Increased LRRK2 kinase activity alters neuronal autophagy by disrupting the axonal transport of autophagosomes

Abstract

Parkinson's disease-causing mutations in the leucine-rich repeat kinase 2 (LRRK2) gene hyperactivate LRRK2 kinase activity and cause increased phosphorylation of Rab GTPases, important regulators of intracellular trafficking. We found that the most common LRRK2 mutation, LRRK2-G2019S, dramatically reduces the processivity of autophagosome transport in neurons in a kinase-dependent manner. This effect was consistent across an overexpression model, neurons from a G2019S knockin mouse, and human induced pluripotent stem cell (iPSC)-derived neurons gene edited to express the G2019S mutation, and the effect was reversed by genetic or pharmacological inhibition of LRRK2. Furthermore, LRRK2 hyperactivation induced by overexpression of Rab29, a known activator of LRRK2 kinase, disrupted autophagosome transport to a similar extent. Mechanistically, we found that hyperactive LRRK2 recruits the motor adaptor JNK-interacting protein 4 (JIP4) to the autophagosomal membrane, inducing abnormal activation of kinesin that we propose leads to an unproductive tug of war between anterograde and retrograde motors. Disruption of autophagosome transport correlated with a significant defect in autophagosome acidification, suggesting that the observed transport deficit impairs effective degradation of autophagosomal cargo in neurons. Our results robustly link increased LRRK2 kinase activity to defects in autophagosome transport and maturation, further implicating defective autophagy in the pathogenesis of Parkinson's disease.

Keywords: JIP4; LRRK2; Parkinson's disease; Rab29; autophagy; axonal transport.

Figure 1.. Overexpression of LRRK2-G2019S disrupts processivity of axonal AV transport.

(A) Kymographs of mCherry-LC3 vesicles in EGFP-LRRK2-WT, −G2019S, and −G2019S-D1994N expressing axons. Asterisks indicate pauses. Arrowheads point to reversals. See also Video S1. (B-C) Bar graph and frequency distribution of (B) pause number per vesicle and (C) pause duration during AV transport in WT, G2019S, and G2019S-D1994N expressing neurons. Frequency distribution does not show pauses > 42 sec (WT: 2.54%, G2019S: 2.87%, G2019S-D1994N: 3.49%). (D) Fraction of time paused (as measured by seconds paused per minute), (E) Number of reversals per AV in WT, G2019S, and G2019S-D1994N expressing axons (mean ± SEM; n = 87-139 AVs from 17-20 neurons from 3 independent cultures; ns, not significant, p=0.44; **p=0.0008; ***p<0.0001; Kruskal-Wallis with Dunn’s multiple comparisons test). See also Figures S1A-C for quantification of AV directionality and density. Figure S3 and Figure S4 show the effect of LRRK2-G2019S overexpression on microtubule and LAMP1-vesicle dynamics, respectively.

Figure 2.. AV transport is disrupted in mouse LRRK2-G2019S knock-in neurons and rescued by LRRK2 kinase inhibition.

(A) Kymographs of axonal EGFP-LC3 vesicles in WT and G2019S KI mouse cortical neurons. See also Video S2. (B-C) Bar graph and frequency distribution of (B) pause number per vesicle and (C) pause duration during AV transport in WT and G2019S KI neurons. Frequency distribution does not show pauses > 48 sec (WT: 2.95%, G2019S: 6.89%). (D) Fraction of time paused, (E) Number of reversals, (F) Δ run length (difference between total run length and net run length) of AVs in WT and G2019S KI neurons (mean ± SEM; n = 112-126 AVs from 26 neurons from 3 independent cultures; ns = not significant, p=0.78; ***p<0.0001; Mann-Whitney test). Figures S1D-G show LRRK2 expression and AV directionality in WT and G2019S KI neurons. (G) Kymographs of axonal EGFP-LC3 vesicles in mouse G2019S KI neurons treated overnight with DMSO or 100 nM MLi-2. See also Video S3. (H) Pause number, (I) Pause duration, (J) Fraction of time paused, (K) Number of reversals, (L) Δ run length of AVs in G2019S KI neurons treated with DMSO or MLi-2 (mean ± SEM; n = 106-116 AVs from 24-28 neurons from 3 independent cultures; ns = not significant, p=0.02; ***p<0.0001; Mann-Whitney test with Bonferroni correction for multiple testing: p<0.0125 was considered statistically significant). Dotted lines indicate the respective average observed in untreated WT neurons. See also Figures S2A-B for phospho-Rab Western blots of WT or G2019S KI neurons +/− MLi-2. Microtubule dynamics in G2019S KI neurons and the effect of MLi-2 on microtubule dynamics are shown in Figures S2 and S3. Figure S4 shows LAMP1-vesicle dynamics in G2019S KI neurons.

