Lipid-mediated motor-adaptor sequestration impairs axonal lysosome delivery leading to autophagic stress and dystrophy in Niemann-Pick type C

SUMMARY

Niemann-Pick disease type C (NPC) is a neurodegenerative lysosomal storage disorder characterized by lipid accumulation in endolysosomes. An early pathologic hallmark is axonal dystrophy occurring at presymptomatic stages in NPC mice. However, the mechanisms underlying this pathologic change remain obscure. Here, we demonstrate that endocytic-autophagic organelles accumulate in NPC dystrophic axons. Using super-resolution and live-neuron imaging, we reveal that elevated cholesterol on NPC lysosome membranes sequesters kinesin-1 and Arl8 independent of SKIP and Arl8-GTPase activity, resulting in impaired lysosome transport into axons, contributing to axonal autophagosome accumulation. Pharmacologic reduction of lysosomal membrane cholesterol with 2-hydroxypropyl-β-cyclodextrin (HPCD) or elevated Arl8b expression rescues lysosome transport, thereby reducing axonal autophagic stress and neuron death in NPC. These findings demonstrate a pathological mechanism by which altered membrane lipid composition impairs lysosome delivery into axons and provide biological insights into the translational application of HPCD in restoring axonal homeostasis at early stages of NPC disease.

eTOC Blurb

Roney et al. reveal a lipid-mediated mechanism underlying the altered axonal homeostasis in Niemann-Pick disease type C (NPC). They show that in NPC neurons, abnormal lipid composition within lysosomal membranes sequesters motor-adaptors, resulting in impaired lysosomal transport into axons and reduced axonal degradation capacity, thus contributing to axonal dystrophy and degeneration.

Graphical Abstract

INTRODUCTION

Lysosomes play key roles in maintaining cellular homeostasis by degrading and recycling substrates received from the endocytic and autophagic pathways (Klionsky and Emr, 2000; Levine and Klionsky, 2004; Mellman, 1996). Degradation of endocytic and autophagic cargos depend on their dynamic interactions with lysosomes, which provide the acid hydrolases required for substrate degradation (Luzio et al., 2007). To carry out this function efficiently, lysosomes must move bidirectionally along microtubules (MTs) throughout the cytoplasm (Bonifacino and Neefjes, 2017; Pu et al., 2016). However, this process is particularly challenging in neurons due to their highly polarized morphology with axons that can extend up to one meter in length. Neurons are therefore highly vulnerable to various defects in lysosome trafficking, positioning, and function, which are associated with both rare and common neurodegenerative diseases (Ferguson, 2019; Marques and Saftig, 2019; Platt et al., 2018; Sharma et al., 2018). While enzymatically active lysosomes are enriched in the cell body (Cai et al., 2010; Gowrishankar et al., 2017; Gowrishankar et al., 2015; Yap et al., 2018), mature lysosomes and lysosome-like organelles are also found in distal axons (Farfel-Becker et al., 2019; Farias et al., 2017; Gowrishankar et al., 2017; Jin et al., 2018; Lee et al., 2011). However, the impact of lysosome anterograde trafficking and positioning on axonal pathology in neurodegenerative diseases remains largely elusive.

Niemann-Pick disease type C (NPC) is a neurodegenerative lysosomal storage disorder (LSD) characterized by the accumulation of multiple lipids in the endolysosomal compartment (Lloyd-Evans et al., 2008). Loss of function mutations in the lysosomal membrane protein NPC1 account for approximately 95% of NPC cases (Carstea et al., 1997), while the other 5% are caused by mutations in the lysosomal luminal protein NPC2 (Naureckiene et al., 2000), though the two forms are clinically indistinguishable (Mengel et al., 2013; Vanier, 2010). One major pathologic feature observed in NPC patients and recapitulated in mouse models of NPC disease is axonal dystrophy, which consists of bulbous swellings containing accumulated organelles (Boland and Platt, 2015; Walkley, 1998; Walkley et al., 2010; Walkley and Suzuki, 2004). These axonal changes precede symptom onset and neurodegeneration in NPC mice, suggesting that defects in the trafficking and degradation of these axonal organelles contribute to early NPC pathology. These clinical implications highlight the importance of understanding the cellular events that precede axonal organelle accumulation and their contribution to neurodegeneration. However, the mechanisms underlying these early pathologic events in NPC neurons remain obscure.

In this study, we reveal a lipid-linked pathological mechanism underlying autophagic stress and organelle accumulation in NPC dystrophic axons. Using live-neuron imaging combined with genetic and pharmacologic interventions, we demonstrate that chronic lysosomal dysfunction in NPC neurons compromises the axonal trafficking and positioning of mature degradative lysosomes, which impairs autophagy-lysosomal clearance, leading to the accumulation of autophagic organelles within the axon. Axonal lysosome delivery is driven by the kinesin-1-SKIP-Arl8 complex, where the small GTPase Arl8b links lysosomes to MT-based kinesin-1 motors through its effector SKIP (Farias et al., 2017). Our findings demonstrate a lipid-mediated disruption to the assembly of the kinesin-1-SKIP-Arl8 transport complex on NPC lysosomes. Abnormal cholesterol accumulation on lysosome membranes contributes to impaired lysosome trafficking and positioning in distal axons of NPC neurons, leading to reduced axonal degradation capacity, thereby contributing to axonal dystrophy and degeneration. Our study highlights the importance of soma-derived axonal lysosomes in the maintenance of axonal homeostasis and reveals a mechanism underlying their defective axonal delivery in NPC. This study also provides biological insights into the translational application of 2-hydroxypropyl-β-cyclodextrin (HPCD) in reducing lysosomal membrane cholesterol and restoring axonal degradation capacity in the early stages of NPC disease.

RESULTS

Axonal Autophagic Stress in Npc1^−/− Mice Starting at Presymptomatic Stages

To characterize the organelle accumulations within Npc1^−/− dystrophic axons, we first examined their composition in dorsal root and cortical axons from symptomatic Npc1^−/− mice corresponding to postnatal day 50–60 (P50–60), representing two regions vulnerable to neurodegeneration (Ohara et al., 2004; Pressey et al., 2012). Using transmission electron microscopy (TEM), we observed a striking buildup of endocytic and autophagic organelles within dystrophic axons consisting primarily of autophagic vacuoles (AVs), multivesicular bodies, and multilamellar vesicles (Figures 1A and 1B). Notably, the majority of AVs in Npc1^−/− swollen axons exhibited features of early stage initial AVs (AVi) characterized by a double-membrane bilayer containing electron-lucent material, while fewer late AVs in the degradative stage (AVd) following fusion with lysosomes (Klionsky et al., 2021) were observed. Importantly, these axonal phenotypes were not found in age-matched wild type (WT) mice. Such striking accumulations represent intermediates of the endo-lysosomal-autophagic pathways, indicating defects in their trafficking, maturation, and degradation in Npc1^−/− distal axons.

Figure 1. Aberrant AV Accumulation in Axons of Npc1^−/− Mice.

(A, B) TEM images showing AV accumulation in DRG (A) and cortical (B) axons in Npc1^−/− mice at symptomatic stages (P50–60). Such accumulations were not readily observed in age-matched WT mice. Colored boxes represent four-fold higher magnification of corresponding regions in upper panels (green: WT; orange: Npc1^−/−).

(C, D) Images (C) and analysis (D) showing a robust increase in AV density in DRG axons isolated from presymptomatic Npc1^−/− mice (P30–40) compared to WT neurons at the same age.

(E, F) Kymographs (E) and analysis (F) showing no significant difference in the motility of axonal AVs between WT and Npc1^−/− neurons. See also Videos S1 and S2. DRG neurons isolated from presymptomatic Npc1^−/− mice (P30–40) were nucleofected with EGFP-LC3 at DIV0, followed by live imaging at DIV3–4.

Data were collected from the total number of axons indicated above bars (D) or vesicles (v) from the number of axons (n) indicated in parentheses (F), presented as mean ± SEM with dots representing individual axons, and analyzed by Student’s t test. Scale bars: 800 nm (A, B), 10 μm (C, E). See also Figure S1.

We next aimed to understand the early dynamic events that might underlie AV accumulation by assessing their distribution and trafficking in neuronal axons from presymptomatic Npc1^−/− mice corresponding to P30–40. We used dorsal root ganglion (DRG) neurons because (1) they display in vivo axonal organelle accumulation; (2) represent a rigorous model amenable to culture from adult mice in our previous studies (Cheng et al., 2015; Farfel-Becker et al., 2019; Lin et al., 2017; Xie et al., 2015); and (3) almost all neurites are axons with uniformly polarized MTs (Maday and Holzbaur, 2012) to allow for directional analyses of organelle transport along axons. DRG neurons isolated from P30–40 mice were nucleofected with AV marker EGFP-LC3 at 0 days in vitro (DIV0), followed by live imaging at DIV3–4. Only distal axons with uniform diameter were selected for live imaging. In WT axons, EGFP-LC3 appeared mostly diffuse as cytosolic LC3I (Figure 1C). However, in Npc1^−/− axons, EGFP-LC3 was more punctate as lipidated LC3II incorporated into AV membranes. Strikingly, Npc1^−/− axons showed a robust increase in AV density compared to control (WT: 1.02 ± 0.18; Npc1^−/−: 9.21 ± 0.69; p < 0.0001; per 100 μm axon length; Figure 1D), reflecting axonal autophagic stress, consistent with our TEM observations in adult Npc1^−/− mice.

Because most axonal autophagosomes are generated distally and exhibit predominant retrograde transport towards the soma for degradation (Cheng et al., 2015; Lee et al., 2011; Maday et al., 2012), we hypothesized that AV accumulation in Npc1^−/− distal axons may reflect defects in their retrograde motility. To test this, we monitored axonal AV trafficking in DRG neurons cultured from P30–40 mice. In WT neurons, EGFP-LC3-labeled AVs moved predominantly (57.63 ± 7.35%) in the retrograde direction (Figures 1E and 1F, Videos S1, S2), consistent with previous studies (Cheng et al., 2015; Khobrekar et al., 2020; Kimura et al., 2008; Lee et al., 2011; Maday et al., 2012). However, contrary to our hypothesis, axonal AVs in Npc1^−/− neurons displayed similar motility patterns to WT neurons, suggesting that increased AV density in Npc1^−/− axons does not result from defects in their retrograde transport. Since axonal AV maturation is linked to their retrograde motility, this prompted us to determine their maturation using the mRFP-EGFP-tagged LC3 reporter (Kimura et al., 2007). AVs in distal axons of Npc1^−/− neurons showed a higher percentage of immature AVs co-labeled by both mRFP and EGFP compared to WT (Figure S1). In contrast, the majority of AVs in WT axons displayed a mature status (only retaining mRFP signal) under our conditions. These findings indicate a defective capacity of axonal AVs to achieve maturation and degradation during their retrograde trafficking in presymptomatic Npc1^−/− axons.

