A Phosphosite Mutant Approach on LRRK2 Links Phosphorylation and Dephosphorylation to Protective and Deleterious Markers, Respectively

The Leucine Rich Repeat Kinase 2 (LRRK2) gene is a major genetic determinant of Parkinson’s disease (PD), encoding a homonymous multi-domain protein with two catalytic activities, GTPase and Kinase, involved in intracellular signaling and trafficking. LRRK2 is phosphorylated at multiple sites, including a cluster of autophosphorylation sites in the GTPase domain and a cluster of heterologous phosphorylation sites at residues 860 to 976. Phosphorylation at these latter sites is found to be modified in brains of PD patients, as well as for some disease mutant forms of LRRK2. The main aim of this study is to investigate the functional consequences of LRRK2 phosphorylation or dephosphorylation at LRRK2’s heterologous phosphorylation sites. To this end, we generated LRRK2 phosphorylation site mutants and studied how these affected LRRK2 catalytic activity, neurite outgrowth and lysosomal physiology in cellular models. We show that phosphorylation of RAB8a and RAB10 substrates are reduced with phosphomimicking forms of LRRK2, while RAB29 induced activation of LRRK2 kinase activity is enhanced for phosphodead forms of LRRK2. Considering the hypothesis that PD pathology is associated to increased LRRK2 kinase activity, our results suggest that for its heterologous phosphorylation sites LRRK2 phosphorylation correlates to healthy phenotypes and LRRK2 dephosphorylation correlates to phenotypes associated to the PD pathological processes.


Introduction
Parkinson's disease (PD) is the most common neurodegenerative motor disease and the gene encoding the Leucine Rich Repeat Kinase 2 (LRRK2) protein is considered to be one of the most important genetic determinants in PD [1]. The importance of LRRK2 in PD was first revealed from genetic linkage studies and several missense mutations have a confirmed link to PD [2][3][4] (reviewed in [5]). For instance, the gene mutations leading to the G2019S amino acid substitution gene is the most frequent and is found in Caucasians in 5% of autosomal dominant forms of the disease and in 1% to 40% of sporadic Parkinson's patients according to the ethnic origin of patients [6]. In addition, genomic variations at the LRRK2 locus are risk factors for sporadic forms of PD [7,8]. The neurodegeneration of PD is caused by environmental and genetic factors that contribute to accumulations of subtle cytotoxic effects leading to the death of dopaminergic neurons in the nigrostriatal system. In surviving neurons, accumulations of macromolecular aggregates rich in alpha-synuclein and lipids called Lewy bodies are observed [9][10][11]. Among other genetic factors associated with PD loci, at least 24 potential susceptibility genes are involved in the trafficking and regulation of endosomal/autophagic proteins, suggesting that there is a need to better understand the deficits of membrane trafficking in PD [12,13]. LRRK2 itself has also been implicated in membrane related processes such as autophagy, the endolysosomal pathway, extracellular vesicle release or synaptic transmission [14,15].
LRRK2 is found at endosomes, lysosomes and amphisomes in the brain of rats [16] and in vesicular structures of the human brain [17]. LRRK2 is reported to regulate different aspects of the lysosomal physiology (reviewed in [15]). For example, the LRRK2 mutant, G2019S causes an increase in the size and decrease in the pH of lysosomes in astrocytes [18]. LRRK2 G2019S also induces a reduction of lysosomal activities, through reduction of proteolytic activity assessed using a colorimetric and fluorescent detection named DQ-Red-BSA, as well as accumulation of alpha-synuclein inclusions. This phenomenon is reversed by LRRK2 kinase inhibition [15]. In primary neuron cultures expressing LRRK2 R1441C, pH of lysosomes is increased and activity of lysosomal enzymes decreased [19]. These cultures also show decreases in protein degradation and autophagosome-lysosome fusion events [20]. Maintaining a pH of 4.5-5 in the lysosomes is crucial for maintenance of lysosomal enzymes and to obtain optimal protein degradation [21]. Induced pluripotent stem cells (iPSCs) expressing LRRK2 G2019S and R1441C are reported to show) decreased activity of GBA (glucocerebrosidase), an enzyme localized in the lysosomes known as a risk factor for PD, through regulation of RAB10, a LRRK2 kinase substrate [22]. This process is reported to be related to the increase in mutant LRRK2 mediated phosphorylation of RAB10 at the LRRK2 dependent site T72 [23]. RAB10 overexpression increases the activity of glucocerebrosidase, but overexpression of phosphomimic RAB10 T72E had no effect on GBA activity [22]. Interestingly, LRRK2 is recruited to lysosomes in cell culture after treatment with lysosomotropic agents [24,25].
LRRK2 interacts with a large number of vesicular proteins including proteins regulating membrane fusion events such as Rabs and SNAREs [26,27]. Recent work shows that LRRK2 collaborates with RAB proteins to operate on endolysosomal homeostasis by regulation of the balance between membrane repair or degradation through lysophagy. LRRK2 recruits RAB8a to the lysosomal membrane in macrophages after lysosomal damage [28], and LRRK2, RAB8a and RAB10 are found to be enriched in chloroquine stressed lysosomes [25]. Another study reports that the partnership between LRRK2 and RAB8a/RAB10 mediates centrosomal and ciliogenesis deficits in several cell types including patient's derived samples [29][30][31].
LRRK2 is a multidomain and multiphosphorylated protein. Its phosphorylation sites can be divided into two categories, autophosphorylation sites and heterologous phosphorylation sites. A heterologous phosphorylation cluster is found between the ankyrin repeat (ANK) domain and leucine-rich repeat (LRR) domain at Serines S860, S910, S935, S955, S973 and S976 [32,33]. Similarly, a cluster of autophosphorylation sites is observed at the level of the GTPase domain of LRRK2 (the Ras of complex(ROC) domain), as well as other sites such as S1292 in the LRR domain [34,35]. Interestingly, the level of phosphorylation at the ANK-LRR interdomain region is found to be decreased in brains tissues of Parkinson's patients [36], while autophosphorylation levels (at S1292) are increased [37] in brains of Parkinson's patients compared to controls.
LRRK2 binds 14-3-3 at LRRK2 phosphorylation sites of both the ANK-LRR interdomain region and ROC domain [38,39]. This binding is proposed to be regulated by the phosphorylation profile of LRRK2, including the S910 and S935 phosphorylation sites [40]. After kinase inhibition with type I kinase inhibitors, LRRK2 phosphorylation is decreased and the binding to 14-3-3 is disrupted. This loss is correlated with the relocalization of LRRK2 into skein-like structures [38,39]. LRRK2, when its localization is associated with membranes, presents greater kinase activity in comparison to cytosolic LRRK2 [41].
While these previous observations point to the importance of phosphorylation in normal LRRK2 functions, as well as in PD, the precise downstream effects of phosphorylation changes in LRRK2 remain poorly known. For these reasons, we aimed to study phenotypic consequences of LRRK2 phosphorylation on LRRK2 s biochemical and cell physiological properties. We used phosphorylation/dephosphorylation mimicking mutants at the ANK-LRR domain of LRRK2 and tested how these changes affect LRRK2 s catalytic activity, interaction and phosphorylation of vesicular partners, recruitment to discrete subcellular compartments and effects on neurite outgrowth.

