Discovery of Autophagy Modulators that Increase I1061T NPC1 Expression and Promote Cholesterol Efflux in Niemann-Pick Type C Patient-Derived Fibroblasts

Abstract

Autophagy, an evolutionarily conserved catabolic process, has been implicated as a potential therapeutic target in Niemann-Pick Type C (NPC) disease, a fatal lysosomal storage disorder. Our goal was to identify autophagy modulators that positively impact NPC-relevant phenotypes to further evaluate the role of autophagy in this disease. Using a phenotypic high-throughput screen for autophagy modulation and a subsequent secondary assay in homozygous I1061T NPC1 patient-derived fibroblasts, two compounds, 1 and 2, were identified that induced autophagy and reduced unesterified cholesterol accumulation to a level comparable to the reduction achieved by 2-hydroxypropyl-β-cyclodextrin. Global protein expression changes were evaluated in I1061T NPC1 fibroblasts following treatment with compounds 1 and 2 to identify affected pathways, and we observed a significant reduction of lysosomal hydrolase levels induced by compound 2. Additional mechanistic studies revealed that the expression of NPC1 protein is required for compound-induced amelioration of cholesterol accumulation and that compound 2 increases expression levels of I1061T NPC1 without broadly inhibiting proteasomal activity or exacerbating ER stress. These results have led to the hypothesis that compound 2 may serve as a proteostasis modulator or small-molecule chaperone that upregulates autophagy to a level that is predominantly cytoprotective and increases the proportion of properly folded mutant NPC1, thus increasing NPC1 expression levels and alleviating cholesterol accumulation and associated phenotypes. This work, along with future mechanistic studies, contributes to the development of novel strategies to modulate autophagy and proteostasis with potentially broad therapeutic applications.

Introduction

Macroautophagy (hereafter autophagy) is a conserved, eukaryotic, catabolic pathway that maintains homeostasis by degrading and recycling intracellular components. , Cellular cargo is sequestered within the double-layered membrane of an autophagosome that fuses with a lysosome to form the autolysosome, where the cargo is degraded by enzymes contained within the acidic lysosome lumen. The resulting products are then recycled and used for other cellular processes. Dysregulated autophagy has been implicated in a wide array of processes including aging, cancer, , and neurodegeneration. In the past decade, modulation of the autophagy pathway has emerged as a promising therapeutic strategy for a wide range of diseases, − including the rare genetic disease Niemann-Pick Type C Disease (NPC).

NPC is an autosomal recessive neurodegenerative disease characterized by an accumulation of unesterified cholesterol in lysosomes that affects multiple tissues, including the brain, liver, and spleen. ,, This accumulation is caused by a mutation in the NPC1 gene in 95% of cases, or in the NPC2 gene in 5% of cases; the NPC1 and NPC2 proteins coordinate cholesterol efflux from late endosomes/lysosomes (LE/Lys). , The most common mutation in NPC1 is the I1061T missense mutation that occurs in 20% of patients, where the endogenous I1061T mutant misfolds in the endoplasmic reticulum (ER) and is degraded by two pathways that function complementarily: selective endoplasmic reticulum autophagy (ER-phagy) and MARCH6-dependent endoplasmic reticulum-associated degradation (ERAD). , The overexpression of I1061T mutant NPC1 has been shown to increase the proportion of nascent I1061T NPC1 that folds correctly, escapes degradation, and is trafficked to the lysosome to clear unesterified cholesterol. Therefore, proteostasis modulators or small-molecule chaperones that prevent mutant NPC1 degradation through the proteasomal and autophagy pathways could increase mutant NPC1 expression levels and ameliorate cellular stress responses, thus making the mutant NPC1 available for cholesterol efflux and ameliorating NPC phenotypes. A similar approach has been highly successful for the development of therapeutics to treat another rare disease, cystic fibrosis, caused by mutations in a specific protein, cystic fibrosis transmembrane receptor (CFTR). − Drugs have been developed that increase the cell surface expression (correctors) or anion conductance (potentiators) of CFTRs harboring specific disease-causing mutations. The small-molecule chaperones, or “correctors”, support folding and improve stability of CFTR mutants that are typically misfolded and degraded and promote trafficking to the cell surface where the mutant proteins can function properly. While it is possible that proteostasis modulators and small-molecule chaperones that increase expression of mutant proteins could be beneficial in NPC disease, autophagy and proteasomal degradation are also important for regulating cellular stress pathways and cell death. NPC1-knockout mouse embryonic fibroblasts (MEFs), NPC1 knockdown human embryonic stem cell (hESC)-derived neurons, NPC1 human dermal fibroblasts heterozygous for P237S and I1061T missense mutations, and the brain of NPC1-deficient mice, , all display an accumulation of autophagosomes. These observations suggest that stalling the autophagy machinery in both NPC1-knockout/downs and cells with disease-relevant mutations is detrimental and that restoration of autophagic flux could potentially resolve NPC symptoms and progression. Autophagy modulators that prevent degradation of mutant proteins and also restore autophagic flux could be especially impactful.

Autophagy induction has emerged as a promising strategy for the treatment of a variety of diseases, and a number of small molecules have been identified and evaluated in disease-relevant assays and models. − Phenotypic high-throughput screens used to identify autophagy inducers typically include GFP-tagged LC3 or dual-tagged (e.g., GFP-mCherry) LC3 puncta formation assays or disease relevant assays followed by mechanistic studies that reveal autophagy induction by prioritized compound(s). − Targeted approaches have also been undertaken to identify kinase inhibitors or protein–protein interaction inhibitors of autophagy-related proteins to induce autophagy. − One of the major strengths of a phenotypic approach is the potential to identify compounds with novel targets or mechanisms that regulate autophagy and ameliorate disease-relevant phenotypes. Library selection for compound screening is also an important consideration to improve the chances of identifying novel chemotypes with potentially new mechanisms. By screening different compound collections, a variety of natural products, synthetic compounds, and FDA-approved drugs have been identified that induce autophagy.

Rapamycin, a macrocyclic polyketide and allosteric mTORC1 inhibitor, is one of the most well-studied autophagy activators. Synthetic ATP-competitive mTOR inhibitors, including AZD8055, an mTORC1-selective inhibitor, and Torin1, a dual mTORC1 and mTORC2 inhibitor, have also been developed to study mTOR signaling pathways and to develop drugs with immunosuppressive, anticancer, and antiaging properties. , A novel mechanism for selective mTORC1 inhibition was discovered by studying EN6, an acrylamide covalent inhibitor of the lysosomal vacuolar ATPase that disrupts the Ragulator-RAG-GTPase complex at the lysosome to induce autophagic flux. Induction of ER stress and activation of the integrated stress response (ISR) is also a well-established strategy to induce autophagy. − Thapsigargin, a sequiterpene lactone natural product, and tunicamycin, a nucleoside antibiotic that inhibits N-linked glycosylation and protein maturation in the ER, are commonly used to study ER stress induction. , However, if ER stress is unresolved, then long-term pathway activation can ultimately lead to apoptosis or autophagy-dependent cell death. , Auxathrol A was also recently identified in a phenotypic screen of natural products to discover compounds with anticancer activity in nonsmall cell lung cancer (NSCLC) cells. This natural product induces autophagy through upregulation of the ISR and production of transcription factor ATF4, resulting in autophagy-dependent cell death. The synthetic PERK activator, CCT020312, also induces the ISR and has been shown to activate autophagy, even at levels lower than required to activate PERK, and has beneficial effects in neurodegenerative disease models. Transcription Factor EB (TFEB), a master regulator of lysosomal biogenesis, has emerged as another target for autophagy induction. Celastrol, a pentacyclic triterpenoid natural product, and the FDA-approved selective-estrogen receptor modulator, clomiphene citrate, enhance TFEB-mediated autophagy through mTORC1 dependent and independent mechanisms, respectively, and improve Alzheimer’s disease relevant phenotypes. , AA-20 is a synthetic small molecule discovered through a screen for lipid droplet clearance that activates autophagy and extends lifespan in C. elegans through a mechanism involving TFEB without inhibiting mTOR. In the context of NPC disease, the natural product sulforaphane activates TFEB and shows benefits in Npc1 ^–/– mice by rescuing loss of Purkinje cells and body weight. Carbamazepine, an FDA-approved anticonvulsant, along with rapamycin and the natural product trehalose, increased cell viability in NPC1-deficient neuronal cultures. Additional synthetic mTOR-independent autophagy activators, SMER28, BRD5631, AUTEN-99, and RH1115, have shown a variety of neuroprotective effects in cell-based or animal disease models. −

