Niemann Pick C1 mistargeting disrupts lysosomal cholesterol homeostasis contributing to neurodegeneration in a Batten disease model

Abstract

Neurodegeneration is a devastating manifestation in most lysosomal storage disorders (LSDs). Loss-of-function mutations in CLN1, encoding palmitoyl-protein thioesterase-1 (PPT1), cause CLN1 disease, a devastating neurodegenerative LSD that has no curative treatment. Numerous proteins in the brain require dynamic S-palmitoylation (palmitoylation-depalmitoylation) for trafficking to their destination. Although PPT1 depalmitoylates S-palmitoylated proteins and its deficiency causes CLN1 disease, the underlying pathogenic mechanism has remained elusive. We report that Niemann-Pick C1 (NPC1), a polytopic membrane protein mediating lysosomal cholesterol egress, requires dynamic S-palmitoylation for trafficking to the lysosome. In Cln1^−/− mice, Ppt1 deficiency misroutes NPC1-dysregulating lysosomal cholesterol homeostasis. Along with this defect, increased oxysterol-binding protein (OSBP) promotes cholesterol-mediated activation of mechanistic target of rapamycin C1 (mTORC1), which inhibits autophagy contributing to neurodegeneration. Pharmacological inhibition of OSBP suppresses mTORC1 activation, rescues autophagy, and ameliorates neuropathology in Cln1^−/− mice. Our findings reveal a previously unrecognized role of CLN1/PPT1 in lysosomal cholesterol homeostasis and suggest that suppression of mTORC1 activation may be beneficial for CLN1 disease.

Misrouting of Niemann Pick C1 disrupts lysosomal cholesterol homeostasis in Batten Disease.

INTRODUCTION

The lysosome (1) has long been considered the terminal organelle for intracellular digestion. However, emerging evidence has remarkably transformed our understanding of the lysosome from a terminal digestive organelle to an important regulator of fundamental metabolic processes (2, 3). Notably, lysosomal dysfunction underlies a family of 70 hereditary metabolic diseases called lysosomal storage disorders (LSDs) (4). Although neurodegeneration is a devastating manifestation in most of the LSDs (5), the underlying pathogenic mechanism(s) have remained elusive. Neuronal ceroid lipofuscinoses (NCLs) (6–8), commonly known as Batten disease (9), constitute a group of the most common neurodegenerative LSDs. These diseases, which mainly affect children and young adults, are both clinically and genetically heterogeneous, although most follow an autosomal recessive pattern of inheritance (6). Although it has been reported that mutations in at least 14 different genes (CLN1–CLN14) underlie various types of NCLs, the CLN9 gene has not yet been identified (10, 11). All types of NCLs share some common clinical and pathological features. These include progressive psychomotor decline, epileptic seizures, visual impairment due to retinal degeneration, and intracellular accumulation of autofluorescent ceroid lipofuscin (6, 11).

The CLN1 disease, formerly called infantile neuronal ceroid lipofuscinosis (12), is caused by loss-of-function mutations in the CLN1 gene (13). Among all NCLs, commonly called Batten disease, CLN1 disease is one of the most devastating neurodegenerative disorders (6, 14), which affect mostly children and young adults. Children afflicted with CLN1 disease are phenotypically normal at birth. However, by 6 to 18 months of age, they manifest psychomotor decline. By 2 years of age, magnetic resonance imaging shows profound cortical atrophy, which is followed by the development of epileptic seizures and blindness due to complete retinal degeneration. Around 4 years of age, an isoelectric electroencephalogram attests to a vegetative state, which lasts for several more years before eventual death (14).

The CLN1 gene encodes palmitoyl-protein thioesterase-1 (PPT1) (15) that catalyzes depalmitoylation of S-palmitoylated proteins, which are the major constituents of ceroid lipofuscin. S-palmitoylation (also called S-acylation) (16) is the only reversible posttranslational lipid modification of proteins. It is a process in which a 16-carbon saturated fatty acid (generally palmitate) is attached to specific cysteine residues in polypeptides via thioester linkage (16–18). Moreover, numerous proteins in the mammalian brain require dynamic S-palmitoylation (palmitoylation-depalmitoylation) (19) for trafficking to their cellular destinations. Localization of proteins to specific cellular compartments is an important determinant of their function. Thus, mistargeting these proteins may impair their function. S-palmitoylation confers hydrophobicity, increases membrane affinity, and promotes protein-protein interactions (17, 18). Recently, an unbiased proteomic study has identified >100 proteins that are putative substrates of PPT1 (20). This study also demonstrated that the levels of S-palmitoylated proteins in the synaptosomes of Cln1^−/− mice were significantly higher compared with those in their wild-type (WT) littermates. We reasoned that inactivating mutations in the CLN1 gene may adversely affect the function(s) of many proteins in the brain, which require dynamic S-palmitoylation for their localization to specific cellular targets. Thus, it is conceivable that more than one pathway may contribute to neuropathology in CLN1 disease. PPT1 deficiency causes intracellular accumulation of S-palmitoylated proteins (constituents of autofluorescent ceroid lipofuscin), a characteristic finding in all CLN diseases. Previously, PPT1 was reported to be exclusively localized to the lysosomal lumen (21). More recent reports have shown that catalytically active PPT1 is also present in the cytoplasm (22). Despite the discovery that inactivating mutations in the CLN1 gene cause CLN1 disease (13), a clear understanding of the pathogenic mechanism(s) underlying this devastating disease has remained elusive for more than two decades.

The lysosome is the major cellular sorting station for dietary cholesterol (23). It maintains cholesterol homeostasis (24). The Niemann-Pick C1 (NPC1) protein, localized to the lysosomal-limiting membrane (25, 26), facilitates cholesterol egress from lysosomal lumen. In contrast, NPC2, a 132–amino acid soluble protein carrying cholesterol, binds to the second, luminally oriented domain of NPC1 and hands off cholesterol to the N-terminal domain (NTD) of NPC1. Docking of the NPC1-NPC2 complex has confirmed a direct cholesterol transfer tunnel between NPC2 and NTD of NPC1, which supported the “hydrophobic handoff” cholesterol transfer model (27). It has been reported that NPC2 in lysosomal lumen facilitates the transfer of cholesterol molecule to the NTD of NPC1 (28). Cholesterol bound to NPC2 is then handed off to NPC1 through a channel within NPC1 spanning the entire glycocalyx layer (24). Under normal physiological conditions, the balance between cholesterol egress by NPC1 and cholesterol input by NPC2 maintains lysosomal cholesterol homeostasis. Moreover, inactivating mutations in either the NPC1 or NPC2 gene cause a fatal hereditary neurodegenerative LSD called NPC disease (29–32).

In this study, we demonstrate that Ppt1 deficiency misroutes NPC1 protein from its normal location on lysosomal membrane. This defect impairs lysosomal cholesterol homeostasis in Cln1^−/− mice. Along with this defect, the increased level of oxysterol-binding protein (OSBP) causes cholesterol-mediated hyperactivation of mechanistic target of rapamycin complex 1 (mTORC1) kinase, which suppresses autophagy contributing to neurodegeneration in CLN1 disease. Pharmacological inhibition of OSBP ameliorates neuropathology in Cln1^−/− mice.

RESULTS

In Cln1^−/− mice both total and lysosomal cholesterol levels are strikingly high

To understand the pathogenic mechanism of CLN1 disease, we used Cln1^−/− mice (33), a reliable animal model of human CLN1 disease (34). Previously, it has been reported that cholesterol levels in the brain of Cln1^−/− mice are significantly higher compared with those in their WT littermates (35). However, in the said study, the cholesterol levels were measured only in total brain homogenates but not in purified lysosomal fractions. This prompted us to measure the cholesterol levels in both total cortical homogenates as well as in purified lysosomal fractions from 2-, 4- and 6-month-old WT and Cln1^−/− mice. The results showed that in all three age groups of Cln1^−/− mice, cholesterol levels in total lysates of cortical tissues were significantly higher compared with those in their WT littermates (Fig. 1A). The cholesterol levels in purified lysosomal fractions of the same tissues from 2-, 4- and 6-month-old Cln1^−/− mice were also significantly higher compared with those from their WT littermates (Fig. 1B). These results were further confirmed by confocal imaging of isolated neurons from the Cln1^−/− mouse brain. The results showed substantially higher level of colocalization of Filipin III–stained cholesterol with that of the lysosomal marker, CellLight lysosomes–green fluorescent protein (GFP), BacMam 2.0 (Fig. 1C). We used CellLight lysosomes-GFP, BacMan 2.0 because fluorescence of lysosomal marker, LAMP-2, was masked by Filipin III cell–based staining of cholesterol. Together, these results showed that cholesterol levels in both total cortical homogenates and in lysosomal fractions from Cln1^−/− mice are significantly higher compared with those in their WT littermates.

Fig. 1. Cholesterol-mediated activation of mTORC1 in Cln1 ^−/− mouse brain.

(A) Cholesterol levels in total cortical homogenates from WT and Cln1^−/− mice (n = 4). (B) Cholesterol levels in lysosomal fractions. (C) Confocal images of cortical neurons from WT (n = 24) and Cln1^−/− mice (n = 24) showing higher colocalization of cholesterol in lysosomes. Scale bar, 20 μm. Also note that in Cln1^−/− mice, a significantly higher level of phosphorylated-S6K1 (pS6K1) (D), p4E-BP1 (E), pULK1 (F), and pTFEB (G) compared with those in their WT littermates (n = 4) suggests hyperactivation of mTORC1 signaling. Immunostaining to confirm the localization of phosphorylated-S6K1 (pS6K1) (H) and p4E-BP1 (I) in the cerebral cortex of WT (n = 12) and Cln1^−/− mice (n = 12). Scale bar, 50 μm. (J) PLA to confirm the interaction between mTOR and lysosome marker, LAMP2 (n = 24). Scale bar, 20 μm. Note a significantly higher colocalization of mTOR with LAMP2 in cortical neurons from Cln1^−/− mice compared with those from their WT littermates. Two-sided permutation t test with complete enumeration was used to calculate the P values (*P < 0.5; **P < 0.01. Asterisks indicate a significant difference relative to corresponding control group). Data are presented as the means ± SD. “n” denotes the number of independent biological replicates for each experiment. In the box plots, the center line is the median, + symbol represents the sample mean, the limits are 25th and 75th percentile, and the whiskers are the minimum and maximum values, respectively. Source data are provided in the source data file.

