Figure 2A compares the levels of endogenous phosphoRab10 and total Rab10 in parental NIH-3T3 cells, parental cells treated with MLi-2 LRRK2 inhibitor, and a pooled NIH-3T3 cell line in which Rab12 has been knocked out. Quantitation of these data confirmed a roughly fivefold decrease in phosphoRab10 levels under these conditions (Figure 2B). This was entirely unexpected as prior studies on Rab29, a protein that can activate apparent LRRK2 activity under conditions of protein overexpression (Liu et al., 2018; Purlyte et al., 2018); loss of Rab29 did not alter phosphoRab10 levels in a Rab29 mouse knockout model in any tissue analyzed or derived mouse embryonic fibroblasts (MEFs) (Kalogeropulou et al., 2020). We did not analyze Rab8A phosphorylation as the available antibody detects multiple phosphorylated Rab proteins (Steger et al., 2017).

To confirm these data in an animal model, we analyzed cells and tissues derived from Rab12 knockout mice generated by the Knockout Mouse Phenotyping Program at The Jackson Laboratory using CRISPR technology (Figure 2—figure supplements 1 and 2). Immunoblotting analysis of MEFs confirmed that the heterozygous and homozygous knockouts expressed the expected 50 or 100% loss of Rab12 protein (Figure 2C). MEFs derived from homozygous knockout animals showed as much as 50% decrease in phosphoRab10 levels as detected by immunoblots from multiple clones (Figure 2D); specificity of the detection method was confirmed upon addition of the MLi-2 LRRK2 inhibitor that abolished all phosphoRab10 signals. PhosphoRab7, the product of LRRK1 action (Hanafusa et al., 2019; Malik et al., 2021), appeared to increase moderately as a function of Rab12 loss (Figure 2E). Various tissues were analyzed for phosphoRab10 changes in LRRK2 heterozygous and homozygous knockout animals. As shown in Figure 2F–H, decreases in phosphoRab10 were detected in the homozygous mouse lung with smaller trends in the large intestine and kidney. Together, these data confirm a role for Rab12 in the LRRK2 signaling pathway that is distinct from that of the previously studied Rab29 protein. We were not able to monitor loss of phosphoRab10 in the brain as phosphoRab10 is more difficult to detect in brain tissue that is enriched in the Rab-specific PPM1H phosphatase (Berndsen et al., 2019). Future work will evaluate the consequences of Rab12 knockout in mouse brain and other organs.

Since loss of Rab12 decreased phosphoRab10 levels, we reasoned that increasing Rab12 should increase phosphoRab10 levels. Indeed, overexpression of GFP-Rab12 in A549 cells led to a tenfold increase in phosphoRab10 levels without changing the levels of LRRK2, PPM1H phosphatase (Berndsen et al., 2019) or total Rab10 (Figure 3A and B). The ability of Rab12 to activate LRRK2 was specific for that GTPase in that exogenous expression of GFP-tagged Rab8A, Rab10, or Rab29 failed to show the same high level of phosphoRab10 increase – Rab29 yielded about a fivefold enhancement while Rab12 was almost twice as effective in HEK293T cells (Figure 3C and D).

The most common, pathogenic, human LRRK2 mutation is LRRK2 G2019S that displays about twofold higher kinase activity than wild type LRRK2; the R1441C mutation activates kinase activity in cells about threefold (Steger et al., 2016). Cells expressing each of these forms showed increased phosphorylation upon Rab12 expression (Figure 3E and F). It is important to note that Rab12 is a more abundant Rab in most tissues than Rab29; for example, A549 cells contain ~134,000 Rab12 molecules and 25,000 Rab29 molecules per cell. This compares with 5000 copies of LRRK2 and 2.5 million copies of Rab10 (https://copica.proteo.info/#/home). Nevertheless, activation was tested at comparable levels of each Rab protein as monitored using anti-GFP antibodies (Figure 3C).

Rab12 activation of LRRK2 did not require Rab12 phosphorylation as the non-phosphorylatable Rab12 S106A was still capable of activation and a phosphomimetic Rab12 S106E failed to increase LRRK2 phospho S1292 (Figure 3G and H). Phosphorylation state Rab mutants must be used with great caution as we have shown previously that Rab8A and Rab10 TA mutants fail to correctly localize and the TE mutants bind phosphoRab effectors with much lower affinity than their correctly phosphorylated counterparts (Dhekne et al., 2018). Nevertheless, the Rab12 S106A mutant was fully capable of LRRK2 activation.