Reduced Lysosome Positioning in Distal Axons from Presymptomatic Npc1^−/− Mice

Mature lysosomes provide the active hydrolases for degradation of endocytic and autophagic cargos. Since the majority of AVs in Npc1^−/− dystrophic axons represent immature AVi-like structures (Figures 1A and 1B), we next asked whether axonal lysosome distribution is altered in DRG neurons from presymptomatic (P30–40) Npc1^−/− mice. Given that the lysosome associated membrane protein 1 (LAMP1) is distributed among endo-lysosomal and autophagic organelles (Cheng et al., 2018), we used LAMP1 combined with activity-based lysosome probes to monitor degradative lysosomes. We first examined axonal lysosome distribution using MDW933, a fluorescent probe that binds specifically to the active form of the lysosomal hydrolase glucocerebrosidase (GCase) under acidic conditions (Westbroek et al., 2016; Witte et al., 2010). Strikingly, Npc1^−/− neurons displayed an approximately six-fold decrease in the number of active lysosomes positioned along axons relative to WT neurons (WT: 15.02 ± 1.50; Npc1^−/−: 2.27 ± 0.48; p < 0.0001; per 100 μm axon length; Figures 2A and 2B), representing reduced degradative capacity in Npc1^−/− distal axons. Since activity-based lysosome probes label enzymatically active hydrolases, we next tested whether this reduction resulted from impaired lysosome positioning in distal axons by examining LAMP1-labeled endolysosomes. Notably, Npc1^−/− axons displayed a two-fold reduction in the number of LAMP1-labeled organelles compared to WT (WT: 17.75 ± 1.37; Npc1^−/−: 8.41 ± 1.35; p < 0.0001; per 100 μm axon length; Figures 2C and 2D), reflecting decreased endolysosome density in Npc1^−/− DRG axons at presymptomatic stages.

Figure 2. Reduced Axonal Density of Lysosomes in Npc1^−/− Neurons.

(A-D) Images (A, C) and analyses (B, D) showing reduced density of active GCase-labeled degradative lysosomes and LAMP1-labeled endolysosomes in adult DRG neuron axons isolated from presymptomatic Npc1^−/− mice (P30–40). DRG neurons were incubated with MDW933 (500 nM) for 1 hour to label active GCase, followed by immunostaining for β3-tubulin (A, B), or co-immunostained for LAMP1 and β3-tubulin (C, D).

(E-H) Images (E, G) and analyses (F, H) showing reduced density of degradative lysosomes and endolysosomes in axons of Npc1^−/− cortical neurons plated in microfluidic devices. At DIV8, neurons were loaded with MDW933 (500 nM) for 1 hour to label active GCase, followed by β3-tubulin staining (E, F), or co-immunostained for LAMP1 and β3-tubulin (G, H).

(I-L) Images (I, K) and analyses (J, L) showing lysosomal accumulation in the soma of Npc1^−/− cortical neurons. Neurons at DIV8 were incubated with MDW933 to label degradative lysosomes followed by MAP2 immunostaining, or co-immunostained for MAP2 and LAMP1 to label endolysosomes (I, J), or incubated with BODIPY-FL-pepstatin A to label active cathepsin D prior to live imaging (K, L). Integrated density was measured in thresholded images using ImageJ, and data are presented as mean integrated density normalized to WT neurons.

(M, N) Immunoblots (M) and quantitative analysis (N) of LAMP1, GCase, and cathepsins B and D (CTSB and CTSD) in WT and Npc1^−/− cortical neurons at DIV7–8. Equal amounts of neuronal lysates (10 μg) were sequentially immunoblotted. Protein intensities of LAMP1, GCase, and both mature and immature forms of CTSB and CTSD were averaged from three repeats and normalized to WT.

Data were collected from the total number of axons (B, D, F, H) or neurons (J, L) indicated below or within bars, presented as mean ± SEM with dots representing individual values, and analyzed by Mann-Whitney test (B) or Student’s t test (D, F, H, J, L, N). Scale bars: 10 μm. See also Figure S2.

We next determined whether decreased lysosome density along axons could also be detected in central nervous system (CNS) neurons. To address this, we examined embryonic cortical neurons cultured in microfluidic devices that physically and fluidically separate axons from cell bodies and dendrites by an array of 450-μm long microgrooves (Taylor et al., 2005). Neurons are plated in the soma/dendritic chamber and by DIV5, only axons that enter the microgrooves extend far enough to grow into the axon chamber, thus allowing analysis of lysosomes in distal axons. Npc1^−/− cortical axons displayed a two-fold reduction in active-GCase labeled lysosomes (WT: 13.49 ± 1.0; Npc1^−/−: 6.32 ± 0.57; p < 0.0001; per 100 μm axon length; Figures 2E and 2F) and LAMP1-labeled endolysosomes compared to WT (WT: 18.18 ± 1.34; Npc1^−/−: 8.64 ± 0.87; p < 0.0001; per 100 μm axon length; Figures 2G and 2H), indicating that reduced axonal lysosome density occurs at early stages of disease progression in embryonic CNS neurons.

We next tested whether such a robust reduction in axonal lysosome density in Npc1^−/− cortical neurons reflected an overall decrease in neuronal lysosomes. Notably, Npc1^−/− neurons at DIV8 exhibited increased integrated density of active GCase (p < 0.0001), LAMP1 (p < 0.0001), and active cathepsin D (p < 0.0001) in the cell body relative to WT (Figures 2I–2L). Total protein levels of GCase, LAMP1, and cathepsins B and D in DIV7 cortical neurons also displayed no reduction in Npc1^−/− compared to WT (Figures 2M and 2N). We also measured mRNA levels of several TFEB targets (Sardiello et al., 2009), and observed increased Ctsd mRNA levels but no significant changes in the transcript levels of the other TFEB targets Ctsb, Lamp1, Atp6v0e1, or Atp6v1h (Figure S2A), suggesting that the increase in Ctsd mRNA may not represent a TFEB-mediated response. In addition, filipin staining to detect unesterified cholesterol showed that Npc1^−/− neurons store substantial levels of cholesterol in the soma compared to WT (Figures S2B and S2C), consistent with a cellular phenotype of NPC. Taken together, these results suggest that the lower number of degradative lysosomes in Npc1^−/− axons and their increased accumulation in the soma may reflect impaired anterograde trafficking and/or positioning of lysosomes into axons.

Impaired Anterograde Transport of Degradative Lysosomes into Npc1^−/− Axons

Our recent study revealed that soma-derived active lysosomes are continuously delivered to axons to maintain distal degradation capacity (Farfel-Becker et al., 2019). Increased lysosome accumulation in cell bodies and reduced lysosome density in distal axons prompted us to ask whether somatic lysosomes were not being delivered properly to distal axons in Npc1^−/− neurons. To address this, we loaded active GCase probe MDW941 or active cathepsin D probe BODIPY-FL-pepstatin A into the soma/dendritic chamber of a microfluidic device for 30 minutes, followed by live imaging the axon chamber for a total period of 90 minutes (Figure 3A). Our previous study confirmed the fluidic restriction of activity-based lysosome probes loaded in the soma/dendritic chamber, and demonstrated that their dynamic influx into the axon chamber represents MT-based anterograde lysosome transport (Farfel-Becker et al., 2019). After soma/dendritic chamber loading with MDW941 (Figures 3B and 3C) or BODIPY-FL-pepstatin A (Figures 3D and 3E), we observed degradative lysosomes delivered to distal axons from the cell body. Strikingly, Npc1^−/− neurons displayed a progressive reduction in the axonal delivery of active lysosomes relative to WT at DIV7 (MDW941: p < 0.0001; BODIPY-FL-pepstatin A: p = 0.0015) and further at DIV10 (MDW941: p < 0.0001; BODIPY-FL-pepstatin A: p < 0.0001).

Figure 3. Impaired Axonal Delivery of Degradative Lysosomes in Npc1^−/− Neurons.

(A) Schematic diagram of soma-restricted labeling of degradative lysosomes and their delivery to distal axons. Neurons are plated in the soma/dendritic chamber (1) of a microfluidic device that provides physical and fluidic separation of axons from cell bodies and dendrites. Only axons that enter the 450-μm long microgrooves (2) are able to grow into the axon chamber (3). The soma/dendritic chamber is loaded with an activity-based lysosome probe for 30 minutes, followed by washes with imaging buffer and live imaging carried out in the axon chamber for 90 minutes.

(B-E) Images (B, D) and analyses (C, E) showing impaired delivery of active GCase- and cathepsin D (CTSD)-labeled degradative lysosomes from the soma chamber to distal axons in Npc1^−/− cortical neurons. At DIV7 and DIV10, soma-restricted labeling of degradative lysosomes was carried out by loading MDW941 (100 nM) (B) or BODIPY-FL-pepstatin A (1 μM) (D) for 30 minutes, followed by washes and live imaging in the axon chamber for 90 minutes. Data were quantified from the number of axons indicated in bars and normalized to WT.

(F-J) Kymographs (F, H) and analyses (G, I, J) showing impaired anterograde motility of degradative lysosomes in distal axonal segments of Npc1^−/− cortical neurons. Neurons plated in microfluidic devices were loaded with MDW941 (100 nM) (F, G) or BODIPY-FL-pepstatin A (1 μM) (H, I) for 30 minutes prior to live imaging at DIV8. Time-lapse images were collected every 2 seconds for 90 frames totaling 3 minutes. The percentage of anterograde (antero), retrograde (retro), or stationary (stat) lysosomes was quantified from the total number of vesicles (v) in the total number of axons (n) as follows: (G) WT: n=28 v=344, Npc1^−/−: n=29 v=226; (I) WT: n=36 v=433, Npc1^−/−: n=35 v=271. Active lysosome motility parameters including flux rate, speed, and the velocity in both directions were analyzed in axons of WT and Npc1^−/− cortical neurons (J). See also Videos S3 and S4.

Data were analyzed by Student’s t test (C, E, I, J) or Mann-Whitney test (G) and presented as mean ± SEM alone (C, E) or with dots representing individual values (G, I, J). Scale bars: 10 μm. See also Figure S3, Videos S5 and S6.

To determine whether this defect in axonal lysosome delivery represents a specific impairment in anterograde transport, we next monitored lysosome motility in distal axons. In WT axons, active GCase-labeled degradative lysosomes exhibited bidirectional transport with motile lysosomes accounting for approximately 60% of the total axonal lysosome pool (Figures 3F and 3G, Videos S3, S4); of which 36.29 ± 2.41% moved in the retrograde direction, and 23.93 ± 2.74% moved anterogradely towards the distal terminal. However, active GCase-labeled lysosomes in Npc1^−/− distal axons displayed reduced anterograde transport (12.59 ± 2%, p = 0.0022) with no significant changes in stationary or retrograde motility. We also monitored lysosome motility in distal axons using BODIPY-FL-pepstatin A. Consistently, Npc1^−/− neurons showed impaired anterograde transport of lysosomes containing active cathepsin D (WT: 33.22 ± 2%; Npc1^−/−: 21.6 ± 3.36%; p = 0.0038) but with a corresponding increase in stationary puncta (WT: 27.92 ± 2.22%; Npc1^−/−: 45.23 ± 3.56%; p < 0.0001) and no significant change in retrograde motility (Figures 3H and 3I). Lysosomes in Npc1^−/− axons displayed reduced flux rates without significant changes in their speed or velocity (Figure 3J). We also observed no changes in axonal mitochondrial density or motility in Npc1^−/− neurons compared to WT (Figures S3A–S3D). These results support the notion that impaired lysosome transport into Npc1^−/− axons does not represent a global defect in axonal organelle transport.

This led us to ask whether decreased lysosome density along axons impacts axonal autophagic clearance in Npc1^−/− neurons. While a robust AV increase was detected in DRG axons from presymptomatic (P30–40) Npc1^−/− mice (Figures 1C and 1D), it is not known if embryonic Npc1^−/− cortical neurons also display axonal autophagic stress. To address this, neurons cultured in microfluidic devices were transduced with EGFP-LC3 at DIV0, followed by live imaging at DIV8 and DIV10. We found no significant difference in the density of AVs in Npc1^−/− axons at DIV8 (Figures S3E and S3F), a time point where lysosome density was already reduced (Figures 2E–2H). However, at DIV10, AV density was significantly increased in Npc1^−/− axons (WT: 2.86 ± 0.44; Npc1^−/−: 4.76 ± 0.62; p = 0.0122; per 100 μm axon length). Notably, this increase in axonal AV density did not reflect changes in their transport as no differences in their motility were observed (Figures S3G and S3H, Videos S5, S6). These results indicate that lysosome transport defects precede axonal AV accumulation in Npc1^−/− cortical neurons, suggesting that impaired axonal lysosome delivery may contribute to autophagic stress in Npc1^−/− axons.