Increased lysosomal cholesterol mediates mTORC1 hyperactivation in Cln1^−/− mice

The transport of cholesterol from the endoplasmic reticulum (ER) to the lysosome is mediated via vesicle-associated membrane protein–associated protein A (VAPA) and VAPB to OSBP on lysosomal-limiting membrane. Intriguingly, increased lysosomal cholesterol mediates the activation of the mTORC1 kinase (36, 37). The mTORC1 integrates intracellular as well as environmental cues to regulate cell growth and metabolism (38, 39). Notably, it has been reported that mTORC1 hyperactivation negatively regulates autophagy (40–42). Moreover, the dysregulation of autophagy has been implicated in the pathogenesis of human diseases (41), including common neurodegenerative diseases like Alzheimer’s and Parkinson’s (43). Furthermore, in most LSDs, dysregulated autophagy has been reported to contribute to neurodegeneration (44). Cumulatively, these findings raised the possibility that dysregulated lysosomal cholesterol homeostasis may activate mTORC1 and suppress autophagy contributing to neuropathology in CLN1 disease.

To determine the status of mTORC1 activation in the brain of Cln1^−/− mice, we evaluated the levels of phosphorylated S6 kinase 1 (pS6K1) and phosphorylated 4E-binding protein 1 (p4E-BP1), which are the canonical substrates of mTORC1 kinase (45). The results showed that in Cln1^−/− mouse brain, the levels of pS6K1 (Fig. 1D) and p4E-BP1 (Fig. 1E), were significantly higher compared with those in their WT littermates. We also determined the levels of phosphorylated unc-51–like autophagy activating kinase 1 (pULK1) (Fig. 1F) and phosphorylated transcription factor EB (pTFEB) (Fig. 1G). These results were further confirmed by immunostaining, which showed substantially higher levels of pS6K1 (Fig. 1H) and p4E-BP1 (Fig. 1I) immunoreactivity in the brain of Cln1^−/− mice compared with that in their WT littermates. Moreover, we performed proximity ligation assay (PLA) to confirm the interaction between mTOR and lysosome marker, LAMP2. The results showed higher PLA signals in the neurons from Cln1^−/−mice, suggesting that increased colocalization between mTOR and LAMP2 compared with those from their WT littermates (Fig. 1J). Together, these findings raised the possibility that in Cln1^−/− mice, high levels of lysosomal cholesterol may mediate the activation of mTORC1 signaling, although the mechanism underlying dysregulation of lysosomal cholesterol homeostasis in these mice remained to be determined.

In Cln1^−/− mice Ppt1 deficiency misroutes NPC1 protein impairing cholesterol homeostasis

The NPC1 and NPC2 proteins mediate lysosomal cholesterol egress and input, respectively, to maintain cholesterol homeostasis (Fig. 2A). Accordingly, we measured the levels of NPC1 and NPC2 proteins using cortical tissues from 2-, 4-, and 6-month-old WT and Cln1^−/− mice. The results showed that the levels of both NPC1 and NPC2 proteins in total cortical homogenates from Cln1^−/− mice were significantly higher compared with those from their WT littermates (fig. S1, A and B). NPC1 protein levels in purified lysosomal fractions from the same tissues were significantly lower compared with those from their WT littermates (Fig. 2B). Confocal imaging studies using isolated cortical neurons confirmed that colocalization of NPC1 with the lysosomal marker, LAMP2, was also significantly lower in Cln1^−/− mice. The treatment of cultured cortical neurons from Cln1^−/− mice with recombinant PPT1 (r-PPT1) significantly increased the colocalization of NPC1 with lysosomal marker, LAMP2, compared with that in untreated control neurons from Cln1^−/− mice (Fig. 2C). One of the most intriguing findings was the presence of high levels of NPC1 protein in plasma membrane fractions from cortical tissues of 2-, 4-, and 6-month-old Cln1^−/− mice compared with those from their WT littermates (Fig. 2D). We analyzed both lysosomal and plasma membrane fractions for purity as cross-contamination between lysosomal and plasma membrane fractions may occur. The results showed that there was only a trace amount of cross-contamination in both lysosomal (fig. S1C) and plasma membrane fractions (fig. S1D). We also analyzed cross-contamination from various organelles by using the ER marker, calreticulin (fig. S1E), nuclear marker, histone H3 (fig. S1F), mitochondrial marker, voltage-dependent anion channel (VDAC) (fig. S1G), and peroxisome-marker, PMP70 (fig. S1H). In addition, we confirmed the enrichment of lysosomal (fig. S1I) and plasma membrane (fig. S1J) fractions and compared them with those of total homogenates.

Fig. 2. Dynamic S-palmitoylation of NPC1 facilitates trafficking to the lysosomal membrane.

(A) Schematic explaining how under physiological conditions cholesterol homeostasis is maintained in WT mice. (B) Levels of NPC1 protein in lysosomes purified from cortical tissues of 2-, 4-, and 6-month-old WT and Cln1^−/− mice (n = 24). (C) Colocalization of NPC1 signal with that of lysosomal marker, LAMP2, in isolated neurons from cortical tissues of WT, Cln1^−/− mice and Cln1^−/− mice treated with r-PPT1 (n = 24). Scale bar, 10 μm. (D) Level of NPC1 in plasma membrane fractions of cortical tissues from 2-, 4-, and 6-month-old WT and Cln1^−/− mice (n = 4). (E) Confocal images showing colocalization of NPC1 with the plasma membrane marker, Na⁺,K⁺-ATPase, in cortical neurons from WT, Cln1^−/− mice and the neurons from Cln1^−/− mice treated with r-PPT1 (n = 24). Scale bar, 10 μm. Flow cytometry sorting of lysosomes (F) and plasma membrane (H) from cortical neurons of WT and Cln1^−/− mice using anti–LAMP1-FITC and anti–Na⁺,K⁺-ATPase–APC antibodies, respectively. The level of NPC1 in purified lysosomal fractions (G) and plasma membrane fractions (I) (n = 4) were then determined by Western blot analysis. Note that in Cln1^−/− mice, higher level of NPC1 is localized on the plasma membrane instead of its normal localization on lysosomal membrane. Two-sided permutation t test with complete enumeration was used to calculate the P values (*P < 0.5; Asterisks indicate a significant difference relative to corresponding control group). Data are presented as the mean ± SD. n denotes the number of independent biological replicates in each experiment. In the box plots, the center line is the median, + symbol representing the sample mean, limits are 25th and 75th percentile, and whiskers are the minimum and maximum values, respectively. Source data are provided as a source data file. r-PPT1, recombinant PPT1.

To confirm these results, we performed confocal imaging using cultured neurons from Cln1^−/− mice and those from their WT littermates. The results showed that in neurons from Cln1^−/− mice, a significantly higher level of NPC1 protein colocalized with the plasma membrane marker, Na⁺- and K⁺-dependent adenosine triphosphatase (Na⁺,K⁺-ATPase). Moreover, treatment of Cln1^−/− cells with r-PPT1 significantly reduced the localization of NPC1 signal with the plasma membrane marker, Na⁺,K⁺-ATPase (Fig. 2E). Furthermore, we transfected cultured neurons from Cln1^−/− mice and those from their WT littermates with Flag-tagged WT-NPC1 construct. We then used Flag antibody for NPC1 to colocalize with the lysosomal marker, LAMP2, and plasma membrane marker, Na⁺,K⁺-ATPase, respectively. Consistent with the previous results, we found a significantly reduced colocalization of Flag-tagged NPC1 with LAMP2 in Cln1^−/− cells (fig. S2A). In Cln1^−/− neurons transfected with WT NPC1-construct showed a significantly higher colocalization of NPC1 with the plasma membrane marker, Na⁺,K⁺-ATPase (fig. S2B).

Evaluation of NPC1 misrouting by flow cytometry and PLA

To confirm that NPC1 was misrouted in Cln1^−/− cells, we performed flow cytometric analysis and PLA (46). The results showed a significantly higher colocalization NPC1 PLA signal compared with that of the plasma membrane marker, Na⁺,K⁺-ATPase (fig. S2C). In Cln1^−/− neurons, a reduced PLA signal for colocalization of NPC1 and LAMP2 (fig. S2D) compared with that in WT neurons. We also performed sorting of lysosomes and plasma membrane fractions by flow cytometry using anti–LAMP1–fluorescein isothiocyanate (FITC) and anti–Na⁺,K⁺-ATPase–allophycocyanin (APC) antibodies, respectively. The sorted samples were then subjected to Western blot analysis. The results showed that the level of NPC1 was significantly lower in the sorted lysosomes from cortical neurons of Cln1^−/− mice compared with that of the WT littermates (Fig. 2, F and G). Similarly, the level of NPC1 was significantly higher in the sorted plasma membrane fractions from cortical neurons of Cln1^−/− mice compared with those from WT neurons (Fig. 2, H and I). These results strongly suggest that in Ppt1-deficient Cln1^−/− mice, NPC1 is mislocalized.

We also determined the levels of NPC2 in lysosomal and plasma membrane fractions from cortical tissues of WT and Cln1^−/− mice. Western blot analysis showed that the level of NPC2 protein was appreciably higher in lysosomal fractions of cortical tissues from Cln1^−/− mice (fig. S3A). The reason for this increase in NPC2 levels in lysosomal fraction from Cln1^−/− mice is unclear. However, the NPC2-protein levels in plasma membrane fractions from WT and Cln1^−/− mouse brain showed no significant differences (fig. S3B). Moreover, in isolated neurons from WT and Cln1^−/− mice, there was no substantial difference in the colocalization coefficient of NPC2 protein with the plasma membrane marker, Na⁺,K⁺-ATPase (fig. S3C). Furthermore, we analyzed the mRNA levels of both NPC1 and NPC2 in cortical tissues from WT and Cln1^−/− mice. The results showed that NPC1-mRNA levels were not significantly altered in cortical tissues from 2-, 4-, and 6-month-old WT and Cln1^−/− mice (fig. S3D). However, the mRNA level of NPC2 was significantly increased in cortical homogenates from 4- and 6-month-old Cln1^−/− mice compared with that of their WT littermates (fig. S3E). We also analyzed the stability of NPC1 protein by treating the normal and CLN1^−/− human embryonic kidney (HEK) 293T cells with various doses of proteasome inhibitor, MG132 (5, 10, and 20 μM). The results showed no significant differences in NPC1-protein levels between treated and untreated groups (fig. S3F). Cumulatively, these results showed that in Cln1^−/− mice, NPC1 protein, which mediates lysosomal cholesterol egress, was mistargeted instead of its normal location on the lysosomal membrane.