Figure 3.. AV transport is disrupted in human iPSC-derived LRRK2-G2019S knock-in neurons and rescued by LRRK2 kinase inhibition.

(A) Kymographs of axonal EGFP-LC3 vesicles in WT and G2019S KI i³Neurons. See also Video S4. (B) Pause number, (C) Pause duration, (D) Fraction of time paused, (E) Number of reversals, (F) Δ run length of AVs in WT and G2019S KI i³Neurons (mean ± SEM; n = 133-189 AVs from 32-33 neurons from 3 independent experiments; ns = not significant, p=0.0299; ***p<0.0001; Mann-Whitney test with Bonferroni correction for multiple testing: p<0.0125 was considered statistically significant). (G) Kymographs of axonal EGFP-LC3 vesicles in G2019S KI i³Neurons treated overnight with DMSO or 100 nM MLi-2. (H) Pause number, (I) Pause duration, (J) Fraction of time paused, (K) Number of reversals, (L) Δ run length of AVs in G2019S KI i³Neurons treated with DMSO or MLi-2 (mean ± SEM; n = 225-279 AVs from 41-43 neurons from 3 independent cultures; ns = not significant, p>0.1368; **p=0.0021; ***p=0.0001; Mann-Whitney test). Dotted lines indicate the respective average in untreated WT i³Neurons.

Figure 4.. LRRK2-G2019S disrupts acidification of axonal AVs.

(A) Kymographs of mCherry-EGFP-LC3 vesicles in the proximal axon of WT or G2019S KI mouse cortical neurons. Magenta arrowheads: mCherry-positive traces; green arrowheads: EGFP-positive traces; white arrowheads: mCherry- and EGFP-positive traces. (B-D) Percentage of acidified (= mCherry-only positive) AVs in (B) the distal (mean ± SEM; n = 17-20 neurons from 4 independent cultures; ns, not significant, p=0.55; Mann-Whitney test), (C) the proximal axon of WT and G2019S KI neurons (mean ± SEM; n = 29-30 neurons from 4 independent cultures; **p=0.0084; Mann-Whitney test), (D) the proximal axon of G2019S KI neurons treated with DMSO or 100 nM MLi-2 overnight (mean ± SEM; n = 24-26 neurons from 3 independent cultures; *p=0.0123; Mann-Whitney test). For acidification of axonal LAMP1 vesicles see Figure S5.

Figure 5.. Overexpression of Rab29 disrupts AV transport.

(A) Time lapse images of axonal mCherry-LC3 and EGFP-Rab29 vesicles in a WT mouse cortical neuron. Scale bar, 5 μm. See also Video S5. (B) Fluorescence recovery after photobleaching Halo-Rab29 signal of an EGFP-LC3 labeled AV. Arrowheads point to the position of the AV. Scale bar, 5 μm. See also Figures S6A-C and Video S6. (C) Proteins associated with the outer membrane of isolated AVs are degraded after treatment with Proteinase K. AV cargo is only degraded by Proteinase K after membrane permeabilization. (D) Western Blot of pT71 Rab29 and total Rab29 from brain lysate, autophagosome fraction, and autophagosome fraction after treatment with Proteinase K. PK, Proteinase K. (E-F) Western Blot quantification of (E) pT71 Rab29 levels (mean ± SEM; n = 8-9 biological replicates; *p=0.022; Mann-Whitney test) and (F) total Rab29 levels (mean ± SEM; n = 7-9 biological replicates; ns, not significant, p=0.25; Mann-Whitney test) in autophagosome fraction of WT and G2019S KI mice. Data shown are normalized to total protein and relative to whole brain lysate. Western Blots of LRRK2, GM130, and LC3 are shown in Figures S6D-F. (G) Kymographs of axonal EGFP-Rab5 or EGFP-Rab29 vesicles, and mCherry-LC3 vesicles in mouse WT cortical neurons. Magenta arrowheads point to tracks of mCherry-LC3 vesicles; green arrowheads highlight EGFP-Rab29 tracks that colocalize with mCherry-LC3 tracks. (H) Pause number, (I) Fraction of time paused, (J) Δ run length of AVs in WT and G2019S KI neurons overexpressing Rab5 or Rab29, and in Rab29 overexpressing WT neurons treated with DMSO or MLi-2 (Overexpression of Rab5 or Rab29: mean ± SEM; n = 107-149 AVs from 27-38 neurons from 3-4 independent cultures; ns, not significant, p>0.06; **p<0.001; ***p<0.0001; Kruskal-Wallis with Dunn’s multiple comparisons test. Treatment of Rab29 overexpressing neurons with DMSO or MLi-2: mean ± SEM; n = 78-109 AVs from 26-27 neurons from 3 independent cultures; ns, not significant, p=0.19; ***p<0.0001; Mann-Whitney test). See also Figures S6G-H.