Similar LRRK2 activation results were obtained using immunofluorescence microscopy to assay phosphoRab10 abundance (Figure 4). The phosphoRab10 generated was present on perinuclear membrane compartments (Figure 4A) as seen previously by many groups (Dhekne et al., 2018; Dhekne et al., 2021; Lara Ordónez et al., 2019). PhosphoRab10 staining disappeared in cells expressing PPM1H but not in cells expressing the catalytically inactive H153D PPM1H (Figure 4A and B). These data were confirmed by immunoblot (Figure 4C and D) and suggest that Rab12 is activating LRRK2 along the same pathway of protein phosphorylation studied previously to date.

It was possible that Rab12 activated a kinase other than LRRK2 to increase Rab10 phosphorylation. This appears not to be the case as GFP-Rab12 expression enhancement of phosphoRab10 levels was not seen in A549 cells lacking LRRK2 expression (Figure 5A and B). It was possible that exogenous GFP-Rab12 inhibited overall Rab phosphatase activity, leading to an apparent increase in phosphoRab10 levels. This was also ruled out as cells lacking PPM1H displayed full Rab12-induced enhancement of phosphoRab10 levels (Figure 5C and D), about fivefold with or without PPM1H.

We showed previously that increased Rab GTPase phosphorylation blocks the formation of primary cilia in cell culture and in specific cell types in the brain (Steger et al., 2017; Dhekne et al., 2018; Sobu et al., 2021). Loss of cilia in cell culture requires Rab10 phosphorylation and its binding to RILPL1 protein (Dhekne et al., 2018). If Rab12 expression increases Rab phosphorylation, it would be expected to interfere with primary cilia formation. We tested this in RPE cells that are well ciliated in culture. As shown in Figure 5E, overexpression of GFP-Rab12 decreased the percent of RPE cells bearing cilia, after 24 hr of serum starvation to trigger cilia formation. Moreover, knockout of Rab12 from A549 cells that poorly ciliate and only ciliate when plated to full confluency, increased the percentage of ciliated cells upon serum starvation, consistent with a decrease in phosphoRab10 (Figure 5F). These experiments show that Rab12 levels regulate primary ciliogenesis downstream of LRRK2 Rab phosphorylation.

Previous work has identified specific residues within the LRRK2 Armadillo domain that enable LRRK2 to be recruited to the Golgi by exogenously overexpressed Rab29; these residues support direct Rab29 binding (McGrath et al., 2021; Vides et al., 2022; Zhu et al., 2022). In particular, R361, R399, and K439 contribute to a Rab binding ‘Site #1’ that supports binding to purified Rab29 (K_D = 1.6 µM; Vides et al., 2022; Figure 6). Rab8A binds this LRRK2 350–550 region with a similar affinity (2.3 µM) but Rab10 binds less well (5.1 µM; Vides et al., 2022). A second site at LRRK2’s N-terminus (Site #2, K17/K18) mediates interaction with phosphorylated Rab8A and Rab10 proteins. Rab GTPase binding to either or both sites contributes to LRRK2 membrane association as Rabs are themselves membrane anchored by two covalently attached, 20 carbon geranylgeranyl groups.

AlphaFold (Jumper et al., 2021) in conjunction with Colabfold in ChimeraX (Mirdita et al., 2022; Pettersen et al., 2004) revealed a third Rab binding site (Site #3) when Armadillo domain residues (1–550) were modeled together with Rab12 (Figure 6B and Figure 6—figure supplement 1). (The Armadillo domain is comprised of residues 1–705; we modeled 1–550 as that portion is biochemically stable and well suited for binding experiments.) The predicted local distance difference test (pLDDT) score (0–100) is a per-residue confidence score, with values greater than 90 indicating high confidence; the top 5 structure models (Figure 6—figure supplement 1) yielded pLDDT scores of 87.6, 87.5, 86.8, 87.4, and 86.4, respectively, consistent with high-accuracy modeling.