Plasmids and Stable Cell Lines
LRRK2 was cloned into the pLV-CSJ vector backbone which is itself constructed by replacing the mCherry sequence digested out with BamHI and NheI from pLV-mCherry (Addgene plasmid #36084) with a multiple cloning site via ligation of a short adaptor sequence containing the restriction digestion recognition sequences for BglII and NheI (New England Biolabs, Ipswich, MA, USA). The LRRK2 sequence cloned into the pLV-CSJ backbone were amplified from the p3flag-CMV10-LRRK2 plasmid (a generous gift from Prof. T. Iwatsubo, University of Tokyo, Japan) and a triple flag tag was introduced by adaptor ligation.

In Vitro Kinase Assay
The kinase activities of purified LRRK2 were measured at 30 • C in Kinase assay buffer consisting of Tris 25 mM (pH 7.5), MgCl 2 15 mM, β-glycerol phosphate 20 mM, sodium fluoride 1 mM, EGTA 1 mM, sodium orthovanadate 1 mM, DTT 2 mM and 0.1 mg/mL BSA. 0.1 µM purified LRRK2 and 75 µM LRRKtide peptides were used in the assay. The reaction was initiated by addition of adenosine triphosphate 25 µM, labeled on the gamma phosphate group with 32 P ( 32 P-γ-ATP) (4 Ci/mmol). At different time points until 30 min of reaction, the reaction mixtures were quenched in 100 mM ice-cold EDTA. The quenched samples were then spotted on P81 phosphocellulose paper discs pre-rinsed with 75 mM ice-cold phosphoric acid and further washed with phosphoric acid to remove free ATP. The paper discs were then air dried before scintillation counting.

GTPase Activity
The GTPase activities of purified LRRK2 were measured at 30 • C in GTPase assay buffer consisting of Tris 50 mM (pH 7.5), NaCl 150 mM, MgCl 2 10 mM, 5% Glycerol, βmercaptoethanol 3 mM and 0.2 mg/mL BSA. 0.1 µM purified LRRK2 was added into 500 µM 32P-γ-GTP (0.15 Ci/mmol) to initiate the reaction. At different time points until 180 min of reaction, the reaction mixtures were quenched in 5% activated charcoal in 20 mM phosphoric acid. The quenched samples were then spun down by table-top centrifuge at 20,000× g for 10 min and the supernatants were aspirated for scintillation counting.
For assessment of binding of LRRK2 with 14-3-3, we used the microscale thermophoresis method. After the expression of mCherry-LRRK2 in HEK293T, cells were lysed in lysis buffer 1 (described above) and 5 µM of recombinant 14-3-3zeta (Abcam, UK) is added to the lysate. The mix is then placed into Monolith NT.115 Premium Capillaries (Nan-oTemper Technologies, Munich, Germany) and analyzed using the Nanotemper Monolith Pico (NanoTemper Technologies, Munich, Germany) at 10% of power using the pico RED excitation color.

