Elevated LRRK2 autophosphorylation in brain-derived and peripheral exosomes in LRRK2 mutation carriers

All protocols were approved by the local Institutional Review Boards. Participants were recruited in the outpatient clinic of the Department of Neurology, St. Olavs hospital, Trondheim, Norway. Urine specimens were collected for this study from 132 participants that included healthy controls (LRRK2- PD-), age-matched G2019S-LRRK2 mutation carriers without manifest PD (LRRK2+ PD-), G2019S-LRRK2 mutation carriers with PD (LRRK2 + PD+), and non-LRRK2 mutation carriers with PD (i.e., idiopathic PD, LRRK2- PD+, see Table 1). CSF specimens were collected from 82 participants (Table 2). Of these subjects, 55 donated both urine and CSF in the same clinic visit and were analyzed here (Table 3). All lumbar punctures were performed between 8 and 9 AM on overnight fasting subjects, and urine samples were collected thereafter to around 10 AM. A small amount of CSF was immediately sent for cell counts and tests for glucose and proteins. Urine and CSF was centrifuged within 15 minutes after the end of collection, aliquoted, and biobanked.

Urine and CSF samples were processed and analyzed by an investigator blinded to their identify. CSF hemoglobin level was determined using ELISA method (Bethyl Labs, #E88-134). Urine was assessed for leukocyte count, pH, glucose, total protein, red-blood cells, and specific gravity via test strips (Fisher). All participants were clinically evaluated for PD severity using the Unified Parkinson’s Disease Rating Scale (UPDRS) part III, Hoehn & Yahr scale, modified Schwab & England scale, Montreal Cognitive Assessment (MoCA), and L-dopa equivalent daily dose (LEDD) calculated as described [8].

Supernatants from cell cultures were centrifuged 10,000 x g at 4 °C for 30 min, supernatant and centrifuged at 100,000 x g for 1 hour at 4°C, and resultant pellets lysed in SDS buffer (2% SDS, 10% glycerol, 60 mM Tris-CL, pH 6.8, and 5% dithiothreitol). HEK293 cells and primary macrophages were cultured and transfected with LRRK2 plasmids as previously described [7, 18]. Urine, CSF, or serum (Supplemental) samples were quick-thawed in a shaking 42 °C water bath and placed on ice after thawing. Samples were centrifuged at 10,000 x g at 4 °C for 30 min, supernatant centrifuged at 100,000 x g for 1 hour at 4 °C, and the resultant exosome-enriched pellet was washed in 1 mL saline and centrifuged at 100,000 x g for one hour at 4 °C. Randomly selected exosome-enriched pellets were analyzed via single-particle tracking Nanosight analysis. Representative vesicle size and concentration traces were recorded over five runs (60 sec per run) each and analyzed with NTA software (Nanosight).

Full-length recombinant Flag-G2019S-LRRK2 protein was purchased from Invitrogen and purity was assessed by Coomassie-stain SDS-PAGE. Concentrations of LRRK2 protein were assessed by BCA assay (Pierce). LRRK2 protein (100 nM) was incubated with 200 μM ATP in buffer containing 150 mM NaCl, 50 mM Tris-HCl, and 10 mM MgCl₂ at 37 °C for up to 60 min. LRRK2 protein standards were digested with endoproteinase Glu-C (Roche) in 50 mM Tris-HCl (pH7.0) at 37 ⁰C overnight, and heavy isotope labeled peptides MGK^LSKIWDLPLDE and MGK^L(pS)KIWDLPLDE (New England Peptide, K^: ¹³C₆ ¹⁵N₂-lysine) were added to samples that were injected into a 300 μm x 5 mm Thermo μ-Precolumn (C18 Pepmap 100, 5 μm, 100 Å pore). After washing with 0.1% formic acid and 2% acetonitrile in water at 20 μL per min, 35 ⁰C, for 4.5 min, samples were separated on a 75 μm x 150 mm Thermo EASY-Spray column (C18 Pepmap, 3 μm particle, 100 Å pore) using a Thermo Ultimate 3000 RSLCnano LC system. Mobile phase A consisted of 0.1% formic acid in water and mobile phase B consisted of 0.1% formic acid in acetonitrile. The peptides were eluted using a linear gradient from 2% to 90% mobile phase B at 400 nL per min, 40 ⁰C, over 15.5 min. The phospho and non-phospho S1292-containing peptides and their internal standards were detected with a Thermo Q Exactive plus Orbitrap mass spectrometer running PRM (parallel reaction monitoring). The electrospray voltage, capillary temperature, and S-lens RF were 2.0 kV, 300 ⁰C, and 50 V, respectively. The b12 and b13 fragments (m/z values as indicated below) of the peptides were summed to generate ion chromatograms, and ratios between the peak areas of the light and heavy peptide were used for quantifying the phospho and non-phospho S1292 peptides from recombinant LRRK2 protein to calculate the degrees of phosphorylation. Detection limit for the unphosphorylated peptide (MGKLSKIWDLPLDE) was estimated to be approximately 100 pg mL^-1 while it was approximately 300 pg mL^-1 for the phosphorylated peptide (MGKLpSKIWDLPLDE).