Aberrant Sequestration of Kinesin-1 and Arl8 on Npc1^−/− Lysosome Membranes

These findings prompted us to examine NPC-linked mechanisms underlying impaired anterograde axonal lysosome transport. A previous study established that the kinesin-1-SKIP-Arl8 complex drives lysosome transport into axons (Farias et al., 2017). Therefore, we hypothesized that this complex might be improperly recruited to Npc1^−/− neuronal lysosomes, and thus underlie impaired lysosome delivery into Npc1^−/− axons. To test this, we applied STED super-resolution microscopy to examine the localization of kinesin-1, SKIP, and Arl8 on somatic lysosomes of cortical neurons. Unexpectedly, we observed a striking phenotype: while both endogenous kinesin-1 and Arl8 were largely dispersed throughout the soma of WT neurons, they appeared accumulated on the surface of LAMP1-labeled endolysosomes in Npc1^−/− neurons (Figures 4A–4D). In contrast, SKIP did not show similarly enhanced association with Npc1^−/− lysosome membranes (Figures 4E and 4F). These super-resolution images suggest that kinesin-1 is not recruited to Npc1^−/− neuronal lysosomes by the GTPase effector SKIP, and therefore may represent a non-functional sequestration of kinesin-1 and Arl8 on Npc1^−/− lysosome membranes.

Figure 4. Aberrant Sequestration of Kinesin-1 and Arl8 on Lysosome Membranes in Npc1^−/− Neurons.

(A-F) STED super-resolution images (A, C, E) and analyses (B, D, F) showing aberrant sequestration of endogenous kinesin-1 and its adaptor Arl8 on the surface of Npc1^−/− lysosome membranes in the soma. Enlarged views of the corresponding boxed regions are shown in (A’), (C’), and (E’). The percentage of LAMP1-labeled areas associated with Arl8 (B), kinesin-1 (D), or SKIP (F) were quantified and normalized to WT. Note that Arl8 and kinesin-1, but not SKIP, display enhanced distribution on the surface of Npc1^−/− lysosomes.

(G-J) STED images (G, I) and analyses (H, J) showing non-specific recruitment of truncated KHC-MD (motor domain, G) and KHC-CBD (cargo-binding domain, I) to Npc1^−/− lysosome membranes. WT and Npc1^−/− cortical neurons were transduced with GFP-tagged KHC-MD or KHC-CBD at DIV4, followed by co-immunostaining for GFP and LAMP1 at DIV8. Enlarged views of the corresponding boxed regions are shown in (G’) and (I’). The percentage of LAMP1-labeled areas associated with KHC-MD (H) or KHC-CBD (J) signals were quantified and normalized to WT.

Data were collected from the total number of neurons indicated within bars (B, D, F, H, J), presented as mean ± SEM with dots representing individual neurons, and analyzed by Mann-Whitney test. Scale bars: 5 μm.

Kinesin-1 family members (known as KIF5) drive the anterograde transport of lysosomes (Farias et al., 2017; Palomo-Guerrero et al., 2019). Each KIF5 heavy chain (KHC) contains an N-terminal motor domain (MD) and a C-terminal cargo-binding domain (CBD) that mediates an association with kinesin light chains or directly interacts with a cargo adaptor (Hirokawa et al., 2010). To test whether KHC may be sequestered non-specifically, we constructed two truncated mutants of KIF5B KHC: EGFP-tagged KHC-MD and KHC-CBD. STED imaging revealed that while both KHC-MD and KHC-CBD were largely dispersed throughout WT neurons, their enrichment was detected on the surface of lysosomes in Npc1^−/− neurons (Figures 4G–4J), suggesting the non-specific sequestration of KHC independent of cargo binding.

We further tested whether kinesin-1 sequestration depends on the normal targeting pathway by knockdown of Arl8 or SKIP (Figures S4A–S4D). STED imaging showed no change in the percentage of LAMP1-labeled areas associated with kinesin-1 signals following depletion of Arl8 (p = 0.7437; Figures 5A and 5B) or SKIP (p > 0.99; Figures 5C and 5D), suggesting that kinesin-1 sequestration occurs independent of the Arl8-SKIP targeting pathway. We next determined whether the sequestration of kinesin-1 and Arl8 is affected by Arl8 GTPase activity. We expressed WT, GTP-locked mutant Q75L (constitutively active), or GDP-locked mutant T34N (constitutively inactive) (Hofmann and Munro, 2006) forms of Arl8b-mCh in Npc1^−/− cortical neurons at DIV4, followed by immunostaining at DIV8. We found that (1) accumulation of kinesin-1 on Npc1^−/− lysosome membranes is independent of Arl8 GTPase activity; and (2) both GTP- and GDP-locked Arl8b mutants display similar levels of accumulation on Npc1^−/− lysosome membranes (Figures 5E and 5F), suggesting that Arl8 may be sequestered in an inactive GDP-bound form. Because SKIP binding to Arl8 is nucleotide dependent (Rosa-Ferreira and Munro, 2011), we further tested possible disruption of Arl8-SKIP interaction in Npc1^−/− and found that an anti-Arl8 antibody co-precipitated less SKIP in Npc1^−/− brain homogenates compared to WT (p = 0.0049; Figures 5G and 5H). Collectively, these results suggest that the aberrant sequestration of kinesin-1 and Arl8 on Npc1^−/− lysosomal membranes impairs the assembly of the motor-adaptor-effector complexes that drive lysosome delivery into axons.

Figure 5. Aberrant Accumulation of Kinesin-1 and Arl8 on Npc1^−/− Lysosome Membranes Is Independent of Arl8 GTPase Activity.

(A-D) STED images (A, C) and analyses (B, D) showing kinesin-1 sequestration on Npc1^−/− lysosome membranes independent of Arl8 and SKIP. Npc1^−/− cortical neurons at DIV4 were transfected with control scrambled (scr), Arl8-siRNA (A, B), or SKIP-siRNA (C, D), followed by co-immunostaining for LAMP1 and kinesin-1 at DIV8. Pseudocolors were applied: magenta for LAMP1; cyan for kinesin-1. Enlarged views of the boxed regions are shown in (A’) and (C’). The percentage of LAMP1-labeled areas associated with kinesin-1 signals in Npc1^−/− cortical neurons transfected with Arl8-siRNA or SKIP-siRNA were quantified and normalized to Npc1^−/− neurons transfected with scr-siRNA. See also Figures S4A–S4D.

(E, F) STED images (E) and analyses (F) showing that kinesin-1 and Arl8 sequestration is not affected by Arl8 GTPase activity in Npc1^−/− neurons. Npc1^−/− cortical neurons were transfected with WT, GTP-locked, or GDP-locked mutant forms of Arl8b-mCh at DIV4, followed by immunostaining for LAMP1, mCherry, and kinesin-1 at DIV8. Pseudocolors were applied. Enlarged views of the boxed regions are shown in (E’). The percentage of LAMP1-labeled areas associated with kinesin-1 or mCherry-tagged Arl8b signals in Npc1^−/− cortical neurons expressing Arl8b mutants were quantified and normalized to Npc1^−/− neurons expressing WT Arl8b.

(G, H) Co-immunoprecipitation (G) and analysis (H) showing reduced interaction of Arl8 and SKIP in Npc1^−/− mouse brain homogenates from E18 embryos. Protein band intensities were quantified using ImageJ and normalized to WT. Data were averaged from three repeats and analyzed by Student’s t test.

Data were collected from the total number of neurons indicated within bars (B, D, F), presented as mean ± SEM with dots representing individual neurons, and analyzed by Mann-Whitney test (B, D) or one-way ANOVA, where p values represent comparisons of each condition against WT Arl8b (F). Scale bars: 5 μm.

Elevated Lysosomal Membrane Cholesterol Sequesters Kinesin-1 and Arl8

A recent study established that the limiting membrane of Npc1^−/− lysosomes in non-neuronal cells shows significant cholesterol accumulation (Lim et al., 2019). This prompted us to examine whether changes in lipid composition in the lysosomal membrane contribute to the aberrant sequestration of kinesin-1 and Arl8. We first determined whether Npc1^−/− neurons show elevated lysosomal membrane cholesterol by applying the cholesterol biosensor GST-D4H*-mCherry to semi-permeabilized cells (Lim et al., 2019; Wilhelm et al., 2017), which allows the D4H* protein to selectively bind cholesterol-rich intracellular membranes with a cholesterol molar content that exceeds 10% (Maekawa and Fairn, 2015). In WT neurons, D4H*-mCherry appeared mostly diffuse and did not bind LAMP1-labeled endolysosomal membranes (Figure 6A), indicating that their membrane cholesterol content is below 10%. However, Npc1^−/− neurons displayed strong vesicular D4H*-mCherry signals overlapping LAMP1-labeled endolysosomes, reflecting elevated cholesterol content within Npc1^−/− lysosomal membranes compared to WT (p = 0.0008; Figure 6B). We also co-labeled free cholesterol with filipin and observed increased luminal cholesterol levels in Npc1^−/− relative to WT (p < 0.0001; Figures 6A and 6C), indicating that Npc1^−/− neuronal lysosomes accumulate both luminal and membrane cholesterol. Notably, the vast majority of vesicular D4H*-mCherry signals in Npc1^−/− neurons overlapped with LAMP1, while its colocalization with SV2-labeled synaptic vesicles was hardly detectable and no significant change in their axonal distribution was observed (Figures S4E–S4H), suggesting that elevated membrane cholesterol may selectively impair lysosome transport in Npc1^−/− neurons.

Figure 6. Reducing Cholesterol Releases Sequestration of Kinesin-1 and Arl8 from Npc1^−/− Lysosome Membranes.

(A-C) Images (A) and analyses (B, C) showing elevated membrane and luminal cholesterol in lysosomes of Npc1^−/− neurons at DIV7–8. Colocalization between D4H* and LAMP1 was measured and expressed as the fraction of LAMP1-labeled organelles that were also labeled by D4H* and normalized to Npc1^−/−. To quantify luminal cholesterol levels, the integrated density of filipin was measured in thresholded images and normalized to Npc1^−/−. See also Figures S4E–S4H.

(D-F) Images (D) and analyses (E, F) showing reduced cholesterol on endolysosome membranes and the lumen of Npc1^−/− neurons at DIV7–8 following HPCD treatment.

(G, H) STED images showing release of sequestrated Arl8 (G) and kinesin-1 (H) from Npc1^−/− endolysosomal membranes by reducing lysosomal cholesterol with HPCD. Pseudocolors were applied: blue for GST-D4H*-mCherry; red for Arl8 (G) or kinesin-1 (H). The edges of cell bodies and proximal processes are outlined with white dashed lines. Enlarged views of the boxed regions are shown in (G’) and (H’).

(I, J) Sequential immunoblots (I) and quantitative analysis (J) showing cholesterol-dependent sequestration of kinesin-1 and Arl8 on Npc1^−/− neuronal lysosomes. Cortical neurons were treated with HPCD (100 μM) or H₂O control for 48 hours, followed by magnetic isolation of lysosomes at DIV8. Equal amounts of whole cell lysates (5 μg) and captured endolysosomes (0.8 μg) were loaded and sequentially immunoblotted with antibodies as indicated. Protein band intensities were quantified and averaged from three repeats, and data were normalized to H₂O-treated WT neurons.

Data were collected from the total number of neurons indicated below bars (B, C, E, F), presented as mean ± SEM with dots representing individual images (B, E) or neurons (C, F), and analyzed by Mann-Whitney test (B, C), Student’s t test (E, F) or one-way ANOVA (J). ** p < 0.01. Scale bars: 10 μm (A, D), 2 μm (G, H).