Immunocytochemistry
SH-SY5Y stably expressing 3xFLAG-LRRK2 WT were grown on coverslips coated with Cell tak solution (292.5 µL of bi-carbonate buffer, 2.5 µL of NaOH 1 M, 5 µL of Cell Tak). Cells were fixed using 4% paraformaldehyde for 15 mn at room temperature, followed by 3 PBS washes. Cells were permeabilized by PBS-Triton-X100 0.1% for 5 mn at room temperature. Cells were blocked for 20 mn with PBS-BSA 0.5% and incubated with primary antibody diluted at 1:1000 in PBS-BSA 0.5% overnight at 4 • C, washed 3 times with PBS and incubated for 2 h at room temperature with secondary antibody diluted at 1:1000 in PBS-BSA 0.5%. Coverslips were mounted in Prolong Gold Antifade mounting medium (Invitrogen, Montigny le Bretonneux, France). Pictures were taken on Zeiss LSM 710 confocal microscope (Zeiss, Oberkochen, Germany).

Lysosomal Isolation
HEK293T cells expressing 3xFLAG-LRRK2 were cultured for 24 h in a medium containing an iron-coupled dextran. This compound accumulates in lysosomes. After a washout period of 24 H in a medium without dextran the cells were centrifuged 60× g for 5 mn then washed with PBS. A mechanical lysis was performed with a Dounce Tight followed by 8 passes in a 23 G needle. After a 400× g centrifugation for 10 min, the lysate was placed on a magnetic column (MS Columns, Miltenyi Biotec, Paris, France). The lysosome-enriched fraction was then collected by detaching the column from the magnetic base in PBS/0.1 mM sucrose buffer. The samples were then analyzed by Western blot.

Image Stream Analysis
Five million PC12 cells are collected by centrifugation (500 g, 5 mn) and resuspended in 1 mL of medium containing 10 µg Hoechst 33258 (Invitrogen, Montigny le Bretonneux, France). After 30 mn incubation at 37 • C, cells were again centrifuged 500 G for 5 mn and incubated in medium containing 75 nM of Lysotracker Deep Red (Thermo Fisher Scientific, Waltham, MA, USA) and 75 µg of FDGlu (Tebu-Bio, Le Perray-en-Yvelines, France) for 1 h at room temperature. Finally, cells are centrifuged and resuspended in 20 µL of PBS + EDTA 0.5 M and analyzed using an ImageStream MKII (AMNIS) at 60X magnification at a low speed with a maximum laser power for 488 nm and 561 nm lasers (200 mW) and with 35 mW for 375 nm laser and 20 mW for 642 nm laser. Compensations were carried out using single stained controls. Around 50,000 cells were recorded in order to have sufficient a number of cells for statistical analysis. Then data were analyzed with IDEAS Software (Millipore, Billerica, MA, USA) using successive masks and features, as described in the Results section.

Incucyte Live Cell Analysis
To analyze the neurite complexity, the Incucyte ®® S3 live-cell imaging system (Sartorius, Minisart, Göttingen, Germany) was used. A total of 4000 PC12 cells stably expressing mCherry were seeded on collagen coated Costar ®® 24-well Clear TC-treated Multiple Well Plates (Corning, Corning, NY, USA). Differentiation was induced with RPMI 1640 medium with 1% fetal horse serum and 0.5% fetal bovine serum containing 50 µg/mL NGF. This medium was refreshed every 48 h until full differentiation at day 10. Plates were scanned using the Incucyte system using a 20× objective and 36 images were taken per well. Assessment of neurite length and branch points was performed using the Incucyte ® Neurotrack analysis software module (Sartorius, Minisart, Göttingen, Germany). Average neurite length and branch points per cell were calculated by dividing total neurite length or branch points per well by the cell confluence expressed in mm 2 .

Statistical Analysis
All data are presented as means ± SEM. N is the number of independent experimental replicates. Statistical significance was determined using a Mann-Whitney analysis for a comparison of two experimental conditions, or a Kruskal-Wallis test followed by a post hoc Dunn test for multiple comparisons or a 1-way ANOVA followed by a Dunnett's test for multiple comparisons when more than 2 experimental conditions were compared. All statistical analyses were performed using GraphPad Prism version 9 (GraphPad Software, San Diego, CA, USA).

