Neurotrophic factors plastically shape neuronal activity by binding to their cognate receptor tyrosine kinases.³² Brain-derived neurotrophic factor (BDNF) is particularly important to modulate neurotransmission in the cortico-striatal circuitry.³³ This circuitry is relevant for PD and presents with the highest LRRK2 expression in comparison with the rest of the brain.^34,35 To address whether LRRK2 Ser935 phosphorylation is stimulated by BDNF, we prepared primary mouse cortical neurons from C57BL/6J wild-type mice and stimulated them with 100 ng/ml of BDNF at days in vitro 14 (DIV14) when synapses are fully functional (Fig. 1A) and LRRK2 expression is detectable.³⁶ Western blot analysis revealed a rapid (5 mins) and transient increase of Ser935 phosphorylation after BDNF treatment (Fig. 1B and 1C). Of interest, BDNF failed to stimulate Ser935 phosphorylation when neurons were pretreated with the LRRK2 inhibitor MLi-2, indicating that either LRRK2 kinase activity controls the upstream kinases activated by BDNF or that the binding with MLi-2, a type I inhibitor, locks the kinase in a conformation that favors dephosphorylation/hinders phosphorylation during the stimulus.³⁷

To gain mechanistic insights into the effects of BDNF stimulation on LRRK2 function, we differentiated neuroblastoma SH-SY5Y cells with 10 µM retinoic acid (RA) treatment for 6 days³⁸ and measured BDNF response by monitoring Tropomyosin receptor kinase B (TrkB) levels and phosphorylation of AKT and ERK1/2, two major downstream effector pathways of BDNF/TrkB signaling (Fig. 1D).³⁹ Expression of TrkB is undetectable in undifferentiated SH-SY5Y cells, whereas RA-differentiation boosts TrkB expression (Fig. 1E). Accordingly, differentiated SH-SY5Y cells respond to 100 ng/ml BDNF stimulation, as evidenced by increased phosphorylation of ERK1/2 and AKT (Fig. 1F-G). Strikingly, BDNF treatment stimulates Ser935 phosphorylation of endogenous LRRK2 with a phosphorylation kinetics overlapping with that observed in primary neurons (Fig. 1H). These findings support RA-differentiated SH-SY5Y as a valuable model to investigate a novel BDNF-LRRK2 pathway.

To clarify the implications of BDNF-mediated Ser935 phosphorylation on LRRK2 neuronal function, we used affinity purification coupled with mass spectrometry (AP-MS/MS). As the yields of endogenous LRRK2 purification were insufficient for AP-MS/MS analysis, we generated polyclonal SH-SY5Y cells stably expressing GFP-LRRK2 wild-type or GFP control (Supplementary Fig. 1). Then we used GFP-trap purification to isolate GFP or GFP-LRRK2 from RA-differentiated cells unstimulated or stimulated with BDNF for 15 mins prior to MS analysis. The unstimulated LRRK2 interactome (GFP-LRRK2 vs. GFP) is large (207 significant hits) and consistent with previous computational analysis of LRRK2 PPIs.^40,41 In particular, we confirmed interactions with known LRRK2 binders including 14-3-3 proteins, cytoskeletal proteins and protein translation factors (Supplementary table 1). Since BDNF is synaptically released and influences pre- and post-synaptic mechanisms, we used SynGO⁴² to identify synaptic LRRK2 interactors. About one-third (35%) of LRRK2 PPI under unstimulated conditions are SynGO annotated genes; among them, 33% are presynaptic proteins and 66% are post-synaptic proteins (Fig. 2B). The most significant enriched categories were pre- and post-synaptic ribosome and post-synaptic actin cytoskeleton (Fig. 2B).

We next compared unstimulated vs. BDNF-stimulated LRRK2 interactome. As shown in Fig. 2C, 15 minutes of BDNF stimulation reshapes LRRK2 PPI, with a group of interactors increasing binding with LRRK2 (blue dots) and another group decreasing binding (orange dots). The inclusion criteria were either P value < 0.05 and fold change FC > 0.5 | FC < −0.5 or FC > 1 | FC < −1 independently from the P value. In this way, we increased the discovery power for gene ontology analysis. SynGO gene ontology revealed that 10 out of 19 hits with increased binding after BDNF (BDNF(+)) and 21 out of 50 hits with decreased binding (BDNF(-)) are synaptic proteins (Fig. 2C inset). Among synaptic interactors, the BDNF(+) binders are enriched in actin-cytoskeleton-related GO:BP categories while the BDNF(-) binders are enriched in protein translation GO:BP categories (Fig. 2D). Furthermore, 80% of synaptic BDNF(+) LRRK2 binders (DBN1, ARPC2, ACTR2, ACTR3, CFL1, CALD1, ACTN1, FLNA) and 62% of synaptic BDNF(-) binders (RPS2, RPS4X, RPS7, RPS9, RPS11, RPS13, RPS15A, RPL13, RPL23, RPL26, RPL27A, TUFM, EEF1A2, PCBP1) form functional and physical interaction networks (Fig. 2F-E).