The observed benefits of autophagy induction in models of neurodegeneration encouraged us to identify small molecules that restore autophagic flux and ameliorate NPC phenotypes, without inhibiting mTOR or activating the ISR, and are amenable to synthetic optimization to enable preclinical efficacy studies. As we are broadly interested in studying the autophagy pathway and its role in different diseases, we decided to use a phenotypic high-throughput screen (HTS) to identify autophagy modulators, followed by evaluation in a disease-relevant secondary assay, specifically filipin staining in I1061T NPC1 fibroblasts to monitor unesterified cholesterol clearance after treatment with the identified autophagy modulators. To potentially identify novel chemotypes and mechanisms, we screened a commercial library of compounds enriched with characteristics that make them more likely to target protein–protein interactions or nonenzymatic targets. Considerations for inclusion in the curated library included number of hydrogen bond acceptors >3, molecular weight >350, LogP = 3–7, fraction of sp³ centers >0.3, no reactive compounds, and no frequent hitters (i.e., PAINs). Implementing a phenotypic approach combined with subsequent mechanism of action studies of prioritized molecules enabled us to study autophagy modulation in the context of NPC disease, which led to the discovery of a proteostasis modulator that prevents degradation of I1061T NPC1, promotes cholesterol efflux, and maintains cellular homeostasis.

Results & Discussion

Upon autophagy induction, ATG3, ATG7, and the ATG5-ATG12-ATG16L1 complex conjugates Light Chain 3 (LC3)-I, which is diffusely located throughout the cytosol, to phosphatidylethanolamine (PE) and generates membrane-associated LC3-II. LC3-II is anchored into the developing autophagosome membrane and promotes expansion and closure of the nascent autophagosome, and it is considered a marker of autophagosome formation. When tagged with a fluorophore, such as green fluorescent protein (GFP), LC3-I and LC3-II can be visualized through fluorescence microscopy and used to assess autophagy induction by measuring autophagosome formation through the production of LC3-II. ,, LC3-I appears as a green cytoplasmic pool, and LC3-II appears like green distinct dots, or puncta, and these puncta can be efficiently quantified using fluorescence microscopy. An important aspect of this assay to take into consideration is that an autophagy inducer and late-stage inhibitor would have the same phenotype, as an inducer would increase the formation of autophagosomes, and a late-stage inhibitor would cause the accumulation of autophagosomes due to inhibition of autophagosome maturation. Both possibilities result in an increase in LC3-II puncta levels, thus additional evaluation is necessary to discern the true nature of newly identified autophagy modulators.

The late-stage autophagy inhibitor, chloroquine (CQ), caused a robust accumulation of autophagosomes in HeLa cells stably expressing eGFP-tagged LC3, as indicated by a significant increase in puncta per cell and Z’ values consistently >0.5 in 384 well plates compared to the negative control, DMSO, and was thus selected as the positive control for the eGFP-LC3 assay (Figure S1). The curated library of 10,000 small molecules (ChemDiv) was screened in duplicate at 20 μM concentration for the ability to increase LC3-II puncta levels, and compounds with a z-score ≥ 2.2 in both replicates were considered hits. Compounds with a z-score ≤ −1.8 for nuclear count were deemed cytotoxic and excluded from subsequent evaluation (Figure S2). From the 10,000 compounds screened in the eGFP-LC3 assay, 312 compounds increased autophagosome/LC3-II puncta levels (Figure S2, Table S1). A 3% hit rate was considered acceptable to avoid excluding compounds that induce autophagy to a modest level. We also planned to screen this set of 312 compounds in multiple disease-relevant secondary assays that would enable us to further prioritize compounds. For example, by investigating structure–activity relationship trends observed within this original screening data, we were able to develop and further optimize an autophagy activator that induces perinuclear clustering of lysosomes in neurons. ,

This set of 312 autophagy modulators was then tested in duplicate at 10 μM concentration in the disease-relevant secondary assay that measures accumulated unesterified cholesterol in homozygous I1061T NPC1 patient-derived skin fibroblasts (Figure A,B). Cholesterol levels were quantified using filipin dye, a naturally fluorescent polyene macrolide antibiotic that binds to unesterified/free cholesterol, but not esterified sterols, and is the current initial method for the diagnosis of NPC disease. , Two of these compounds, termed compounds 1 and 2 (Figure C), were selected for additional studies based on chemical structures and properties and the observation that these two compounds reduced unesterified cholesterol to a similar level as the positive control, 2-hyroxypropyl-β-cyclodextrin (2HPβCD). 2HPβCD clears accumulated cholesterol through binding unesterified cholesterol and potentially additional mechanisms, including facilitating cholesterol efflux via lysosomal-associated membrane protein (LAMP1) and enhancing lysosomal function. , Interestingly, of all of the autophagy modulators, only three compounds out of the 312 compounds decreased filipin intensity by more than 55% in patient-derived cells and induced a greater than 2.2-fold increase in LC3-II puncta levels (Figure D). A comparison of the average data for each compound from two replicates in the eGFP-LC3 assay and the filipin assay, revealed a significant correlation (p = 0.0026) between activity in the two assays, indicating an inverse relationship between autophagosome formation and reduction of unesterified cholesterol. However, most of the autophagy modulators did not enhance cholesterol efflux; therefore, autophagy modulation does not appear to generally impact cholesterol levels in the NPC patient cells, indicating that the specific mechanism of action of each compound must be critical for cholesterol clearance.

1.

Discovery of small-molecule autophagy modulators that reduce accumulation of unesterified cholesterol in NPC patient-derived fibroblasts with I1061T NPC1 mutation. (A) The cherry-picked autophagy modulators from the eGFP-LC3 HTS were evaluated in the filipin assay for 24 h in I1061T NPC1 fibroblasts in duplicate and their activity assessed using z-score, where 1 and 2 (10 μM, purple and orange circles, respectively) had z-scores comparable to 2HPβCD (1 mM, green circles). Filipin intensity is measured as the pit integrated intensity, which is the total pixel intensity over all the pit areas in the image. (B) Normalized filipin intensity following treatment with cherry-picked 1 (5 and 10 μM) and 2 (5 and 10 μM) compared to CQ (20 μM), Rap (4 μM), and 2HPβCD (1 mM) in in I1061T NPC1 fibroblasts. Data are presented as the mean ± SD of biological duplicates. (C) Structure of hit compound 1 and 2. (D) Correlation between eGFP-LC3 HTS results and filipin secondary assay. (E) Activity of resynthesized 1 and 2 in eGFP-LC3 assay at multiple doses (0.63–80 μM) compared to the average of CQ (20 μM) (dashed line) and DMSO (solid line). (F) Activity of resynthesized 1 (10 and 20 μM), and 2 (20 and 40 μM), DMSO, 2HPβCD (1 mM and 300 μM) in filipin assay in I1061T NPC1 fibroblasts. (G) Quantitative effect of 1 (5–80 μM) on average number of autolysosomes vs autophagosomes compared to DMSO and CQ (20 μM) in dual reporter assay. (H) Quantitative effect of 2 (5–80 μM) on average number of autolysosomes vs autophagosomes compared to DMSO and CQ (20 μM) in dual reporter assay. Quantified data (E–H) are presented as mean ± SEM of three independent experiments each with duplicate biological replicates. Significance in (B,F) was determined by one-way ANOVA using GraphPad Prism 10.6.1, where *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Only significant values (p ≤ 0.05) are indicated.