LRRK2 Phosphorylation Is Affected by Its Phosphorylation Mutants
First, to determine if the phosphorylation motif of LRRK2 could induce new phosphorylation profile we mutated the Serine (S) residues of position 910, 935, 955, 973, and 1292 to Alanine (A) to mimic non-phosphorylated Serines and to Aspartate (D) to mimic phosphorylated Serine. A compound mutant was also generated to mimic phosphorylated/non-phosphorylated LRRK2 presenting six mutations of Serines to Alanines or Aspartates in the heterologous phosphorylation cluster noted 6xA or 6xD (mutation of Serines 860/910/935/955/973/976 to Alanines (6xA) or Aspartates (6xD)).
Phosphodead mutants S910A and S935A, but not S955A and S973A, changed the phosphorylation of neighboring Serines (Figure 1), matching previous observations [38,40]. By comparison, phosphomimicking mutants of these same sites presents the same effect on neighboring Serines as the phosphodead mutants, except S910D which does not display altered S935 phosphorylation. Of note, S910D is the only phospho-mimetic mutant to be recognized by the corresponding anti-phospho antibody, suggesting that the aspartate substitution in this position may effectively resemble phosphorylation.
We reported no significant effect on the cellular localization of LRRK2 mutants (Supplementary Figure S1). Of note, the phosphodeficient mutant LRRK2 6xA does not present skein-like structures in basal conditions. It is also noteworthy that we observe the presence of skein-like structure with MLi-2 treatment in all cases. Interestingly, mimicking the dephosphorylation of six sites in the ANK-LRRK domain does not result in a loss of binding with 14-3-3 in the experimental conditions described in materials and methods (Supplementary Figure S2).

Heterologous Phosphorylation of LRRK2 Affects Its Autophosphorylation
Individually, S910 S935, S955, and S973 had no significant effect on the phosphorylation rate of the S1292 autophosphorylation site (Figure 2). In addition, while our mutants LRRK2 6xA and 6xD presented 30% lower average values for phosphorylation at S1292, these were likewise not statistically different compared to the WT control ( Figure 2A). To assess a potential deficit in kinase activity of these two particular mutants, we measured in total cell lysate the phosphorylation of two known substrates of LRRK2, RAB8a and RAB10. Phosphorylation of both RAB8 T72 and RAB10 T73 was significantly decreased when LRRK2 6xD was expressed but in the case of LRRK2 6xA ( Figure 2B,C). We reported no significant effect on the cellular localization of LRRK2 mutants (Supplementary Figure S1). Of note, the phosphodeficient mutant LRRK2 6xA does not present skein-like structures in basal conditions. It is also noteworthy that we observe the presence of skein-like structure with MLi-2 treatment in all cases. Interestingly, mimicking the  [46]. The ratio of phosphorylated site over total LRRK2 signal is shown in (B) for S910, (C) for S935, in (D) for S955 and in (E) for S973. The blot image for total LRRK2 is reused for illustration purposes in (B-E). The data represents the mean ± SEM from four independent experiments. The quantification of the signals was done by using Typhoon FLA 9500 and ImageQuant software (ImageQuant Tl (IQTL) 8.1, GE Healthcare, Chicago, IL, USA). The data were analyzed by One-way-ANOVA followed by a Dunnett's test, using the WT condition as control. (*** p < 0.001), (**** p < 0.0001).
LRRK2 6xA and 6xD presented 30% lower average values for phosphorylation at S1292, these were likewise not statistically different compared to the WT control (Figure 2A). To assess a potential deficit in kinase activity of these two particular mutants, we measured in total cell lysate the phosphorylation of two known substrates of LRRK2, RAB8a and RAB10. Phosphorylation of both RAB8 T72 and RAB10 T73 was significantly decreased when LRRK2 6xD was expressed but in the case of LRRK2 6xA ( Figure 2B,C). In cellulo LRRK2 kinase activity assays of LRRK2 phosphorylation mutants. Western blot analysis of HEK293T cell lysates after transfection of LRRK2 phosphomutants. The ratio of phosphorylated site over total protein signal is shown, (A) for LRRK2 autophosphorylation site S1292. Endogenous phosphorylation level of RAB8a T72 (B) and RAB10 T73 (C) were also quantified. The data represents the mean ± SEM from four independent experiments (A) and 3 independent experiments for (B,C). The quantification of the signals was done by using Typhoon FLA 9500 and Im-ageQuant software. The data were analyzed by One-way-ANOVA followed by Dunnett's test using the WT condition as control. (* p < 0.05), (** p < 0.01)).
We then tested whether or not LRRK2 6xD could trigger dephosphorylation of RAB29 (Alias RAB7L1), a PD risk factor that has previously been reported as an interactor and substrate of LRRK2 [47,48]. Expression of the LRRK2 phosphorylation mutants 6xA and 6xD did not induce changes in RAB29-T72 phosphorylation compared to LRRK2 WR (Supplementary Figure S3). In this experiment, we decided to also test for effects of RAB29 on LRRK2, given previous reports that RAB29 overexpression activates autophosphorylation of LRRK2 at S1292 [49]. In this experiment, we found that expression of RAB29 led to an activation of LRRK2-S1292 phosphorylation of LRRK2 6xA that was 6-fold higher compared RAB29 induced activation of LRRK2-WT or LRRK2-6xD ( Figure 3).