We next sought to reduce lysosomal membrane cholesterol levels. Cyclodextrin derivatives have commonly been used to extract cholesterol from membranes (Kilsdonk et al., 1995; Ohvo and Slotte, 1996; Zidovetzki and Levitan, 2007). Interestingly, the cyclodextrin derivative HPCD was shown to reduce luminal cholesterol, delay disease onset, and extend lifespan in Npc1^−/− mice (Davidson et al., 2009; Liu et al., 2008; Liu et al., 2009), supporting its rationale for evaluation in a clinical trial with NPC patients (Ory et al., 2017). Treatment with low doses of 100 and 1000 μM HPCD for 24 hours mobilized luminal cholesterol from endolysosomes without toxicity to Npc1^−/− mouse neurons (Peake and Vance, 2012). However, it remains to be determined whether HPCD can effectively remove membrane cholesterol from Npc1^−/− lysosomes. Strikingly, treatment of Npc1^−/− cortical neurons with HPCD (100 μM for 48 hours) nearly abolished D4H*-mCherry signals on lysosome membranes (p < 0.0001) and also significantly reduced luminal cholesterol (p < 0.0001) in lysosomes (Figures 6D–6F). STED imaging further revealed that treatment with HPCD released Arl8 and kinesin-1 sequestration from the surface of Npc1^−/− lysosomes, returning to a cytosolic distribution similar to that observed in WT neurons (Figures 6G and 6H).

As a complementary approach, we magnetically isolated endolysosomes to further examine the altered distribution of the kinesin-1-SKIP-Arl8 complex on Npc1^−/− lysosomes. Neuronal lysosomes were loaded with iron-dextran and separated by magnetic chromatography (Bilgin et al., 2017; Diettrich et al., 1998), followed by immunoblotting to assess relative protein levels within endolysosomal membranes. LAMP1, but not GAPDH, TOM20, or Sec61β, was readily detected in captured organelles, indicating an enriched endolysosome population following magnetic chromatography. Consistent with results observed by STED imaging, lysosomal membranes showed increased levels of Arl8 and kinesin-1, but not SKIP, in Npc1^−/− neurons compared to WT, further supporting their non-functional accumulation in the absence of the GTPase effector SKIP. Importantly, HPCD treatment reduced the abnormal accumulation of Arl8 and kinesin-1 on endolysosome membranes of Npc1^−/− neurons (Figures 6I and 6J), further suggesting cholesterol-mediated sequestration. These changes occurred without noticeable differences in the total protein levels in whole cell lysates, suggesting that the observed alterations occur locally at Npc1^−/− lysosomal membranes. Collectively, these STED imaging and lysosome purification analyses consistently suggest that altered lipid composition in Npc1^−/− lysosomal limiting membranes drives the aberrant sequestration of kinesin-1 and Arl8.

Reduction of Lysosomal Membrane Cholesterol Rescues Axonal Lysosome Delivery and Reduces Autophagic Stress in Npc1^−/− Neurons

We next determined whether releasing kinesin-1/Arl8 sequestration by reducing lysosomal membrane cholesterol has an impact on axonal lysosome density in DRG neurons isolated from presymptomatic (P30–40) Npc1^−/− mice. Neurons were incubated with HPCD for 48 hours, followed by co-immunostaining at DIV3–4. In WT neurons, HPCD treatment did not alter axonal lysosome density compared to H₂O control (p > 0.9999; Figures 7A and 7B). However, Npc1^−/− neurons treated with HPCD showed a striking increase in the number of axonal lysosomes (Npc1^−/− H₂O: 8.54 ± 1.02; Npc1^−/− HPCD: 15.20 ± 1.44; p = 0.0014; per 100 μm axon length; Figures 7A and 7B), suggesting that reducing lysosomal membrane cholesterol effectively rescues lysosome transport into axons from presymptomatic Npc1^−/− mice.

Figure 7. Cholesterol Reduction and Arl8b Expression Rescue Lysosome Transport into Axons of Npc1^−/− Neurons.

(A, B) Images (A) and analysis (B) showing rescued axonal lysosome density in presymptomatic Npc1^−/− DRG neurons by reducing membrane cholesterol with HPCD treatment. (C, D) Images (C) and analysis (D) showing enhanced delivery of soma-labeled degradative lysosome to distal axons following HPCD treatment in Npc1^−/− cortical neurons cultured in microfluidic devices. At DIV8, MDW933 (500 nM) was loaded to the soma/dendritic chamber for 15 minutes. Neurons were then washed and fixed after 0, 1, 2, or 3 hours, followed by immunostaining for β3-tubulin. Axonal terminals are outlined with white dashed lines. See also Videos S7 and S8.

(E, F) Images (E) and analyses (F) showing reduced density of axonal AVs and increased density of lysosomes with HPCD treatment in DRG axons from P30–40 Npc1^−/− mice.

(G, H) Images (G) and analysis (H) showing rescued axonal lysosome density with elevated Arl8b expression in DRG axons from P30–40 Npc1^−/− mice.

(I, J) Images (I) and analysis (J) showing reduced axonal autophagic stress with elevated Arl8b expression in DRG neurons from presymptomatic Npc1^−/− mice at P30–40.

(K, L) Images (K) and analysis (L) showing ameliorated Npc1^−/− neuron death following Arl8b overexpression. Cortical neurons isolated from presymptomatic mice (P30–40) were transduced at DIV0 with Arl8b-mCh or mCh. TUNEL assays were performed at DIV7 and DIV10, and the percentage of TUNEL-positive cells to total DAPI staining was calculated in thresholded images (705 × 705 μm) using ImageJ.

Data were collected from the total number of axons (B, F, H, J) or images (L) as indicated, or from > 30 axon terminals for each time point (D), presented as mean ± SEM with dots representing individual axons, and analyzed by one-way ANOVA (B, H, L) or Student’s t test (D, F, J). ** p < 0.01, *** p < 0.001, and **** p < 0.0001. Scale bars: 10 μm (A, E, G, I), 5 μm (C), and 50 μm (K). See also Figures S5 and S6.

We directly tested this possibility by examining the anterograde transport of lysosomes from the soma to distal axons in Npc1^−/− neurons following HPCD treatment. Neurons were cultured in microfluidic devices, and only the soma/dendritic chamber was briefly loaded with MDW933 for 15 minutes. Neurons were then washed and fixed after 0, 1, 2, or 3 hours, followed by imaging growth cones within the axon chamber. Lysosomes visualized in distal axon terminals following this soma-restricted labeling represent those transported there from the cell body. Remarkably, as post-wash time increased from 0 to 3 hours, HPCD treatment resulted in a more rapid accumulation of active GCase signal at growth cones compared to H₂O-treated Npc1^−/− neurons (Figures 7C and 7D, Videos S7, S8), reflecting enhanced axonal delivery of lysosomes from the cell body.

Lysosome transport to distal axons is critical for maintaining efficient autophagic flux (Farfel-Becker et al., 2019). Given that lysosome transport defects precede axonal AV accumulation in Npc1^−/− neurons (Figures 3B–3J, S3E and S3F), we next determined whether rescuing axonal lysosome delivery with HPCD treatment has an impact on autophagic stress in Npc1^−/− axons. To test this, DRG neurons from P30–40 Npc1^−/− mice were nucleofected with EGFP-LC3 at DIV0, treated with HPCD for 48–72 hours, then incubated with MDW941 prior to live imaging at DIV3–4. Treatment of Npc1^−/− neurons with HPCD resulted in increased axonal lysosome density (H₂O: 5.22 ± 0.60; HPCD: 12.47 ± 1.24; p < 0.0001; per 100 μm axon length) and reduced axonal AV density (H₂O: 8.32 ± 0.62; HPCD: 3.28 ± 0.45; p < 0.0001; per 100 μm axon length) (Figures 7E and 7F), suggesting that reducing lysosomal membrane cholesterol and therefore enhancing axonal lysosome delivery facilitates autophagic flux in presymptomatic Npc1^−/− DRG axons.

Elevated Arl8b Expression Facilitates Lysosome Transport into Axons and Reduces Autophagic Stress and Neuronal Death in Presymptomatic Npc1^−/− Neurons

Kinesin-1 requires SKIP to drive lysosome transport into axons (Farias et al., 2017), and SKIP requires Arl8 in its active GTP-bound state for binding (Rosa-Ferreira and Munro, 2011). Because SKIP did not show enhanced lysosomal membrane association and displayed reduced interaction with Arl8 in Npc1^−/− mouse brains (Figures 5G and 5H), Arl8 may be sequestered in an inactive GDP-bound form. We tested whether elevated Arl8b expression in Npc1^−/− neurons is able to overcome the cholesterol-linked sequestration of endogenous Arl8 and recruit SKIP to rescue lysosome transport into axons. DRG neurons isolated from presymptomatic (P30–40) Npc1^−/− mice were nucleofected with Arl8b-mCh at DIV0, followed by co-immunostaining at DIV3–4. Strikingly, elevated Arl8b expression rescued lysosome density in Npc1^−/− axons (Npc1^-/: 8.58 ± 0.91; Npc1^−/− with Arl8b-mCh: 29.73 ± 3.18; p < 0.0001; per 100 μm axon length) and also further increased the number of lysosomes in WT axons (WT: 18.80 ± 1.50; WT with Arl8b-mCh: 32.11 ± 2.56; p = 0.0156; per 100 μm axon length) (Figures 7G and 7H). The effect of Arl8b-mCh on axonal lysosome density was nucleotide dependent, as expressing the GTP-locked, but not GDP-locked, form of Arl8b-mCh increased lysosome density in both WT and Npc1^−/− axons (Figures S5A and S5B). Concordantly, elevated Arl8b-mCh expression in Npc1^−/− neurons reduced lysosomal accumulation in the soma (Figures S5E and S5F). Notably, high levels of Arl8b overexpression resulted in axonal organelle accumulation (Figure S6), highlighting that maintaining a steady-state distribution of axonal lysosomes is critical for axonal homeostasis (Gowrishankar et al., 2017).

We next asked whether facilitating lysosome transport to distal axons by elevating Arl8b expression also reduces AV accumulation in Npc1^−/− axons. DRG neurons cultured from presymptomatic Npc1^−/− mice were nucleofected with EGFP-LC3 alone or together with Arl8b-mCh at DIV0, followed by live imaging at DIV3–4. Consistently, elevated Arl8b expression reduced AV density in Npc1^−/− axons (Npc1^−/−: 9.29 ± 0.57; Npc1^−/− + Arl8b-mCh: 4.93 ± 0.35; p < 0.0001; per 100 μm axon length) (Figures 7I and 7J). Importantly, this reduction was also dependent on Arl8 GTPase activity (Figures S5C and S5D). Together, these experiments suggest that facilitating lysosome transport into Npc1^−/− axons is sufficient to support the maturation and clearance of axonal AVs at this early presymptomatic stage.

We further examined whether elevated Arl8b expression has an impact on the survival of Npc1^−/− cortical neurons isolated from presymptomatic adult mice (P30–40) by performing terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assays. Npc1^−/− neurons showed increased TUNEL-positive neuron death compared to WT at DIV7 (WT: 18.81 ± 0.64%; Npc1^−/−: 26.96 ± 0.80%; p < 0.0001) and a further increase at DIV10 (WT: 22.55 ± 0.45%; Npc1^−/−: 35.62 ± 0.67%; p < 0.0001), indicating progressive apoptosis with age in culture. However, elevated Arl8b expression reduced Npc1^−/− neuron death at DIV7 and DIV10, restoring neuron survival to levels observed in WT neurons (Figures 7K and 7L). Taken together, these results support a model in which a cholesterol-linked impairment of lysosome transport to distal axons disrupts autophagic maturation and clearance, thus contributing to axon degeneration in Npc1^−/− neurons.