Mutagenesis across this putative Site #3 binding interface yielded full-length LRRK2 proteins with decreased overall activity as monitored by phosphoRab10 levels in HEK293 cells expressing the mutant proteins (Figure 7A and Figure 7—figure supplement 1). Note that in these experiments the cells rely only on endogenous Rab12 protein. Mutation of E240 and S244 had the greatest impact on LRRK2 activity; remarkably, mutation of F283 to A increased kinase activity twofold. These data demonstrate that Site #3 sequences are important for overall LRRK2 activity.

Mutation of LRRK2 Site #3 E240R and S244R predicted to be important for Rab12 binding blocked the ability of exogenous Rab12 to enhance phosphoRab10 levels (Figure 7B and Figure 7—figure supplement 1). Moreover, F283A LRRK2 had twofold higher basal activity but was not activated by exogenous Rab12 significantly more than wild type LRRK2 protein. These data strongly suggest that Rab12 activates LRRK2 by binding to Site #3 within the Armadillo domain.

Extensive previous mutagenesis defined Site #1 as being critical for exogenous Rab29-dependent relocalization of LRRK2 to the Golgi complex and apparent activation (Vides et al., 2022). It was therefore important to assess whether Rab29’s ability to increase phosphoRab10 levels upon overexpression relies upon Site #3. As expected, exogenous expression of Rab29 increased phosphoRab10 levels (albeit to a lower extent than exogenous Rab12 expression; Figure 3C and D). However, mutation of Site #3 residues critical for Rab12-mediated LRRK2 activation (E240 and S244) had no effect on the ability of Rab29 to activate LRRK2 kinase (Figure 7C). Similarly, mutation of Site #1 residues preferentially decreased the ability of Rab29 to activate LRRK2 with little if any change in Rab12 activation (Figure 7D). These experiments show that Rab29 interacts preferentially with Site #1 and demonstrate the Rab12 selectivity of Site #3 for LRRK2 activation.

These experiments strongly suggest that Rab29 and Rab12 activate LRRK2 by two different routes: Rab29 via binding to LRRK2 Site #1 and Rab12 via binding to Site #3. We validated Rab12 direct binding to Site #3 using purified Rab12 and Armadillo domain proteins mutated at either Site #1 (K439E) or Site #3 (E240R). As shown in Figure 8, Rab12 bound as well to the wild type Armadillo domain (Figure 8A, 1.4 µM) as to an Armadillo domain construct bearing a Site #1 mutation (Figure 8B, 1.6 µM) as determined by microscale thermophoresis. In contrast, the Site #3 E240R mutation abolished the interaction, yielding a K_D of >40 µM (Figure 8C). Thus, Rab12 binds tightly and directly to Site #3 in vitro and does not appear to interact with Site #1. Interestingly, the LRRK2 Site #3 F283A mutation that increases kinase activity in cells did not influence Rab12 binding significantly, displaying a K_D of 1.2 µM (Figure 8D).

Binding of Rab12 to LRRK2 Site #3 was also detected in cell extracts in co-immunoprecipitation experiments. As shown in Figure 8E and F, HA-tagged Rab12 and endogenous Rab12 proteins co-precipitated with FLAG-LRRK2 upon transfection in HEK293T cells. In contrast, significantly less co-precipitation was seen with LRRK2 Site #3 mutant E240R and S244R proteins, with or without exogenous HA-Rab12 expression. Rab12 bound F283A LRRK2 as well as wild type LRRK2 protein, consistent with its binding affinity in vitro.

We showed previously that phosphoRab binding to Rab binding Site #2 is critical for cooperative LRRK2 membrane recruitment and apparent activation (Vides et al., 2022). Thus, it was important to investigate whether Rab12 acts via this feed-forward process. If true, such activation would be predicted to rely on LRRK2 Lys17 and Lys18. As shown in Figure 8G, mutation of Lys17 and 18 had no effect on the ability of Rab12 to increase phosphoRab10 levels in HEK293T cells co-expressing exogenous LRRK2 and GFP-Rab proteins. Once again, Rab12 activation was dramatic and K17/K18 containing-LRRK2 was activated to the same overall level as the K17A/K18A mutant LRRK2 protein. These data are consistent with our finding that non-phosphorylatable Rab12 S106A is still capable of LRRK2 activation (Figure 3G and H).