To validate these findings, we performed western blot analysis on the AP samples analyzed by MS. We selected drebrin, encoded by DBN1, because it sits at the center of the BDNF(+) network (Fig. 2E, inset SynGO) and, in DIV14 primary neurons, it colocalizes with PSD95, a postsynaptic marker (Fig. 2G), confirming previous observations.⁴³ As predicted, BDNF treatment increases LRRK2:drebrin interaction by ∼2 fold (Fig. 2H-I and Supplementary Fig. 2). We conclude that BDNF stimulation engages LRRK2 into actin cytoskeleton dynamics through physical association with actin-related proteins. This stimulus-dependent function may be relevant at the postsynaptic site (Fig. 2B and 2G). Our findings also suggest that a pool of LRRK2 binds protein translation factors under unstimulated conditions and translocates to the actin-cytoskeleton upon BDNF-TrkB signaling.

We recently investigated the molecular environment surrounding LRRK2 specifically by constructing the protein-protein interaction (PPI) interactome around LRRK2 based on PPIs derived from peer reviewed literature. After filtering the LRRK2 PPIs for reproducibility (keeping only those interactions that have been reported at least twice in literature or at least replicated with 2 different interaction detection methods) we obtained the interactions linking across the LRRK2 interactors thus effectively building the LRRK2 protein interaction network (LRRK2_net) which describes the protein milieu around LRRK2 in the cellular context (see Zhao et al.⁴⁴ for details). The LRRK2_net was topologically analyzed to identify the portions (clusters) of the network that are more densely connected locally in comparison with the overall connectivity of the entire network as these topological clusters are likely to represent functional units.⁴⁵ Among the 11 topological clusters identified in Zhao et al.⁴⁴, we further elaborated on one cluster in particular, containing 41 genes (Fig. 3A) and found that it exhibited significantly GO-BP enriched terms into semantic categories related to: actin polymerization (16 GO-BP terms), synaptic vesicles (5 GO-BP terms), postsynaptic actin-cytoskeleton (2 GO-BP terms), lamellipodium (6 GO-BP terms), cell-neuron projection/size (2 GO-BP terms), DNA recombination/repair (4 GO-BP terms) and uncategorized (6 GO-BP terms) (Fig. 3B and Supplementary table 2). Out of the 41 proteins forming the cluster, 23 are proteins associated with the actin cytoskeleton including the 4 larger nodes (where the dimension of the nodes is directly proportional to the node degree/connections within the network): DBN1/drebrin, CAPZA2 and LIMA whose association with LRRK2 was identified in Meixner et al.⁴⁶ and IQGAP1 identified in Liu et al.⁴⁷ (Fig. 3C). Given the overrepresentation of synapse-related GO-BP terms, we further performed GO functional enrichment using SynGO and mapped 16 synaptic proteins within the cluster (Fig. 3D). Strikingly, DBN1/drebrin is a key protein in the cluster, as it engages in multiple interactions (16 edges) and sits at the interface between the nodes involved in GO:BPs functions related to the actin cytoskeleton (blue-colored nodes) and to the synapse (red-colored nodes) (Supplementary Fig. 3). Taken together, these findings corroborate our experimental interactome network (Fig. 2), supporting the existence of an actin-synaptic LRRK2 network, which appears dynamically shaped by BDNF stimulation.

Based on Ser935 phosphorylation and AP-MS/MS analysis indicating that LRRK2 responds to BDNF (Fig. 1 and 2), we next asked whether LRRK2 knockout could affect BDNF/TrkB downstream cascades. To this end, we generated LRRK2 knockout (KO) SH-SY5Y cells with CRISPR/Cas9 editing (Supplementary Fig. 4), differentiated them with RA and assessed their response to BDNF stimulation. While application of BDNF increases AKT and ERK1/2 phosphorylation in naïve SH-SY5Y, as expected, LRRK2 KO cells exhibit a remarkably weaker response (Fig. 4A-B). These results indicate that LRRK2 acts downstream of TrkB and upstream of AKT and ERK1/2, in line with its rapid and transient phosphorylation kinetic upon BDNF stimulation (Fig. 1B-C).

Likewise, Lrrk2 WT and KO primary cortical neurons at DIV14 were stimulated with BDNF at different timepoints. Similar to differentiated SH-SY5Y cells, Lrrk2 WT neurons responded to BDNF by rapidly increasing Akt and Erk1/2 phosphorylation, whilst KO neurons exhibited a significantly reduced phosphorylation of Akt and Erk1 (Fig. 4D-E). Of note, 90 mins of Lrrk2 inhibition (MLi-2) prior to BDNF stimulation did not prevent phosphorylation of Akt and Erk1/2, suggesting that LRRK2 participates in BDNF-induced phosphorylation of Akt and Erk1/2 independently from its kinase activity but dependently from its ability to be phosphorylated at Ser935 (Fig. 4C-D and Fig. 1B-C).