Figure 6.. LRRK2-G2019S recruits JIP4 to the AV membrane and activates kinesin.

(A-D) Representative Western Blot and quantification of (A) JIP4 (mean ± SEM; n = 8 biological replicates; **p=0.0047; Mann-Whitney test), (B) phospho Threonine Rabs (mean ± SEM; n = 6-7 biological replicates; *p=0.0135; Welch’s t test), (C) p150^Glued (mean ± SEM; n = 7-8 biological replicates; *p=0.61; Mann-Whitney test), (D) KHC (mean ± SEM; n = 6-7 biological replicates; *p=0.014; Mann-Whitney test) in the autophagosome fraction of WT and G2019S KI mouse brain lysates. Data shown are normalized to total protein and relative to whole brain lysate. (E-F) Microtubule pelleting assay (E) with lysates of WT or G2019S KI MEFs incubated with GMPCPP-stabilized microtubules (5 μM) in the presence of 10 mM AMP-PNP and quantification (F) of KHC bound to the microtubule pellet (mean ± SEM; n = 4 biological replicates; *p=0.0216; Welch’s t test). (G-H) Microtubule pelleting assay (G) with lysates of G2019S KI MEFs treated with 200 nM MLi-2 overnight and quantification (H) of KHC bound to the microtubule pellet (mean ± SEM; n = 4 biological replicates; *p=0.0286; Welch’s t test). S, supernatant; P, pellet. Figure S7 shows additional Western Blots and IF staining.

Figure 7.. Overexpression of JIP4 disrupts AV transport.

(A) Time lapse images of axonal EGFP-LC3 and Halo-JIP4 vesicles in a mouse WT cortical neuron. Scale bar, 5 μm. See also Video S7. (B) Kymographs of axonal EGFP-LC3 and Halo-JIP4 vesicles. Arrowheads highlight comigration. (C) Directionality of EGFP-LC3 vesicles in Halo-Tag only and Halo-JIP4 overexpressing WT neurons. Antero + bidir., anterograde + bidirectional; retro., retrograde; stat., stationary (mean ± SEM; n = 24-25 neurons from 3 independent cultures; ***p<0.0001; Two-way ANOVA). (D) Pause number, (E) Fraction of time paused, (F) Number of reversals, (G) Δ run length of AVs in Halo-Tag only and Halo-JIP4 overexpressing neurons (mean ± SEM; n = 68-87 AVs from 24-25 neurons from 3 independent cultures; ***p<0.0001; Mann-Whitney test). Dotted lines indicate the respective average observed in mouse G2019S KI neurons. (H-I) Model depicting the effect of increased LRRK2 kinase activity on axonal AV transport and maturation. (H) Normal LRRK2 activity allows for processive retrograde AV transport, facilitating efficient fusion en route with lysosomal vesicles. Motor adaptors inhibit kinesin and promote dynein activity, resulting in processive retrograde transport. (I) LRRK2 hyperactivation by LRRK2-G2019S mutation or Rab29 overexpression disrupts processive AV transport, leading to inefficient autophagosome-lysosome fusion and impaired AV acidification. Hyperactive LRRK2 enhances recruitment of JIP4 to the AV membrane via binding to phospho-Rabs, resulting in abnormal kinesin activation and a tug-of-war between anterograde and retrograde motors.