Loss-of-function mutations of the Cln1 gene cause misrouting of NPC1 protein

How might NPC1 protein in Cln1^−/− mice be misrouted away from the lysosome? Previously, we reported that a critical subunit of vATPase, V0a1, required dynamic S-palmitoylation for trafficking to the lysosomal membrane (47). We also reported that the loss-of-function mutations in the Cln1 gene misrouted V0a1 to the plasma membrane instead of its normal location on lysosomal membrane. Although PPT1 was originally reported to be localized exclusively in the lysosome (21), recent evidence indicates that this thioesterase is not exclusively localized in the lysosomal lumen as catalytically active PPT1 is detectable in the cytoplasm (22). Thus, we reasoned that if NPC1, like V0a1, is S-palmitoylated and require dynamic S-palmitoylation for endosomal trafficking to the lysosomal membrane, Ppt1 deficiency in Cln1^−/− mice may cause misrouting of NPC1 protein to other membranes like the plasma membrane. Accordingly, we first analyzed the peptide sequence of NPC1 protein by CSS-palm (48), a computer program that predicts the cysteine residue(s) in a polypeptide that is likely to undergo S-palmitoylation. The results of this analysis of mouse and human NPC1 peptide sequence predicted that Cys¹⁶, Cys⁹⁷, and Cys⁶⁴⁵ in mouse NPC1 and Cys¹⁶ and Cys⁹⁷, in human NPC1 are potential S-palmitoylation sites (fig. S4A). To confirm whether the predicted cysteine residue(s) in mouse are indeed S-palmitoylated, we transfected HEK293T cells with either WT or mutant (Cys¹⁶Ala, Cys⁹⁷Ala, and Cys⁶⁴⁵Ala) mouse NPC1-cDNA constructs. In the mutant constructs, the predicted S-palmitoylated Cys residues were mutated to Ala, as this amino acid does not undergo S-palmitoylation. To confirm the identity of the S-palmitoylated Cys residue(s) in NPC1, we performed Acyl-Rac assay (49) using WT and each of the mutant (Cys¹⁶Ala, Cys⁹⁷Ala, and Cys⁶⁴⁵Ala)–NPC1 proteins expressed in HEK293T cells. We found that only Cys⁹⁷Ala mutation of NPC1 abrogated S-palmitoylation (Fig. 3A). Notably, Cys⁹⁷ in NPC1 is evolutionarily conserved (fig. S4B). No S-palmitoylation site(s) was predicted for NPC2 protein by CSS-Palm. However, we performed Acyl-Rac assay of NPC2 generated by transfecting HEK293 cells and the results confirmed that NPC2 is not an S-palmitoylated protein (fig. S5A). The results of the Acyl-Rac assay using NPC2 protein confirmed that it does not undergo S-palmitoylation although it was predicted by CSS-Palm (fig. S4A). Together, these results suggested that while NPC1 is S-palmitoylated on Cys⁹⁷, NPC2 does not undergo S-palmitoylation.

Fig. 3. Ppt1 deficiency misroutes S-palmitoylated-NPC1 to the plasma membrane.

(A) Acyl-RAC assay using lysates from HEK293T cells transfected with WT-NPC1 and mutant (Cys⁹⁷Ala)–NPC1 constructs. Note that Cys⁹⁷Ala mutation abrogates S-palmitoylation of NPC1 (n = 4), confirming that Cys⁹⁷ is the S-palmitoylation site in this protein. (B) Various APs were pulled down by NPC1 antibody (n = 4) and immunoblotted using AP antibodies. (C) NPC1 pull-down assays (n = 4) were performed using various AP antibodies and immunoblotted using NPC1antibody. (D) Level of S-palmitoylated NPC1 in plasma membrane fractions of cortical tissues from WT and Cln1^−/− mice (n = 4) by Acyl-RAC assay. Colocalization of NPC1 with Rab5 (E), EEA1 (F), Rab9 (G), and Rab11 (H) in cortical neurons from WT and Cln1^−/− mice (n = 24). Scale bar, 10 μm. (I) Endosomal trafficking of NPC1. Note that in WT cells (left), trafficking of NPC1 protein is facilitated by dynamic S-palmitoylation on Cys⁹⁷, requiring thioesterase activity of Ppt1. Depalmitoylation of NPC1 allows handover of NPC1 bound to AP-2 to AP-3. The AP-3 bound NPC1 is then transported to the limiting membrane of late endosome/lysosome. In Cln1^−/− cells (right), the lack of Ppt1 activity impairs dynamic S-palmitoylation of NPC1. Consequently, S-palmitoylated NPC1 is transported via recycling endosome to the plasma membrane instead of its normal location on lysosome. The two-sided permutation t test with complete enumeration was used to calculate the P values (*P < 0.5; **P < 0.01. Asterisks indicate a significant difference relative to corresponding control group. NS, non-significant). Data are presented as mean values ± SD. n denoting the number of independent biological replicates. In the box plots, the center line is the median, + symbol representing the sample mean, limits are 25th and 75th percentile, and whiskers are the minimum and maximum values, respectively. Source data are provided as a source data file. PM, plasma membrane, PBS-C, PBS control, HA⁻, hydroxylamine unbound; HA⁺, hydroxylamine bound.

Ppt1 deficiency in Cln1^−/− mice impairs endosomal trafficking of NPC1 to the lysosome

Thus far, our results have revealed that in Cln1^−/− mice, Ppt1 deficiency causes misrouting of NPC1 away from its normal location on the lysosomal membrane. We therefore sought to determine whether endosomal trafficking of NPC1 protein to the late endosome/lysosome is dysregulated in Cln1^−/− mice. Recent reports indicate that the cytoplasmic tail of NPC1 interacts with the clathrin-adaptor protein, AP-1, containing an acidic dileucine motif, which facilitates its trafficking to the late endosomal/lysosomal compartment (50). The sorting of lysosomal membrane proteins from the ER to the late endosomal/lysosomal surface may occur either directly, or they are first transported to the plasma membrane followed by endocytosis to the early endosome by AP-2, and from there via AP-3 to the late endosomal/lysosomal-limiting membrane (51). In this scheme, NPC1 must undergo depalmitoylation by a thioesterase like Ppt1 to detach from AP-2 for its handoff from AP-2 to AP-3. The NPC1 bound to AP-3 is then transported to its destination on late endosomal/lysosomal membrane.

To evaluate the interaction of NPC1 with various adaptor proteins (i.e., AP-1, AP-2, and AP-3), we performed immunoprecipitation studies and PLA (46), respectively, using antibodies to NPC1 and to those of various APs. First, we performed pull-down assays of AP-1 using NPC1-antibody. The results showed that the binding of NPC1 with AP-1 in WT and Cln1^−/− mice was virtually identical (Fig. 3B). However, when we performed pull-down assays with AP-2 using NPC1 antibody, we found a significantly increased level of NPC1-AP-2 complex in brain tissues from Cln1^−/− mice compared with that from WT littermates (Fig. 3B). Most notably, pulldown of AP-3 with NPC1 antibody showed a significantly lower level of NPC1-AP-3 complex in Cln1^−/− mice compared with that in their WT littermates (Fig. 3B). We also obtained the same result when we pulled down NPC1 with antibodies to AP-1, AP-2, or AP-3 (Fig. 3C). To further confirm these findings, we performed PLA reaction (46) using antibodies to NPC1-, AP-2, and AP-3, respectively, using isolated cortical neurons from WT and Cln1^−/− mice. The results showed that compared with the neurons from WT mice, those from Cln1^−/− littermates showed a significantly higher interaction between NPC1 and AP-2 (fig. S5B) and a significantly less interaction between NPC1 and AP-3 (fig. S5C). These results confirmed that in Cln1^−/− mouse brain as well as in isolated cortical neurons from these mice, PPT1 deficiency prevented the handover of NPC1 from AP-2 to AP-3. Since the handover of NPC1 from AP-2 to AP-3 is essential for its trafficking to the late endosomal/lysosomal membrane, its lysosomal level was significantly lower in Cln1^−/− mice compared with that in their WT littermates.

NPC1 is S-palmitoylated on cysteine-97 for endosomal trafficking to the lysosome

S-palmitoylation has been reported to increase protein-protein interactions (17, 18). Moreover, dynamic S-palmitoylation (19), which requires the action of thioesterases, like the Ppt1, is suggested to facilitate endosomal trafficking of proteins that are S-palmitoylated (47). Thus, we sought to determine whether Ppt1 deficiency in Cln1^−/− mice impairing depalmitoylation of S-palmitoylated NPC1 might dysregulate the handover of NPC1 from AP-2 to AP-3, which is essential for its trafficking from early endosome to the late endosome/lysosome. Accordingly, we transfected HEK293T cells with cDNA constructs of Flag-tagged WT-NPC1 or mutant (Cys⁹⁷Ala)–NPC1 and performed immunoprecipitation assays using Flag antibody for NPC1 [since both WT-NPC1 and mutant (Cys⁹⁷Ala)–NPC1 constructs were Flag tagged], AP-2 and AP-3 antibodies. When we pulled down AP-2 and AP-3 proteins using Flag antibodies, we found that the binding of mutant (Cys⁹⁷Ala)–NPC1 to AP-2 was significantly lower (fig. S6A) compared with that of WT-NPC1 and AP-2. Most notably, the binding of AP-3 with mutant (Cys⁹⁷Ala)–NPC1 was also significantly lower compared with that of WT-NPC1 (fig. S6B). We also observed similar results when we performed pulldown of WT-NPC1 and mutant (Cys⁹⁷Ala)–NPC1 proteins by AP-2- (fig. S6C) and AP-3 antibodies, respectively (fig. S6D). To determine the level of S-palmitoylated NPC1 in the plasma membrane fractions, we performed Acyl-Rac assay. We found that a significantly higher level of S-palmitoylated NPC1 was associated with the plasma membrane fractions of cortical tissues from Cln1^−/− mice compared with that of their WT littermates (Fig. 3D). These results suggested that dynamic S-palmitoylation of NPC1 is essential for its binding to AP-2 and for its handover from AP-2 to AP-3. However, in Cln1^−/− cells, Ppt1 deficiency most likely impaired NPC1 depalmitoylation, which disrupted its dynamic S-palmitoylation, required for its handover from AP-2 to AP-3, essential for its trafficking from early endosome to the late endosome/lysosome. Consequently, in PPT1-deficient cells, NPC1 in early endosome entered the recycling endosome, and from there, it was routed to the plasma membrane.