DISCUSSION

In this study, we reveal a lipid-mediated mechanism contributing to axonal dystrophy in NPC neurons. Abnormal cholesterol accumulation within NPC lysosome membranes in the soma impairs their axonal delivery and positioning by sequestering kinesin-1 and Arl8 independent of SKIP, thus disrupting axonal degradation capacity contributing to dystrophy and degeneration. The biological insights revealed in this study advance our understanding of early axonal dystrophy in NPC and shed light on the recent translational application of HPCD in NPC patients (Ory et al., 2017).

Coordinated Bidirectional Transport of Lysosomes Maintains Axonal Homeostasis

Axonal dystrophy and autophagic stress are two major pathological hallmarks of common neurodegenerative diseases and LSDs associated with lysosome dysfunction (Beard et al., 2017; Bordi et al., 2016; Gowrishankar et al., 2015; Haidar and Timmerman, 2017; Lee et al., 2011; Micsenyi et al., 2009; Tagliaferro and Burke, 2016; Walkley et al., 2010; Xie et al., 2015; Yang et al., 2013; Zigdon et al., 2017). These studies raised a fundamental question as to whether reduced clearance of endo-lysosomal-autophagic organelles contributes to axon pathology and dystrophy. Since degradative lysosomes have been thought to be confined to the neuronal cell body, it was unclear why a disturbance in the lysosomal system would be associated with axonal pathology in neurodegenerative LSDs such as NPC (Walkley et al., 2010). In agreement with previous reports (Boland and Platt, 2015; Walkley et al., 2010; Walkley and Suzuki, 2004), dystrophic axons in NPC did not contain the characteristic lysosome inclusions representative of NPC lysosomes. Rather, we observed a predominant accumulation of autophagic vacuoles, multivesicular bodies, and multilamellar vesicles, largely representing intermediates of endocytic and autophagic pathways. It has been well established that the retrograde transport of axonal AVs is linked to their maturation and degradation (Lee et al., 2011; Maday et al., 2012). One of our most unexpected observations was that despite increased axonal autophagic stress, AVs in NPC axons displayed similar motility patterns compared to control axons. Given global disruptions to axonal transport would lead to accumulation of a more diverse set of organelles, these findings suggested that impaired maturation and clearance of lysosomal substrates likely underlies early axonal pathology in NPC.

Our recent study in microfluidic devices demonstrated that soma-labeled degradative lysosomes are dynamically delivered to distal axons, where they target to AVs, facilitating their maturation into autolysosomes by supplying active lysosomal hydrolases (Farfel-Becker et al., 2019). We propose that both the anterograde delivery of lysosomes from the soma to distal axons, as well as the retrograde trafficking of AVs generated at axon terminals, are essential to achieve effective autophagic flux in axons. Consistent with this concept, we observed that NPC neurons display a reduced number of lysosomes delivered into distal axons, thus impairing maturation and turnover of AVs during their retrograde transport route. This concept is supported by several studies showing that impairments in either transport process lead to axonal autophagic stress and neurodegeneration (Cheng et al., 2015; Farfel-Becker et al., 2019; Farias et al., 2017; Tammineni et al., 2017; Wong and Holzbaur, 2014). Together, our findings conceptually advance current knowledge and highlight that efficient autophagic clearance in axons requires coordination between their retrograde transport and the anterograde delivery of soma-derived degradative lysosomes to distal axons. This bidirectional transport model facilitates efficient degradation and removal of axonal cargos during their opposing trafficking routes.

Altered Membrane Lipid Composition Impairs Lysosome Delivery to NPC Axons

Cholesterol accumulation in the lumen of NPC lysosomes was reported to alter lysosome positioning through the Rab7/RILP/ORP1L (Rocha et al., 2009) or the TMEM55B/JIP4 pathway (Willett et al., 2017) as well as impair Rab7 and kinesin motor function (Lebrand et al., 2002), highlighting the pathological role of cholesterol accumulation in altering endolysosomal transport. However, the mechanisms and pathological relevance in NPC neurons and its contribution to axonal dystrophy and autophagic stress remained largely unknown. Here we observed that axonal lysosomes in NPC display a selective defect in anterograde transport without affecting their retrograde motility, resulting in a net reduction in the lysosome density in NPC axons that impeded efficient axonal cargo clearance.

Defective axonal lysosome delivery suggested a disruption to the kinesin-1-SKIP-Arl8 transport complex in NPC. To elucidate mechanisms underlying impaired lysosome transport into axons, we investigated recruitment of this complex to lysosomes using super-resolution imaging and biochemical endolysosomal isolations. Consistent with our notion that elevated lipid content within lysosomal membranes alters anterograde lysosome transport, we observed a disordered accumulation of kinesin-1 and Arl8 on NPC lysosome membranes, which occurs independent of the kinesin cargo-binding domain, the motor adaptor effector SKIP, and the nucleotide status of Arl8b. Such lipid-mediated Arl8 sequestration disrupted formation of the motor-adaptor-effector complex, thus impairing lysosome transport to NPC distal axons. Notably, the Ragulator complex negatively regulates the BORC complex in non-neuronal cells (Filipek et al., 2017; Pu et al., 2017) and functions upstream of the kinesin-1-SKIP-Arl8 complex (Pu et al., 2015). Disrupting Ragulator by Lamtor1 knockdown in P30–40 WT DRG neurons did not result in an increase in axonal lysosomes (data not shown), suggesting that other neuron-specific complexes may regulate the BORC-dependent axonal trafficking of endolysosomes. However, the lipid-mediated disruption of the kinesin-1-SKIP-Arl8 transport complex in NPC was further supported by application of HPCD, which reduced membrane cholesterol and released Arl8 and kinesin-1 sequestration, thereby rescuing axonal lysosome delivery and reducing autophagic stress in presymptomatic NPC axons.

Cyclodextrin was reported to reduce lysosomal sphingolipid accumulation (Hoque et al., 2020) and reduce cholesterol storage through lysosomal exocytosis (Chen et al., 2010). It is possible that elevated membrane sphingolipid content may also contribute to the lipid-mediated motor-adaptor sequestration on NPC lysosomes. Levels of cholesterol and sphingolipids in membranes influence their properties and organization into liquid-ordered and -disordered lipid phases (Elson et al., 2010). Recent structural analyses of the yeast orthologue of NPC1 revealed a hydrophobic tunnel within it that moves cholesterol and other sterols from the lysosomal lumen to the inner leaflet of the limiting membrane (Winkler et al., 2019). While it is well established that cholesterol stores in the lumen of NPC lysosomes, it was recently reported that NPC lysosomes also accumulate ER-derived cholesterol at the limiting membrane (Lim et al., 2019), which may be distinguished from cholesterol from an endocytic source.

At physiological levels, cholesterol-enriched membrane domains play an important role in supporting various signaling and membrane trafficking events (Epand, 2006; Mukherjee and Maxfield, 2004). Proteins may preferentially associate and sequester into cholesterol-enriched regions of the membrane for their normal function (Ge et al., 2018; Rahbek-Clemmensen et al., 2017). However, abnormally elevated cholesterol levels were reported to alter SNARE complex assembly (Fraldi et al., 2010) and interfere with the GDP-GTP cycle of small GTPases (Ganley and Pfeffer, 2006; Lebrand et al., 2002), indicating a pathological role for excess membrane cholesterol in driving aberrant protein sequestration. Because cholesterol is an important regulator of membrane organization, cells must maintain cholesterol levels within a narrow range. Therefore, treatment with cyclodextrin derivatives may require careful dosing, as excessive depletion of cholesterol from cellular membranes may alter cholesterol-dependent homeostatic events (Peake and Vance, 2012; Sarkar et al., 2013; Yang et al., 2017).

Lysosomal Dysfunction and Axonal Autophagic Stress in NPC Neurons

Reducing lysosomal degradation capacity by leupeptin or bafilomycin A1 treatment results in AV accumulation in axons, neurites, and/or the soma (Boland et al., 2008; Lee et al., 2011; Maday and Holzbaur, 2016). These studies suggest that reduced lysosomal proteolytic function observed in NPC cells (Elrick et al., 2012) may also contribute to autophagic stress in NPC axons. Impaired AV-lysosome fusion was also shown to induce AV accumulation in NPC1-deficient cells (Maetzel et al., 2014; Sarkar et al., 2013). Studies in Npc1^−/− mouse brains demonstrated AV accumulation (Ko et al., 2005) revealed to be mTOR-independent (Boland et al., 2010), suggesting that autophagic stress may not result from increased autophagy induction in the brain. Interestingly, a recent study in an NPC cat model reported accumulation of LC3-positive puncta in axons and presynaptic terminals but not in the neuronal cell body or dendrites (Gurda et al., 2018), further highlighting the importance of the axonal trafficking and positioning of mature lysosomes to maintain autophagic flux and local degradation capacity. These studies support our bidirectional transport model in which axonal AVs achieve effective maturation and clearance during their retrograde trafficking route along axons, where they accept active lysosomal hydrolases through fusion with mature lysosomes moving anterogradely from soma-derived pools.

While progressive lipid storage in NPC alters lysosome degradation capacity with disease progression, we unexpectedly observed a reduced density of degradative lysosomes in axons but an expanded lysosomal system in the cell body in embryonic NPC cortical neurons at DIV7–8, which preceded axonal AV accumulation. Subsequently, in adult DRG neurons isolated from adult presymptomatic NPC mice, decreased axonal lysosome density was associated with a further reduction in active lysosomes and a robust increase in axonal AVs. Consistent with these observations, rescuing lysosome delivery to distal axons with elevated Arl8b expression reduced, but did not eliminate, autophagic stress in presymptomatic NPC axons. Therefore, proper lysosome positioning and degradation capacity are both important to maintain axonal homeostasis.

Limitations of the Study

While our study supports a cholesterol-dependent sequestration of kinesin-1 and Arl8 on the surface of NPC lysosomes, we do not rule out that changes in other membrane lipids may also contribute to this abnormal protein sequestration. Further development of tools for visualizing diverse lipid species in live cells at high spatiotemporal resolutions may help determine the lipid environment leading to protein sequestration on NPC lysosomes and further advance our understanding of the dynamic interplay between lipids and proteins in mediating cellular events at the lysosomal membrane.

STAR Methods

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Zu-Hang Sheng (shengz@ninds.nih.gov).

Materials Availability

Plasmids newly generated in this study are listed in the key resources table and available on request.

Data and Code Availability

The datasets supporting the current study are available from the corresponding author on request.

Experimental Model and Subject Details

Npc1 Mouse Model and Care

Npc1 null (BALB/cNctr-Npc1^m1N/J; Npc1^−/−) and WT (Npc1^+/+) mice were generated from heterozygote matings, and breeder pairs were kindly provided by Dr. William Pavan (Loftus et al., 1997). Animals were maintained in the NINDS Animal Facility, where they were housed in groups of 3–5 mice per cage on a 12-hour light-dark cycle (6 am – 6 pm) with access to food and water ad libitum. Primary neurons were obtained from mouse embryos at embryonic day 18–19 or adult mice at postnatal day 30–40, and animals of both sexes were used. All animal care and procedures were performed in accordance with NIH guidelines and approved by the NIH, NINDS/NIDCD Animal Care and Use Committee.

Primary Neuron Cultures

For embryonic cortical neurons, embryos from timed heterozygote breedings were obtained at embryonic day 18–19 and stored in ice-cold Hibernate medium supplemented with 2% B27 and 0.1X antibiotic-antimycotic while genotyping was performed to identify WT and Npc1^−/− embryos. Following identification, cortical neurons were prepared as previously described (Kang et al., 2008). Cortical neurons were cultured in Neurobasal media supplemented with 2% B27, 0.5 mM GlutaMAX, and 55 μM 2-mercaptoethanol and maintained at 37°C and 5% CO₂. Media was one-half changed at DIV1 and one-fourth changed every three days after. For adult cortical neurons, cultures were prepared from mice at postnatal day 30–40 (P30–40) using the Adult Brain Dissociation Kit according to the manufacturer’s instructions.