To measure the amount of total LRRK2 and pS1292-LRRK2 in clinical samples, a pool was first created by combining 1/10 (w/v) of all samples from the experiment used in each analytical run. The amount of pS1292-LRRK2 and total LRRK2 protein in the pool sample was determined by comparing the pool together with recombinant protein analyzed by mass spectrometry to determine absolute values of total LRRK2 and pS1292-LRRK2 via immunoblot method (see Additional file 1: Figure S1 and S2). The amount of pS1292-LRRK2 and total LRRK2 protein in the HEK293 cell and exosome samples was determined similarly, with WT-LRRK2, pathogenic mutations G2019S, R1441G and double LRRK2 mutation R1441C/G2019S samples that covers the entire signal range of LRRK2 and comparing with recombinant protein previously analyzed by mass spectrometry. Samples were electrophoresed on 4–20% TGX gradient gels (BioRad) and transferred onto PVDF membranes (Immobilon-FL, Millipore). Membranes were blocked using 5% non-fat milk in TBST and cut in half for analysis of LRRK2 in the top half and exosomal housekeeper proteins Tsg101 or flotillin-1 in the bottom half. Digital signal intensities were recorded with the ChemiDoc Touch platform and ImageLab software (BioRad). Linearity of digital signals were confirmed across the possible dynamic range (10-600pg depending on antibody conditions, see Additional file 1: Figure S1 and S4).

The following antibodies were used: N241A/34 anti-LRRK2 (Antibodies Inc), MJFF2 c41-2 anti-LRRK2 (Abcam), UDD3 anti-LRRK2 (Abcam), UDD2-10 anti-pS935 LRRK2 (Abcam), MJFR-19-7-8 anti-pS1292-LRRK2, UDD3 anti-LRRK2 (Abcam), D2V7J anti-flotillin-1 (Cell Signaling), and polyclonal antibodies to phospho-threonine (P-Thr-Poly 9381, Cell Signaling) and Tsg101 (ab30871, Abcam).

For some immunoblots, after transfer of protein was completed, membranes were blocked with 3% BSA (Jackson ImmunoResearch) in TBS-T and treated with or without 250 units mL^-1 calf intestinal alkaline phosphatase (CIP, New England BioLabs) in CutSmart buffer (New England BioLabs) for 2 hours at 37 ⁰C. Quantification of proteins on the membranes then proceeded as described above.