In preparation for additional mechanistic studies with compounds 1 and 2, we developed convergent synthetic routes with multiple branching points to facilitate analogue generation and to expediently access each of the hit compounds (Schemes S1 and S2). Activity of the prioritized compounds, 1 and 2, was evaluated in the eGFP-LC3 assay along with CQ and the mTOR inhibitor/autophagy activator, rapamycin (Rap), which both increased LC3-II puncta levels, and 2HPβCD, which did not increase puncta levels (Figure S3A,B). The inability of 2HPβCD to significantly increase LC3-II puncta levels highlights that not all molecules that enhance cholesterol efflux also modulate autophagy, complementing the earlier observation that not all autophagy modulators decrease unesterified cholesterol levels. The two resynthesized hits were then evaluated in the eGFP-LC3 assay, and it was determined that compounds 1 and 2 had similar potencies (EC₅₀ of 10.97 and 14.33 μM, respectively) (Figure E). The resynthesized compounds were also tested in the filipin assay at multiple doses, and compounds 1 and 2 decreased filipin intensity, as did 2HPβCD (Figure F).

To discern whether compounds 1 and 2 were autophagy activators or late-stage inhibitors, a dual reporter assay was conducted. The dual reporter assay relies on the pH-sensitive GFP fluorophore and pH-insensitive mCherry red fluorophore to differentiate autophagy activators from inhibitors in HeLa cells stably expressing tandem mCherry and eGFP tags on LC3 (mCherry-eGFP-LC3). , Once the autophagosome fuses with the lysosome to form the autolysosome, the pH-sensitive GFP is quenched and only mCherry fluorescence is detected. Quantitatively, an autophagy inducer would increase the number of autophagosomes and increase or maintain levels of autolysosomes, whereas a late-stage inhibitor would significantly increase the number of autophagosomes (AP) but decrease the number of autolysosomes (AL) as compared to DMSO due to inhibition of autophagic flux. The mCherry-eGFP-LC3 HeLa cells were treated with compounds for 24 h in the dual reporter assay, and DMSO-treated cells exhibited a ratio of 2:2 AP:AL and CQ treatment caused an increase in the number of AP and a decrease in the number of AL to give a ratio of 6:1 AP:AL. (Figure G,H). Across several concentrations (5–80 μM), compound 1 had an approximate ratio of 4:2.4 AP:AL, which means the levels of AP and AL increased compared to DMSO treatment, indicative of an autophagy inducer (Figure G). Compound 2 (40 μM) elicited a ratio of 4.5:1 AP:AL, which means the levels of AP increased, and the levels of AL decreased compared to DMSO treatment, similar to CQ, and consistent with an autophagy late-stage inhibitor (Figure H). However, at a lower concentration (10 μM), compound 2 had an AP:AL ratio of 3.8:2.5, which is more similar to the profile of an autophagy activator. Therefore, it is possible that this compound is activating autophagy at concentrations close to its EC₅₀ and that administration at higher concentrations could create an abundance of autophagosomes in excess of the lysosomes available for fusion.

To test if compounds 1 and 2 affect lysosomal protease activity, the Dye Quenched-Bovine Serum Albumin (DQ-BSA) assay was implemented. , DQ-BSA is labeled with a self-quenching dye that only fluoresces once it is endocytosed and digested, which creates isolated fluorophores that are no longer quenched and become brightly fluorescent, indicative of proteolytic activity. When performed in HeLa cells with a wild-type (WT) NPC1 protein for 6 h, a method typically used for evaluating inhibition of lysosome acidification or function by late-stage autophagy inhibitors, CQ and the vacuolar ATPase inhibitor, bafilomycin A1 (BafA1), drastically decreased proteolytic activity, as expected (Figure A). Interestingly, compounds 1 and 2 did not significantly affect DQ-BSA processing in HeLa cells at concentrations near the EC₅₀ in the eGFP-LC3 assay (EC₅₀ of 10.97 and 14.33 μM, respectively); however, at higher concentrations, compounds 1 and 2 slightly reduced DQ-BSA processing, but not the same degree as CQ and BafA1. Next, we evaluated proteolytic processing of DQ-BSA in NPC patient-derived fibroblasts treated with compounds 1 and 2 to determine how they impact this process in the context of I1061T NPC1 (Figure B). The I1061T mutation in NPC1 affects protein folding, stability, and trafficking, and the resulting cholesterol accumulation in lysosomes can affect lysosomal function and the proteolytic activity of enzymes found in the lysosome, which would impact DQ-BSA processing. CQ decreased proteolytic activity in I1061T NPC1 fibroblasts, whereas 2HPβCD and BafA1 increased proteolytic activity. Compounds 1 and 2 elicited a 3-fold increase in proteolytic activity after 24 h as compared to DMSO, indicating that compounds 1 and 2 significantly enhance proteolytic activity in the presence of I1061T NPC1 compared to the vehicle control, which signifies potential restoration of lysosome function. Viability in I1061T NPC1 patient-derived fibroblasts after 24-h compound treatment was measured using the CellTiter-Glo assay, and both compounds were noncytotoxic at concentrations that induce beneficial effects in the patient-derived fibroblasts (Figure C).

2.

Mechanisms of compounds 1 and 2 are mTOR-independent and dependent on expression of NPC1. (A) Puncta/cell counts in DQ-BSA assay following compound treatment (6 h) in HeLa cells. (B) Puncta/cell counts in DQ-BSA assay following compound treatment (24 h) in I1061T NPC1 fibroblasts. (C) Viability of I1061T NPC1 fibroblasts following 24 h treatment with DMSO, 1 and 2 (10–80 μM), 2HPβCD (1 mM), CQ (40 μM), BafA1 (0.1 μM) in the CellTiter-Glo 2.0 assay. (D) Representative P-p70S6K and p70S6K immunoblots following 24 h treatment of I1061T NPC1 fibroblasts with resynthesized compounds 1 (20 μM) and 2 (80 μM), CQ (20 μM), and BafA1 (0.2 μM). (E) Representative NPC1 immunoblot of I1061T NPC1 fibroblasts following 24 h treatment with DMSO, 1 (10 and 20 μM), 2 (10 and 20 μM), and 2HPβCD (1 mM). (F) Quantified NPC1 immunoblots presented as mean ± SEM of three independent experiments normalized to β-actin. (G) Filipin assay in NPC1 null HeLa cells following 24 h treatment with DMSO, 1 (40 μM), 2 (80 μM), and 2HPβCD (0.3 and 1 mM). Data are presented as mean ± SEM of three independent experiments each with duplicate biological replicates. Significance was determined by one-way ANOVA using GraphPad Prism 10.6.1, where *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Only significant values (p ≤ 0.05) are indicated.

Due to its role in many essential pathways, inhibition of mTOR can cause undesirable off-target effects. To support our efforts to identify autophagy modulators with novel mechanisms of action, the effect of compounds 1 and 2 on mTOR kinase activity was also evaluated in I1061T NPC1 fibroblasts after 24-h treatment to determine if the hit compounds induce autophagosome accumulation in an mTOR-dependent or mTOR-independent manner (Figure D and Figure S4A). Inhibition of the mTOR serine/threonine protein kinase prevents phosphorylation of p70S6 kinase (p70S6K), a substrate of mTOR that has a role in translation, and thus compounds that inhibit the kinase activity of mTOR will decrease phosphorylation of p70S6K. Rap inhibits mTOR and prevented p70S6K phosphorylation, but CQ, BafA1, and compounds 1 and 2 did not significantly decrease formation of P-p70S6K, suggesting they do not significantly inhibit mTOR activity (Figure D).

Next, the effects of compounds 1 and 2 on I1061T NPC1 protein expression levels were evaluated after 24-h treatment, and both compounds increased NPC1 protein levels (Figure E,F and Figure S4B). An increase in NPC1 expression levels could improve unesterified cholesterol clearance by providing excess protein to facilitate cholesterol efflux from the lysosome and late endosome. To test if the decrease in unesterified cholesterol is dependent on the NPC1 protein, compounds 1 and 2 were evaluated in the filipin assay using NPC1 null HeLa cells to assess cholesterol clearance in the absence of NPC1 protein (Figure G). As expected, 2HPβCD retained the ability to clear unesterified cholesterol, whereas compounds 1 and 2 were unable to clear unesterified cholesterol, indicating that the activity of compounds 1 and 2 is dependent on the presence of NPC1 protein.