As mentioned earlier, under conditions of lysosomal damage, LRRK2 is recruited to lysosomes and participates in the repair of damaged endomembranes (Eguchi et al., 2018; Herbst et al., 2020; Bonet-Ponce et al., 2020). Such stress greatly increases LRRK2 kinase activity (Kalogeropulou et al., 2020). Figure 9A–C show that Rab12 is required for the modest increase in LRRK2 activity seen upon lysosomal damage triggered by 1 mM LLOME addition for 2 hr in NIH-3T3 cells. In MEFs (Figure 9D and E), loss of Rab12 dampened but did not abolish the increase in phosphoRab10 levels, especially at later times. Upon treatment of NIH-3T3 cells with nigericin that also causes mitochondrial stress and is a potent activator of the NLRP3 inflammasome (Figure 9F–H), Rab12 knockout diminished Rab10 phosphorylation to control levels. These findings point to the contribution of Rab12 in regulating LRRK2 activity in lysosome repair.

HEK293T, HEK293, NIH-3T3, RPE, A549, and A549 CRISPR knockout lines for LRRK2 and PPM1H (Berndsen et al., 2019) were cultured in high-glucose DMEM supplemented with glutamine, sodium pyruvate, and penicillin-streptomycin. All cells were regularly tested for Mycoplasma PCR products using a Lonza Mycoplasma kit. Before the screen, cells were cultured in the presence of plasmocin as prophylaxis against Mycoplasma infection.

Generation of Cas9 expressing NIH-3T3 cells is described in full detail on protocols.io (Dhekne et al., 2022b). Briefly, NIH-3T3-Flpin cells were from Thermo Fisher. Early passage cells were transduced with lentivirus carrying HA-Cas9 (Addgene). Cells were selected with blasticidin and single-cell sorted onto a 96-well plate. After 2 wk of culture, 20 individual colonies were picked, expanded, and 5 were analyzed for Cas9 expression and phosphoRab10, LRRK2, and good growth. The two best clones were tested along with a known positive control lentiviral sgRNA, selected with puromycin, and immunoblotted on day 5 to estimate knockout efficiency.

Wild type, heterozygous, and homozygous Rab12 knockout MEFs were isolated from littermate matched mouse embryos at day E12.5 resulting from crosses between heterozygous Rab12 KO/WT mice using the protocol described in Tonelli, 2023a. Genotypes were verified via allelic sequencing and immunoblotting analysis. Cells were cultured in DMEM containing 10% (v/v) FBS, 2 mM L-glutamine, penicillin-streptomycin 100 U/mL, 1 mM sodium pyruvate, and 1× non-essential amino acid solution (Life Technologies, Gibco).

The BRIE library from Addgene was expanded according to protocols.io (Dhekne et al., 2022a). Briefly, the DNA library was electroporated into Lucigen Endura Duos bacteria and the cells plated onto large format Luria broth agar plates to obtain single colonies across the plate. These plates were grown for 14 hr at 37°C and plasmid extracted using a Machery-Nagel mega-prep kit. Expanded library was PCR amplified using Illumina barcoded PCR primers as described on Addgene and are part of Supplementary file 1. PCR products were sequenced with Miseq to confirm uniform distribution of the gRNA sequences across the population. Aliquots of the plasmid library were frozen at –80°C for future use.

The detailed protocols can be found on protocols.io (Dhekne et al., 2022a and Dhekne et al., 2022b).