Figure 2.
In cellulo LRRK2 kinase activity assays of LRRK2 phosphorylation mutants. Western blot analysis of HEK293T cell lysates after transfection of LRRK2 phosphomutants. The ratio of phosphorylated site over total protein signal is shown, (A) for LRRK2 autophosphorylation site S1292. Endogenous phosphorylation level of RAB8a T72 (B) and RAB10 T73 (C) were also quantified. The data represents the mean ± SEM from four independent experiments (A) and 3 independent experiments for (B,C). The quantification of the signals was done by using Typhoon FLA 9500 and ImageQuant software. The data were analyzed by One-way-ANOVA followed by Dunnett's test using the WT condition as control. (* p < 0.05), (** p < 0.01).
We then tested whether or not LRRK2 6xD could trigger dephosphorylation of RAB29 (Alias RAB7L1), a PD risk factor that has previously been reported as an interactor and substrate of LRRK2 [47,48]. Expression of the LRRK2 phosphorylation mutants 6xA and 6xD did not induce changes in RAB29-T72 phosphorylation compared to LRRK2 WR (Supplementary Figure S3). In this experiment, we decided to also test for effects of RAB29 on LRRK2, given previous reports that RAB29 overexpression activates autophosphorylation of LRRK2 at S1292 [49]. In this experiment, we found that expression of RAB29 led to an activation of LRRK2-S1292 phosphorylation of LRRK2 6xA that was 6-fold higher compared RAB29 induced activation of LRRK2-WT or LRRK2-6xD ( Figure 3). The ratio of phosphorylated site over total protein signal is shown, LRRK2 autophosphorylation site S1292 was quantified. The data represent the mean ± SEM from 3 independent experiments. The quantification of the signals was done by using Imager 600 and ImageQuant software. The data were analyzed by One-way-ANOVA followed by a Tukey's test. (* p < 0.05).

In Vitro Assay Is Not Affected by the Phosphorylation Profile of LRRK2
Next, we assayed LRRK2 phosphomutants for in vitro kinase activity (LRRKtide peptide phosphorylation in the presence of 32 P-γ-ATP) and GTPase activity (hydrolysis of 32 Pγ-GTP). Mutants of phosphorylation site S1292 (S1292A, S1292D) had no effect on either Figure 3. In cellulo LRRK2 activation by RAB29. Western blot analysis of HEK293T cell lysates after transfection of LRRK2 phosphomutants and 2myc-RAB29. The ratio of phosphorylated site over total protein signal is shown, LRRK2 autophosphorylation site S1292 was quantified. The data represent the mean ± SEM from 3 independent experiments. The quantification of the signals was done by using Imager 600 and ImageQuant software. The data were analyzed by One-way-ANOVA followed by a Tukey's test. (* p < 0.05).

In Vitro Assay Is Not Affected by the Phosphorylation Profile of LRRK2
Next, we assayed LRRK2 phosphomutants for in vitro kinase activity (LRRKtide peptide phosphorylation in the presence of 32 P-γ-ATP) and GTPase activity (hydrolysis of 32 P-γ-GTP). Mutants of phosphorylation site S1292 (S1292A, S1292D) had no effect on either activity ( Figure 4A,B). However, in the multi-mutant LRRK2 6xD, kinase activity showed a trend to decrease (p = 0.0549) ( Figure 4A). As we identified LRRK2 6xD to present in cellulo a decreased phosphorylation activity on RAB8a and RAB10, we tested LRRK2 multi-mutants for in vitro kinase assay on recombinant RAB8a. In addition, we verified the thermal stability of purified LRRK2 proteins and found that WT LRRK2 is of comparable stability to the LRRK2 6xA and LRRK2 6xD mutants (Supplementary Figure S4, while confirming a subtle improvement in thermal stability in the Hepes pH 8.2 buffer compared to the Tris buffers [44]. The losses of pT72-RAB8 and pS1292-LRRK2 phosphorylation observed in cellulo for the LRRK2-6xD mutant were not replicated in vitro and are not explained by changes in LRRK2 stability in vitro ( Figure 4C-E).

NGF Induced Neurite Length and Branch Points in PC12 Cells Are Not Affected by Expression of LRRK2 WT or Phosphomutants
LRRK2 mutants are reported to affect neurite length, for instance, LRRK2 mutant G2019S and R1441G cause neurite length reductions in primary cultured hippocampal neurons [50]. We generated PC12 cell line stably expressing mCherry-LRRK2 WT and compound phosphomutants to measure the neurite length of each construct, after 10 days of differentiation no change in neurite length or neurite branch points was shown between our phosphorylation mutants ( Figure 5A-C).

NGF Induced Neurite Length and Branch Points in PC12 Cells Are Not Affected by Expression of LRRK2 WT or Phosphomutants
LRRK2 mutants are reported to affect neurite length, for instance, LRRK2 mutant G2019S and R1441G cause neurite length reductions in primary cultured hippocampal neurons [50]. We generated PC12 cell line stably expressing mCherry-LRRK2 WT and compound phosphomutants to measure the neurite length of each construct, after 10 days of differentiation no change in neurite length or neurite branch points was shown between our phosphorylation mutants ( Figure 5A-C).