Considering that (i) BDNF induces dendritic spine formation, maturation and structural plasticity,^48,49 (ii) BDNF(+) LRRK2 interactors are actin-related proteins enriched at the postsynapse (Fig. 2) and (iii) loss of LRRK2 causes reduced phosphorylation of BDNF/TrkB downstream kinases AKT and ERK1/2 (Fig. 4), we next evaluated BDNF-induced spine formation in the presence or absence of Lrrk2. Primary cortical neurons were transfected with GFP filler at DIV4 and treated with BDNF for 24 h at DIV13 (Fig. 4E). Consistent with previous findings,^50,51 BDNF treatment significantly increased the density of dendritic spines in WT cultures (P < 0.001) (Fig. 4F-G). In contrast, BDNF failed to induce spinogenesis in Lrrk2 KO neurons. These data support a crucial role for LRRK2 in spine morphogenesis in response to BDNF.

Based on the in vitro results indicating that BDNF stimulates the recruitment of LRRK2 through Ser935 phosphorylation and binding to actin-cytoskeleton to promote spinogenesis, we next investigated whether loss of Lrrk2 in vivo results in dendritic spine defects. During development, dendritic spines are highly dynamic, with a rate of extension and retraction that allows proper and accurate synapse formation and neuronal circuit assembly.⁵² By the adulthood, dendritic spines formation, motility, and pruning decrease reaching a relatively stable number of protrusions.⁵³ Since spines dynamics vary according to the stage of brain development, we performed a longitudinal analysis of young (1 month-old), adult (4 month-old) and aged (18 month-old) Lrrk2 WT and KO mice in order to capture any age-dependent defect (Fig. 5A). We focused on the dorsal striatum, a region highly enriched in SPNs that receives excitatory afferents from the cortex and dopaminergic modulation from the SNpc. This region is relevant in the prodromal stages of PD neurodegeneration⁵⁴ and expresses high levels of LRRK2^34,55. Moreover, several studies showed a key role of LRRK2 in shaping the structure and function of excitatory synapses in the striatum.^29,30,56,57 We evaluated the number and morphology of dendritic spines using Golgi-Cox staining (Fig. 5B). The total number of protrusions varied across ages, but there were no differences between the two genotypes (Fig. 5C). Instead, the width and the length of the protrusions were reduced in 1 month-old Lrrk2 KO striata with respect to WT. Specifically, the average neck height was 15% shorter and the average head width was 27% smaller, meaning that spines are smaller in Lrrk2 KO brains. No differences were observed at 4 and 18 months of age (Fig. 5C). We then classified the morphology of spines into four categories: filopodia and thin protrusions (immature spines) and mushroom and branched protrusions (mature spines). One month-old Lrrk2 KO animals exhibit a reduced number of filopodia and an increased quantity of thin protrusions, suggesting that loss of Lrrk2 may favor this transition. Instead, the proportion of mature spines (mushroom and branched) remained unaltered. No differences were observed in older mice (Fig. 5D).

Since the overall width/height of Lrrk2 KO protrusions is reduced (Fig. 4C), we next analyzed the postsynaptic ultrastructure using electron microscopy. To increase the experimental power, we used a separate group of animals. We measured the postsynaptic density (PSD) length as it directly correlates with the amount of postsynaptic receptors and signaling proteins, thus providing an indication of the synaptic strength (Fig. 5E). PSD is significantly shorter in 1 month-old Lrrk2 KO animals compared to WT (P < 0.001), while no differences in PSD length are observed in 18 month-old KO mice (Fig. 5F). Taken all these data together, we conclude that subtle postsynaptic maturation defects are present in the striatum of post-natal Lrrk2 KO mice.

To establish whether loss of Lrrk2 in young mice causes a reduction in dendritic spines size by influencing BDNF-TrkB expression, we measured Bdnf and TrkB transcripts, as well as Psd95 and Shank3, two major postsynaptic scaffolds. Neither Bdnf nor TrkB expression is altered in the striatum, cortex or midbrain of 1-month old Lrrk2 KO as compared to WT mice (Fig. 5G), supporting a mechanism whereby LRRK2 is influenced by BDNF signaling, but the absence of LRRK2 does not affect Bdnf or TrkB expression. Instead - and as expected - Psd95 and Shank3 expression is lower in Lrrk2 KO mice and, interestingly, the effect is not limited to the striatum but is also found in the cortex and midbrain.

Given (i) the central role of Dbn1/drebrin in the LRRK2 synaptic network (Fig. 2-3), (ii) its key function as an actin regulatory protein and (iii) its expression highly enriched at dendritic spines (Fig. 2I), we assessed drebrin protein levels at the 3 stages of life in whole brain lysates. Of interest, drebrin expression is significantly lower in 1 month-old Lrrk2 KO mice but not in older animals (Fig. 5H), further linking defective spine maturation with improper actin-dynamics in the absence of LRRK2.

Taken together, our results indicate that LRRK2 influences dendritic spines maturation in the developing brain.