Briefly, the screen was performed maintaining a 300× fold representation of guides in the transduced cells (Pusapati et al., 2018). For ~79,500 gRNAs, NIH-3T3-Cas9 cells were plated in 20, 15 cm dishes at 5 × 10⁶ cells per dish. Lentiviral gRNAs were infected at an MOI of 0.2 (for ~100 × 10⁶ cells, ~20 × 10⁶ virus particles). After 48 hr, cells were passed into 60, 15 cm dishes with 1 µg/ml puromycin for selection. After 72 hr, cells in the control plate that did not receive the virus were dead. Puromycin-resistant NIH-3T3-Cas9-BRIE cells were pooled and frozen in cryovial aliquots. Four days before the flow cytometry assay, 40 × 10⁶ cells were thawed and plated into 10, 15 cm dishes and allowed to attach and grow for 3 d. On the fourth day, cells were trypsinized, resuspended to a cell density of 5 × 10⁶ cells/mL, passed through a 40 µm cell strainer and fixed with 3% PFA for 30 min, permeabilized with 0.2% Saponin for 30 min, and stained overnight at 4°C with rabbit anti-phosphoRab10 antibody at 1 µg/mL. Cells were then washed and stained with goat anti-rabbit 647 antibody diluted 2 µg/mL for 1 hr at room temperature (RT). Cells were washed, resuspended to 2 × 10⁶ cells/mL, and injected into a Sony SH800 sorter with FSC of 1, FL4 PMT with a gain of 40%, and sample pressure maintained at level 6. MLi-2-treated and secondary antibody-alone samples were used as negative controls to identify cell population gates. Cells treated with 4 µM nigericin for 3 hr were positive controls for the detection of high level of phosphoRab10.

Cells were sorted based on the histogram of Alexa-647 fluorescent signal. The lowest 5% and highest 7.5% signal containing gates were sorted into two 5 mL collection tubes until each had at least 2 × 10⁶ cells. To control for total distribution of gRNAs across the population, 10 × 10⁶ unsorted cells were reserved as input sample. This exercise was performed on two independent sorts from two independent stainings. Sorted cells were pelleted and stored at –80°C for genomic DNA isolation.

For genomic DNA extraction, frozen cells were thawed, uncrosslinked, and genomic DNA (gDNA) extracted according to Dhekne et al., 2022c. All primers used for PCR amplification for next-generation sequencing (NGS) were ordered as Polypak cartridges purified from the Protein and Nucleic Acid facility, Stanford University. Those used for cloning were ordered unpurified. Primer sequences can be found in Supplementary file 1.

Variable sequences were incorporated into forward primer sequences to improve diversity in the NGS run, and eight such primers were pooled in equimolar ratio (Addgene-P5-F [0–8]). Reverse primers were incorporated with TrueSeq indices. PCR was performed as described in protocols.io (Dhekne et al., 2022a). Briefly, input plasmid library and each of the genomic DNA libraries were amplified using Titanium-Taq polymerase. PCR products were cleaned up and size selected using Ampure magnetic beads and concentrated by eluting in small volume, quantified with Qubit high-sensitivity dsDNA assay, and finally amplicon size confirmed on an Agilent Bioanalyzer. Each PCR amplicon library (two replicates each of unsorted, low phosphoRab10 and high phosphoRab10) was mixed at equimolar ratio and sequenced at Novogene Co, CA, using their 150 × 2 HiSeq platform.

Raw sequencing reads were mapped to sgRNA sequence guides in the BRIE library using a modified version of count_spacer.py script (Joung and Gootenberg, 2016; Joung et al., 2017), which provided the count of each sgRNA in each sample. For quality control, evenness of the sgRNA representation was visually assessed by plotting the cumulative distribution of sgRNA representation and quantified using the Gini Index. All samples had a Gini Index lower than 0.42. Consistency between replicates was measured using the Spearman correlation of the sgRNA counts. These quality metrics were computed using Python in a Jupyter Notebook available on GitHub (Limouse, 2023).

Large-scale lentiviral preparation for generating pooled lentiviral gRNA libraries was performed according to a modified protocol from Joung et al., 2017 and is published on protocols.io (Dhekne et al., 2022a). Briefly, low-passage HEK293T cells were transfected with BRIE library along with the packaging plasmids and viral supernatant was collected 48 hr (day 2) and 72 hr (day 3) post-transfection. These two separate days of supernatants were pooled, filtered through 0.45 µm, and frozen at –80°C. An aliquot of the frozen virus was used for titration such that <30% of the cells were transduced and showed puromycin resistance. An estimate of the number of virus particles/µL was made. For small-scale preparations of lentiviruses to express individual gRNAs or GFP-tagged Rab GTPases, a standard lentiviral protocol was used as is published in protocols.io (Dhekne and Pfeffer, 2022a).