To further analyze changes induced by these compounds in patient-derived fibroblasts, we used tandem mass tagging and mass spectrometry to quantify protein expression levels following 24-h compound treatment in I1061T NPC1 fibroblasts. Compound 1 had a minimal impact on protein expression levels in these cells compared to DMSO treatment (Table S2 and Figure S5), and future experiments will be performed to determine the mechanism of this compound. Compound 2 significantly decreased various lysosomal hydrolase levels (Table S3), including the cysteine endopeptidase legumain (LGMN); members of the cathepsin family, such as cathepsin Z (CTSZ), cathepsin B (CTSB), and cathepsin D (CTSD), which are primarily found in the endosomal/lysosomal system and are the most abundant lysosomal proteases; and the lysosomal glycosyl hydrolase, β-hexoaminidase (HexA and HexB) (Figure A). − Changes in protease expression levels induced by compound 2 treatment were validated using Western blotting and tested alongside the late-stage inhibitors CQ and BafA1 (Figure B–F and Figure S6). Compound 2 significantly decreased expression of the cysteine endopeptidase LGMN, CTSZ, CTSB, and CTSD without affecting pro-LGMN or pro-CTSB levels. CQ and BafA1 similarly reduced lysosomal hydrolase levels in I1061T NPC1 fibroblasts, as expected for compounds that affect lysosomal function. Compound 2 likely modulates expression levels of lysosomal enzymes through a different mechanism as it did not have a similar effect on lysosomal degradative capacity compared to CQ and BafA1 in the DQ-BSA assay. LAMP1 expression and glycosylation were also evaluated as LAMP1 binds cholesterol and aids in lysosomal cholesterol export, and the state of LAMP1 glycosylation may have a role in NPC disease. Compound 2 increased levels of glycosylated and nonglycosylated LAMP1 to the greatest degree compared to the two late-stage inhibitors (Figure F and Figure S6E,J). It is possible that excess LAMP1 could also contribute to enhanced cholesterol efflux from the LE/Lys induced by compound 2 through an NPC1-dependent mechanism.

3.

Analysis of proteome changes elicited by compound 2. (A) Fold change of protein expression after 24 h treatment with 2 (80 μM) in I1061T NPC1 fibroblasts compared to the vehicle, DMSO. Bars are labeled with p-values. (B) Left: Representative immunoblot images evaluating levels of pro-LGMN and LGMN with β-actin control after 24-h treatment with DMSO, 2 (80 μM), CQ (10 μM), and BafA1 (0.1 μM) in I1061T NPC1 fibroblasts. Right: Protein levels were normalized to the β-actin loading controls and data quantified from three independent experiments using ImageJ and presented as mean ± SEM. (C−F) follow the same format as (B), except evaluating CTSZ, CTSB, CTSD, and LAMP1 levels, respectively. Significance was determined by unpaired two-tailed t test using GraphPad Prism 10.6.1, where *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. Only significant values (p ≤ 0.05) are indicated. (G) Significantly altered proteins, identified via mass spectrometry, were entered into WebGestalt for relationship mapping. Genes involved in protein catabolic processes are highlighted in red. Protein changes detected in this data set are represented by large circles. Small circles represent top-ranking neighbors assigned by WebGestalt that were not detected to have significant changes. Red boxes indicate proteins that mediate ubiquitination and degradation.

Next, WEB-based Gene SeT AnaLysis Toolkit (WebGestalt) was used to identify affected pathways and potential mechanisms of action based on the acquired proteomic data which confirmed that compound 2 affects metabolic processes, including protein catabolic processes (Figure S7A). , Included among the ten ‘top-ranking neighbor” genes that were not detected to have significant expression changes were three E3 ligases that mediate ubiquitination of proteins, really interesting new gene (RING) finger protein 4 (RNF4), tripartite motif-containing protein 25 (TRIM25), and F-box only protein 6 (FBXO6) (Figure G and Figure S7B). RNF4 is a small ubiquitin-like modifier (SUMO)-targeted ubiquitin ligase (STUbL) that ubiquitinates SUMOylated proteins to facilitate proteasomal degradation and plays an important role in maintaining genomic integrity. TRIM25 is involved in innate immunity by mediating K63-linked ubiquitination of a caspase recruitment domain (CARD) within retinoic acid-inducible gene 1 (RIG-I), a cytosolic pattern recognition receptor (PRR) that recognizes viral RNA and promotes induction of type 1 interferon expression, in addition to promoting proteasomal degradation of protein targets. , FBXO6 is an SKP1-CUL1-F-box protein (SCF) E3 ligase that is involved in the ERAD pathway for misfolded luminal proteins by promoting their ubiquitination and degradation, with particular affinity for high mannose N-linked glycoproteins (Figure G and Figure S7). , Although the expression levels of these proteins did not change, it is possible that directly or indirectly inhibiting the activity or protein–protein interactions of E3 ligases that target mutant NPC1 for degradation could enable misfolded NPC1 protein to escape degradation and remain available to facilitate unesterified cholesterol efflux from lysosomes. Considering previous experiments using an NPC1-deficient mouse model revealed increased amounts of CTS Z, B, and D, the observed changes in protein expression trends after treatment with compound 2 could be especially impactful for ameliorating phenotypes associated with NPC1 mutations. We observed significant decreases in expression levels of these cathepsins following treatment of I1061T NPC1 fibroblasts with compound 2. Furthermore, CTSZ and LGMN expression levels were also significantly elevated in late-stage NPC disease in the Npc1 ^–/– mouse model, so it is possible that compound 2 could reduce the impact of an NPC1 deficit by decreasing expression levels of cathepsins and LGMN, which could alleviate neuronal toxicity. ,

The results from initial mechanistic studies encouraged further evaluation of the effects of compound 2 on proteostasis pathways to determine how modulation of these pathways impacts NPC disease phenotypes. To assess the effects of compound 2 on protein degradation through the ubiquitin-proteasome system (UPS) and autophagy, we measured the levels of several different markers for proteasomal (p21, p27) and autophagic (p62, LC3-II/LC3-I ratio) degradation following 24-h treatment with autophagy modulators, CQ, Rap, and PIK-III, or the proteasome inhibitior, MG132 (Figure A–D and Figure S8). ,− As expected, MG132 treatment caused an accumulation of proteins that are primarily degraded by the proteasome, p21 (Figure A) and p27 (Figure B), but did not cause an accumulation of p62 (Figure C), which is primarily degraded by autophagy. Autophagy inhibitors, CQ and PIK-III, caused a significant accumulation of p62 (Figure C). By contrast, compound 2 did not cause an accumulation of p21, p27, or p62 (Figure A–C), indicating that this compound does not broadly inhibit proteasomal activity and providing further support that it is not inhibiting autophagy.

4.

Evaluation of proteostasis modulation by compound 2. (A–D) Immunoblot analysis of proteasomal and autophagic markers p21 (A), p27 (B), p62 (C), and LC3 (D). I1061T NPC1 fibroblasts were treated with CQ (20 μM), MG132 (10 μM), Rap (0.2 μM), PIK-III (10 μM), or 2 (40 or 80 μM) for 24 h prior to lysis and analysis. Data in (A–D) are presented as mean ± SEM of four independent experiments. (E) Representative immunoblot for glycan digestion assay in I1061T NPC1 fibroblasts. Glycan digestion assays were performed in fibroblasts treated for 24 h with DMSO, HPβCD (1 mM), BafA1 (0.1 μM), or 2 (20 μM). Lysates were subjected to no treatment (NT), Endo H treatment (E), or PNGase treatment (P) for 3 h prior to Western blot analysis. (F) Quantified immunoblots for Endo H-treated samples with bands at 180 kDa (E-Res) and 140 kDa (E-Sen). (G) Quantified immunoblot data for glycan digestion assays performed with WT NPC1 fibroblasts. No treatment (NT), Endo H treatment (E). Data in (F,G) are presented as mean ± SEM of three independent experiments. Significance was determined by unpaired two-tailed t test comparing control (DMSO) E-Sen band intensity with E-Sen band intensity for each of the compounds in (F) and by unpaired two-tailed t test comparing control (DMSO) NT and E band intensity with the NT and E band intensity for compound 2 in (G) using GraphPad Prism 10.6.1. (H) Quantified immunoblot data for ATF4 performed with I1061T NPC1 fibroblasts treated with CQ (40 μM), BafA1 (0.2 μM), MG132 (10 μM) or 2 (40 or 80 μM) for 24 h prior to lysis and analysis. Data are presented as the mean ± SEM of four independent experiments each with duplicate biological replicates. In (A–D) and (H), significance was determined by one-way ANOVA using GraphPad Prism 10.6.1. For all comparisons, *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, and only significant values (p ≤ 0.05) are indicated.