LRRK2 and Its Phosphorylation Mutants Are Localized to the Lysosomes
LRRK2 is reported to stabilize RAB8a and RAB10 on lysosomes when phosphorylated [25]. In addition, LRRK2 can relocalize to enlarged lysosomes to phosphorylate its substrates RAB8a and RAB10 [25]. To test if the reduced phosphorylation of RAB8a and RAB10 observed in cells above is due to a mislocalization of LRRK2 relative to RAB8a/RAB10 compartments, we tested colocalization of LRRK2 and RAB8a in SH-SY5Y stably expressing 3xFLAG-LRRK2 under basal conditions, as well as after chloroquine treatment, reported to recruit LRRK2 to RAB positive lysosomes [25]. While we confirmed the recruitment of LRRK2 to RAB8a positive compartments after choloroquine treatment, we found no changes in the subcellular distribution of LRRK2 6xA or 6xD compared to WT in any of the conditions tested ( Figure 6B).

LRRK2 and Its Phosphorylation Mutants Are Localized to the Lysosomes
LRRK2 is reported to stabilize RAB8a and RAB10 on lysosomes when phosphorylated [25]. In addition, LRRK2 can relocalize to enlarged lysosomes to phosphorylate its substrates RAB8a and RAB10 [25]. To test if the reduced phosphorylation of RAB8a and RAB10 observed in cells above is due to a mislocalization of LRRK2 relative to RAB8a/RAB10 compartments, we tested colocalization of LRRK2 and RAB8a in SH-SY5Y stably expressing 3xFLAG-LRRK2 under basal conditions, as well as after chloroquine treatment, reported to recruit LRRK2 to RAB positive lysosomes [25]. While we confirmed the recruitment of LRRK2 to RAB8a positive compartments after choloroquine treatment, we found no changes in the subcellular distribution of LRRK2 6xA or 6xD compared to WT in any of the conditions tested ( Figure 6B). To further investigate LRRK2 relocalization to the lysosomes, we measured abundance of LRRK2 in purified lysosomes (Figure 7). In our experiment, we were able to identify 3xFLAG-LRRK2 and its phosphorylation mutant in the isolated lysosomes ( Figure 7C) without chloroquine treatment. Chloroquine treatment significantly increased the proportion of 3xFLAG-LRRK2 WT in the isolated lysosomes, but not for the conditions LRRK2-6xA and LRRK2-6xD ( Figure 7C) while Rab8a phosphorylation remained low in the presence of LRRK2 6xD expression and chloroquine treatment (Supplementary Figure S5). In imaging flow cytometry, we were able to detect LRRK2 at the same position as lysotracker spots ( Figure 8A). To further investigate LRRK2 relocalization to the lysosomes, we measured abundance of LRRK2 in purified lysosomes (Figure 7). In our experiment, we were able to identify 3xFLAG-LRRK2 and its phosphorylation mutant in the isolated lysosomes ( Figure 7C) without chloroquine treatment. Chloroquine treatment significantly increased the proportion of 3xFLAG-LRRK2 WT in the isolated lysosomes, but not for the conditions LRRK2-6xA and LRRK2-6xD ( Figure 7C) while Rab8a phosphorylation remained low in the presence of LRRK2 6xD expression and chloroquine treatment (Supplementary Figure S5). In imaging flow cytometry, we were able to detect LRRK2 at the same position as lysotracker spots ( Figure 8A). To assess if LRRK2 phosphorylation influences the number of lysosomes in cells, we generated PC12 expressing LRRK2 and its 6xA/D mutants and control the number of lysosomes by imaging flow cytometry (ImageStream MKII, luminex). Chloroquine treatment did not increase the number of lysotracker spots in PC12 cells ( Figure 8B). Expression of mCherry-LRRK2-6xA or 6xD reduced the number of lysotracker spot per cell and chloroquine treatment is able to restore the loss of lysotracker spots induced by expression of LRRK2 phosphomutants.
Lysosomal glucocerebrosidase activity in live PC12 cells was measured using the quenchable substrate FDGlu that allows us to measure in real time the activity of this enzyme [51]. In our experiments, LRRK2 overexpression or its phosphorylation had no effect on lysosomal glucocerebrosidase activity ( Figure 8C).