For individual cell lines, RPE and A549 cells were transduced with the relevant virus (GFP, GFP-Rab12, wtPPM1H-mApple, PPM1H H153D-mApple, PPM1H-D288A mApple) and 5 µg/mL polybrene. After 72 hr, cells were either selected for protein expression with puromycin or sorted for the relevant fluorescent protein expression. Sorted cells were tested for protein expression by immunoblot.

HEK293 cells were seeded into 10 cm plates and transiently transfected at 70–80% confluence using Lipofectamine 2000 transfection reagent with FLAG-tagged LRRK2 wild type or variant plasmids and HA-Rab12 or HA-empty. Cells were lysed 24 hr post-transfection in ice-cold lysis buffer containing 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EGTA, 270 mM sucrose, supplemented with 1× phosSTOP phosphatase inhibitor cocktail (PhosSTOP tablet: Roche, REF# 04906837001), 1× protease inhibitor cocktail (cOmplete EDTA-free protease inhibitor cocktail tablet: Roche, REF# 1187358000) and 0.1% (v/v) NP40-Alternative. 1 mg of whole-cell lysate was used to immunoprecipitate LRRK2 with 25 µL anti-FLAG M2 resin for 1 hr at 4°C. Immunoprecipitates were washed three times with 50 mM Tris–HCl pH 7.4, 150 mM NaCl, and eluted by adding 25 µL of 2× lithium dodecyl sulfate (LDS) loading buffer to the resin. A detailed method can be found on protocols.io (Tonelli, 2023b).

The Rab12 knockout mouse strain used for this research project, C57BL/6N-Rab12em1(IMPC)J/Mmucd (RRID:MMRRC_049312-UCD) was obtained from the Mutant Mouse Resource and Research Center (MMRRC) at the University of California at Davis and was donated to the MMRRC by the KOMP Repository, University of California, Davis (originating from Stephen Murray, The Jackson Laboratory). Mice selected for this study were maintained under specific pathogen-free conditions at the University of Dundee (UK). All animal studies were ethically reviewed and carried out in accordance with the Animals (Scientific Procedures) Act 1986 and regulations set by the University of Dundee and the U.K. Home Office. Animal studies and breeding were approved by the University of Dundee ethical committee and performed under a U.K. Home Office project license. Mice were housed at an ambient temperature (20–24°C) and humidity (45–55%) and were maintained on a 12 hr light/12 hr dark cycle, with free access to food and water. For the experiments described in Figure 2 and Figure 2—figure supplement 1, 3-month-old littermate or age-matched mice of the indicated genotypes were injected subcutaneously with vehicle (40% [w/v] (2-hydroxypropyl)-β-cyclodextrin; Sigma-Aldrich #332607) or MLi-2 dissolved in the vehicle at a 30 mg/kg final dose. Mice were killed by cervical dislocation 2 hr following treatment, and the collected tissues were rapidly snap frozen in liquid nitrogen.

For individual gene knockout validation by microscopy, NIH-3T3-Cas9 cells were transduced with sgRNA lentiviruses for 48 hr, then selected for 3 d with 1 µg/mL puromycin. On day 6, cells were plated at 30% confluency (75,000 cells) on coverslips in a 24-well plate. After 24 hr, cells were washed and fixed with 3% paraformaldehyde for 30 min at RT, permeabilized with 0.1% Saponin for 30 min, blocked with 2% BSA, and stained with rabbit anti-phosphoRab10 and mouse anti-p115 polyclonal antibody for 2 hr at RT.

A549 cells stably expressing GFP-Rab12 and PPM1H-mApple were co-plated with parental A549 cells on coverslips for 24 hr. Cells were then fixed, stained, and imaged for phosphoT73 Rab10 as described below. Cells were washed and stained with DAPI (0.1 µg/mL), donkey anti-mouse 488, and donkey anti-rabbit 568 (1:2000) for 1 hr at RT. After washing the secondary antibody, coverslips from all wells were mounted on slides using Mowiol. Staining of cells for immunofluorescence is described in the protocol (Dhekne and Pfeffer, 2022b). After the coverslips dried, unbiased multi-position images were obtained using a spinning disk confocal microscope (Yokogawa) with an electron multiplying charge coupled device (EMCCD) camera (Andor, UK) and a 100 ×1.4 NA oil immersion objective. Image acquisition was performed using the multidimensional acquisition using Metamorph. All images were analyzed using an automated pipeline built using Cell Profiler. Whole-cell intensities of phosphoRab10 were extracted as median and mean intensities of phosphoRab10 across the cell. Given the non-uniform nature of the phosphoRab10 dispersal inside cells, median intensity across cell was used for plotting graphs. Images histograms were adjusted on Fiji and are presented as maximum intensity projections.