To further explore the connection between LRRK2 and BDNF activity in a cell model more relevant for human disease, we differentiated human induced pluripotent stem cells (hiPSCs) into cortical neurons following established protocols.⁵⁸ Wild-type and isogenic LRRK2 knockout hiPSCs⁵⁹ were differentiated in parallel for 70 days (3 independent culture rounds per genotype), prior to subjecting them to patch clamp to assess spontaneous neurotransmitter release in the absence or presence of acute exposure to BDNF (24 hrs, 50 ng/ml) (Fig. 6A). Acute BDNF treatment led to a significant increase in the mean frequency of miniature post-synaptic excitatory currents (mEPSC) in WT cultures (vehicle: 0.16 ± 0.03 Hz, n=15, N=3; BDNF: 1.05 ± 0.25 Hz, n=10, N=3) (Fig. 6B, left). This is corroborated by the significant leftward shift in cumulative probability curves of interevent intervals (IEIs) between BDNF-treated neurons versus untreated neurons (Fig. 6B, right). In contrast, no significant change in mEPSC frequency of IEI was observed when KO neurons were stimulated with BDNF (pre-BDNF: 1.09 ± 0.21 Hz, n=11, N=3; post-BDNF: 1.11 ± 0.25 Hz, n=12, N=3) (Fig. 6C). BDNF treatment had instead no effect on the average peak amplitude in both genotypes (WT - vehicle: 11.00 ± 1.85 pA (n=14, N=3); BDNF 7.4 ± 0.77 pA (n=10, N=3); KO - vehicle: 10.42 ± 1.53 pA (n=11, N=3); BDNF 11.12 ± 1.76 pA (n=12, N=3)) (Fig. 6D-E). Of note, LRRK2 KO cultures appear to exhibit consistently higher mEPSC frequency (not amplitude) with respect to WT cultures (Fig. 6B-C, Supplementary Table1). As Lrrk2 KO in mouse display maturation abnormalities of dendric spines during development, it is possible that anomalous synaptic maturation also occurs in LRRK2 KO hiPSCs. In this experimental paradigm, we conclude that BDNF mainly affects the presynapse rather than the postsynapse (unchanged mEPSC amplitude), likely increasing neurotransmitter release in WT neurons. Indeed, confocal imaging of synapses shows no increase in the total number of synapses under BDNF stimulation (Supplementary Fig. 5), thus pointing directly to an increase in release probability (P_r) of presynaptic neurotransmitter. Importantly, loss of LRRK2 blunts the positive effect of BDNF on synaptic transmission.

Our data so far indicate that LRRK2 influences synaptic morphology and BDNF-dependent function by engaging in complex formation with actin-cytoskeleton elements. We then asked whether the kinase activity of LRRK2 may also play a role in these processes. To this end, we purified synaptosomes from WT and hyperactive Lrrk2^G2019S knockin (KI) mice and performed phosphopeptide enrichment and MS analysis (Fig. 7A). Synaptosomal preparations are highly enriched in pre- and postsynaptic components.²⁹ We identified 2957 phosphopeptides common between WT and Lrrk2^G2019S synaptosomes. Among them, 343 phosphopeptides showed a significant decrease or increase phosphorylation in Lrrk2^G2019S with respect to WT (Fig. 7B and Supplementary Table 3). The LRRK2 kinase substrates Rabs are not present in the list of significant phosphopeptides, likely due to the low stoichiometry and/or abundance. Next, by overlapping the list of proteins containing at least 1 significant phosphopeptide (267 proteins) (and with a human orthologue) with the list of AP-MS LRRK2 interactors identified in SH-SY5Y cells (207 proteins) (Fig. 7A-B), we found 5 proteins in common (CEP170, HSP90AB1, MAP1B, TUBA1A, and DBN1) highlighted with colored dots in the volcano plot (Fig. 7B). One hundred thousand random lists containing 207 proteins each (to mimic the AP-MS list) were overlapped with the 267 significant genes in the phosphoproteomic dataset resulting in a distribution of overlaps peaking around 2, indicating that 5 overlaps are not a random event. Four (CEP170, HSP90AB1, MAP1B and TUBA1A) were previously reported LRRK2 interactors.^16,60–63 Strikingly, DBN1/drebrin is a significant hit also in this dataset. Drebrin binding with actin is regulated by Cdk5-mediated phosphorylation at Ser142.⁶⁴ In our dataset however, the phosphorylation of this site is not affected by Lrrk2^G2019S activity, instead we observed a significant reduction of a peptide containing two phosphorylated residues (S385 and S387 in mouse and S337 and S339 in human drebrin), previously reported by multiple high throughput studies (https://www.phosphosite.org/homeAction). The lower phosphorylation of this drebrin peptide in Lrrk2^G2019S implies that the effect is indirect. Possible kinases phosphorylating these sites are Cdk5 and PKA (NetPhos 3.1 Server)^65,66, providing clues on regulatory mechanisms underlying LRRK2:drebrin binding.

