43 Finally, LRRK2 protein levels and kinase activity (pSer1292 and pSer935 levels) were measured in striatal and cortical tissue lysates and synaptosomes, and target engagement of LRRK2 inhibitors assessed.

| Animals
Experiments were performed in accordance with the ARRIVE guidelines. Experimenters were blinded to treatments. Three-month-old effects on glutamate release, which highlights their positive potential for the treatment of PD, especially of LRRK2 mutation carriers.

| Synaptosome preparation
Synaptosomes were prepared as previously described. 42

| Release studies
Synaptosomes were incubated at 36.5°C with 50 nmol L −1 [ 3 H]-DA (specific activity 40 Ci mmol −1 ; Perkin-Elmer, Boston, MA) for 25 minutes, after which 12 mL of preoxygenated Krebs were added. 24,40 One milliliter aliquots of the suspension (~0. 35  Jasco, Tokyo, Japan) with a mobile phase containing 0.1 M sodium acetate, 10% methanol and 2.2% tetrahydrofuran (pH 6.5). GLU was detected by means of a fluorescence spectrophotometer FP-2020 Plus (Jasco, Tokyo, Japan) with the excitation and the emission wavelengths set at 370 and 450 nm, respectively. The limit of detection for GLU was ~1 nmol L −1 . Images were acquired and quantified using the ChemiDoc MP System and the ImageLab Software (Bio-Rad). Membranes were then stripped and re-probed with rabbit anti-LRRK2 (Abcam, ab133474, 1:1000).

| Western blot analysis
Data were analyzed by densitometry; the optical density of pSer935 LRRK2 and pSer1292 LRRK2 was normalized to total LRRK2 whereas total LRRK2 was normalized to α-tubulin levels. To minimize experimental variability, each blot was replicated twice, and data averaged.

| Data presentation and statistical analysis
The data and statistical analysis comply with the recommendations on experimental design and analysis in pharmacology. 45 Neurochemical data are means ± SEM (standard error of the mean) of n determinations per group. Tritium efflux ( Figure 1 (Figures 3 and 4) was expressed as pmol mg protein −1 . Western blot data have been presented as ratio between pSer935 or pSer1292 and total LRRK2, or between total LRRK2 and α-tubulin (housekeeper). Statistical analysis on neurochemical and biochemical data was performed (Prism software; San Diego, CA) by two-way or one-way ANOVA followed the Bonferroni test for multiple comparisons. When only two groups were compared, the Student's t test, 2-tailed for unpaired data, was used.
Outliers were identified using the Outlier calculator freely available in Graphpad Prism software. P-values < 0.05 were considered to be statistically significant.

| Materials
IN-1 and GSK2578215A were purchased from Tocris Bioscience (Bristol, UK). Both compounds were dissolved in DMSO to 10 mmol L −1 , then diluted with Krebs at the tested concentrations.

| Nomenclature of Targets and Ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guide topha rmaco logy. org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY, 46 and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16. 47

| Striatal DA release
To investigate whether LRRK2 regulates DA release, we first   (Table 1), and 15 mmol L −1 KCl tended to be more effective in KO than WT mice ( Figure 2C). We therefore used a milder stimulation and found that 10 mmol L −1 KCl was significantly more effective   (Figure 2A-B).