For DRG neurons, WT and Npc1^−/− littermates at P30–40 were anesthetized with intraperitoneal injection of 2.5% avertin then transcardially perfused with ice-cold perfusion buffer (125 mM NaCl, 25 mM NaHCO₃, 2.5 mM KCl, 1.25 mM NaH₂PO₄, 10 mM D-glucose, 2 mM MgCl₂, 2 mM CaCl₂ bubbled with 95% O₂ and 5% CO₂). After bisecting the spinal canal to expose the spinal cord, DRGs were removed and transferred to HBSS supplemented with 10 mM HEPES and 0.1X antibiotic-antimycotic. After excess dorsal roots were trimmed off, DRGs were digested in 2.5 units/mL dispase II and 200 units/mL collagenase at 37°C for 30 minutes followed by 35 minutes gently rotating at room temperature. DRGs were triturated in Neurobasal-A media supplemented with 2% B27, 0.5 mM GlutaMAX, and 2.5% fetal bovine serum. After filtering debris with a 70-μm nylon strainer, neurons were collected and plated on 12-mm coverslips coated with 20 μg/mL poly-L-ornithine and 10 μg/mL laminin. Where indicated and before plating, DRGs were nucleofected with 0.4 μg plasmid DNA using a Basic Neuron Small Cell Number Nucleofector Kit according to the manufacturer’s instructions. DRG neurons were maintained at 37°C and 5% CO₂, and media was one-half changed three hours after plating.

Method Details

Lentivirus Production

Lenti-X 293T cells were transfected with the desired lentiviral construct, the packaging plasmid psPAX2 (Addgene #12260), the envelope plasmid pMD2.G (Addgene #12259), and pAdVAntage for enhanced protein expression at a ratio of 11:7.5:2.5:1. DMEM media was changed every 24 hours and collected at 48- and 72-hours post-transfection. Following centrifugation to remove cell debris, lentivirus-containing media was aliquoted and stored at −80°C until use.

Microfluidic Device Preparation and Culture

Microfluidic devices were prepared as previously described (Zhou et al., 2016). Briefly, SYLGARD 184 silicone elastomer base was combined with its corresponding curing agent at a ratio of 10:1. After mixing and de-foaming, the silicone elastomer was poured into a silicone wafer template made out of SU-8 by photolithography. Following desiccation for 2.5 hours to remove air bubbles, the wafer was heated at 80°C for 2 hours to cure. Then, the silicone elastomer was carefully removed from the wafer, and microfluidic devices were punched out following template guides. The devices were then washed in an ultrasonic cleaner consecutively for 10 minutes each in 50% ethanol followed by double-distilled water. Once dry, microfluidic devices were adhered to coverslips prior to neuron culture.

For culture in microfluidic devices, 2 × 10⁵ freshly dissociated cortical neurons in a volume of 10 μl were added directly to the soma/dendritic transwell. If necessary, for lentiviral transduction, 5 μl of the desired lentivirus was combined with 2 × 10⁵ freshly dissociated neurons in a volume of 5 μl, and this 10 μl volume was then added directly to the soma/dendritic transwell. To equilibrate, 10 μl of media was added directly to the axon transwell. Following a 10-minute incubation, 200 μl of media was added to the axon chamber followed by 180 μl of media to the soma/dendritic chamber. After a one-fourth media change at DIV1, media was one-fourth changed every three days.

Labeling Degradative Lysosomes with Activity-based Lysosome Probes

For labeling active glucocerebrosidase (GCase), green (MDW933) and red (MDW941) fluorescent activity-based probes specific for GCase (Farfel-Becker et al., 2019; Witte et al., 2010) were kindly provided by Dr. Ellen Sidransky (Westbroek et al., 2016). Unless otherwise stated, live neurons were incubated with 500 nM MDW933 or 100 nM MDW941 for 30 minutes or 1 hour at 37°C and 5% CO₂ followed by live-neuron imaging or fixation in Bouin’s solution with 4% sucrose for immunostaining. For labeling active cathepsin D (CTSD), live neurons were incubated with 1 μM BODIPY-FL-pepstatin A for 30 minutes at 37°C and 5% CO₂ followed by live-neuron imaging.

Live Neuron Imaging and Motility Analysis

For live-cell imaging, neurons were rinsed three times and transferred to low fluorescence Hibernate medium supplemented with 2% B27 and 0.5 mM GlutaMAX. Cells were visualized with a 40x/1.3 NA oil immersion objective on a Zeiss LSM880 confocal microscope. Time-lapse imaging was performed in an incubation chamber maintained at 37°C. To quantify motility, kymographs were generated using ImageJ as previously described (Kang et al., 2008). Organelles were considered stationary if the net displacement was ≤ 10 μm during the entire acquisition period. If the net displacement was ≥ 10 μm throughout this period, the organelle was considered motile in its corresponding direction.

Immunofluorescence

Neurons were plated on poly-L-ornithine-coated 12-mm coverslips in 24-well plates at a density of 0.2 × 10⁶ cells per well or in microfluidic devices as indicated above. At DIV7–8, neurons were quickly rinsed in PBS then fixed with either 4% paraformaldehyde (PFA) with 4% sucrose or Bouin’s solution with 4% sucrose for 15 minutes at room temperature. Cells were then washed three times with PBS, permeabilized and blocked with 0.4% saponin, 1% BSA, and 5% goat serum in PBS for 1 hour, and incubated with primary antibodies in 0.1% saponin, 1% BSA, and 5% goat serum in PBS overnight at 4°C. Alternatively, after fixation, cells were washed three times with PBS, permeabilized with 0.1% Triton X-100 for 10 minutes, washed three times with PBS, blocked with 2% BSA and 5% goat serum in PBS for 1 hour, and incubated with primary antibodies in blocking buffer overnight at 4°C. After primary antibody labeling, cells were washed three times with PBS then incubated with Alexa Fluor conjugated secondary antibodies for 30 minutes at room temperature. After washing four times with PBS, coverslips were mounted with anti-fade mounting medium while neurons in microfluidic devices were transferred to low fluorescence Hibernate medium prior to imaging.

Western Blot

Neurons were plated on poly-L-ornithine-coated 35-mm dishes at a density of 2 × 10⁶ cells per dish. At DIV7–8, cells were rinsed with cold PBS and lysed in RIPA buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 0.5% DOC) supplemented with a protease inhibitor cocktail tablet. Cell lysates were centrifuged at 15,000 x rpm for 15 minutes at 4°C and soluble proteins were collected. Equal amounts of proteins (10 μg) were resolved on NuPAGE 4–12% Bis-Tris protein gels and transferred to PVDF membranes under 250 mA constant current for 1.5 hours at 4°C. After blocking in 10% milk in TBS with 0.1% Tween 20 (TBST) for 1 hour at room temperature, membranes were incubated with primary antibodies overnight at 4°C. After washing four times with TBST, membranes were incubated with horseradish peroxidase (HRP) conjugated secondary antibodies (1:5000) in 5% milk in TBST for 1 hour at room temperature. After washing four times with TBST, proteins were detected using SuperSignal chemilumescent substrates and analyzed in ImageJ.

Electron Microscopy

Samples for EM were prepared as previously described (Xie et al., 2015) with minor modifications. Briefly, mice were anesthetized and transcardially perfused with freshly made EM fixative (2% glutaraldehyde and 2% paraformaldehyde in 0.1 N cacodylate buffer). Tissues were removed and post-fixed in EM fixative overnight, then sent to the NINDS EM Facility for staining and silver enhancement. After dehydration, embedding, and sectioning, images were taken using an electron microscope (1200EX; JEOL).

STED Super-resolution Imaging

STED super-resolution imaging was performed as previously described (Lin et al., 2017). Neurons were plated on 25-mm poly-L-ornithine-coated coverslips in 6-well plates and processed for GST-D4H-mCherry labeling and/or immunostaining. An inverted STED microscope with a resolution of ~50–90 nm (TCS SP8 STED 3X, Leica, Germany) was used. Alexa Fluor 488 was excited by a tunable white light laser (70% of maximum power) at 470 nm (10%) with the STED depletion laser at 592 nm (30% of the maximum power); its fluorescence at 480–560 nm was collected using time-gated detection (1.5–6.5 nsec). Alexa Fluor 594 was excited by the tunable white light laser at 594 nm (10%) with the STED depletion laser at 775 nm (30% of the maximum power); its fluorescence between 600–640 nm was collected using time-gated detection (1.5–6.5 nsec). GST-D4H*-mCherry was excited by the tunable white light laser at 568 nm (30%) with the STED depletion laser at 660 nm (30% of the maximum power); its fluorescence between 580–610 nm was collected using time-gated detection (1.5–6.5 nsec). Alexa Fluor 647 was excited by the tunable while light laser at 647 nm (20%) with the STED depletion laser at 775 nm (30% of the maximum power); its fluorescence between 660–750 nm was collected using time-gated detection (1.5–6.5 nsec). For dual-color and triple-color imaging, channels at the higher wavelength range were imaged first to avoid bleaching with a frame-scanning mode. Images were subject to deconvolution processing using Huygens software.

Co-immunoprecipitation

To detect the interaction between SKIP and Arl8, an immunoprecipitation assay was performed. Brain tissues from WT and Npc1^−/− E18 embryos were collected and homogenized in lysis buffer (25 mM Tris pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.1% NP-40, 5% glycerol) supplemented with a protease inhibitor cocktail tablet and incubated on ice for 30 minutes. The crude homogenates were centrifuged at 13,000 x g at 4°C for 15 minutes. One milligram of cleared brain homogenates was incubated with anti-Arl8-coated Protein A Sepharose beads at 4°C for 3 hours, followed by centrifugation and three washes with cold lysis buffer. Samples were eluted by heating at 95°C in sample buffer supplemented with 20 mM DTT, followed by western blotting analysis. The VeriBlot for IP detection reagent was used at 1:2000 as the secondary antibody to test IP signals without interference from the IP antibodies in the eluted samples.

Recombinant GST-D4H*-mCherry Protein Purification

Recombinant GST-D4H*-mCherry proteins were purified as previously described (Lim et al., 2019) with minor modifications. GST-D4H*-mCherry (Addgene #134604) was expressed in BL21(DE3) Escherichia coli and grown to an optical density at 600 nm (OD₆₀₀) of ~0.6. Protein production was then induced with IPTG at a final concentration of 0.4 mM and grown for 20 hours at 18°C. The bacterial pellet was resuspended in lysis buffer (20 mM Tris-Cl pH 8.0, 0.1 M NaCl, 1 mM DTT) supplemented with a protease inhibitor cocktail tablet and incubated with end-over-end rotation at 4°C for 30 minutes. The bacterial lysate was sonicated on an ice-cold water bath with a 5 second pulse at 30% amplitude followed by a 10 second break for a total of 15 minutes. Cleared lysate containing soluble protein was incubated with pre-equilibrated Glutathione Sepharose 4B beads for 2 hours at 4°C with end-over-end rotation. GST-D4H*-mCherry proteins were eluted with 25 mM reduced L-glutathione in 50 mM Tris-Cl pH 8.8 and 200 mM NaCl, then sequentially filtered using 3kDa and 30kDa molecular weight cutoff centrifugal filter units to concentrate the protein in 50 mM Tris-Cl pH 8.0 and 5 mM DTT. Following supplementation with sucrose to a final concentration of 0.5 M, the GST-D4H*-mCherry protein solution was aliquoted, frozen in liquid nitrogen, and stored at −80°C until use.