To determine how compound 2 impacts NPC1 expression and trafficking, we performed glycan digestion experiments (Figure E–G and Figure S9). Glycans are added to NPC1 in the ER, and the glycosylated protein is trafficked to the Golgi where the glycans are modified and become resistant to endoglycosidase H (Endo H) digestion but remain sensitive to digestion by peptide-N-glycosidase F (PNGase F). Because I1061T NPC1 is recognized as misfolded in the ER and sent for degradation before reaching the Golgi, its glycans remain sensitive to Endo H digestion. Following treatment with compound 2, we observed a significant increase in the amount of Endo H-sensitive (E-Sen) I1061T NPC1 (Figure E,F and Figure S9A), but there was no increase in WT NPC1 levels in control fibroblasts (Figure G and Figure S9B), indicating that the primary effect is likely prevention of the degradation of the E-sen NPC1 rather than increasing overall NPC1 expression levels to improve cholesterol clearance. In fact, it has previously been shown that increasing the cellular population of E-sen, mutant NPC1 results in late endosomal localization of the mutant protein, likely due to the ability of a proportion of this protein to fold correctly and escape ERAD quality control checkpoints, which promotes clearance of cholesterol from lysosomes despite the trafficking defect.

To determine how compound 2 affects ER stress, the unfolded protein response (UPR), and the ISR, we measured changes in the expression levels of Activating Transcription Factor 4 (ATF4) in I1061T NPC1 fibroblasts following treatment with DMSO control, compound 2, MG132, CQ, or BafA1 (Figure H and Figure S10). ATF4 is upregulated in response to ER stress and activates transcription of adaptive genes to endure periods of cellular stress or promote apoptosis under irremediable stress conditions. ,, In I1061T NPC1 fibroblasts, we observed ATF4 expression with DMSO treatment, indicating that this stress response pathway is activated in these cells. Treatment with late-stage autophagy inhibitors or a proteasomal inhibitor significantly reduced ATF4 protein expression, whereas treatment with compound 2 did not affect ATF4 expression levels compared to the DMSO control. It is possible that maintenance of higher ATF4 levels could enhance the transcription of adaptive genes and production of chaperone proteins to support protein folding. Combined with the activation of autophagy, compound 2 may induce an adaptive response and overall protective effect in cells that are stressed due to accumulation of proteins or protein aggregates.

To explore the structure–activity relationships (SAR) of this scaffold and to identify which moieties are required to maintain activity, we prepared a variety of compound 2 analogues using our optimized synthetic route (Schemes S2 and S3). Identification of clear SAR trends that are dependent on specific structural modifications and not just chemical properties, like lipophilicity, provide support that changes in activity are due to molecular interactions with a specific target(s), rather than lysosomotropic effects. In addition, information obtained through SAR studies facilitate medicinal chemistry optimization efforts in preparation for in vivo efficacy studies. These initial sets of analogues modified the piperazine (R₁) position and the amide (R₂) position to assess the importance of these regions (Figure and Figures S11 and S12). Even conservative modifications of the amino substituent at R₂ resulted in a significant reduction or complete loss of autophagy activation activity (Figure and Figure S11), so we shifted our focus to the R₁ position. Modification of R₁ from methyl piperazine to ethyl piperazine, methylsulfonyl piperazine, diethyl amine, or pyrrolidine significantly reduced or completely abolished activity. However, the propyl-substituted piperazine (AA2123) displayed slightly improved activity in the autophagy assay and similar activity to compound 2, in the filipin assay (Figure and Figures S11 and S12). Replacement of the propyl group with a larger, branched sec-butyl group (AA2124) further improved activity in both assays (Figure and Figures S11 and S12). These results reveal that even subtle changes in the structure of compound 2 can improve activity, indicating that modulation of key interactions with a specific target(s), rather than modulation of lysosomotropic effects, is likely responsible for observed changes in activity. The encouraging information obtained from these initial SAR studies along with our established, efficient synthetic strategies will facilitate further optimization of potency, chemical properties, and stability to enable in vivo evaluation.

5.

Activity of compound 2 analogues in eGFP-LC3 assay and filipin assay. Average puncta/cell values for the eGFP-LC3 assay were calculated from an 8-point dose curve (0.63–80 μM) after 4 h compound treatment (Figure S8). eGFP-LC3 assay data are presented as mean ± SEM of three independent experiments each with duplicate biological replicates. Average pit integrated intensity values for the filipin assay were calculated from an 8-point dose curve (0.63–80 μM) after 24 h compound treatment (Figure S9). Filipin assay data are presented as mean ± SEM of three independent experiments each with duplicate biological replicates.

Conclusions

We identified 312 autophagy modulators using an eGFP-LC3 HTS, and two of these modulators significantly reduced unesterified cholesterol accumulation in NPC patient-derived fibroblasts and were prioritized for further evaluation. The two modulators were unable to clear cholesterol in NPC1 null cells, indicating a requirement for NPC1 for the observed activity. Although compound 1 had a profile consistent with an autophagy inducer, there were not many significant changes in protein expression levels, so additional experiments are necessary to elucidate the mechanism of action of this compound. Compound 2 induced a significant reduction in the expression of several lysosomal proteases and an increase in the expression levels of glycosylated LAMP1 and I1061T NPC1 proteins. Pathway analysis confirmed that this molecule has significant impacts on protein catabolic processes and revealed E3 ligases that could contribute to the observed phenotypic outcomes. The mechanism of action of compound 2 is clearly distinct from the mechanism of 2HPβCD, because 2HPβCD does not increase I1061T NPC1 protein expression levels, and its activity is not dependent on the presence of NPC1 protein. Additional mechanistic studies were performed, and we found that compound 2 does not broadly inhibit proteasomal or autophagic degradation. Glycan digestion assays revealed that compound 2 specifically increases expression levels of Endo H-sensitive I1061T NPC1, which would typically misfold and be recognized for degradation through ERAD. While the inability to remove misfolded mutant NPC1 protein through proteasomal degradation could increase ER stress, we did not observe an increase in ATF4 expression, as is typically observed upon induction of ER stress. Taken together, these studies support the hypothesis that compound 2 may serve as a proteostasis modulator or small-molecule chaperone that effectively increases the proportion of properly folded and localized mutant NPC1 to ameliorate cholesterol accumulation and upregulates autophagic flux to a level that does not induce cytotoxicity and may predominantly help restore homeostasis. Future work will focus on further evaluation of the targets of compounds 1 and 2 and continued development of proteostasis modulators that could have broad therapeutic benefits and applications in a wide range of rare diseases caused by mutations that impact protein folding and stability.

Materials and Methods

Control Compounds

DMSO (Corning, 25–950-CQC), CQ (Sigma-Aldrich, C6628), Rap (LC Laboratories, R5000), BafA1 (LC Laboratories, B-1080) were purchased from commercial vendors. 2HPβCD was provided by Roquette (Kleptose HPB).