NPC1 requires dynamic S-palmitoylation for endosomal trafficking to the lysosomal membrane

To delineate whether dynamic S-palmitoylation of NPC1 on Cys⁹⁷ is essential for its lysosomal targeting, we transfected WT cortical neurons with either Flag-tagged WT- or mutant (Cys⁹⁷Ala)–NPC1 constructs. The results of confocal microscopic analyses showed that while the WT-NPC1 readily colocalized with the lysosomal marker, LAMP2 (fig. S7A, top), such colocalization of mutant (Cys97Ala)–NPC1 was significantly lower (fig. S7A, bottom). Moreover, we have performed colocalization assays using WT-NPC1 as well as mutant (Cys⁹⁷Ala)–NPC1 with the ER marker, calreticulin, and plasma membrane marker, Na⁺,K⁺-ATPase. The results showed that there was a significantly higher colocalization of mutant (Cys⁹⁷Ala)–NPC1 with calreticulin compared with WT-NPC1 (fig. S7B). These results suggested that dynamic S-palmitoylation facilitates the forward transport of NPC1 from the ER toward the late endosome/lysosome. However, we could not find any significant colocalization of WT- or mutant (Cys97Ala)–NPC1 with the plasma membrane marker, Na⁺,K⁺-ATPase (fig. S7C). Furthermore, we found a significantly reduced level of colocalization of mutant (Cys⁹⁷A)–NPC1 with early endosomal marker, EEA1 (fig. S7D) compared with that of WT-NPC1. Cumulatively, these results strongly suggested that dynamic S-palmitoylation of Cys⁹⁷ in NPC1 is critical for its endosomal trafficking to the lysosomal membrane.

To further confirm that Ppt1 deficiency in Cln1^−/− mice impaired trafficking of NPC1 to the late endosome/lysosome, we performed confocal imaging using isolated neurons from WT and Cln1^−/− mice to determine the colocalization of NPC1 with various endocytic markers. The results showed a significantly increased colocalization of NPC1 with the sorting endosomal marker, Rab5, in cultured neurons from Cln1^−/− mice compared with those from their WT littermates (Fig. 3E). The colocalization of NPC1 with early endosomal marker, EEA1, was also significantly increased in neurons from Cln1^−/− mice (Fig. 3F). However, the colocalization of NPC1 with late endosomal marker, Rab9, was substantially less pronounced in neurons from Cln1^−/− mice (Fig. 3G). In isolated cortical neurons from Cln1^−/− mice, colocalization of recycling endosomal marker, Rab11, with NPC1 was significantly higher compared with that from their WT littermates (Fig. 3H). These results suggested that the increased interaction of S-palmitoylated-NPC1 with AP-2 and failure of AP-2 bound NPC1 to be handed over to AP-3 in Cln1^−/− mice promoted its trafficking to Rab5- and Rab11-positive recycling endosomes. From here, NPC1 was then recycled back to the plasma membrane. These results suggested that in WT mice, Ppt1 most likely facilitated depalmitoylation of S-palmitoylated NPC1, which detached it from AP-2 for its handover to AP-3 (Fig. 3I, left). However, in Cln1^−/− mice, Ppt1 deficiency blocked the handover of NPC1 bound to AP-2 to AP-3, thereby preventing its trafficking to the lysosomal-limiting membrane (Fig. 3I, right). Together, these results suggest that in Cln1^−/− mice, Ppt1 deficiency dysregulated endocytic trafficking of NPC1 protein misrouted it from reaching its normal location on lysosomal membrane.

Increased lysosomal OSBP level promotes cholesterol-mediated mTORC1 activation in Cln1^−/− mice

It has been reported that an important class of cholesterol carriers, OSBP, establishes a high cholesterol pool on lysosomal-limiting membrane (37). Intriguingly, in NPC1-null cells increased cholesterol on lysosomal membrane plays a critical role in promoting cholesterol-mediated mTORC1 activation (36, 37). Moreover, it has been reported that the transport of cholesterol from the ER to OSBP on lysosomal membrane is mediated via the VAPA and VAPB (36, 37). These proteins are anchored to the ER membrane and function as conduits for the transfer of cholesterol from the ER to OSBP localized on lysosomal-limiting membrane (Fig. 4A). Therefore, we first determined the levels of OSBP in total cortical homogenates as well as in lysosomal fractions from WT and Cln1^−/− mice by Western blot analysis. We found that the OSBP-protein levels in total cortical lysates from Cln1^−/− mice and those from their WT littermates were virtually identical (fig. S8A). However, those levels in purified lysosomal fractions from Cln1^−/− mice were significantly higher compared with those from their WT littermates (Fig. 4B). Next, we sought to determine the levels of VAPA and VAPB as these proteins, anchored to the ER membrane, which transport cholesterol to OSBP on lysosomal membrane (37). Our results showed that the levels of both VAPA (fig. S8B) and VAPB (fig. S8C) were significantly higher in total cortical lysates from Cln1^−/− mice compared with those from their WT littermates. To further confirm these results, we performed confocal imaging, using isolated cortical neurons from Cln1^−/− mice, which showed a significantly higher level of colocalization of OSBP with that of the lysosomal marker, LAMP2. Treatment of the cells with r-PPT1 significantly reduced the colocalization of OSBP on lysosome (Fig. 4C). Moreover, the lysosomal fractions from cortical tissues of Cln1^−/− mice contained significantly higher levels of VAPA and VAPB. Confocal imaging further confirmed the colocalization of both VAPA and VAPB with the lysosomal marker, LAMP2. Notably, treatment of the cells with r-PPT1 substantially reduced the colocalization of VAPA (Fig. 4D) and VAPB (Fig. 4E) on lysosomal membrane.

Fig. 4. Dysregulation of cholesterol homeostasis in Cln1^−/− mice.

(A) Schematic representation of cholesterol transport from the ER to the lysosome; (B) the levels of OSBP, VAPA, and VAPB in lysosomal fractions from cortical tissues of WT and Cln1^−/− mice (n = 4). (C) Colocalization of OSBP with lysosomal marker, LAMP2, in cortical neurons from WT, Cln1^−/− mice, and Cln1^−/− mice treated with r-PPT1 (n = 24). Scale bar, 10 μm. (D) Colocalization of VAPA with lysosomal marker, LAMP2, in cortical neurons from WT, Cln1^−/− mice, and Cln1^−/− mice treated with r-PPT1 (n = 24). Scale bar, 10 μm. (E) Colocalization of VAPB with lysosomal marker, LAMP2, in cortical neurons from WT, Cln1^−/− mice, and Cln1^−/− mice treated with r-PPT1 (n = 24). Scale bar, 10 μm. (F) TEM analysis of cortical tissues to determine the ER-lysosome contacts in cortical cells from WT and Cln1^−/− mice (n = 3). Scale bar, 0.5 μm. (G) TEM analysis of cortical tissues to determine ER-lysosome contacts in WT and Cln1^−/− mice. Note that intracellular accumulation of GRODs, characteristically found in Cln1^−/− mice, interferes with clear visualization of ER-lysosome contacts. Despite this disadvantage, the ER-lysosome contacts in Cln1^−/− mice appear to be tighter than those in their WT littermates (n = 3). Scale bar, 5 μm. Two-sided permutation t test with complete enumeration was used to calculate the P values (*P < 0.5; **P < 0.01. Asterisks indicate a significant difference relative to corresponding control group). Data are presented as mean values ± SD. n denoting the number of independent biological replicates. In the box plots, the center line is the median, + symbol representing the sample mean, limits are 25th and 75th percentile, whiskers are the minimum and maximum values, respectively. Source data are provided in the source data file. L, Lysosomes; G, GRODS.

Recently, it has been reported that NPC1 plays an important role in ER contacts with endocytic organelles to mediate direct cholesterol transport from the lysosome to the ER (52). To evaluate the ER-lysosome contacts, we performed transmission electron microscopic (TEM) analysis of cortical tissues from WT and Cln1^−/− mice. The results showed that there was an increased ER-lysosome contact in cortical tissues from Cln1^−/− mice compared with those from their WT littermates (Fig. 4F). Because the Ppt1-deficient cells in Cln1^−/− mice accumulate the electron-dense granular osmiophilic deposits (GRODS), it is difficult to clearly visualize ER-lysosome contacts (Fig. 4G). Therefore, we performed confocal microscopic analyses and PLA reaction (46) using antibodies to the ER marker, calreticulin, and lysosomal marker, LAMP2. The PLA reaction yields positive signals when two proteins are within 10 to 30 nm apart. Our results showed a significantly increased colocalization of calreticulin with LAMP2 in cortical neurons from Cln1^−/− mice compared with those from their WT littermates (fig. S8D). Similarly, the results of this test showed increased number of PLA signals in cortical neurons from Cln1^−/− mice compared with those from their WT littermates (fig. S8E), suggesting that in Cln1^−/− mice ER-lysosome contacts are significantly tighter than those in their WT littermates. Together, these results show that in the brain of Cln1^−/− mice, misrouting of NPC1 to the plasma membrane in conjunction with elevated OSBP level on lysosomal membrane promoted cholesterol-mediated hyperactivation of mTORC1 signaling.

Pharmacological inhibition of OSBP suppresses mTORC1 activation and rescues autophagy

Elevated level of lysosomal OSBP has been reported to cause cholesterol-mediated mTORC1 activation in NPC1-null cells (36, 37). Therefore, we sought to determine whether pharmacologically inhibiting OSBP in cultured cells from human CLN1 disease and in Cln1^−/− mice may suppress mTORC1 activation and rescue autophagy. Accordingly, we first treated cultured lymphoblasts from patients with CLN1 disease with a pharmacological inhibitor of OSBP, OSW1 (53), and evaluated the levels of pS6K1 and p4E-BP1. The results showed that the levels of pS6K1 and p4E-BP1 in OSW1-treated cells from a patient with CLN1 disease were significantly lower compared with those in their untreated counterparts (Fig. 5A). To test whether these in vitro effects are replicable in vivo, we treated Cln1^−/− mice with OSW1 (10 μg/kg body weight) by gavage feeding every day for 2 weeks. The results showed a significantly lower levels of pS6K1 and p4E-BP1 in OSW1-treated mice compared with those in the untreated controls (Fig. 5B). It has been reported that dysregulated mTORC1 activation in a neuronal model of NPC disease suppresses autophagy, which contributes to neurodegeneration (40–42). Since our results showed that OSW1 treatment significantly suppressed the activation of mTORC1, we determined whether this treatment rescued autophagy. The results showed that the level of autophagosome marker, LC3-II, and the marker of autophagic flux, p62, were significantly decreased in OSW1-treated cells from human patient with CLN1 disease (Fig. 5C) compared with those in untreated control cells. Similarly, we observed a significant decrease in the levels of both LC3-II and p62 levels in OSW1-treated Cln1^−/− mice compared with those in untreated controls (Fig. 5D).

Fig. 5. Pharmacological inhibition of OSBP suppresses mTORC1 activation rescuing autophagy.