Detecting Luminal and Membrane Cholesterol with Filipin and GST-D4H*-mCherry

Luminal and membrane cholesterol was labeled as previously described (Lim et al., 2019; Wilhelm et al., 2017) with minor modifications. Neurons were plated on poly-L-ornithine-coated 12-mm coverslips in 24-well plates at a density of 0.2 × 10⁶ cells per well. At DIV7–8, cells were quickly rinsed in PBS, then fixed with 4% PFA with 4% sucrose for 15 minutes at room temperature. The coverslips were washed three times with PBS, then submerged in liquid nitrogen for 30 seconds to permeabilize the plasma membrane while leaving intracellular membranes intact. Cells were blocked with 1% BSA in PBS for 1 hour, then incubated with recombinant GST-D4H*-mCherry (1:100) in 1% BSA in PBS for 2 hours at room temperature. The cells were quickly rinsed twice in PBS, re-fixed with 4% PFA with 4% sucrose for 10 minutes at room temperature, blocked with 1% BSA in PBS for 1 hour at room temperature, and incubated with a LAMP1 primary antibody overnight at 4°C. Cells were then washed three times with PBS, and incubated with an Alexa Fluor 488 conjugated secondary antibody for 30 minutes at room temperature. After washing three times with PBS, cells were incubated with 50 μg/mL filipin in PBS for 40 minutes at room temperature followed by three more washes in PBS. The coverslips were then mounted with anti-fade mounting medium.

Magnetic Isolation of Lysosomes

Lysosomes were isolated as previously described (Bilgin et al., 2017; Diettrich et al., 1998) with minor modifications. Neurons were plated on poly-L-ornithine- and laminin-coated 10-cm culture dishes at a density of 10 × 10⁶ cells per dish. The day before isolation, conditioned neuronal media was collected by removing 10 mL media from each 10-cm dish. Cells were then treated with 1 mL DexoMAG iron-dextran solution (10 mg/mL) for 20 hours, washed with PBS, and cultured for an additional 2 hours in conditioned neuronal media collected from the day before. Cells were gently scraped and washed three times in cold PBS, then lysed using a 25 G needle in cold SCA buffer (20 mM HEPES-KOH, pH 7.5, 10 mM KCl, 1.5 mM MgCl₂, 1 mM EDTA, 1 mM EGTA, 250 mM sucrose) supplemented with 1 mM DTT, 1 mM Pefabloc, benzonase nuclease, and a protease inhibitor cocktail tablet. Ruptured cells were centrifuged at 800 x g for 5 minutes at 4°C, then the supernatant containing cytosol and organelles was collected and this process repeated two more times, followed by loading onto pre-equlibrated MS columns fitted into the magnetic field of a MiniMACS separator attached to a MACS multistand. Columns were washed five times with SCA buffer, removed from the separator unit, and lysosomal membrane fractions were eluted with 200 μL buffer containing 5 mM Tris-HCl (pH 7.5), a protease inhibitor cocktail, and 1% Triton X-114. Protein concentration was determined using a BCA Protein Assay Kit according to the manufacturer’s instructions. For immunoblots, eluted proteins were precipitated in methanol/chloroform, resuspended in LDS sample buffer with 100 mM DTT, and resolved on NuPAGE Bis-Tris protein gels.

RNA Extraction and Quantitative Real-Time PCR

Neurons were plated on poly-L-ornithine-coated wells of a 6-well plate at a density of 2 × 10⁶ cells per well. At DIV7, total RNA was isolated using the RNeasy Mini Kit, and cDNA was synthesized using the iScript cDNA Synthesis Kit according to the manufacturer’s instructions. Quantitative real-time PCR was performed using the iTaq Universal SYBR Green Supermix and a Bio-Rad CFX384 Real-Time system. Relative gene expression was normalized to β-actin. Oligonucleotide sequences were previously described (Chen et al., 2017; Mocholi et al., 2018) and are listed in the Key Resources Table.

TUNEL Assay and Analysis

Neurons were plated on 12-mm poly-L-ornithine-coated coverslips in 6-well plates. Where indicated and before plating, neurons were transduced with lentiviruses encoding Arl8b-mCh or mCh. At DIV7 and DIV10, cells were fixed and immunostained for MAP2 followed by TUNEL staining. TUNEL-positive neurons were detected using the Fluorescein In Situ Cell Death Detection Kit according to the manufacturer’s instructions. The coverslips were then mounted with anti-fade mounting medium with DAPI. Neurons were visualized with a 40x/1.3 NA oil immersion objective on a Zeiss LSM880 confocal microscope. Images of entire coverslips were acquired using the Tiles acquisition mode setting in Zeiss ZEN black software. The percentage of TUNEL-positive cells to total DAPI staining was calculated in thresholded, watershed segmented images using ImageJ.

Image Acquisition and Quantification

Confocal images were acquired using a Zeiss LSM880 confocal microscope with a 40x/1.3 NA oil immersion objective. All images of endogenous proteins from a given experiment were taken on the same day using the same settings and analyzed using the same threshold. Image analyses were performed using ImageJ (Schindelin et al., 2012). For quantification of axonal autophagosome density, the number of EGFP-LC3 puncta was manually counted and normalized by axon length. For quantification of axonal lysosome density, the lysosome channel was segmented to identify bona fide lysosome vesicles. The fill holes and watershed functions were then applied, and the number of lysosomes was detected using the analyze particles function and normalized by axon length. For quantification of lysosomal integrated density in the cell body and axonal growth cones, the lysosome channel was segmented, and raw fluorescent intensity values were retrieved by redirecting the binary mask to the original 16-bit grayscale image. Mean integrated density values were calibrated to the area of the cell body or axonal growth cone and normalized to a reference group as indicated in the corresponding figure legend. For endolysosomal membrane cholesterol quantification, the Just Another Colocalization Plugin (JACoP) (Bolte and Cordelieres, 2006) was used. A set threshold was applied to each LAMP1 and GST-D4H*-mCh channel to identify authentic signals, and the Manders’ overlap coefficient representing the fraction of LAMP1-vesicles that were also positive for GST-D4H*-mCh was calculated. Data were normalized to a reference group as indicated in the corresponding figure legend.

Quantification and Statistical Analysis

Statistical analyses were carried out using GraphPad Prism 8 software. For comparison between two groups, the Student’s t test (sample size n ≥ 30) or Mann-Whitney test (sample size n < 30) was used. For comparison between three or more groups, one-way ANOVA was used. All data are expressed as mean ± SEM. Differences were considered statistically significant at a value of p < 0.05.

Supplementary Material

2

Video S1. Axonal Transport of Autophagic Vacuoles in Distal Axons of WT DRG Neurons, Related to Figure 1E

WT DRG neurons isolated from P30–40 mice were nucleofected with EGFP-LC3 at DIV0, followed by live imaging at DIV3–4. Time-lapse images were collected every 2 seconds for 150 frames totaling 5 minutes. Cell body is toward the left.

3

Video S2. Axonal Transport of Autophagic Vacuoles in Distal Axons of Npc1^−/− DRG Neurons, Related to Figure 1E

Npc1^−/− DRG neurons isolated from presymptomatic P30–40 mice were nucleofected with EGFP-LC3 at DIV0, followed by live imaging at DIV3–4. Time-lapse images were collected every 2 seconds for 150 frames totaling 5 minutes. Note the increased number of autophagic vacuoles in presymptomatic Npc1^−/− axons with similar motility patterns compared to control WT axons. Cell body is toward the left.

4

Video S3. Axonal Transport of Active Lysosomes in Distal Axons of WT Cortical Neurons, Related to Figure 3F

WT cortical neurons cultured in microfluidic devices were incubated with MDW941 (100 nM) for 30 minutes to label active GCase-positive lysosomes before live imaging at DIV7–8. Time-lapse images were collected every 2 seconds for 90 frames totaling 3 minutes. Distal axon terminal is toward the right.

5

Video S4. Axonal Transport of Active Lysosomes in Distal Axons of Npc1^−/− Cortical Neurons, Related to Figure 3F

Npc1^−/− cortical neurons cultured in microfluidic devices were incubated with MDW941 (100 nM) for 30 minutes to label active GCase-positive lysosomes before live imaging at DIV7–8. Time-lapse images were collected every 2 seconds for 90 frames totaling 3 minutes. Note that Npc1^−/− axons show reduced anterograde lysosome motility compared to control WT axons. Distal axon terminal is toward the right.

6

Video S5. Axonal Transport of Autophagic Vacuoles in Distal Axons of WT Cortical Neurons, Related to Figure S3G

WT cortical neurons cultured in microfluidic devices were transduced with a lentivirus encoding EGFP-LC3 at DIV0, followed by live imaging at DIV10. Time-lapse images were collected every 2 seconds for 90 frames totaling 3 minutes. Cell body is toward the left.

7

Video S6. Axonal Transport of Autophagic Vacuoles in Distal Axons of Npc1^−/− Cortical Neurons, Related to Figure S3G

Npc1^−/− cortical neurons cultured in microfluidic devices were transduced with a lentivirus encoding EGFP-LC3 at DIV0, followed by live imaging at DIV10. Time-lapse images were collected every 2 seconds for 90 frames totaling 3 minutes. Note that autophagic vacuoles in Npc1^−/− axons show similar motility patterns compared to control WT axons. Cell body is toward the left.

8

Video S7. Axonal Lysosome Delivery from the Soma to Distal Axons of Npc1^−/− Cortical Neurons, Related to Figure 7C

Npc1^−/− cortical neurons were cultured in microfluidic devices. At DIV8, MDW941 (5 nM) was applied only to the soma/dendritic chamber for 1 hour to label active GCase-positive lysosomes in the cell body. Time-lapse images of axon bundles in microgrooves were collected every 1 second for 180 frames totaling 3 minutes. Axon chamber is toward the right.

9

Video S8. Axonal Lysosome Delivery from the Soma to Distal Axons of Npc1^−/− Cortical Neurons Treated with HPCD, Related to Figure 7C