Human Subjects and Cell Lines

Fibroblasts from NPC patients (I1061T NPC1) were obtained from individuals enrolled in NIH protocol 06-CH-0186 (NICHD, PI: F.D. Porter) and Rush University Medical Center (RUMC) ORA L05040701 (RUMC, PI: Elizabeth Berry Kravis). Written informed consent was obtained from guardians, and assent was obtained when applicable. Clinical protocols were approved by the NICHD Institutional Review Board with continuing review provided by the NIH clinical Center IRB and the RUMC Institutional Review Board, respectively. NPC1 mutant fibroblasts and control human fibroblasts were obtained from skin biopsy specimens.

eGFP-LC3 Assay

HeLa cells stably expressing LC3 labeled with green fluorescent protein were a gift from Ramnik Xavier from Massachusetts General Hospital. 50 μL of eGFP-LC3 HeLa cells were plated at a cell density of 3,000 cells/well in a black 384-well plate (Corning, 3764) in DMEM (Corning, 15–013-CV) supplemented with 10% FBS (Sigma-Aldrich, 12306C), 1x pen-strep (Corning, 30–002-Cl), and 1.0% l-glutamine (Corning, 25–005-Cl), and incubated overnight at 37 °C and 5% CO₂. After 24 h, a Biomek NX^P liquid handler (Beckman-Coulter) was used to transfer compounds from a compound plate into the 384-well assay plate using a 96-well pin tool head (V&P Scientific), where compounds had a final concentration of 20 μM, and controls included CQ (20 μM), Rap (4 μM), and 2HPβCD (1 mM). The assay plate was incubated at 37 °C and 5% CO₂ for 4 h, and then the MultiFlo FX (BioTek) was used to aspirate the media and add 25 μL of 4% paraformaldehyde (PFA) (Electron Microscopy Sciences, 15710) to each well to fix the cells. The cells were incubated in the PFA for 12 min at RT (rt) in the dark, after which the solution from each well was aspirated and 25 μL of PBS was added and then aspirated to wash the wells. Hoechst 33342 nuclear stain (Thermo, H3570) was diluted in PBS and 25 μL dispensed into each well for a final concentration of 2 μg/mL, and the plate was incubated in the dark at rt for 10 min. The Hoechst solution was aspirated, 25 μL of PBS added to each well, and the plate was sealed using the PlateMax semiautomatic plate sealer (Axygen) and imaged at 10× magnification on the ImageXpress Micro (IXM) XLS automated fluorescent microscope (Molecular Devices) using the DAPI and FITC filters. The images were analyzed using MetaXpress software where an acceptable Z’ (Z factor) for each plate was between 0.4 and 0.6 for cell-based assays, and the percent coefficient of variation (% CV) of less than 20%. A z-score ≥ 2.2 in two replicates was used to identify hits in the HTS, and 312 autophagy modulators were identified and cherry-picked from the ChemDiv library to use in subsequent experiments. Prioritized hits were then synthesized and validated in additional eGFP-LC3 experiments.

Filipin Assay

I1061T NPC1 patient-derived fibroblasts were provided by Forbes Porter (NICHD/NIH) and Elizabeth Berry Kravis (Rush University). Fibroblasts were plated at 35,000 cells/mL in a black 384-well assay plate (PerkinElmer, 6057300) at 50 μL/well using DMEM (Corning, 15–013-CV) supplemented with 20% FBS (Sigma-Aldrich, 12306C), 1x pen-strep (Corning, 30–002-Cl), and 1.0% l-glutamine (Corning, 25–005-Cl) and incubated at 37 °C and 5% CO₂ for 24 h. Cells were then compound treated with cherry-picked compounds (10 μM), Rap (4 μM), and 2HPβCD (1 mM) using the Biomek NX^P liquid handler (Beckman-Coulter) and 96-well pin tool head (V&P Scientific) and incubated for an additional 24 h at 37 °C and 5% CO₂. Media was aspirated from each well using the MultiFlo FX (BioTek) and replaced with 25 μL of 4% PFA for 1 h at rt. After 1 h, the PFA solution was aspirated, and the fibroblasts were rinsed three times with PBS containing 50 mM NH₄Cl. Cells were then stained with 25 μL of 0.05 mg mL^–1 filipin (Sigma-Aldrich, F9765) where the filipin was dissolved in PBS that was mixed with 10% FBS (Sigma-Aldrich, 12306C). The assay plate was then left at rt in a drawer covered in foil for 2 h. To conclude the assay, cells were rinsed three times with PBS without ammonium chloride, 25 μL of PBS dispensed into each well, the plate sealed using the PlateMax semiautomatic plate sealer (Axygen) and imaged immediately using the DAPI filter at 20× magnification with the IXM XLS automated fluorescent microscope (Molecular Devices). The images were analyzed using MetaXpress software and quantified as pit integrated intensity, the total pixel intensity divided by all the pit areas in the image.

Filipin Assay in NPC1 Null HeLa cells

NPC1 null HeLa cells were generated as described previously and plated at 38,000 cell/mL in a black 384-well plate (PerkinElmer, 6057300) in warmed media (same media as eGFP-LC3 assay), and all remaining steps for the assay and analysis were performed as described above with the I1061T NPC1 fibroblast cell line.

mCherry-GFP-LC3 (Dual Reporter) Assay

HeLa cells stably expressing mCherry-GFP-LC3 were a gift from Ramnik Xavier at Massachusetts General Hospital. The dual reporter assay was performed following the same protocol as the eGFP-LC3 assay, except the cells underwent 24 h compound treatment and the plates were imaged using 3 filters: DAPI for Hoechst, FITC for GFP, and Texas Red for mCherry. CellProfiler 3.1.9, rather than MetaXpress, was used to analyze the images by determining the number of autophagosomes and autolysosomes to gain insight into autophagic flux. −

Dye Quenched-Bovine Serum Albumin (DQ-BSA) in HeLa cells

HeLa cells were plated in a black 384-well plate (Corning, 3764) at 2,500 cells/well in 40 μL of warmed media (same media as eGFP-LC3 assay) and incubated overnight at 37 °C and 5% CO₂. The next day, DQ-BSA stock of 1 mg mL^–1 (Invitrogen, D12051) was diluted in warmed media, and 10 μL dispensed into each well for a final concentration of 10 μg/mL. The plate was incubated at 37 °C and 5% CO₂ for 1 h, and then the media was removed using the MultiFlo FX. The wells were washed with 1x PBS twice and 40 μL of media was added. DMSO, 2HPβCD (1 mM), CQ (20 μM), BafA1 (0.1 μM), compound 1 (10 and 20 μM), compound 2 (20 and 40 μM) were pin-transferred into the assay plate using the Biomek NX^P liquid handler (Beckman-Coulter) and incubated at 37 °C and 5% CO₂ for 6 h. Hoechst 33342 was diluted in warmed media and 10 μL of this solution was dispensed into each well using the Multi Flo FX for the last 30 min of incubation for a final concentration of 2 μg/mL. The media was then aspirated using the MultiFlo FX and replaced with 40 μL of 1x HBSS. The plate was sealed using the PlateMax semiautomatic plate sealer (Axygen) and imaged at 20× magnification with the IXM XLS automated fluorescent microscope (Molecular Devices) using the DAPI and Texas Red filters. The images were analyzed using MetaXpress software.

Dye Quenched-Bovine Serum Albumin (DQ-BSA) in I1061T NPC1 Fibroblasts

I1061T NPC1 patient-derived fibroblasts were plated in a 384-well black plate (Corning, 3764) at 28,000 cells/mL using the same media as the filipin assay and incubated overnight at 37 °C and 5% CO₂. The next day, DQ-BSA was added as with the DQ-BSA in HeLa cells and then removed using the MultiFlo FX. The wells were washed twice with PBS and replaced with 40 μL of warmed media. DMSO, 2HPβCD (1 mM), CQ (20 μM), BafA1 (0.1 μM), compound 1 (10 μM), compound 2 (20 μM) were pin-transferred into the assay plate using the Biomek NX^P liquid handler (Beckman-Coulter) and incubated at 37 °C and 5% CO₂ for 24 h. Hoechst 33342 (2 μg/mL) was then used to stain the nuclei and replaced with 40 μL of 1x HBSS. The plate was sealed and imaged as described for the HeLa cells.

Cytotoxicity Assay

I1061T NPC1 patient-derived fibroblasts were plated in a white 384-well plate (Corning, 3765) at 28,000 cells/mL with 50 μL/well and incubated at rt for 1 h, and then placed in the 37 °C incubator in a container containing a water reservoir for 24 h. Cells were then compound treated with DMSO, 2HPβCD (1 mM), CQ (20 μM), BafA1 (0.1 μM), compound 1 (10–80 μM), and compound 2 (10–80 μM) and incubated in 37 °C and 5% CO₂ for 24 h. The next day, the assay plate and CellTiter-Glo 2.0 reagent (Promega, G9242) were both incubated at rt for 30 min, and then 25 μL of CellTiter-Glo 2.0 reagent was added to each well for a total volume of 75 μL/well. The assay plate was covered in foil and placed on the orbital shaker for 10 min at 150 rpm. The luminescence reading was taken on the SpectraMax i3x (Molecular Devices) using the Softmax Pro 6.5.1 software where the integration time was set to 500 ms and the read height set to 6.50 mm from the plate.