Western blot analyses: (A) pS6K1 and p4E-BP1 in normal (1), untreated human CLN1-disease lymphoblasts (2), and those treated with OSBP-inhibitor, OSW1 (5 nM) (n = 4) (3). (B) Quantitation of pS6K1 and p4E-BP1 levels in WT (1), untreated Cln1 ^−/− mice (2) and those treated with OSW1 (10 μg/kg body weight), (n = 5) (3). (C) Levels of LC3-II and p62 in normal (1), untreated human CLN1 disease-lymphoblasts (2), and those treated with OSBP-inhibitor, OSW1 (5 nM)] (n = 4) (3). (D) Quantitation of LC3-II and p62 in the cortical tissue of WT (1), Cln1^−/− mice (n = 4) untreated (2) or treated with OSBP inhibitor, OSW1 (10 μg/kg body weight), (n = 4) (3). (E) Level of CD68 and GFAP in cortical tissues from WT (1), untreated-Cln1^−/− mice (2), and from those treated with OSW1 (10 μg/kg body weight) (n = 4) (3). Immunostaining of mTORC1-activation markers, (F) pS6K1 and (G) p4E-BP1 in cortical tissues from OSW1-treated Cln1^−/− mice compared with those from untreated mice (n = 12). Scale bar, 50 μm. Two-sided permutation t test with complete enumeration was used to calculate the P values (*P < 0.5. Asterisks indicate a significant difference between treated and untreated control group). Data are presented as mean values ± SD. n denotes the number of independent biological replicates for each experiment. In the box plots, the center line is the median, + symbol representing the sample mean, limits are 25th and 75th percentile, whiskers are the minimum and maximum values, respectively. Source data are provided in source data file.

It has been demonstrated that treatment of NPC1-null cells from patients with NPC with cyclodextrin significantly decreased intracellular cholesterol levels (54). Accordingly, we treated cultured lymphoblasts from patients with CLN1 disease and determined the levels of pS6K1 and p4E-BP1. The results showed that cyclodextrin treatment does not significantly reduce the levels of pS6K1 and p4E-BP1. Since dysregulation of cholesterol occurred in PPT1-deficient cells, we treated these cells with rPPT1 as a positive control. It has been demonstrated that cultured cells from patients with LSD can take up the enzyme that is deficient in such cells from conditioned culture media via endocytosis and correct the genetic defect (55). Therefore, we treated the cultured cells from patients with CLN1 disease with rPPT1 as a control. Our results showed that while treatment of the cells with cyclodextrin did not significantly reduce the levels pS6K1 (fig. S9A) and p4E-BP1 (fig. S9B), these canonical markers of mTORC1 activation were substantially lower in cells that were treated with rPPT1. Moreover, the cyclodextrin treatment did not significantly alter the colocalization of OSBP on the lysosome in cultured neurons from cortical tissues of Cln1^−/− mice compared with those from their WT littermates (fig. S9C). These results suggested that cyclodextrin treatment is ineffective in suppressing cholesterol-mediated mTORC1 activation in cells from patients with CLN1 disease.

Levels of TPC2 mRNA and TPC2 protein in the brain of WT and Cln1^−/− mice

It has been reported that Ca⁺⁺ permeable two-pore channel 2 (TPC2) ameliorates the cellular phenotypes associated with cholesterol and lipofuscin in LSDs (56). It has been shown that the activation of TPC2 in neurons derived from iPSC model of CLN3 disease ameliorated the disease phenotype. To determine whether TPC2 is involved in the pathogenesis of CLN1 disease, we determined the TPC2-mRNA and TPC2-protein levels in cortical tissues from WT and Cln1^−/− mice and in cultured fibroblasts from control and patients with CLN1 disease. We found that there is no significant difference in the levels of TPC2 mRNA and TPC2 protein between WT and Cln1^−/− mice as well as in fibroblasts from control and patients with CLN1 disease (fig. S10, A to D). Moreover, we treated cultured fibroblasts from a patient with CLN1 disease with TPC2 agonist, TPC2-A1-P, and determined the lysosomal cholesterol levels. The results showed that there was no reduction in lysosomal cholesterol levels in CLN1 disease fibroblasts (fig. S10E). In addition, we checked the levels of the canonical markers of mTOR activation, pS6K1 and p4E-BP1, in untreated and TPC2-agonist–treated fibroblasts from a patient with CLN1 disease. The results showed that the level of pS6K1 and p4E-BP1 were not significantly altered in the TPC2-agonist–treated fibroblasts of a patient with CLN1 compared with that in untreated control fibroblasts (fig. S10F). Together these results suggest that the inactivity of TPC2 may not be one of the causes of high lysosomal cholesterol mediating mTORC1 activation in Cln1^−/− mice and in fibroblasts from a patient with CLN1 disease.

Treatment of Cln1^−/− mice with an OSBP inhibitor, OSW1, ameliorates neuroinflammation

In common neurodegenerative diseases like Alzheimer’s and Parkinson’s, neuroinflammation contributes to neurodegeneration (57). Moreover, we (58) and others (59) have previously reported that in Cln1^−/− mice, the activation of innate immune cells of myeloid origin in the brain (e.g., microglia) contributes to neuroinflammation. Further, NPC1-mutations dysregulating cholesterol homeostasis have been reported to cause early-onset neuroinflammation, which contributes to neurodegeneration in NPC disease (60). Thus, we measured the levels of microglial marker, CD68, and astrocyte marker, glial fibrillary acidic protein (GFAP), in untreated and OSW1-treated Cln1^−/− mice. The results showed significantly decreased levels of both CD68, and GFAP, in OSW1-treated Cln1^−/− mice (Fig. 5E). To further confirm these results, we performed immunostaining of brain sections from OSW1-treated mice, which showed significantly decreased level of CD68 (fig. S11A), GFAP (fig. S11B), and mTORC1-activation markers, pS6K1 (Fig. 5F) and p4E-BP1 (Fig. 5G), compared with those in untreated mice. In addition, we evaluated whether OSW1 treatment ameliorated brain atrophy in Cln1^−/− mice by quantifying the number of neurons in specified areas by using Nissl staining. Nissl-positive cells were counted from four comparable sections from each group in an identical area of all cortical tissue sections. Moreover, we estimated the cortical thickness after hematoxylin and eosin (H&E) staining of the sections from the WT, untreated- and OSW1-treated-Cln1^−/− mice. Briefly, sagittal sections (5 μm thick) from 6-month-old WT, untreated- and OSWI-treated Cln1^−/− mice were deparaffinized, hydrated, and stained with H&E staining kit (ab245880, Abcam) using supplier’s protocol. At least four comparable sections from each mouse were used to measure the cortical thickness and a similar area from all the sections of whole brains. The results showed that compared with WT mice, in untreated Cln1^−/− mice, there was a significantly lower number of neurons (fig. S11C) and reduced cortical thickness (fig. S11D). OSW1 treatment of these mice showed a substantially increased number of neurons (fig. S11C) and higher cortical thickness (fig. S11D) compared with those in the untreated Cln1^−/− controls. Moreover, we found that OSW1 treatment also suppressed mTORC1 activation and rescued autophagy in cultured cells from a patient with CLN1 disease as well as in neurons from Cln1^−/− mice. Together, these results suggest that inhibition of cholesterol-mediated mTORC1 activation may be a targetable pathway for CLN1 disease.

DISCUSSION

In this study, we show that increased lysosomal cholesterol mediates hyperactivation of mTORC1 in Cln1^−/− mice (33), a reliable animal model of human CLN1 disease (34). We also demonstrate that NPC1, which mediates cholesterol egress from the lysosomal lumen, requires dynamic S-palmitoylation (palmitoylation-depalmitoylation) on Cys⁹⁷ for its endosomal trafficking to the late endosomal/lysosomal membrane. Intriguingly, the loss of Cln1/Ppt1 function misroutes NPC1 to the plasma membrane instead of its normal location on lysosomal-limiting membrane. Notably, the level of NPC1 protein in lysosomal fractions from Ppt1-deficient Cln1^−/− mice was significantly lower compared with that in their WT littermates. Moreover, the level of S-palmitoylated NPC1 was substantially higher in the plasma membrane of Cln1-null cells. These results suggest that enzymatically active, extra lysosomal PPT1 (22) may catalyze depalmitoylation of S-palmitoylated NPC1, essential for its endosomal trafficking to the lysosomal membrane. Notably, NPC2, which binds cholesterol and transfers it to NPC1 for egress, does not require S-palmitoylation for trafficking to the lysosome. Thus, in Cln1^−/− mice impaired lysosomal cholesterol egress by NPC1 and input of cholesterol by NPC2 results in dysregulated lysosomal cholesterol homeostasis.

It has been reported that VAPA and VAPB anchor OSBP to the ER, facilitating delivery of cholesterol across the ER-lysosome contacts and promote mTORC1 activation (36). Moreover, NPC1 has been reported to tether ER-endocytic organelles like the lysosome to membrane contact sites regulating cholesterol egress (61). Consistent with this finding, the results of TEM analysis of cortical tissues from WT and Cln1^−/− mice showed that ER-lysosome contacts are substantially tighter in Cln1^−/− mouse brain compared with those in their WT littermates. Moreover, the deficiency of Cln1/Ppt1 causes intracellular accumulation of S-palmitoylated proteins (constituents of ceroid lipofuscin), which becomes consolidated to generate electron-dense GRODS (7). As stated earlier, the presence of these GRODS makes it difficult to unequivocally determine the ER-lysosome contacts in Cln1^−/− cells. To circumvent this problem, we performed PLA reaction (46) using antibodies to calreticulin, an ER marker, and LAMP2, a lysosomal marker. The results confirmed that ER-lysosome contacts are tighter in Cln1^−/− brain than those in their WT littermates. Our results also showed that in Cln1^−/− mice, the significantly higher levels of OSBP, VAPA, and VAPB, most likely caused cholesterol-mediated hyperactivation of mTORC1. However, we do not have a clear understanding of the linkage between the trafficking defect of NPC1 and the increased OSBP level in Cln1^−/− mice. We propose that aberrant lysosomal cholesterol–mediated hyperactivation of mTORC1 signaling, which suppresses autophagy, at least in part, contributes to neuropathology in this mouse model of CLN1 disease.