Npc1^−/− cortical neurons cultured in microfluidic devices were treated with HPCD at 100 μM for 48 hours. At DIV8, MDW941 (5 nM) was applied only to the soma/dendritic chamber for 1 hour to label active GCase-positive lysosomes in the cell body. Time-lapse images of axon bundles in microgrooves were collected every 1 second for 180 frames totaling 3 minutes. Axon chamber is toward the right.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Chicken polyclonal anti-mCherry	Novus	Cat#NBP2–25158; RRID: AB_2636881
Goat polyclonal anti-Cathepsin B	R&D Systems	Cat#AF965; RRID: AB_2086949
Mouse monoclonal anti-β3-tubulin	Sigma-Aldrich	Cat#T8578; RRID: AB_1841228
Mouse monoclonal anti-β-actin	Novus	Cat#NB600–501; RRID: AB_10077656
Mouse monoclonal anti-GAPDH	Millipore	Cat#CB1001; RRID: AB_2107426
Mouse monoclonal anti-Kinesin Heavy Chain	Millipore	Cat#MAB1614; RRID: AB_94284
Mouse monoclonal anti-LAMP1	Cell Signaling Technology	Cat#15665; RRID: AB_2798750
Mouse monoclonal anti-MAP2	BD Biosciences	Cat#556320; RRID: AB_396359
Mouse monoclonal anti-Sec61β	Santa Cruz Biotechnology	Cat#sc-393633; RRID: AB_2885065
Mouse monoclonal anti-SV2	Developmental Studies Hybridoma Bank	Cat#SV2; RRID: AB_2315387
Rabbit polyclonal anti-Arl8	Proteintech	Cat#13049–1-AP; RRID: AB_2059000
Rabbit polyclonal anti-GFP	Abcam	Cat# ab6556; RRID: AB_305564
Rabbit polyclonal anti-Glucocerebrosidase	Sigma-Aldrich	Cat#G4171; RRID: AB_1078958
Rabbit polyclonal anti-LAMP1	Abcam	Cat#ab24170; RRID: AB_775978
Rabbit monoclonal anti-Niemann-Pick C1	Abcam	Cat#ab134113; RRID: AB_2734695
Rabbit polyclonal anti-SKIP (PLEKHM2)	Abcam	Cat#ab91581; RRID: AB_2050192
Rabbit polyclonal anti-SKIP (PLEKHM2)	Thermo Fisher Scientific	Cat#PA5–20850; RRID: AB_11153975
Rat monoclonal anti-Cathepsin D	R&D Systems	Cat#MAB1029; RRID: AB_2292411
Rat monoclonal anti-LAMP1	Developmental Studies Hybridoma Bank	Cat#1D4B; RRID: AB_2134500
Donkey polyclonal anti-rabbit IgG, highly cross-adsorbed secondary antibody, Alexa Fluor 488	Thermo Fisher Scientific	Cat#A-21206, RRID: AB_2535792
Goat polyclonal anti-chicken IgY, cross-adsorbed secondary antibody, Alexa Fluor 594	Thermo Fisher Scientific	Cat#A-11042; RRID: AB_2534099
Goat polyclonal anti-mouse IgG, cross-adsorbed secondary antibody, Alexa Fluor 488	Thermo Fisher Scientific	Cat#A-11017; RRID: AB_2534084
Goat polyclonal anti-mouse IgG, cross-adsorbed secondary antibody, Alexa Fluor 546	Thermo Fisher Scientific	Cat#A-11018; RRID: AB_2534085
Goat polyclonal anti-mouse IgG, cross-adsorbed secondary antibody, Alexa Fluor 594	Thermo Fisher Scientific	Cat#A-11005; RRID: AB_2534073
Goat polyclonal anti-rabbit IgG, cross-adsorbed secondary antibody, Alexa Fluor 555	Thermo Fisher Scientific	Cat#A-21428; RRID: AB_2535849
Goat polyclonal anti-rabbit IgG, highly cross-adsorbed secondary antibody, Alexa Fluor 594	Thermo Fisher Scientific	Cat#A-11037; RRID: AB_2534095
Goat polyclonal anti-mouse IgG, highly cross-adsorbed secondary antibody, Alexa Fluor 647	Thermo Fisher Scientific	Cat#A-21236; RRID: AB_2535805
Goat polyclonal anti-rabbit IgG, highly cross-adsorbed secondary antibody, Alexa Fluor 647	Thermo Fisher Scientific	Cat#A-21245; RRID: AB_2535813
Goat polyclonal anti-rat IgG, cross-adsorbed secondary antibody, Alexa Fluor 488	Thermo Fisher Scientific	Cat#A-11006; RRID: AB_2534074
Donkey anti-rabbit IgG, HRP-linked	GE Healthcare	Cat#NA934; RRID: AB_772206
Rabbit anti-goat IgG, HRP-linked	Thermo Fisher Scientific	Cat#81–1620; RRID: AB_2534006
Sheep anti-mouse IgG, HRP-linked	GE Healthcare	Cat#NA931; RRID: AB_772210
Bacterial and Virus Strains
BL21 (DE3) Escherichia coli	Invitrogen	C600003
Chemicals, Peptides, and Recombinant Proteins
MDW933, MDW941	A gift from Dr. Ellen Sidransky (NHGRI, NIH) (Westbroek et al., 2016)	N/A
BODIPY-FL-pepstatin A	Thermo Fisher Scientific	Cat#P12271
MitoTracker Green	Thermo Fisher Scientific	Cat#M7514
Filipin	Sigma-Aldrich	Cat#F9765
2-hydroxypropyl-β-cyclodextrin	Sigma-Aldrich	Cat#H107
Critical Commercial Assays
Basic Neuron Small Cell Number Nucleofector Kit	Lonza	Cat#VSPI-1003
Fluorescein In Situ Cell Death Detection Kit	Roche	Cat#11684795910
BP Clonase II Enzyme Mix	Thermo Fisher Scientific	Cat#11789020
LR Clonase II Enzyme Mix	Thermo Fisher Scientific	Cat#11791020
RNeasy Mini Kit	QIAGEN	Cat#74104
iScript cDNA Synthesis Kit	Bio-Rad	Cat#1708891
iTaq Universal SYBR Green Supermix	Bio-Rad	Cat#1725120
Experimental Models: Cell Lines
Lenti-X 293T	Takara	Cat#632180
Experimental Models: Organisms/Strains
Mouse: WT/Npc1-/-: BALB/cNctr-Npc1m1N/J	(Loftus et al., 1997)	JAX:003092; RRID: IMSR_JAX:003092
Oligonucleotides
Ctsb forward: CTTAGGAGTGCACGGGAGAG	(Chen et al., 2017)	N/A
Ctsb reverse: CTTGTCATGGGCACTGGTCA	(Chen et al., 2017)	N/A
Ctsd forward: TACTCCATGCAGTCATCGCC	(Chen et al., 2017)	N/A
Ctsd reverse: GACGACTGTGAAACACTGCG	(Chen et al., 2017)	N/A
Atp6v0e1 forward: GCATACCACGGCCTTACTGT	(Chen et al., 2017)	N/A
Atp6v0e1 reverse: GAGGATTGAGCTGTGCCAGA	(Chen et al., 2017)	N/A
Atp6v1h forward: ACTCCCCGAGGCTATCCAG	(Chen et al., 2017)	N/A
Atp6v1h reverse: CACGAACTTCAGCAGCCTTG	(Chen et al., 2017)	N/A
Lamp1 forward: GCCCACAAACCCCACTGTAT	(Chen et al., 2017)	N/A
Lamp1 reverse: TTTGGGCTGATGTTGAACGC	(Chen et al., 2017)	N/A
Actb forward: GTGACGTTGACATCCGTAAAGA	(Mocholi et al., 2018)	N/A
Actb reverse: GCCGGACTCATCGTACTCC	(Mocholi et al., 2018)	N/A
Arl8a siRNA #1 ID MSS290361	Thermo Fisher Scientific	Cat#1320001
Arl8b siRNA #2 ID MSS289173	Thermo Fisher Scientific	Cat#1320001
SKIP (Plekhm2) siRNA #1 ID MSS290835	Thermo Fisher Scientific	Cat#1320001
SKIP (Plekhm2) siRNA #2 ID MSS290836	Thermo Fisher Scientific	Cat#1320001
Recombinant DNA
pCMV-EGFP-LC3	(Cheng et al., 2015)	N/A
pLEX-PGK-EGFP-LC3	(Farfel-Becker et al., 2019)	N/A
ptf-LC3	(Kimura et al., 2007)	Addgene plasmid # 21074
pCMV-Arl8b-mCh	A gift from Dr. Juan Bonifacino (NICHD, NIH) (Farias et al., 2017)	N/A
pCMV-Arl8b-mCh-T34N	This study	N/A
pCMV-Arl8b-mCh-Q75L	This study	N/A
pLEX-PGK-Arl8b-mCh	This study	N/A
pLEX-PGK-mCh	This study	N/A
pHAGE-CMV-EGFP-KHC-(1–514)	This study	N/A
pHAGE-CMV-EGFP-KHC-CBD	This study	N/A
psPAX2	A gift from Dr. Didier Trono (EPFL)	Addgene plasmid # 12260
pMD2.G	A gift from Dr. Didier Trono (EPFL)	Addgene plasmid # 12259
pAdVAntage	Promega	Cat#E1711
pGEX-KG-D4H*-mCherry	(Lim et al., 2019)	Addgene plasmid # 134604
Software and Algorithms
Fiji 2.0.0	(Schindelin et al., 2012)	http://fiji.sc; RRID: SCR_002285
Huygens Software	Scientific Volume Imaging	https://svi.nl/HuygensSoftware; RRID: SCR_014237
GraphPad Prism 8	GraphPad	http://www.graphpad.com/; RRID: SCR_002798
ZEN Digital Imaging for Light Microscopy	Zeiss	https://www.zeiss.com/microscopy/us/products/microscope-software/zen.html; RRID: SCR_013672
Other
Poly-L-ornithine	Sigma-Aldrich	Cat#P4957
Laminin	Roche	Cat#11243217001
Hibernate-E	Thermo Fisher Scientific	Cat#A1247601
B27	Thermo Fisher Scientific	Cat#17504–044
Antibiotic-antimycotic	Thermo Fisher Scientific	Cat#15240062
Neurobasal	Thermo Fisher Scientific	Cat#21103–049
GlutaMAX	Thermo Fisher Scientific	Cat#35050–61
2-mercaptoethanol	Thermo Fisher Scientific	Cat#21985–023
HBSS, no calcium, no magnesium	Thermo Fisher Scientific	Cat#14170120
HEPES	Thermo Fisher Scientific	Cat#15630080
Dispase II	Roche	Cat#4942078001
Collagenase	Worthington Biochemical	Cat#LS004176
Neurobasal-A	Thermo Fisher Scientific	Cat#10888–022
Fetal Bovine Serum	HyClone	Cat#SH30071.03
DMEM	Thermo Fisher Scientific	Cat#11995–065
Dow Corning SYLGARD 184 Silicone Encapsulant Clear 0.5 kg kit	Ellsworth Adhesives	Cat#184 SIL ELAST KIT 0.5KG
Low Fluorescence Hibernate E	BrainBits	Cat#HELF
Low Fluorescence Hibernate A	BrainBits	Cat#HALF
Bouin’s solution	Sigma-Aldrich	Cat#HT10132
Fluoro-Gel with Tris Buffer mounting media	Electron Microscopy Sciences	Cat#17985–10
Prolong Diamond antifade mounting media with DAPI	Thermo Fisher Scientific	Cat#P36971
VECTASHIELD antifade mounting media with DAPI	Vector Laboratories	Cat#H-1200–10
Glutathione Sepharose 4B beads	GE Healthcare	Cat#17–0756-01
Protein A-Sepharose CL-4B	GE Healthcare	Cat#17–0780-01
cOmplete EDTA-free protease inhibitor cocktail tablets	Roche	Cat#4693159001
DexoMAG 40	Liquids Research Limited	Cat#40
Pefabloc SC	Roche	Cat#PEFBSC-RO
Benzonase nuclease	Millipore	Cat#E1014–5KU
MS Columns	Miltenyi Biotec	Cat#130–042-201
Adult Brain Dissociation Kit	Miltenyi Biotec	Cat#130–107-677
VeriBlot for IP Detection Reagent (HRP)	Abcam	Cat#ab131366

Highlights:

• Endocytic and autophagic organelles accumulate in NPC dystrophic axons

• Elevated membrane cholesterol sequesters motor-adaptors of lysosome transport

• Impaired lysosome transport into NPC axons leads to altered axonal homeostasis

• Reducing lipid levels rescues lysosome delivery and autophagic stress in NPC axons

Abbreviations:

AV

autophagic vacuole

AVd

degradative autophagic vacuole

AVi

initial autophagic vacuole

CBD

cargo-binding domain

DIV

days in vitro

DRG

dorsal root ganglion

GCase

glucocerebrosidase

HPCD

2-hydroxypropyl-β-cyclodextrin

KHC

kinesin heavy chain

KIF5

kinesin-1 family

LAMP1

lysosome associated membrane protein 1

LSD

lysosomal storage disorder

MD

motor domain

NPC

Niemann-Pick disease type C

SKIP

SifA and kinesin-interacting protein

STED

stimulated emission and depletion

TEM

transmission electron microscopy

TUNEL

terminal deoxynucleotidyl transferase dUTP nick end labeling

WT

wild type