Immunoblotting

NPC1 Expression Immunoblots

I1061T NPC1 patient-derived fibroblasts were plated in a 24-well plate (Corning, 353047) at a seeding density of 75,000 cells/well in the same media as the filipin assay and left at rt for 1 h to allow cells to adhere to plate bottom, and then the plate was placed into the incubator at 37 °C and 5% CO₂ for 24 h. DMSO, 2HPβCD (1 mM), compound 1 (10 and 20 μM), and compound 2 (10 and 20 μM) were added to wells by hand and incubated at 37 °C and 5% CO₂ for 24 h. Cells were then lysed using NP-40 Lysis Buffer made in a 15 mL tube comprised of 10 mL of 1xTrisec-Buffered Saline (TBS) (Corning, 46–012-CM), one Pierce protease and phosphatase inhibitor tablet (Thermo Scientific, A32959), and 1% IGEPAL (Sigma-Aldrich, 56741). Lysates were centrifuged (Thermo Scientific, Sorvall Legend Micro 21R) at 18,000 g for 30 min at 4 °C to remove debris, the supernatant was separated by 10% SDS-PAGE (120 V, 1.5 h), and protein was transferred onto 0.45 μm PVDF membrane (EMD Millipore, IPFL00010) using a mini blot module (Invitrogen) at 25 V for 1 h. The membrane was blocked with 5% Blotting-grade Blocker (BioRad, 1706404) in TBS (Corning, 46–012-CM) supplemented with Tween-20 (VWR, 500–018–3) (TBS-T) for 1 h at rt and then incubated overnight with primary antibody for NPC1 (1:1000) and β-actin (1:2000) in 5% Blotting-grade Blocker in TBS-T. The blots were washed three times with TBS-T and incubated with HRP-conjugated secondary antibody in 5% blotting-grade blocker for 1 h at rt. Membranes were washed again with TBS-T and proteins visualized using antirabbit IgG with SuperSignal West Pico Plus Stable Peroxide Solution (Thermo Scientific, 1863097) and luminol/enhancer solution (Thermo Scientific, 1863096) using the c Series Capture Software on the Azure Imaging System.

P70S6K and Phospho-P70S6K Immunoblots

Same as NPC1 immunoblot protocol, except 1% BSA (Sigma-Aldrich, 9048–46–8) was used for the blocking step for the phospho-P70S6K antibody. Additionally, after visualizing the phosphorylated p70S6K (1:500) and β-actin (1:2000) on the Azure Imaging System, the membrane was stripped using stripping buffer (Thermo Scientific, 21059) and reprobed with the nonphosphorylated p70S6K antibody (1:1000).

LGMN, CTS Z, B, D, and LAMP1 Immunoblots

I1061T NPC1 fibroblasts were plated at a density of 132,000 cells/well in a 6-well plate (Corning, 353046) in the same media as the filipin assay and grown for 24 h. Cells were treated with DMSO, compound 2 (80 μM), CQ (10 μM), and BafA1 (0.1 μM) and incubated for 24 h. Lysate preparation, SDS-PAGE separation, transfer, blotting, and imaging were the same as for NPC1 immunoblots, except primary antibodies were used that recognize LGMN, CTSZ, CTSB, CTSD, and LAMP1 and 1% BSA (Sigma-Aldrich, 9048–46–8) was the blocking buffer for CTSZ and CTSD. Membranes were stripped with stripping buffer (Thermo Scientific, 21059) and reprobed for β-actin.

Proteasomal and Autophagic Degradation Immunoblots

I1061T NPC1 fibroblasts were plated at a density of 100,000 cell/well in a 12-well plate (Corning, Cat 353043) in the same media as the filipin assay and grown for 24 h. Cells were treated with DMSO, CQ (20 μM), PIK-III (10 μM), MG-132 (10 μM), Rap (0.1 μM and 0.2 μM), compound 2 (40 μM and 80 μM) and incubated at 37 °C and 5% CO₂ for 24 h. Cells were lysed using Pierce IP Lysis Buffer (Thermo Scientific, 87787) with one Pierce protease and phosphatase inhibitor tablet (Thermo Scientific, PIA32959). Lysate was centrifuged at 12,000 rpm for 30 min at 4 °C to remove debris. Twenty-six μL of lysate was transferred to a 1.5 microcentrifuge tube containing 10 μL of 4x Laemmli Sample Buffer (Bio-Rad, 1610747) and 4 μL of 10x Reducing Agent (Thermo Scientific, B0009), and the solutions were heated at 95 °C for 5 min and separated by SDS-PAGE using 4–15% Mini-PROTEAN TGX Precast Protein Gels (Bio-Rad) for 30 min at 200 V. After separation by SDS-PAGE, proteins were transferred onto a 0.2 μm Nitrocellulose membrane using the Trans-Blot Turbo Transfer System (Bio-Rad). The membrane was blocked with Intercept (TBS) Blocking Buffer (Li-Cor, 927–60001) and incubated for 30 min at rt. Membranes were incubated overnight at 4 °C with primary antibodies for p62, p21, p27, and LC3 (1:1000) and β-actin (1:4000) in Intercept (TBS) Blocking Buffer and then washed three times with 1x PBS spiked with 0.1% Tween-20 (Sigma, P7949) (PBS-T). Membranes were incubated with IRDye 680RD Goat anti-Mouse IgG and IRDye 800CW Goat anti-Rabbit IgG secondary antibodies (1:10,000) in Intercept (TBS) Blocking Buffer for 1 h at rt and then washed three times with PBS-T. Blots were imaged with Odyssey CLx imaging system using 700- and 800 nm channels and visualized using ImageStudio software version 5.2. Images were quantified using ImageJ.

ATF-4 Immunoblots

I1061T NPC1 fibroblasts were plated at a density of 100,000 cell/well in a 12-well plate (Corning, Cat no 353043) in the same media as the filipin assay and grown for 24 h. Cells were treated with DMSO, CQ (40 μM), MG-132 (10 μM), BafA1 (0.2 μM), compound 2 (40 μM and 80 μM) and incubated for 24 h. Cells were then lysed using 1x RIPA (Thermo Scientific, A32959) with one Pierce protease and phosphatase inhibitor tablet (Thermo Scientific, PIA32959). Lysate preparation, SDS-PAGE separation, transfer, blotting, imaging, and image analysis were the same as for proteasomal and autophagic degradation immunoblots, except primary antibodies were used that recognize ATF4 and β-actin loading control.

Tandem Mass Tag and Mass Spectrometry Analysis

Sample Prep

I1061T NPC1 patient-derived fibroblasts were cultured in the same media as the filipin assay at 1,500,000 cells/mL in six 10 cm tissue culture dishes (Corning, 353003) and grown overnight (22–24 h) in 37 °C and 5% CO₂. Three plates were then treated with DMSO and three plates were treated with either compound 1 (40 μM) or compound 2 (80 μM) for 24 h. Cells were then harvested, rinsed three times with 1x PBS (Corning, 21–040-CM), and flash frozen in a 1.5 mL microcentrifuge tube. Cells were resuspended in 70 μL Lysis Buffer (10% SDS in 100 mM TEAB, 1 mM protease inhibitor, and phosphatase inhibitor [10 mM Na₄O₇P₂, 1 mM PMSF, 1 mM Sodium Orthovanadate, 1 mM β-Glycerophosphate disodium salt, 1 mM Sodium Fluoride]) and sonicated (QSonica Sonicators, model CL-18) for 3 sonication cycles where 1 cycle consisted of 10 s (1 s on and 1 s off). After sonication, samples were centrifuged (Eppendorf, Centrifuge 5424 R) at 14,000 rpm for 10 min at 20 °C and the supernatant transferred to a new microcentrifuge tube.