Previously, we demonstrated that heterozygous G2019S-LRRK2 mutations increased the ratio of pS1292-LRRK2 to total LRRK2 protein ~4-fold in urinary exosomes of G2019S carriers as compared to non-carriers [6]. These and other measurements from model systems have not revealed the proportion of LRRK2 protein phosphorylated at pS1292, as only relative ratios of signal have been presented in past studies [6, 8, 14, 23]. Further, it is not known how different LRRK2 mutations and different cellular and exosome sources may differ in the amount of pS1292-LRRK2 with respect to other LRRK2 kinase substrates like Rab10 [26]. In measuring pS1292-LRRK2 from in vitro LRRK2 kinase reactions, we observed a low-starting level of pS1292-LRRK2 in recombinant G2019S-LRRK2 protein that increased over time, as expected (Fig. 1a). In contrast, the constitutive phosphorylation site pS935 that is not a LRRK2 autophosphorylation site did not change over time. Other threonine LRRK2 autophosphorylation sites have been described in the LRRK2 ROC GTPase domain [10, 12, 31]. Cumulatively, these increased in abundance over time like pS1292 as revealed by pan-phospho-threonine detection (Fig. 1a). Thus, in vitro, pS1292-LRRK2 levels correlate closely with overall autophosphorylation. Using a novel mass spectrometry approach to quantify pS1292-LRRK2 over time in recombinant protein, we could determine that 45% of the LRRK2 protein was pS1292-LRRK2 after one-hour (Fig. 1b). Cellular levels of immunoprecipitated pS1292-LRRK2 (before kinase assays) demonstrated only ~2% of total LRRK2 is pS1292-LRRK2 (Fig. 1a, b). To verify the specificity of the pS1292-LRRK2 antibody for a phosphorylated peptide, we introduced a Ser1292Ala mutation into the G2019S-LRRK2 backbone and found that the Ser1292Ala mutation blocks pS1292-LRRK2 antibody binding (Fig. 1c).

We next compared the effect of pathogenic LRRK2 mutations on pS1292-LRRK2 levels in both cell extracts from transfected HEK293 cells as well as extracellular exosomes purified from their cell culture media. As expected, all the pathogenic LRRK2 mutations increased cellular levels of pS1292-LRRK2 (Fig. 1d), with the highest levels induced by a double-mutation R1441C / G2019S, also known to enhance neurotoxicity in primary neurons [23]. pS1292-LRRK2 levels were similar between WT-LRRK2, the prevalent Far-East Asian G2385R-LRRK2 risk variant, and the non-pathogenic I2012A-LRRK2 variant. Through comparison to recombinant pS1292-LRRK2 standards defined by mass spectrometry (and in linear range, see Additional file 1: Figure S2), we deduced the percent of pS1292-LRRK2 in cells in the double-LRRK2 mutant as ~27% of available LRRK2 protein (Fig. 1d,e).

Exosome protein levels do not usually reflect the levels of proteins and their post-translational modifications found in cell cytosols due to the mechanisms that select for exosome cargo and transport in the endolysosomal system [28]. However, exosomes from the transfected HEK293-derived exosomes revealed similar pS1292-LRRK2 levels as compared to the parental cellular lysates (Fig. 1d, e). The effects of the double-mutation G2019S/R1441C appeared slightly exaggerated compared to other mutations, and in general, pS1292-LRRK2 levels were slightly lower on average (Fig. 1e). Analysis of the exosome pellets from these cells via single-particle nanotracking revealed a typical microvesicle population that had an average size of 134 nm (Fig. 1f). Thus, in these cells, exosomes reported pS1292-LRRK2 levels that were consistent with cellular levels that were increased by pathogenic LRRK2 mutations.

In humans, LRRK2 expression is highest in circulating leukocytes in the blood according to RNAseq tissue expression profiles [32]. Leukocytes like macrophages and monocytes may secrete exosomes to control some aspects of immune reactions [15]. To measure endogenous LRRK2 protein in macrophages in culture, we isolated exosomes and cellular lysates from wild-type or G2019S-LRRK2 expressing macrophages. In cellular lysates, we could not reliably detect any pS1292-LRRK2 in wild-type macrophages, whereas pS1292-LRRK2 levels were robust with G2019S-LRRK2 expression (Fig. 1g). pS1292-LRRK2 signal, as well as pRab10 signal (a trans LRRK2 kinase substrate), was ablated with one-hour treatment of a LRRK2 kinase inhibitor. However, we were unable to detect pS1292-LRRK2 or total LRRK2 protein in exosomes isolated from primary macrophages, irrespective of LPS stimulation. These results suggest that leukocytes like macrophages that express high LRRK2 levels may not be an important source for LRRK2 protein in exosomes.