Discussion
The question of the phenotypic consequences of LRRK2 phosphorylation or dephosphorylation has remained mostly unanswered [34,52,53]. Here, we present results that show that changes at the phosphosites in the ANK-LRR interdomain region affect LRRK2 activity in cells, including cellular kinase activity and RAB29 induced activation of LRRK2.
Using a phosphomutant approach, we first made a survey of LRRK2 phosphomutant variants, including modifications of individual sites compared to modifications of multiple sites. While some of the phosphodead mutants (Serine (S) to Alanine (A)) have been reported before, few reports have tested phosphomimicking mutants ((Serine (S) to Aspartate (D)) of LRRK2 and our report is also the first to include side by side comparisons To assess if LRRK2 phosphorylation influences the number of lysosomes in cells, we generated PC12 expressing LRRK2 and its 6xA/D mutants and control the number of lysosomes by imaging flow cytometry (ImageStream MKII, luminex). Chloroquine treatment did not increase the number of lysotracker spots in PC12 cells ( Figure 8B). Expression of mCherry-LRRK2-6xA or 6xD reduced the number of lysotracker spot per cell and chloroquine treatment is able to restore the loss of lysotracker spots induced by expression of LRRK2 phosphomutants.
Lysosomal glucocerebrosidase activity in live PC12 cells was measured using the quenchable substrate FDGlu that allows us to measure in real time the activity of this enzyme [51]. In our experiments, LRRK2 overexpression or its phosphorylation had no effect on lysosomal glucocerebrosidase activity ( Figure 8C).