Next, we performed GO analysis of differentially phosphorylated proteins using SynGO. Biological processes (BP) significantly enriched include synaptic vesicle cycle/exocytosis and postsynaptic structural organization (Fig. 7C). Enrichment by cellular components (CC) similarly indicates synaptic vesicles and postsynaptic density/cytoskeleton as enriched terms (Fig. 7D). This analysis confirms the established role of LRRK2 as a crucial regulator of synaptic vesicles cycle²² and further support a previously underestimated function of LRRK2 in regulating postsynaptic structural plasticity.

C57Bl/6J Lrrk2 knock-out (KO) and G2019S-Lrrk2 knock-in mice were provided by Dr. Heather Melrose (Mayo Clinics, Florida, USA). Housing and handling of mice were done in compliance with national guidelines. All animal procedures were approved by the Ethical Committee of the University of Padova and the Italian Ministry of Health (license 1041/2016-PR and 105/2019) and followed the guidelines approved by the Northwestern University Animal Care and Use Committee. Approximately equal numbers of males and females were used for every experiment.

Human neuroblastoma-derived SH-SY5Y cells (naïve, LRRK2 KO, expressing GFP or GFP-LRRK2 wild type) were cultured in a mixture (1:1) of Dulbecco’s Modified Eagle’s Medium (DMEM, Biowest) and Ham’s F-12 Nutrient Mixture (F12, Biowest), supplemented with 10% (v/v) Fetal Bovine Serum (FBS, Thermo Fisher Scientific) and 1% Penicillin/Streptomycin solution (PS, GIBCO Life Technologies).

To promote N-type (neuronal-like cells) cell differentiation, cells were plated in DMEM/F12 containing 1% PS, 1% FBS and 10μM of all-trans-retinoic acid (RA, Sigma-Aldrich). At regular intervals of 48 hours, RA was newly provided to the medium. Cells were differentiated for 6 days and then subjected to the treatments.

SH-SY5Y cells were starved for 5 hours in DMEM/F12 supplemented with 1% PS and then stimulated with 100ng/mL BDNF (50240-MNAS, Sino Biological) for different time periods (5, 15, 30 and 60 minutes) in serum-free medium. LRRK2 kinase activity was inhibited pre-treating cells with 0.5μM MLi-2 (ab254528, Abcam) for 90 minutes.

Primary cortical neurons from Lrrk2 WT and KO C57BL/6J mice were derived from postnatal mouse (P0) exploiting the Papain Dissociation System (Worthington Biochemical Corporation). Cortices were incubated in Papain solution (Papain and DNase solution in Earle’s Balanced Salt Solution) for 40 minutes at 37°C. Subsequently, tissue was triturated and centrifugated for 5 minutes at 200g. The supernatant was discarded and the pellet was resuspended in Stop solution [DNase solution and Trypsin inhibitor solution (15,5mg/mL) in Earle’s Balanced Salt Solution]. For 10 minutes the tissue was allowed to precipitate and then the supernatant was pipetted drop-by-drop on 5mL of 10/10 solution (10 µg/mL Trypsin inhibitor and 10 µg/mL BSA in Earle’s Balanced Salt Solution) and centrifugated for 10 minutes at 100g. The pellet was resuspended in Neurobasal A medium (GIBCO Life Technologies) supplemented with 2% B27 supplement (GIBCO Life Technologies), 0.5 mM L-glutamine (GIBCO Life Technologies), 100 units/mL penicillin, and 100 μg/mL streptomycin (GIBCO Life Technologies). Cells were diluted and counted in 0.4% Trypan blue. Neurons were plated at 1000-1500 cells/mm² onto 6-well plates or at 200 cells/mm² on 12mm glass coverslips in 24-well plates and maintained in culture for 14 days prior to the treatments. After 7 days, 50% of the Neurobasal medium was removed and replaced with fresh one.

At DIV14, Lrrk2 WT and KO mature neurons cultured in the 6-well plates for western blot analysis were treated with MLi-2 (0.5μM) for 90 minutes and with BDNF (100ng/mL) for 5, 30, 60 and 180 minutes in Neurobasal completed medium. At DIV4, Lrrk2 WT and KO primary neurons plated in 24-well plates were transfected with lipofectamine (Lipofectamine 2000, Invitrogen) in a 1:2 ratio with DNA (v/w). The transfection was carried in Opti-MEM (GIBCO Life Technologies) for 45 minutes. Transfected neurons were then maintained in culture for additional 9-10 days. At DIV14, neurons were treated for 24 hours with 100ng/mL BDNF to study the spinogenesis process.

Neuronal maintenance media (NMM): 1:1 mixture of N2 media (DMEM/F12 GlutaMax supplemented with 1X N2, 50 units/mg/ml Penicillin/Streptomycin solution, 1X MEM Non-Essential Amino Acids Solution (100X), 100µM 2-Mercaptoethanol (ThermoFisher Scientific), 1mM sodium pyruvate and 5 µg/ml insulin (Merck)) and B27 media (Neurobasal supplemented with 1X B27 and 2mM L-Glutamine (ThermoFisher Scientific).