Although the implications of S-palmitoylation on Cys⁹⁷ in NPC1 remain unclear, structural studies have revealed that 15 disulfide bonds are present in NPC1 protein (62). Among these, Cys⁹⁷ was found to form a disulfide bond with Cys²³⁸. Disulfide bonds like this have been suggested to stabilize the Psi-loop in NPC1 (63). Moreover, these disulfide bonds promote binding of the C-terminal and N-terminal domains, and this interaction has been shown to be important for cholesterol transport (62). Notably, in a recent unbiased proteomic study using synaptosomes from Cln1^−/− mice, at least 100 proteins have been identified as putative substrates of Ppt1 (20). Moreover, this study also suggested that depalmitoylation plays a role in disulfide bond formation. Furthermore, it has been reported that introduction of a single disulfide bond constraining luminal/extracellular domains or shortening the cytoplasmic loop may abolish the cholesterol egress activity of NPC1 (63). Thus, dynamic S-palmitoylation of NPC1 is not only essential for its endosomal trafficking to the lysosomal membrane but also may be important for interdomain dynamics in this protein. Further investigations may reveal whether mutation in Cys⁹⁷ in NPC1 has any deleterious effects. Whether the loss-of-function mutations in the CLN1 gene interferes with the formation of these disulfide bonds and promotes the misrouting of NPC1 protein to the plasma membrane remains to be determined.

Lysosomes are known to perform diverse cellular functions (38). Although it has been demonstrated that NPC1 mediates lysosomal cholesterol egress to maintain homeostasis (36, 37), the precise mechanisms of endosomal trafficking of NPC1 and NPC2 proteins to the lysosome, until now, had remained elusive. How might Cln1/Ppt1 deficiency impair dynamic S-palmitoylation of NPC1, which is localized on lysosomal-limiting membrane? Previously, it was reported that enzymatically active Ppt1 is present within the lysosomal lumen (21). However, recently, enzymatically active Ppt1 has been found in the cytoplasm (22). Intriguingly, the loss of Cln1/Ppt1 in Cln1^−/− cells prevented the handoff of NPC1 from AP-2 to AP-3 during endosomal trafficking causing NPC1 to be recycled back to the plasma membrane disrupting its normal localization on lysosomal membrane. We previously reported that the lysosomal acidification in Cln1^−/− mice and in cultured cells from patients with CLN1 disease is impaired, causing the pH of lysosomal lumen to significantly increase compared with that in their WT counterparts (47). Consonant with this finding, recent structural characterizations by cryo–electron microscopy revealed a critical requirement of low pH for the handover of cholesterol from NPC2 to the transmembrane domain of NPC1 (63). These defects impair lysosomal cholesterol egress in PPT1-deficient neurons and other cell types in CLN1 disease causing cholesterol-mediated mTORC1 hyperactivation, which suppresses autophagy. It has been reported that mTOR is at the nexus of nutrition, growth, ageing, and disease (64). It has been clearly demonstrated that in mice, loss of autophagy by activated mTORC1 in the central nervous system causes neurodegeneration (65).

The pathogenic mechanism(s) of most neurodegenerative diseases is highly complex. This is partly because only a handful of these diseases can be traced to monogenic inheritance. Intriguingly, the pathogenic mechanism of NPC disease and that of the CLN1 disease bears some apparent similarities. For example, the loss-of-function mutations in the NPC1 or NPC2 genes cause NPC disease. Similarly, our results show that NPC1 mislocalized to the plasma membrane in Cln1^−/− mice instead of its normal location on lysosomal membrane. Moreover, like Cln1/Ppt1-deficient cells, NPC1-null cells in patients with NPC disease have been reported to manifest increased lysosomal cholesterol accumulation causing an elevated level of OSBP on lysosomal-limiting membrane, which promotes lysosomal cholesterol–mediated mTORC1 activation (36). Furthermore, neuroinflammation contributes to neurodegeneration in both CLN1 disease model (58) and NPC disease (60). Notably, both NPC disease and CLN1 disease appear to impair NPC1 gene function and contribute to pathogenesis. Similarly, in CLN1 disease–mistargeted NPC1 also disrupts its function of lysosomal cholesterol egress impairing cholesterol homeostasis. Moreover, hyperactivation of mTORC1, which suppresses autophagy, may be a major contributor of neuropathology in most LSDs like the CLN1 disease. Our results show that treatment of Cln1^−/− mice with a pharmacological inhibitor of OSBP, OSW1, suppressed mTORC1 activation, restored autophagy, reduced neuroinflammation, and ameliorated neuropathology. Cumulatively, the results of our study reveal a previously unrecognized role of CLN1/PPT1 in regulating lysosomal cholesterol homeostasis and suggest that pharmacological inhibition of cholesterol-mediated mTORC1 hyperactivation is a targetable pathway for patients with CLN1 disease.

Complex mechanisms underlie pathogenesis of most neurodegenerative diseases. Thus, it is expected that the pathogenic mechanism of only a handful of neurodegenerative diseases with monogenic inheritance have been studied in detail. Unraveling the mechanism of pathogenesis in rare, monogenic neurodegenerative LSDs, like the CLN1 disease, may provide a window to uncover the pathogenic mechanism(s) of more common neurodegenerative diseases. Despite the discovery that loss of CLN1/PPT1 causes CLN1 disease, the molecular mechanism of pathogenesis has remained elusive for more than two decades. The significance of the results of our present study is the finding that NPC1 protein requires dynamic S-palmitoylation for its endosomal trafficking to the lysosomal membrane. Another more significant finding in this study is that PPT1 deficiency impaired dynamic S-palmitoylation of NPC1, causing dysregulation of lysosomal cholesterol homeostasis. This defect along with elevated OSBP promoted cholesterol-mediated hyperactivation of mTORC1 kinase suppressing autophagy, which contributed to neuropathology in our CLN1 disease model. Our results provide a direct link between inactivation of CLN1/PPT1 and hyperactivation of mTORC1 leading to pathogenesis of CLN1 disease. Further investigations may culminate in the development of previously unknown therapeutic strategies for this devastating neurodegenerative lysosomal storage disease. Recently, it has been reported that TPC2 rescues elevated lysosomal cholesterol and mTOR activation in mucolipidosis type IV, CLN3 disease, and NPC1 (56). However, in Cln1^−/− mice and in fibroblasts from a patient with CLN1 disease, TPC2-mRNA and TPC2-protein levels did not seem to be altered. Moreover, treatment of human CLN1 disease fibroblasts with a TPC2 agonist, TPC2-A1-P (66) did not reduce lysosomal cholesterol levels. These results may suggest that TPC2 inactivation may not be involved in the pathogenesis of CLN1 disease.

Complex mechanisms underlie pathogenesis of most neurodegenerative diseases including neurodegenerative LSDs. While genetic links for common neurodegenerative diseases like Alzheimer’s and Parkinson’s are often difficult to establish, neurodegenerative LSDs are monogenic. Thus, delineating the pathogenic mechanism(s) of rare neurodegenerative LSDs, like the CLNs, may provide a window for understanding the pathogenic mechanism(s) of more common neurodegenerative disorders. As we continue to learn more about the physiological functions of the mutant genes underlying pathogenesis of neurodegenerative LSDs, the development of mechanism-based treatment may be possible.

Potential limitations of the study

The current study has some potential limitations. First, we used a mouse model of CLN1 disease (33) and cultured fibroblasts from patients with CLN1 disease. Although Cln1^−/− mice are a reliable animal model of CLN1 disease (34), the results from animal models are not always replicable in humans. The patient fibroblasts are also not true representative of the brain cells although both the brain and the fibroblasts are of ectodermal origin and CLN1/PPT1 is expressed in all tissues and neuron-like cells (67). Second, the isolation of pure lysosome and plasma membrane fractions from brain tissues can be difficult. Although we have used additional measures like sorting of lysosome and the plasma membrane by flow cytometry, these measures can sometimes be influenced by uncontrollable sorting-related issues, such as the length of time, damage due to harsh isolation conditions, and random variation during sorting (68). Nevertheless, the flow cytometric purification of the lysosomes and the plasma membranes seem to have enhanced the purity of these cellular components.

MATERIALS AND METHODS

Chemicals

Phosphate-buffered saline (PBS, catalog no. 10010023) and protease and phosphatase inhibitor cocktail (catalog no. 78447) were purchased from Thermo Fisher Scientific, Rockford, IL. β-Mercaptoethanol (catalog no. M6250), Na₂CO₃, hydroxylamine, N-ethylmaleimide, β cyclodextrin, acetone, methanol, and NaHCO₃ were from Sigma-Aldrich, St. Louis, MO. Paraformaldehyde (catalog no. 15710) was purchased from Electron Microscopy Sciences, Hatfield, PA. Proteasome inhibitor, MG132 was purchased from MilliporeSigma (catalog no. 474790).

Animals and treatments

All procedures were carried out under an animal protocol (ASC#19-012) approved by the Animal Care and Use Committee of the Eunice Kennedy-Shriver National Institute of Child Health and Human Development (NICHD), National Institutes of Health (NIH). The Ppt1^−/− mice were generated by targeted disruption of the last exon in the Cln1 gene encoding Ppt1 in embryonic stem cells of 129 mouse strain as previously reported (33). These gene-targeted mice were subsequently backcrossed for 10 generations with WT C57BL/6J mice to obtain congenic C57 background. A breeding pair of these mice was provided to us by M. S. Sands (Department of Medicine, Washington University School of Medicine, St Louis, MO) to start our Ppt1^−/− mouse colony. Both male and female mice were used in this study. Mice were housed and maintained in a pathogen-free facility under a 12-hour light and 12-hour dark cycle with room temperature of 65° to 75°F (~18° to 23°C) and 60 to 70% humidity. The animals were provided with food and water ad libitum. Normal diet (catalog no. LABDIET NIH-31-3006898-203, PMI Nutrition International, Arden Hills, MN, USA) was always provided. For OSW1 treatment, we used approximately five and a half–month-old Cln1^−/− mice treated with OSW1 (10 μg/kg body weight) prepared in saline every day for 2 weeks by gavage feeding. The animals were euthanized at 6 months of age for evaluating the biochemical and histological parameters.

Plasmids and transfection of HEK293T cells

FLAG-tagged NPC1 (WT) plasmid (catalog no. Ex-Mm30039-M13) was purchased from GeneCopoeia, Rockville, MD. NPC1 mutant constructs were generated by Bioinnovatise (Rockville, MD). WT HEK293T cells were purchased from American Type Culture Collection (ATCC, catalog no. CRL-11268, Lot# 62312975) and transfected with the plasmids using Lipofectamine 3000 (catalog no. L3000015, Invitrogen) as per the manufacturer’s protocol.