Discussion
The question of the phenotypic consequences of LRRK2 phosphorylation or dephosphorylation has remained mostly unanswered [34,52,53]. Here, we present results that show that changes at the phosphosites in the ANK-LRR interdomain region affect LRRK2 activity in cells, including cellular kinase activity and RAB29 induced activation of LRRK2.
Using a phosphomutant approach, we first made a survey of LRRK2 phosphomutant variants, including modifications of individual sites compared to modifications of multiple sites. While some of the phosphodead mutants (Serine (S) to Alanine (A)) have been reported before, few reports have tested phosphomimicking mutants ((Serine (S) to Aspartate (D)) of LRRK2 and our report is also the first to include side by side comparisons of a broad range of phosphodead with phosphomimicking mutants. For those constructs that have been previously reported, such as the S910A or S935A mutants, our observations were similar to published observations, namely that these mutants affect the phosphorylation rate at other sites in the S910/S935 cluster, underscoring the robustness of our results [40,54]. The phosphomutant approach was thus implemented to verify which known phenotypes of LRRK2 are modified by LRRK2 phosphorylation.
Previous data indicates that several grey zones persist as to whether LRRK2 dephosphorylation is linked to its toxicity. LRRK2 is dephosphorylated in PD patient brains [36], and in cells with LRRK2 pathogenic mutations [34]. A special note can be made for LRRK2-G2019S, the most common LRRK2 mutation, where heterologous phosphorylation is unchanged in cell culture but reduced in brains, lungs and kidney of G2019S knockin mice [55][56][57]. Dephosphorylation also occurs in cells/tissues treated with LRRK2 type I inhibitors, that have also shown in several studies the ability to abolish adverse phenotypes induced by LRRK2 pathogenic mutations such as LRRK2 mutant induced neurite shortening [50]. Dephosphorylation of the N-TER region does not occur with Type II LRRK2 inhibitors, stabilizing LRRK2 in an open conformation [58]. Induction of filamentous accumulations of LRRK2, where LRRK2 binds to microtubules, is possible with LRRK2 stabilized in a closed conformation [58,59]. In our experiment, LRRK2 6xA and 6xD can also present filamentous cellular accumulations after treatment with MLi-2, a type I LRRK2 kinase inhibitor, suggesting that immobilization of LRRK2 in a closed conformation by MLi-2 is independent of the phosphorylation state of LRRK2 ANK-LRR domain. Interestingly, our previous work showed that cellular treatment with type I inhibitors of LRRK2 kinase leads to rapid recruitment of phosphatases to the LRRK2 complex (subunits of both protein phosphatase 1 and 2A holoenzymes), suggestive of a conformational change in LRRK2 [60]. We postulated that such a conformational change enables LRRK2 dephosphorylation, and given our observation that LRRK2 phosphomutants remain sensitive to type I inhibitor induced filamentous accumulations, we postulate that this phenomenon is related to a conformational change in LRRK2 rather than to dephosphorylation of LRRK2 per se.
Looking at cellular kinase activity, when we tested effects of the individual LRRK2 phosphosite mutants on phosphorylation of the LRRK2 substrates RAB8a and RAB10 in cells, we had no or low changes, while a significant decrease was observed for the 6xD mutant, consistent with a decreased kinase activity of LRRK2 in cells when it is phosphorylated. Intriguingly, when tested in vitro this result was not as pronounced as the in cellulo experiment, despite the observation that thermal stability of purified LRRK2 is not affected by phosphosite mutations. It is possible that, in cellulo¸yet unidentified partners act as a sensor for the phosphorylation motif at the ANK-LRR and affects phosphorylation of its substrates in cells. We hypothesized that the decreased dephosphorylation of RAB8a and S1292 preferentially occurs when LRRK2 6xD is found in a cellular context, perhaps via the aid of additional proteins as a part of a LRRK2 complex.
While we showed in cellulo that LRRK2 s phosphorylation profile specifically regulates the phosphorylation of RAB8a and RAB10, we did not detect any effect on RAB29 phosphorylation. It has been reported that RAB29 binds to LRRK2 via the ankyrin domain and increased RAB29 expression is a trigger to activate LRRK2 kinase activity [49]. While the exact mechanisms of RAB29 induced activation of LRRK2 remains to be elucidated, our observation that the phosphodead variant of LRRK2 is significantly more activated than the phosphomimicking variant of LRRK2 suggests that LRRK2 phosphorylation is a modulator of LRRK2 activation. Further work would be warranted to explore the link between RAB29 mediated activation of LRRK2 and LRRK2 phosphorylation. For instance, a recent study has modeled binding between RAB38 and ARM domain, using a high resolution structure of LRRK2 [46], a similar analysis could be performed for LRRK2 and RAB29. Additionally, RAB29 overexpression recruits LRRK2 to the trans Golgi network and this is assumed to modulate kinases/phosphatases and stimulate LRRK2 s activity [49].
Another phenotypic effect of LRRK2 mutations is its ability to affect neurite outgrowth; however, the role of LRRK2 phosphorylation in this phenotype is not known. LRRK2 G2019S is reported to reduce stable contact between growth-cones [61] and G2019S impair neurite outgrowth in mice [62]. LRRK2 G2019S also increases the phosphorylation of RAB8a/RAB10 that are crucial factors required in rat hippocampal neurons to control neurite outgrowth [63,64]. In brief, a protein from the trans Golgi network, TRIO, regulates the axonal guidance by interacting with RABIN8, a guanine exchange factor (GEF) that activates RAB8/RAB10 [65]. RAB8a interaction with RABIN8 is dependent of the phosphorylation of RAB8a's S111 phosphorylation site [66] the Rab8-RABIN8 interaction is also blocked with RAB8a T72 phosphorylation [67] and this interaction promotes neurite outgrowth [68]. Despite the reduced RAB8a and RAB10 phosphorylation in the presence of LRRK2-6xD, no significant effect on neurite complexity and length was detected in the tested conditions of NGF induced neurite outgrowth in PC12 cells. Our results suggest that other mechanisms may be at play besides Rab phosphorylation in LRRK2-mediated regulation of neurite complexity and further work is warranted, such as testing different experimental conditions (for instance testing other neurite forming cell types) and/or analyzing finer phenotypes, such as growth-cone formation deficits.
LRRK2 kinase activity has been associated with the regulation of lysosomal properties. Enhanced kinase activity results in enlarged lysosomes and reduced GBA activity and reduced lysosomal degradative properties [18,22,69,70]. Our results show that LRRK2 phosphorylation at S860/910/935/955/973/976 is not associated with modification of glucocerebrosidase activity. Interestingly, our results suggest that this cluster of Serine residues is involved in determining cellular lysosome numbers, as both the phosphodead and the phosphomimicking variants showed reduced numbers on lysosomes in PC12 cells ( Figure 8B). Further work is needed to investigate the link between phosphorylation change and lysosomal dysfunction. Our results also suggest that the recruitment of LRRK2 to damaged lysosomes is not dependent on its ANK-LRR phosphorylation status. In the study by Eguchi et al. [25], LRRK2, in absence of chloroquine treatment, was not detected in the lysosomal compartment after lysosomal capture with iron-dextran beads, while our results show that we detect LRRK2 with and without chloroquine treatment in lysosomes. LRRK2 6xD abundance in lysosomes is not statistically different from the WT condition, with or without chloroquine treatment. The difference with the previous finding by Eguchi and colleagues could be explained by a higher expression level of LRRK2 resulting in a bigger proportion of LRRK2 in the lysosomes at the basal level. It is important to correlate these results with microscopy imaging where, without chloroquine, LRRK2 and RAB8a were not present in the same vesicular structures. Therefore, we conclude that LRRK2 can be recruited to lysosomes independently of its phosphorylation status, in the conditions tested.
Reduced phosphorylation of RAB8a and RAB10 observed in the presence of LRRK2 6xD phosphomimicking could be accompanied by reduced RAB35 phosphorylation, however this would need to be verified. Recent resolution of LRRK2 structure in its inactive state indicate that RAB proteins may bind to the ARM domain of LRRK2 [46]. In its inactive conformation, the ARM is remotely positioned relative to the kinase domain that is also covered by the LRR domain that wraps around the ATP binding cleft and block its access. The coordination of the RAB with the ARM domain suggests that LRR and ARM would shift positions in the active state to allow access of the RAB to the kinase.

Conclusions
Considering the postulates that enhanced kinase activity and increase of RABs phosphorylation are detrimental processes, our study results suggest that the activity of the phosphorylation mutant 6xD of LRRK2 at the ANK-LRR phosphosites corresponds to a protective state for LRRK2 while dephosphorylation mutant 6xA at these sites correspond to a deleterious state. Future research should pursue the investigations on this hypothesis to confirm a specific deleterious profile of phosphorylation, notably to understand the precise role of LRRK2 phosphorylation on disease related processes, such as LRRK2 effects on cytoskeleton, endolysosomal dysfunction, or cell viability.