Neuronal induction media (NIM): NMM media supplemented with 100nM LDN193189 (Sigma Aldrich) and 10µM SB431542 (Bio-Techne).

Isogenic human induced-pluripotent stem cell lines (hiPSC) of WT and LRRK2 KO, made and kindly provided by Mark R Cookson’s laboratory,⁵⁹ were cultured using E8 media (Thermo Fisher Scientific) in a 6-well plate coated with Cultrex. Cortical neuronal differentiation was carried out according to Shi, et al.⁵⁸ Briefly, upon reaching 100% confluency, media was changed to neuronal induction media (NIM) and is marked as 0 days in vitro (DIV0). Daily NIM media change was carried out till DIV12, upon which neuronal precursor cells (NPCs) are generated. On DIV12, NPCs were expanded via passage at 1:2 ratio, supplemented with neuronal maintenance media (NMM) media and 20 ng/ml FGF2 (Bio-Techne). Neurogenesis was initiated on DIV18 with the removal of FGF2 and the cells supplemented with NMM only changing every 2 days. At DIV25-30, neurons were cryopreserved in CryoStor CS10 solution (Stem Cell technologies).

For electrophysiological recordings and immunocytochemistry/immunofluorescence (ICC/IF), thawed neurons were plated on Cultrex coated 10mm coverslips at a density of 60k/cm². All cells were treated with 1 ug/ml Mitomycin C (Sigma-Aldrich) for 1 hour at 37°C, 48-72 hours after plating to remove any proliferating cells. Neurons were maintained in NMM until used for experiments. BDNF treatment was carried out with cells incubated with 50 ng/ml BDNF (Bio-Techne) for 24 hours prior to use in electrophysiological recordings and ICC/IF.

Whole-cell patch-clamp recordings were performed on cortical neurons at DIV70-75 (mentioned as DIV70 from here on) in voltage clamp at Vh −70 mV and the membrane test function was used to determine intrinsic membrane properties ∼1 min after obtaining whole-cell configuration, as described previously.⁹⁷ Briefly, coverslip containing neurons were transferred to warner bath (RC-26G) and maintained at 35 ± 2 °C through constant perfusion using in-line heater and controller (Warner Instruments) with extra cellular solution (ECS) containing (in mM unless stated): 145 NaCl, 3 KCl, 2 MgCl₂, 2 CaCl₂, 10 glucose, 10 HEPES, pH 7.4, 310 mOsm. Tetrodotoxin (TTX 0.2μM), and picrotoxin (PTX 100μM), were added before use to block sodium and GABA_A currents. Pipette resistance (Rp) of glass electrodes was 4–6 MΩ when filled with (in mM): 130 Csmethanesulfonate, 5 CsCl, 4 NaCl, 2 MgCl2, 5 EGTA, 10 HEPES, 5 QX-314, 0.5 GTP, 10 Na2-phosphocreatine, and 5 MgATP, 0.1 spermine, pH 7.2, 300 mOsm. Data was acquired by Multiclamp700B amplifier and digidata 1550B and signals were filtered at 2kHz, digitized at 10 kHz, and analyzed in Clampfit 10 (MolecularDevices). Tolerance for series resistance (Rs) was <35 MΩ and uncompensated; ΔRs tolerance cut-off was <20%. mEPSCs were analyzed using Clampfit10 (threshold 5pA); all events were checked by eye. Data are presented as mean ± s.e.m. where n is cells from a minimum of 3 separate cultures (culture N in brackets).

Immortalized cells, neurons and mouse brain tissues were lysed for 30 minutes on ice in appropriate volume of cold RIPA lysis buffer (Cell Signaling Technologies) supplemented with protease inhibitor cocktail. Protein concentration was assessed by performing BCA assay (Thermo Fisher Scientific). Proteins were solubilized in sample buffer (200mM Tris-HCl pH 6.8, 8% SDS, 400mM DTT, 40% glycerol, q.s. Bromophenol Blue) and resolved by SDS-PAGE on 8% Tris polyacrylamide homemade gels in Tris-glycine-SDS running buffer for the AP-MS experiments, and on 4-20% polyacrylamide gels (GenScript® Express Plus PAGE) in Tris-MOPS-SDS running buffer (GenScript® Running Buffer Powder) for the other cases. Proteins were transferred on PVDF (polyvinylidene difluoride, Bio-Rad) membranes using a semi-dry transfer system (Trans-Blot® Turbo Transfer System, Bio-Rad). Non-specific binding sites were blocked incubating membranes with 5% non-fat dry milk diluted in 0.1% Tween-20 Tris-buffered saline (TBS-T) for 1 hour at room temperature under agitation. Membranes were subsequently incubated overnight at 4°C with primary antibodies in 5% non-fat dry milk in TBS-T or in 5% BSA in TBS-T. After three washes in TBS-T at room temperature, membranes were incubated for 1 hour at room temperature with horseradish peroxidase (HRP)-conjugated goat anti-mouse or anti-rabbit IgG. After three more washes in TBS-T, proteins were visualized using chemiluminescence (Immobilon ECL western HRP substrate, Millipore). Densitometric analysis was carried out using the Image J software. The antibodies used for western blotting are as follows: rabbit α-LRRK2 (MJFF2 c41-2, ab133474, Abcam, 1:300); rabbit α-phospho-Ser935 LRRK2 (ab133450, Abcam, 1:300); mouse α-phospho-Ser473-AKT (sc-293125, Santa Cruz Biotechnology, 1:500); rabbit α-AKT (9272S, Cell Signaling Technology, 1:1000); rabbit α-phospho [(Thr202/Tyr204, Thr185/Tyr187)-ERK1/2 (12-302, Millipore, 1:1500); rabbit α-ERK1/2 (4695, Cell Signaling Technology, 1:1000); mouse α-GAPDH (CSB-MA000195, 1:5000); mouse α-βIII tubulin (T8578, Sigma-Aldrich, 1:40000); mouse α-DREBRIN (MA1-20377, Thermo Fisher Scientific, 1:500); α-Flag M2-HRP (A8592, Sigma-Aldrich, 1:5000); α-mouse IgG-HRP (A9044, Sigma-Aldrich, 1:80000); α-rabbit IgG-HRP (A9169, Sigma-Aldrich, 1:16000).