Bicinchoninic acid (BCA) Assay (Thermo Scientific, 23225) was performed in a 96-well plate (Fisher Scientific, 12565501) and analyzed on a VERSA max tunable microplate reader (Molecular Devices) to determine protein concentration in accordance with the manual. 75 μg protein after treatment with compound 1, and 100 μg protein after treatment with compound 2, was transferred into a new microcentrifuge tube, where proteins were reduced with DL-Dithiothreitol (Sigma-Aldrich, D0632) and alkylated with iodoacetamide (Sigma, I1149). The samples were then trypsin digested (Thermo Scientific, 90058) using a micro S-trap spin column (Protifi, C02 mini/micro-80) following provided protocol with minor modification. After loading the acidified protein mixture onto the S-Trap, captured proteins were rinsed 7 times with s-trap buffer, and 1:50 trypsin to protein ratio was used for overnight digestion. After elution of digested peptides, samples were dried down in vacuo (Labconco Acid-resistant CentriVap Concentrator). Samples were then labeled using the TMT6plex Isobaric Reagent Set (Thermo Scientific, 90061) in accordance with the provided protocols. Briefly, 0.8 mg of TMT reagent was resuspended in 41 μL MeCN and added to each of the six samples. The mixture was incubated at rt for 1 h and then quenched with 8 μL 5% hydroxylamine for 15 min. The six samples (three DMSO replicates, and three compound-treated replicates) were combined into one microcentrifuge tube and dried down in vacuo.

A series of MeCN solutions with percentages ranging from 1 to 80% MeCN were prepared. The protein sample was then resuspended via vortexing in 800 μL of 1% MeCN solution and fractionated manually into twelve 1.5 mL microcentrifuge tubes using Oasis HLB cartridge (Waters, WAT094225), nitrogen gas, and the series of MeCN solutions. After fractionation, samples were dried down in vacuo and stored at −20 °C. Samples were then resuspended in 100 μL of 0.1% formic acid (FA) to yield an estimated final concentration of 0.5 μg/μL in each fraction. The samples were then vortexed 40 times with pulsing, centrifuged at 14,000 g for 10 min, and 50 μL supernatant transferred to LCMS vials (Thermo Scientific, C4011–13).

LC-MS Data Acquisition

Approximately 250 ng of protein from each of the twelve samples was then injected into an Agilent 1260 Infinity nanoLC system (Agilent Technologies) coupled with a Q Exactive mass spectrometer (Thermo Fisher Scientific). Samples were loaded onto a Thermo NanoViper trap column (75 μm x 20 mm, 3 μm C18, 100 Å) (Thermo Fisher Scientific) and washed with solvent A (0.1% FA in water) for 10 min at 2 μL/min flow rate and then loaded onto an Agilent Zorbax 300SB-C18 column (0.075 × 150 mm, 3.5 μm 300 Å) at 5% B (0.1% FA in MeCN). Separation was carried out using a 60 min gradient going from 5 to 60% B with a flow rate of 0.25 μL/min. The system was then increased from 60 to 90% B in 0.1 min, then maintained at 90% B for 10 min prior to a 15 min re-equilibration segment at 5% B prior to the next run. Mass spectra were collected using data-dependent acquisition (DDA) with a capillary temperature of 250 °C and spray voltage of 1.5 kV. Full MS scans were collected at a mass resolution of 70,000 with a scan range of 375–1600 m/z. Automatic gain control (AGC) target was set at 1 × 10⁶ for a maximum injection time (IT) of 100 ms. The top ten most intense peaks were selected for MS/MS analysis, with an isolation width of 1.5 m/z. MS/MS spectra were acquired at a resolution of 17,500, ACG target 1 × 10⁵, maximum IT of 50 ms. The first fixed mass was set at 100 m/z. Parent ions were fragmented at a normalized collision energy (NCE) of 27%. Dynamic exclusion was set for 20 s. Parent ions with charges of 1 and larger than 6 were excluded.

Data Analysis

Raw files were analyzed using Thermo Proteome Discoverer 2.3.0.523 (Thermo Fisher Scientific) Using the Sequest HT search engine, data were searched against the UniProt Homo Sapien database (42,368 gene sequence; downloaded June 12, 2019) with 10 ppm precursor mass tolerance. A maximum missed cleavage of 2 by trypsin was set for amino acid sequence between 6 and 144 residues in length. Fragment masses were searched with a tolerance of ± 0.02 Da. Dynamic modifications included Oxidation (+15.995 Da; M), TMT6plex (+229.163 Da; S, T), Deaminated (+0.984 Da; N, Q, R), and Acetylation (+42.011 Da; N-terminus). Carbamidomethylation (+57.021 Da; C) and TMT6plex (+229.163 Da; K, N-terminus) were set as static modifications. Both peptides and PSMs were set to a target false discovery rate (FDR) of ≤ 0.05 for peptide-spectrum matches with moderate confidence and ≤ 0.01 for matches with high confidence. All protein hits contain two or more peptide matches. Quantification was performed using relative abundance of TMT6plex reporter ions intensities, based on signal-to-noise values. Protein ratios were calculated based on group protein abundances. For statistical analysis, t tests were performed based on the abundances of individual proteins.

Glycan Digestion Assays

I1061T NPC1 patient-derived fibroblasts were plated in a 6-well tissue culture treated plate (Corning, 353046) at a density of 200,000 cells/mL with 2 mL of the cell suspension in each well. The cells were left outside of the incubator in the dark at 25 °C for 30 min to allow the cells to settle and begin adhering to the bottom of the plate before incubation at 37 °C and 5% CO₂ for 24 h. Then the cells were treated with DMSO (2 μL), HPβCD (1 mM), BafA1 (0.1 μM), or compound 2 (20 μM) and incubated at 37 °C and 5% CO₂ for 24 h. After 24 h, the cells were lysed with 100 μL of M-PER Mammalian Protein Extraction Reagent (Thermo Scientific, 78501) supplemented with a Pierce protease and phosphatase inhibitor tablet (Thermo Scientific, A32959). Lysates were centrifuged at 18,000 g for 30 min at 4 °C to remove debris and the supernatant was transferred to a clean microcentrifuge tube. Protein concentration was measured using the Pierce BCA Protein Assay Kit (Thermo Scientific, 23225). For the PNGase F, Endo H, and NT (not treated) samples, a 10 μL reaction volume was made by combining 10 μg of glycoprotein, 1 μL of 10x Glycoprotein Denaturing Buffer (NEB, B1704SVIAL) and ultrapure H₂O if necessary to bring total volume to 10 μL. Samples were denatured for 10 min at 100 °C, cooled on ice for 5 min, then briefly spun down. For the PNGase F reaction, 2 μL 10x GlycoBuffer 2 (NEB, B3704SVIAL), 2 μL 10% NP-40 (NEB, B2704SVIAL), and 6 μL H2O were added to the reaction vial to make a final volume of 20 μL. The samples were briefly spun down, then 0.5 μL of PNGaseF stock (NEB, P0704SVIAL) was added to the reaction and incubated at 37 °C for 3 h. For the Endo H and NT samples, 2 μL of 10x GlycoBuffer 3 (NEB, B1720SVIAL), 0.75 μL Endo H (NEB, P0702SVIAL), and ultrapure H₂O were added to the reaction vial to bring total volume to 20 μL. Endo H was not added to the NT reaction vial. These samples were incubated at 37 °C for 3 h. Reactions were stopped by adding 7.7 μL of 4x LDS Sample Buffer (Thermo Scientific, NP0007). These samples were run on a NuPAGE 4 to 12% Bis-Tris Gel (Thermo Scientific, NP0335BOX) following manufacturer’s instructions for gel running buffer and transfer buffer, and protein was transferred onto 0.45 μm PVDF membrane (EMD Millipore, IPFL00010) using a mini blot module (Invitrogen) at 25 V for 1 h. For experiments with WT NPC1 fibroblasts, cells were plated in a 6-well tissue culture plate (Corning, 353046) at a density of 200,000 cells/mL with 2 mL of the cell suspension in each well. The cells were treated with DMSO (2 μL), HPβCD (1 mM), BafA1 (0.2 μM), or compound 2 (20 μM) and lysed after 24 h with 100 μL of M-PER Mammalian Protein Extraction Reagent (Thermo Scientific, 78501) supplemented with a Pierce protease and phosphatase inhibitor tablet (Thermo Scientific, A32959). 10 μg of protein was used for the PNGase F, Endo H, and NT reactions.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acschembio.6c00039.

• Supplemental figures, tables, and synthetic schemes in addition to detailed synthetic methods, compound characterization data, and H¹NMR and C¹³NMR spectra for all isolated, synthetic compounds (PDF)

‡.

M.S. and A.A.S. have equal contribution.

The authors declare the following competing financial interest(s): L.N.A., M.S., A.A.S., and Q.G. are named as inventors on patents owned by UIC and the U.S. government.