Figures were made in Adobe illustrator. Graphs and statistical analyses were performed in GraphPad Prism.

His-Rab12 Q101L, His-LRRK2 Armadillo WT, K439E, E240R, and F283A were purified after expression in Escherichia coli BL21 (DE3 pLys). Detailed protocols can be found in Gomez et al., 2019, Gomez et al., 2020, Dhekne et al., 2021 and Vides and Pfeffer, 2021. Bacterial cells were grown at 37°C in Luria Broth and induced at A600 nm = 0.6–0.7 by the addition of 0.3 mM isopropyl-1-thio-β-d-galactopyranoside (Gold Biotechnology) and harvested after growth for 18 hr at 18°C. The cell pellets were resuspended in ice-cold lysis buffer (50 mM HEPES, pH 8.0, 10% [vol/vol] glycerol, 500 mM NaCl, 10 mM imidazole, 5 mM MgCl₂, 0.2 mM tris(2-carboxyethyl) phosphine [TCEP], 20 μM GTP, and EDTA-free protease inhibitor cocktail [Roche]). The resuspended bacteria were lysed by one passage through an Emulsiflex-C5 apparatus (Avestin) at 10,000 lbs/in² and centrifuged at 40,000 rpm for 45 min at 4°C in a Beckman Ti45 rotor. Cleared lysate was filtered through a 0.2 µm filter (Nalgene) and passed over a HiTrap TALON crude 1 mL column (Cytiva). The column was washed with lysis buffer until absorbance values reached pre-lysate values. Protein was eluted with a gradient from 20 to 500 mM imidazole containing lysis buffer. Peak fractions were analyzed by 4–20% SDS-PAGE to locate protein. The eluate was buffer exchanged and further purified by gel filtration on Superdex-75 (GE Healthcare) with a buffer containing 50 mM HEPES, pH 8, 5% (vol/vol) glycerol, 150 mM NaCl, 5 mM MgCl₂, 0.2 mM tris(2-carboxyethyl) phosphine (TCEP), and 20 μM GTP.

A detailed method can be found on protocols.io (Vides and Pfeffer, 2021).

Protein–protein interactions were monitored by microscale thermophoresis using a Monolith NT.115 instrument (NanoTemper Technologies). His LRRK2 Armadillo (1–552) WT, K439E, E240R, and F283A were labeled using RED-NHS 2nd Generation (Amine Reactive) Protein Labeling Kit (NanoTemper Technologies). For all experiments, unlabeled Rab12 was titrated against a fixed concentration of the fluorescently labeled LRRK2 Armadillo (100 nM); 16 serially diluted titrations of the unlabeled protein partner were prepared to generate one complete binding isotherm. Binding was carried out in reaction buffer (50 mM HEPES pH 8, 150 mM NaCl, 5 mM MgCl₂, 0.2 mM tris(2-carboxyethyl) phosphine [TCEP], 20 μM GTP, 5% [vol/vol] glycerol, 5 μM BSA, 0.01% Triton-X) in 0.5 mL Protein LoBind tubes (Eppendorf) and allowed to incubate in the dark for 30 min before loading into NT.115 premium-treated capillaries (NanoTemper Technologies). A red LED at 20% excitation power (red filter, excitation 605–645 nm, emission 680–685 nm) and IR-laser power at 60% was used for 30 s followed by 5 s of cooling. Data analysis was performed with NTAffinityAnalysis software (NanoTemper Technologies) in which the binding isotherms were derived from the raw fluorescence data and then fitted with both NanoTemper software and GraphPad Prism to determine the Kd using a nonlinear regression method. The binding affinities determined by the two methods were similar. Shown are averaged curves of Rab GTPase-binding partners from two independent experiments, with averaged replicates from each run.