Previously we detected total LRRK2 protein in exosome-enriched fractions purified from post-mortem remnant CSF [7]. To determine whether we can detect pS1292-LRRK2 in biobanked clinical CSF samples verified to have low or no detectable hemoglobin, we analyzed three samples from neurologically normal controls provided by the BioFIND repository. Guided by Nanosight analysis of microvesicle fractions, we applied a modified differential ultracentrifugation strategy to isolate an exosome fraction that harbors vesicles with an average size of ~125 nm (Fig. 2a, b). These exosomes were similar in characteristics to those isolated from HEK293 cells. Using a reference pS1292-LRRK2 protein, we detected appreciable signal for total LRRK2 and pS1292-LRRK2, but only in the exosome-enriched fraction and not in supernatants or low speed (10,000 x g) pellets (Fig. 2c). The signal elicited by the pS1292-LRRK2 antibody in the human CSF exosome pellet was destroyed by brief treatments of the membranes with calf intestinal alkaline phosphatase (CIP), indicating that the cross-reactive band is a phospho-protein and of the exact molecular weight as our full-length recombinant protein standard (~280 kDa, Fig. 2d). As opposed to HEK293, macrophage, and urinary exosome lysates, a significant immunoglobulin signal due to the species secondary antibody is apparent at ~90 kDa, as well as a low-intensity band (compared to pS1292-LRRK2) at ~70 kDa. In these CSF specimens, hemoglobin levels were below our ELISA detection limit (<2 pg mL^-1), suggesting LRRK2 protein in the CSF is probably not from blood product contamination as our lower detection limit for pS1292-LRRK2 is higher than 2 pg mL^-1.

In a recent past study, a ‘pooled’ exosome fraction was created that is composed of urinary exosome lysates from ~160 subjects with and without PD [8]. Using our analytical process described here, we can estimate ~5% of total LRRK2 in this pools sample is pS1292-LRRK2 (Fig. 2e). In comparing the BioFIND CSF exosome lysates, total LRRK2 protein was lower but pS1292-LRRK2 was much higher and calculated at ~80% on average (Fig. 2e). This level of pS1292-LRRK2 was higher than that we recorded from urinary exosomes from LRRK2 mutation carriers, higher than that we could achieve by in vitro kinase assays with recombinant protein, and higher than from HEK293 exosomes transfected with our double-LRRK2 mutation construct. Cryo-electron microscopy analysis of the CSF exosome pellets revealed a similar microvesicle constituency as our past observations with urinary exosomes [7] and single-particle tracking measurements here (Fig. 2f).

Previously, we found an elevated ratio of pS1292-LRRK2 normalized to total LRRK2 protein and the control exosome protein Tsg101 in urinary exosomes isolated from male G2019S-LRRK2 mutation carriers from the MJFF LRRK2 Cohort [6]. Here, we sought to determine whether a similar relationship exists in a Norwegian cohort of LRRK2 mutation carriers, with and without PD, and extended the analysis to include females as well as an absolute quantification of pS1292-LRRK2. The cohort we analyzed consisted of 132 subjects that donated urine (Table 1), 82 subjects that agreed to donate CSF (Table 2), and 55 subjects that donated both CSF and urine in the same clinic visit (Table 3).

Measurements of pS1292-LRRK2 levels in urinary exosomes, normalized to the abundance of the control exosome protein Tsg101, revealed elevated pS1292-LRRK2 in G2019S-LRRK2 mutation carriers compared to non-carriers (~4.8 fold on average, p<0.0001, Fig. 3a1). In breaking groups according to sex, male G2019S-LRRK2 mutation carriers with PD had higher pS1292-LRRK2/Tsg101 levels than in carriers without PD (~8.9 fold versus ~3.8 fold, p=0.04). However, in the female group, this trend was reversed (~3.6 fold versus ~5.4 fold, p=0.012, Fig. 3a3). Thus, the female mutation carriers with PD had similar levels of pS1292-LRRK2 as the male carriers, but the age-matched female LRRK2 mutation carriers without PD also had high levels. We did not detect significant correlations between pS1292-LRRK2/Tsg101 levels and any urine characteristic we measured in the samples (leukocyte count, pH, glucose, total protein, red-blood cells, and specific gravity), or any demographic or clinical data (Spearman R values, all p values >0.1, Table 1).