For immunocytochemistry DIV70 cells on coverslips were fixed in 4% paraformaldehyde (PFA) for 10 minutes. Cells were then permeabilized and blocked using 0.4% Tween with 10% normal goat serum (NGS), in PBS for 1 hour. Primary antibodies were incubated overnight at 4°C in PBST plus 10% NGS. Cells were washed 3× with PBST before 1 hour incubation at room temperature with secondary antibodies (α-guinea pig Alexa-488, α-rabbit Alexa 568, α-chicken 647 along with DAPI) from Molecular probes and Jackson Laboratories). Primary antibodies were chicken anti-microtubule associated protein 2 (MAP2) (Antibodies A85363, dil. 1:1000), rabbit anti-homer 1 (Synaptic Systems 160003, dil. 1:500), and guinea pig anti-bassoon (Synaptic Systems 141004, dil. 1:500). Coverslips were slide mounted with FluorSave (Sigma-Aldrich) and all images were acquired on Leica confocal microscope as 0.45 μm z-stacks at 60× magnification (flattened with the max projection function for cluster analysis). Single 60× images were used for analyses. Excitation and acquisition parameters were constrained across all experimental paired (culture) acquisitions. All images were analyzed using IMARIS microscopy image analysis software (v10.0.0 Oxford Instruments). Data are presented as mean ± s.e.m.

Total RNA was isolated from midbrain, striatum, and cortex of 1 month-old animals (n=6 animals per group, Lrrk2 WT and Lrrk2 KO) using Animal Tissue RNA Purification Kit (Norgen), according to manufacturer’s instruction. After extraction, RNA concentration was quantified with NanoDrop 2000C spectrophotometer (Thermo Fisher Scientific). After DNAse treatment (Thermo Fisher Scientific, according to manufacturer’s protocol), complementary DNA (cDNA) was generated using qRT SuperMix (Bimake). The cDNAs were used for quantitative PCR (qPCR) exploiting iTaq Universal SYBR® Green Supermix and CFX96 RealTime System (BioRad) for 40 cycles. All samples were triplicated, and transcripts levels normalized for RPL27 relative abundance (housekeeping gene). Data shown were produced using Bio-Rad CFX Manager software and analyzed according to ddCt algorithm. Primers (5’-3’) used are: Bdnf FW: GGCTGACACTTTTGAGCACGTC and Bdnf REV: CTCCAAAGGCACTTGACTGCTG; TrkB FW: TGAGGAGGACACAGGATGTTGA and Trkb REV: TTCCAGTGCAAGCCAGTATCTG; Shank3 FW: ACCTTGAGTCTGTAGATGTGGAAG and Shank3 REV: GCTTGTGTCCAACCTTCACGAC; Psd95 FW: GGTGACGACCCATCCATCTTTATC and Psd95 REV: CGGACATCCACTTCATTGACAAAC; housekeeping Rpl27 FW AAGCCGTCATCGTGAAGAACA and Rpl27 REV: CTTGATCTTGGATCGCTTGGC.

For the affinity purification (AP) protocol, SH-SY5Y GFP and SH-SY5Y GFP-LRRK2 cells were plated onto 100mm dishes and differentiated for 6 days. SH-SY5Y GFP-LRRK2 cells were treated with BDNF (100ng/mL) or with an equal volume of vehicle for 15 minutes, while SH-SY5Y OE-GFP cells were left untreated. Cells were harvested in lysis buffer (20mM Tris-HCl pH 7.5, 150mM NaCl, 1mM EDTA, 1% Tween 20, 2.5mM sodium pyrophosphate, 1mM β-glycerophosphate, 1mM sodium orthovanadate) supplemented with protease inhibitor cocktail (Sigma-Aldrich). Lysates were incubated end-over-end with GFP-TrapA beads (ChromoTek) overnight at 4°C. Immunocomplexes were subsequently washed ten times with buffers containing a decreasing amount of NaCl and Tween20 and incubated with sample buffer 2X. The eluted proteins were resolved by SDS-PAGE and processed for Western Blot analysis or Mass Spectrometry (MS).