Primary neuron culture and treatment

Primary neurons from cortical tissues were isolated using neural dissociation kit (130-094-802, Miltenyl Biotec) and neuron isolation kit, mouse (130-115-389, Miltenyl Biotec) from P2 pups using the standard manufacturer’s protocol with minor modifications. These neurons were then cultured in chamber slides using neurobasal media containing B27 supplement and glutamine. For determining the effects whether β-cyclodextrin or r-PPT1 (Creative Biomart, catalog no. PPT1-367H), we treated the neurons for 48 hours with 10 mM of β-cyclodextrin or of r-PPT1 protein (5 μg/ml) and dimethyl sulfoxide used as a control. The doses were selected by a dose-dependent study and cell viability assay. Before harvesting, the cells were washed twice and replaced with serum-free media containing lipid-free bovine serum albumin (BSA) and inhibitors. The duration of this treatment was for 3 hours to deplete all PPT1 activity in the media.

Lymphoblast culture and treatment

Immortalized lymphoblasts from a healthy subject (C9955) and those from a patient with CLN1 disease (C11796) carrying nonsense mutations (R151X) in the CLN1 gene were a gift from the late K. E. Wisniewski. These cells were maintained with RPMI 1640 containing 15% fetal bovine serum (FBS) and penicillin-streptomycin. The lymphoblasts were treated with OSW1 (5 nM) for 48 hours. For β-cyclodextrin or r-PPT1 treatment, we used β-cyclodextrin (10 mM) or r-PPT1 (5 μg/ml) for 48 hours. The doses were selected by a dose-dependent study and cell viability assay. Before harvesting, the cells were washed twice with serum-free RPMI 1640, and the media were replaced with RPMI 1640 containing lipid-free BSA and inhibitors. The treatment continued for 3 hours to deplete any PPT1 activity in the media.

Culture and treatment of human CLN1 disease fibroblasts

Normal fibroblasts (GM00498) were obtained from Coriell Institute of Medical Research, Camden, New Jersey. PPT1-deficient, human CLN1 disease fibroblasts were derived from skin biopsy of a patient with CLN1 admitted to a “bench-to-bedside” clinical protocol (#01-CH-0086) approved by the Institutional Review Board of NICHD, NIH. Normal fibroblasts and fibroblasts from a patient with CLN1 were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Thermo Fisher Scientific, catalog no. 11965-092) with 10% FBS and 1% penicillin-streptomycin (Thermo Fisher Scientific, catalog no. 15140-122). The cells were maintained at 37°C in a humidified 5% CO₂ atmosphere. The fibroblasts were treated with TPC2 agonist, TPC2-A1-P (10 μM) (SML-3700, Sigma-Aldrich) for 48 hours. Before harvesting, the cells were washed twice with serum-free DMEM, and the media were replaced with DMEM containing lipid-free BSA and agonist. The treatment continued for 3 hours to deplete any PPT1 activity in the media.

Total RNA isolation and RT-PCR

Total RNA was isolated from cortical tissues of WT and Cln1^−/− mice and fibroblasts using the RNA-easy Mini Kit (QIAGEN) followed by cDNA synthesis using the Superscript III First-Strand Synthesis kit (Invitrogen) according to the manufacturer’s instructions. The primers sets used for NPC1-mRNA are: Forward-5′ AACCGTGACACTGCAGGACAT-3′ and Reverse-5′-CTCATAATGGTGCAGTTCTT-G TTG-3′, for NPC2-mRNA are: Forward-5′-CATTGTCCCCCGAGATAGCC-3′ and Reverse-5′-CCATCCTGT CTGGTGGAACC-3′, human specific TPC2 are: Forward-5′-CGTTGTCCT GCTGGT TTTGG-3′ and Reverse-5′-GATGATACGCAGGAAGCGGA-3′, and mouse-specific TPC2 are: Forward-5′- CCCCTG AGTTAGTTGGGGTG-3′ and Reverse-5′- GGGTTTTGTTCAGCTGCGTT-3′. The ΔCt values were calculated using glyceraldehyde-3-phosphate dehydrogenase as control. The levels of mRNA expression were quantified and compared by real-time reverse transcription polymerase chain reaction (PCR) with SYBR Green PCR mix using ABI Prism 7000 Sequence detection system and analyzed by using the ABI Prism Software Version 1.01 (Applied Biosystems).

Preparation of cortical homogenate for phosphoprotein analysis

For the preparation of cortical tissue homogenates, the mouse cerebral cortex was homogenized using radioimmunoprecipitation assay (RIPA) buffer containing 1× Halt protease and phosphatase inhibitors (Thermo Fisher Scientific, catalog no.78447) in Dounce homogenizer. To inhibit the protease and phosphatase activities, protease and phosphatase inhibitors were present throughout the procedure. The homogenates were centrifuged at 1000g for 10 min to pellet out the nuclear fraction. An aliquot (10 μl) of the supernatant was kept for protein estimation. The supernatant was immediately diluted with 4× LDS sample buffer (Thermo Fisher Scientific, catalog no. NP0007) and heated to 95°C for 10 min and stored in aliquots (100 μl) at −80°C. For cell lysate preparation, cultured cells were harvested and washed with PBS and the cell pellet was sonicated in RIPA buffer containing 1× Halt protease and phosphatase inhibitors. The cell lysates were immediately diluted using 4× LDS sample buffer and heated at 95°C for 10 min and stored in aliquots (100 μl) at −80°C to avoid multiple freezing and thawing.

Protein stability assay for NPC1

Control HEK293T cells were purchased from ATCC (catalog no. CRL11268, Lot# 62312975). CLN1^−/− HEK293T cells were generated by CRISPER-Cas9–mediated gene ablation (58) and cultured in DMEM supplemented with 10% FBS, penicillin (100 U/ml), and streptomycin (100 μg/ml). For MG132 treatment, both normal and CLN1^−/− HEK293T cells were treated with various doses of proteasome inhibitor, MG132 (5, 10, and 20 μM) for 12 hours. Before harvesting, the last 3 hours, the cells were washed twice and replaced with serum-free media containing lipid-free BSA and inhibitor. For cell lysate preparation, the harvested cells were washed with PBS and the cell pellet was sonicated in RIPA buffer containing 1× Halt protease and phosphatase inhibitors. The cell lysates were immediately diluted using 4× LDS sample buffer and heated at 95°C for 10 min and stored in aliquots (100 μl) at −80°C to avoid multiple freezing and thawing. Proteins were resolved by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) and subjected to Western blot analysis.

Isolation of plasma membrane and lysosomal fractions

For the isolation of proteins from plasma membrane of cortical tissues and cultured cells, a plasma membrane isolation kit (Abcam, catalog no. ab65400) was used following the manufacturer’s protocol. Approximately 150 mg of brain tissue (cortex) was homogenized using recommended buffer, and the rest of the isolation was performed according to the manufacturer’s protocol. Lysosomal fractions were isolated from mouse brain cortex using Optiprep density gradient media using the lysosome isolation kit (Sigma-Aldrich; catalog no. LYSISO1) as per the manufacturer’s protocol. The enrichment of plasma membrane and lysosomes from total plasma membrane and lysosomal fractions were assessed by Western blot analysis using antibody to plasma membrane marker, Na⁺,K⁺-ATPase, and lysosomal membrane marker, LAMP2, respectively. The purity of the fractions was evaluated by using various organelle markers: ER marker, calreticulin; nuclear marker, histone H3; mitochondrial marker, VDAC; and peroxisomal marker, peroxisomal membrane marker-70 (PMP70). To evaluate the purity and cross-contamination between each fractionation, we loaded total homogenates, enriched plasma membrane, and lysosomal fractions. Proteins were resolved by SDS-PAGE and subjected to Western blot analysis using antibodies to the membrane marker, Na⁺,K⁺-ATPase and lysosomal marker, LAMP2, respectively.

Sorting of lysosome and plasma membrane by flow cytometry

Lysosome and plasma membrane fractions were enriched from cultured neurons by using lysosomes enrichment kit (Sigma-Aldrich, catalog no. LYSISO1) and plasma membrane enrichment kit (Abcam, catalog no. ab284937), respectively, according to manufacturer’s protocol. Cell samples were washed with fluorescence-activated cell sorting (FACS) buffer (Corning, catalog no. 21-022-CV) without calcium and magnesium containing 4% FBS (GeminiBio, catalog no. 100-500). The samples were centrifuged at 20,000g (Microfuge 20R, Beckman Coulter) for 30 min at 4°C. For staining, lysosome pellets were resuspended in 100 μl of FACS buffer and incubated with FITC-conjugated anti-CD107a (LAMP-1) monoclonal antibody (Thermo Fisher Scientific, catalog no. 11-1079-42) for 1 hour at 4°C. For staining, plasma membranes pellets were resuspended in 100 μl of FACS buffer and incubated with anti–Na⁺,K⁺-ATPase a-1 monoclonal antibody (Millipore, catalog no. 05-369) for 1 hour at 4°C followed by APC-conjugated rat anti-mouse IgG1 secondary antibody (R&D Systems, catalog no. F0108; dilution: 1:50) for 1 hour at 4°C. The samples were washed and resuspended in FACS buffer and sorted on a FACS Aria flow cytometer (BD Biosciences). LAMP-1–positive and Na⁺,K⁺-ATPase a-1–positive populations were established by back-gating using unstained isotype control sample.

TEM analysis

Both WT and Cln1^−/− mouse brain tissue samples were fixed with 2.5% glutaraldehyde in PBS for 2 hours. Tissues were then kept in Millonig’s phosphate buffer. The tissue sections were stained with lead citrate and uranyl acetate. They were then examined by a LEO 912 electron microscope by JFE Enterprises, Beltsville, Maryland.

Determination of cholesterol level

The cholesterol level in the total homogenate and lysosomal fractions was estimated by colorimetric assay kit from Sigma-Aldrich (catalog no. MAK043) according to the manufacturer’s protocol. Briefly, mouse brain tissue and lysosomal fraction were extracted with 200 μl of chloroform: isopropanol: IGEPAL CA-630 (7:11:0.1). The samples were centrifuged at 13,000g for 10 min to remove insoluble material. Then, the organic phase was transferred to a new tube and air dried at 50°C to remove chloroform and samples under speed-vacuum for 30 min to remove any residue organic solvent. The dried lipids were dissolved with 200 μl of the cholesterol assay buffer and sonicated or vortex until mixture was homogenous. In a 96-well plate, the samples were added to a final volume of 50 μl with cholesterol assay buffer 50 μl of the appropriate reaction mix to each of the samples. The samples were mixed by using a horizontal shaker or by pipetting, and the reaction was incubated for 60 min at 37°C. The absorbance was measured at 570 nm.