In CSF exosome isolations from the cohort, we could not reliably detect the exosome protein Tsg101 whereas we could reliably measure the exosomal housekeeping protein flotillin-1 in the samples. Analysis of CSF for pS1292-LRRK2 levels, as normalized to the abundance of flotillin-1, revealed similar amounts in the groups irrespective of PD diagnosis, LRRK2 mutation status, or sex (Fig. 3b). In 60 of 81 CSF samples analyzed, hemoglobin levels were below the limits of reliable detection using our ELISA platform (< 2 pg mL^-1). Of the remaining samples with measured hemoglobin, there was no correlation between pS1292-LRRK2 / flotillin-1 or other protein measurements (Spearman R <0.1, all p>0.4). Two CSF specimens had particularly high hemoglobin levels >200 pg mL^-1, but these samples had average pS1292-LRRK2, total LRRK2, and flotillin-1 protein levels. These results demonstrate that we could readily measure pS1292-LRRK2 in CSF exosome fractions in a large biobanked series, although levels were not different in LRRK2 mutation carriers despite the robust differences we could observe in urine collected in the same clinic visit in the same subjects.

We hypothesized that the reason pS1292-LRRK2/flotillin-1 levels in CSF could not identify LRRK2 mutation carriers was related to the very high proportion of pS1292-LRRK2 in CSF and thus possible ceiling effects inherent to limited substrate (i.e., total LRRK2 protein). We therefore defined the percent of LRRK2 phosphorylated at the 1292 residue in each urine and CSF exosome sample. As expected, we found an elevated proportion of pS1292-LRRK2 in urinary exosomes in LRRK2 mutation carriers compared to non-carriers (7.3% of total LRRK2 protein in mutation carriers versus 4.2% in non-carriers, p<0.0001, Fig. 4a). Breaking these groups according to sex, consistent with pS1292-LRRK2/Tsg101 measurements, male carriers with PD had significantly higher phosphorylation at the 1292 residue than carriers without PD (11.5% versus 6.4%, respectively, p=0.0239 Fig. 4a2). Samples from females with a LRRK2 mutation and PD again showed a slightly lower percent pS1292-LRRK2 than mutation carriers without PD, although the difference was not significant (6.5% versus 9.4%, respectively, p=0.26, Fig. 4a3).

Evidence both in vitro in primary cultured cells, as well as in rodents and monkeys treated with LRRK2 kinase inhibitors, suggests that LRRK2 kinase activity may enhance protein stability and decrease LRRK2 protein turnover [9, 23, 26, 32]. In considering the absolute amounts of protein on a volumetric basis with no normalization to other exosome proteins, in the average mL of CSF, there is ~120 pg of LRRK2 harbored in exosome fractions, with only a slightly lower amount of pS1292-LRRK2, whereas in urine there is much less pS1292-LRRK2 protein but similar total LRRK2. In both urine and CSF, a strong positive correlation exists between pS1292-LRRK2 and total LRRK2 (Spearman’s rho 0.76 and 0.38, respectively, p<0.0001 for both), giving further confidence that the pS1292-LRRK2 signal, always measured on different immunoblots than total LRRK2, is authentic in both urine and CSF.

While more LRRK2 protein positively predicts more pS1292-LRRK2 protein in both CSF and urine exosomes, we wondered whether more exosomes in general in the biofluids predicts more LRRK2 protein that we can measure. In urine, there was a positive correlation between LRRK2 protein concentration and Tsg101 levels (Spearman’s rho 0.52, p<0.0001), showing the Tsg101 may be in the same exosome population as LRRK2, consistent with previous observations [7]. In addition, there is an overall increase in LRRK2 protein in males compared to females (2.61 in males versus 1.64 in females, p = 0.003), consistent with recent observations in a cohort from Birmingham, Alabama [8]. In contrast, in CSF exosomes, where Tsg101 protein was not reliably detected, we found that flotillin-1 poorly correlated with LRRK2 protein levels (Fig. 5). These results suggest that the majority of the pool of flotillin-1 positive exosomes may not be LRRK2 positive. Further, levels of LRRK2 normalized to Tsg101 in urine and levels of LRRK2 normalized to flotiiln-1 in CSF did not correlate with one another in subjects from our cohort (Fig. 5c, d). Irrespective of other exosomal proteins, absolute levels of both total LRRK2 protein as well as pS1292-LRRK2 protein in CSF and urine exosome fractions also did not correlate (Fig. 5e, f). These results suggest that the parental cells shedding LRRK2-positive exosomes within an individual have differential regulation of LRRK2 expression and regulation of autophosphorylation with respect to LRRK2 mutation status and PD diagnosis.