| Ser1292 LRRK2 and Ser935 LRRK2 phosphorylation
In the attempt to demonstrate the engagement of synaptosomal LRRK2 by kinase inhibitors, preliminary analysis of endogenous LRRK2 levels and kinase activity was performed in striatum and cerebral cortex. Kinase activity was evaluated by quantifying pSer1292 and pSer935 LRRK2 levels, first in tissue lysates, to allow a comparison with published studies, then in synaptosomes to specifically investigate presynaptic LRRK2. In striatal lysates, LRRK2 levels were found to be lower in G2019S KI (−30%) and KD (−68%) mice compared to WT mice (F 2,15 = 31.50, P < 0.0001), and undetectable in KO mice (Supplementary Figure 1A). pSer1292 levels were 18-fold elevated compared to WT mice (t = 11.69, df = 10) but undetectable in both KD and KO mice ( Figure S1B). Conversely, pSer935 levels were reduced by 60% in G2019S KI mice and by 89% in KD mice (F 2,15 = 141.9 P < 0.0001; Figure S1C). The overall pattern in cerebral cortex lysates showed notable differences compared to striatum.
Given the very low levels of pSer1292 in cerebrocortical synaptosomes from WT mice, we therefore decided to evaluate LRRK2 engagement by measuring pSer935 levels in striatal and cortical synaptosomes acutely treated with IN-1 and GSK2578215A. A 3 minutes  (Figure 6). Both concentrations reduced pSer935 levels in striatal synaptosomes by ~40% (F 2,24 = 13.62, P < 0.0001; Figure 6A). To draw a complete concentration-response curve, lower profile. Indeed, after 3 minutes application it reduced by 28% pSer935 levels in striatum at 1 μmol L −1 (F 2,36 = 3.519 P = 0.0402; Figure 8A) but was ineffective on pSer935 levels in cortex at any concentrations ( Figure 8C). Moreover, no effect of GSK2578215A on LRRK2 levels in striatal or cortical synaptosomes was observed ( Figure 8B,D). Longer application times of GSK2578215A were evaluated ( Figure S2), and similar to IN-1, GSK2578215A failed to alter striatal ( Figure S2A) and cortical ( Figure S2C) pSer935 levels after 12 minutes or 30 minutes. Different from IN-1, GSK2578215A did not alter LRRK2 protein levels at any concentrations ( Figure S2B,D).

| D ISCUSS I ON
A synaptosomal preparation was used to investigate the role of  activity measured with an enzymatic assay (13 nmol L −1 ), 39 much higher than that reported to inhibit pSer935 in cell lines (1-3 μmol L −1 ). 38,39 Dephosphorylation at Ser935 has been validated as a readout of LRRK2 kinase activity in cells or ex-vivo 36,39,43,54 but has never been measured in a specific neuronal compartment (ie the nerve terminal), and particularly in a preparation of nerve terminals subjected to various preparative steps. Moreover, we should consider that Ser935 is a heterophosphosite, being phosphorylated by other kinases activated by LRRK2, or dephosphorylated by protein phosphatase 1 (PP1), which, in turn, is inhibited by LRRK2. 55 Therefore, the potencies of Ser935 dephosphorylation in the different areas might depend on the levels and availability of endogenous LRRK2 and any other components of the pathways regulating Ser935 LRRK2 phosphorylation (which are a property of a specific nerve terminal) but also on the protocol of synaptosomes preparation, during which such components might be lost.
Nonetheless, the ineffectiveness of GSK2578512A on cortical pSer935 levels corroborates a study where systemic GSK2578215A was unable to dephosphorylate brain LRRK2 at Ser935, as measured in a whole brain preparation (where the cerebrocortical tissue is predominant). 38 In conclusion, IN-1 and GSK2578215A exert differential effects on exocytosis and pSer935 levels in synaptosomes, which are specifically mediated by LRRK2 kinase activity but vary depending on the concentration, the nerve terminal and brain area considered.
This suggests that LRRK2 inhibitors might possess unique patterns of neurochemical effects, which might rely on different modes of interaction with presynaptic LRRK2 (and associated interactors) or with different, nerve-specific presynaptic pools of LRRK2. Expression of the G2019S mutation increases the sensitivity of DA nerve terminal to the favorable, DA-release promoting action of LRRK2 inhibitors, at the same time attenuating their impact over GLU nerve terminals.
Although these data need to be translated in vivo, they might predict a beneficial effect of LRRK2 inhibitors on DA release and therefore on motor symptoms, in G2019S LRRK2 PD patients. Likewise, the GLUinhibiting action of GSK2578215A in cerebrocortical synaptosomes of G2019S KI mice might also translate into a therapeutic action in view of the pathogenic role of abnormal cortical transmission in nonmotor symptoms, for example, pain and depression, 56,57 and levodopa pharmacotherapy, for example, dyskinesia, 58 associated with PD.

ACK N OWLED G EM ENTS
This work was supported by grants from the Telethon Foundation