Confocal imaging

Cln1^−/− mouse neurons and age- and sex matched normal control subjects were cultured in eight-well chamber slides with neurobasal media with B27 supplement and glutamine. Before fixing, the cells were washed twice and replaced with serum free media containing lipid-free BSA for 3 hours to deplete any PPT1 activity in the media. The cells were washed two to three times with PBS, and cells were fixed using 100% methanol. Then, washed with PBS and blocked with 10% normal goat serum for 1 hour., the cells were incubated with primary antibodies overnight at 4°C for 1 hour at room temperature. The primary antibodies used are listed in table S1, followed by Alexa Fluor–conjugated secondary antibodies (Invitrogen). The cells were mounted using 4′,6-diamidino-2-phenylindole (DAPI)–Fluoromount G (Thermo Fisher Scientific, 010020), and fluorescence was visualized with the Zeiss LSM 710 Inverted Meta confocal microscope (Carl Zeiss). For confocal imaging of cholesterol in the lysosome of neurons and patient fibroblasts, we have used Filipin III cell–based cholesterol kit (Abcam, catalog no. ab133116) and CellLight Lysosomes-GFP, BacMam 2.0 (Invitrogen, C10507) as a lysosomal marker according to the manufacturer’s instructions since the fluorescence of lysosomal marker LAMP2 was masked using Filipin III cell–based cholesterol kit. The image was processed with the LSM Image Software (Carl Zeiss 710) and calculated the colocalization coefficient using the Zen Desk Software from Zeiss.

Acyl-RAC assay

Acyl-Rac assay (49) to determine S-palmitoylation was performed according to the method previously reported with minor modifications (fig. S12). To determine which cysteine residue in NPC1 protein is S-palmitoylated, we transfected HEK293T cells with each of the flag tagged-cDNA constructs of WT- and each of the NPC1-mutants (i.e., Cys¹⁶Ala, Cys⁹⁷Ala, and Cys⁶⁴⁵Ala). The cells expressing the proteins were homogenized and used for the Acyl-Rac assay. For the level of S-palmitoylated NPC1 in the plasma membrane, we used the isolated plasma membrane fractions of cortical tissues from WT and Cln1^−/− mice. Total homogenates from the cortical tissues were used to determine whether NPC2 undergoes S-palmitoylation (fig. S5A). Briefly, samples (1 mg of protein) were diluted in 1 ml of blocking buffer [100 mM Hepes, 1.0 mM EDTA, 2.5% SDS, and 50 mMN-ethylmaleimide (pH 7.5)] and incubated at 50°C for 60 min with frequent vortexing (after each 15 min). Three volumes of ice-cold acetone were then added and kept at −20°C for 20 min. This was followed by centrifugation at 5,000g for 10 min, and the pellet was extensively washed with 70% acetone at least four times. Thereafter, the pellet was resuspended in 250 μl of binding buffer [100 mM Hepes, 1.0 mM EDTA, and 1% SDS (pH 7.5)]. Approximately 50 μl of resuspended protein was saved as the total input. Diluted protein samples (200 μl each) were added to approximately 80 μl of prewashed thiopropyl Sepharose beads (GE-Amersham, #17-0420-01). Half of this mixture (140 μl) was treated with 100 μl of freshly prepared 2 M NH₂OH [hydroxylamine (HA) (pH 7.5)] and the other half with PBS (for control). Binding reactions were carried out on a rotator at room temperature for around 4 hours. The unbound fraction was collected from both PBS- and HA-treated samples to a separate tube, which was a non–S-palmitoylated protein fraction (PBS and HA unbound fractions). Thiopropyl Sepharose beads were washed at least four times with binding buffer, considered as S-palmitoylated protein fraction (PBS and HA bound fractions). For immunoblot analysis, input, unbound, and bound fractions were boiled (95°C for 5 min) with 4× loading buffer (NuPage 4× LD, Thermo Fisher Scientific, catalog no. NP0007). Input, unbound, and bound fractions were separated via SDS-PAGE (Bolt 4 to 12% bis-tris Plus, #NW04120BOX) on a mini-gel apparatus (Invitrogen).

Western blot analysis

For Western blot, protein samples (20 μg) were resolved by electrophoresis using 4 to 12% SDS–polyacrylamide gels (Invitrogen) under denaturing and reducing conditions and blotted to nitrocellulose or polyvinylidene difluoride membranes (Invitrogen). The membranes were blocked with 5% nonfat dry milk (Bio-Rad) and then subjected to immunoblot analysis using standard methods. The primary antibodies used for the immunoblots are listed in table S1. The blots were then probed with horseradish peroxidase–conjugated secondary antibodies (Santa Cruz Biotechnology) followed by detection using SuperSignal west femto or pico solution (Thermo Fisher Scientific) according to the manufacturer’s instructions. Image quant 4000 mini (GE Healthcare Lifesciences) was used to capture the chemiluminescent signals, and the immunoblots were quantified using the image quant TL software (GE Healthcare Lifesciences). Experiments were repeated at least four times to confirm the reproducibility. For the analysis of protein bands from Western blots of cortical tissues from 2-, 4- and 6-month-old WT and Cln1^−/− mice, we first normalized the intensity to those of the WT mice. For phosphoprotein analysis, Halt protease and phosphatase inhibitors were used and were present throughout sample preparation. The samples were further processed by boiling with SDS sample buffer on the same day of the sample collection. Uncropped and unprocessed scans of the blots are provided in the source data file.

Proximity ligation assay

Duolink PLA (46) was used to study interaction between different proteins. Briefly, the primary neurons from WT and Cln1^−/− mice were grown in chamber slides, fixed with paraformaldehyde, and permeabilized with 0.1% Triton X-100 in PBS. The cells were covered with PLA blocking buffer and incubated with respective antibodies overnight. Following three washes with PBS, the cells were incubated with anti–rabbit-MINUS (Sigma-Aldrich, DUO92005) and anti–mouse-PLUS (Sigma-Aldrich, DUO92001) PLA probes and subjected to ligation and amplification reaction using Duolink In situ Detection Reagents Orange (Sigma-Aldrich; DUO80102) or Duolink In Situ Detection Reagents Green (Sigma-Aldrich; DUO9201) according to the manufacturer’s protocol. The cells were mounted with DAPI-Fluoromount G (Thermo Fisher Scientific, 010020) and visualized with Zeiss 710 inverted confocal microscope. Z stack images were taken to include all the signals at different focal planes, and the images were merged using maximum intensity projection. The Duolink image tool was used to count the PLA signals as previously reported (69).

Immunohistochemistry

Frozen whole-brain sections were prepared with help of Histoserv Inc., Germantown, USA and mounted with Optimal Cutting Temperature compound and the specimens were kept at −80°C for at least 12 hours before sectioning. Mounted samples (10 μm thick) were sectioned using a cryostat. For immunohistochemical analyses, tissue sections were washed with PBS and then all the sections were kept in methanol at −20°C for 20 min. The tissue sections were then washed with 1× PBS three times. The paraffin embedded sections (10 μm thick) were deparaffinized and hydrated. Then, antigen retrieval was performed by using 5 mM Hepes buffer with 1 mM EDTA and 0.05% Triton X-100 at pH 8 and incubated at 80°C for 10 min and blocked with 10% normal goat serum for 1 hour. The sections were then incubated with TrueBlack Lipofuscin Autofluorescence Quencher (Biotium, CA, USA. catalog no. 23007) for 30 s and washed three times with 1× PBS to quench any background autofluorescence. The slides were then incubated with primary antibodies, listed in table S1, overnight at 4°C. Then, the slides were washed three times with 1× PBS and incubated with Alexa Fluor–conjugated secondary antibodies (AF488 and AF555, diluted 1:200) (Invitrogen) for 2 hours at room temperature in the dark. Then, the slides were washed with 1× PBS (three times) and mounted with DAPI–Fluoromount-G mounting medium (Southern Biotech, Birmingham, USA. catalog no. 0100-20) (69, 70).

Immunoprecipitation

Cortical tissues were homogenized using 0.32 M sucrose, 1 mM EDTA, and 10 mM Hepes with proteinase inhibitor in a Dounce homogenizer and centrifuged at 1000g for 10 min. The nuclear fractions were discarded, and the supernatants were suspended in RIPA buffer, centrifuged at 10,000g, and the supernatants were collected. The HEK293T cells were lysed in RIPA buffer and centrifuged at 10,000g for 10 min. The supernatants were collected and used for immunoprecipitation studies. An aliquot of each sample was kept as input. The samples were used to pull down the NPC1 and various adaptor proteins (AP-1, AP-2, and AP-3) using specific antibodies (table S1). Briefly, the samples were incubated with specific antibodies for overnight at 4°C, and then, the Rockland true blot ip-specific beads were added and incubated for 1 hour with constant agitation on a rocker at 4°C. The beads were washed with PBS and resuspended in electrophoresis sample buffer (Santa Cruz Biotechnology, catalog no. sc-24945). Immunoprecipitates were resolved by SDS-PAGE and analyzed by Western blotting.

H&E staining

Hematoxylin-eosin staining was performed according to a previously published protocol (69). Whole-brain sagittal sections (5 μm thick) were deparaffinized, hydrated and stained with hematoxylin and eosin kit (ab245880, Abcam) using manufacturer’s protocol. At least 4 comparable sections from each brain were used for measurement of cortical thickness and a similar area for all the sections was used.

Cresyl Violet staining

Cresyl violet staining was performed according to the previously published protocol (69, 70). Briefly, 5-μm-thick whole-brain sagittal sections were deparaffinized, hydrated, and stained with 0.1% Cresyl Violet stain solution (Abcam, catalog no. ab246817) using the manufacturer’s protocol. Nissl-positive cells were counted from four comparable sections from each group from similar area of the cortex for all the sections.

Statistical analysis

The data are displayed using box-whisker diagrams with “n” denoting the number of biological replicates for each experiment unless otherwise stated. These diagrams display the standard composition of minimum, first quartile, median, third quartile, and maximum, respectively, from bottom to top, with the + symbol representing the sample mean. Alternatively, some of the data are represented as bar diagrams with means ± SD. For confocal imaging in isolated neurons, n is the total number of images of cells used for analysis are presented as the means ± SD. For imaging of cortical sections, n numbers are the total number of fields used for analysis, which were acquired from four biological replicates. Plots were created using GraphPad prism 8. We performed two-sample permutation t tests to examine the differences between two independent groups using R version 4.2.1. Group means and tests with P values <0.05 are considered statistically significant are also depicted in the plots as (*P < 0.05, **P < 0.01).

Supplementary Materials

The PDF file includes:

Figs. S1 to S12

Table S1

Legends for data S1 and S2

Other Supplementary Material for this manuscript includes the following:

Data S1 and S2