Three biological replicates of GFP-trap purifications of SH-SY5Y cells were processed for the proteomics experiments. Gels were stained with Colloidal Coomassie Brilliant Blue (0.25% Brilliant Blue R-250, 40% ethanol, 10% acetic acid in milli-Q water) for at least 1 hour and then rinsed in destaining solution (10% isopropanol, 10% acetic acid in milli-Q water). Gel slices were cut into small pieces and subjected to reduction with dithiothreitol (DTT 10 mM in 50 mM NH4HCO3, for 1 h at 56 °C), alkylation with iodoacetamide (55 mM in 50 mM NH4HCO3, for 45 min at RT and in the dark), and in-gel digestion with sequencing grade modified trypsin (Promega, 12.5 ng/μL in 50 mM NH4HCO3). Samples were analyzed with a LTQ-Orbitrap XL mass spectrometer (Thermo Fisher Scientific) coupled to a HPLC UltiMate 3000 (Dionex – Thermo Fisher Scientific) through a nanospray interface. Peptides were separated at a flow rate of 250 nL/min using an 11-cm-long capillary column (PicoFrit, 75-μm ID, 15-μm tip, New Objective) packed in house with C18 material (Aeris Peptide 3.6 μm XB C18; Phenomenex). A linear gradient of acetonitrile/0.1% formic acid from 3 to 40% was used for peptide separation and the instrument operated in a data dependent acquisition mode with a Top4 method (one full MS scan at 60,000 resolution in the Orbitrap, followed by the acquisition in the linear ion trap of the MS/MS spectra of the four most intense ions). Two technical replicates were acquired for each biological replicate. Raw data files were analyzed using MaxQuant and Andromeda software package¹⁰⁰ and searched against the human section of the UniProt database (version September 2020, 75093 entries) concatenated with a database of common contaminants found in proteomic experiments. Trypsin was selected as digesting enzyme with up to two missed cleavages allowed, carbamidomethylation of cysteine residues was set as a fixed modification and methionine oxidation as a variable modification. A search against a randomized database was used to assess the false discovery rate (FDR), and data were filtered to remove contaminants and reverse sequences and keep into account only proteins identified with at least two peptides and a FDR ≤ 0.01, both at peptide and protein level.

Intensity values were retrieved for 350 proteins. A single step of data QC was applied to the intensity values: proteins whose intensity value was 0.00 for both the 2 technical replicates in at least 1 of the 3 pull-down experiments performed with LRRK2-GFP were removed. This is a stringent QC step, but it was implemented to remove proteins with no triplicate data available for the LRRK2 pull-down as this could have occurred because of genuine low levels of pull-down (values close to the limit of detection, missing not at random values) or because of technical issues with the single sample (loss/poor quality of sample, missing at random values). QC reduced the number of proteins in the analysis from 350 to 269 entries.

Intensity data were then processed according to Aguilan et al.¹⁰¹ Briefly, intensity data were log2 transformed and normalized to the median of all proteins within the same experiment. Missing data (0.00 intensity values) were considered missing not at random thus they were imputed via probabilistic minimum imputation (random values within less than 2.5 standard deviations from the distribution of intensity values per the single experiment, 0.3 max variability).^101,102

Finally, fold change (GFP-LRRK2 vs GFP and GFP-LRRK2 BDNF treated vs GFP-LRRK2 untreated) and associated p-value (two-tailed, paired t-test) were calculated and visualized.

The LRRK2 interactome was constructed following the procedure reported in Zhao et al.¹⁰³ Briefly, protein interactions reported for LRRK2 in peer-reviewed literature were considered if the interaction was replicated in at least 2 different papers or identified with 2 different interaction detection methods. The obtained LRRK2 protein interaction network was analysed in Zhao et al.⁴⁴ to identify topological clusters using the FastGreed R Package (based on degree centrality). Here we report one of the identified clusters that is enriched for synaptic functions. Functional enrichment was performed using g:Profiler, g:GOSt (https://biit.cs.ut.ee/gprofiler/gost)¹⁰⁴ with Gene Ontology Biological Processes (GO-BPs) and SynGO (https://www.syngoportal.org/).

Statistical analyses and plotting were performed with GraphPad-Prism9. Unpaired, two-tailed t-test or Shapiro-Wilk test were used for experiments comparing two groups; one-way ANOVA was used for experiments comparing three or more groups; two-way ANOVA was used when multiple groups and two factors (i.e., genotype and treatment) were to be compared. Šídák’s multiple comparisons test was used when determining statistical significance for comparisons between groups.