The Cytosolic Adaptor AP-1A Is Essential for the Trafficking and Function of Niemann-Pick Type C Proteins


  • Steve Poirier,

    1. Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, Montréal, QC, Canada
    2. Laboratory of Molecular Cell Biology, Montreal Heart Institute, Montréal, QC, Canada
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Gaétan Mayer,

    1. Laboratory of Molecular Cell Biology, Montreal Heart Institute, Montréal, QC, Canada
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Stephanie R. Murphy,

    1. Department of Biochemistry, Dartmouth Medical School, Hanover, NH, USA
    Search for more papers by this author
    • These authors contributed equally to this work.

  • William S. Garver,

    1. Department of Biochemistry and Molecular Biology, The University of New Mexico, Albuquerque, NM, USA
    Search for more papers by this author
  • Ta Yuan Chang,

    1. Department of Biochemistry, Dartmouth Medical School, Hanover, NH, USA
    Search for more papers by this author
  • Peter Schu,

    Corresponding author
    1. Center for Biochemistry and Molecular Cell Biology, Georg-August-Universität Göttingen, Göttingen, Germany
    • Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, Montréal, QC, Canada
    Search for more papers by this author
  • Nabil G. Seidah

    Corresponding author
    • Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, Montréal, QC, Canada
    Search for more papers by this author

Corresponding author: Nabil G. Seidah, and Peter Schu,


Niemann-Pick type C (NPC) disease is a fatal neurodegenerative disorder characterized by over-accumulation of low-density lipoprotein-derived cholesterol and glycosphingolipids in late endosomes/lysosomes (LE/L) throughout the body. Human mutations in either NPC1 or NPC2 genes have been directly associated with impaired cholesterol efflux from LE/L. Independent from its role in cholesterol homeostasis and its NPC2 partner, NPC1 was unexpectedly identified as a critical player controlling intracellular entry of filoviruses such as Ebola. In this study, a yeast three-hybrid system revealed that the NPC1 cytoplasmic tail directly interacts with the clathrin adaptor protein AP-1 via its acidic/di-leucine motif. Consequently, a nonfunctional AP-1A cytosolic complex resulted in a typical NPC-like phenotype mainly due to a direct impairment of NPC1 trafficking to LE/L and a partial secretion of NPC2. Furthermore, the mislocalization of NPC1 was not due to cholesterol accumulation in LE/L, as it was not rescued upon treatment with Mβ-cyclodextrin, which almost completely eliminated intracellular free cholesterol. Our cumulative data demonstrate that the cytosolic clathrin adaptor AP-1A is essential for the lysosomal targeting and function of NPC1 and NPC2.

Low-density lipoproteins (LDL) are mainly endocytosed by cell surface LDL receptor (LDLR), wherein the acidic environment of late endosomes/lysosomes (LE/L) allows the detachment of the complex [1]. In LE/L, cholesterol esters within LDL particles are hydrolyzed by the acid lipase to generate a pool of free cholesterol [2], which is then sequentially transferred to NPC2 and NPC1 allowing the efflux of cholesterol from the LE/L to other cellular compartments such as the trans-Golgi network (TGN), endoplasmic reticulum (ER), plasma membrane and mitochondria [3-9]. Consequently, human mutations at NPC1 or NPC2 loci are associated with a severe and fatal lysosomal storage disorder, commonly known as Niemann-Pick type C disease, characterized by a systemic and abnormal accumulation of lipids primarily due to a defect in LDL-derived cholesterol efflux from lysosomes [10-12]. Recently, a new biological function was assigned to NPC1 in a context that differs from that of cholesterol homeostasis. NPC1 was recognized as a permissive factor for the entry of filoviruses such as Ebola [13-15]. Once cleaved by cysteine proteases, the activated Ebola viral glycoprotein directly binds to the second luminal domain of NPC1, which in turn allows the escape of the virus from lysosomes into the host's cytoplasm.

Both NPC1 and NPC2 are resident lysosomal proteins that must be properly transported to their final destination in order to correctly fulfill their functions [16]. NPC2 is a 151-amino acid (aa) soluble protein that traffics from the TGN to LE/L by the mannose 6-phosphate receptors (MPRs) [17, 18]. In contrast, proper lysosomal targeting of the 1278-aa transmembrane protein NPC1 requires a di-leucine motif at the C-terminus of its cytosolic tail (LLNF1278), a typical recognition sequence for cytosolic adaptor protein (AP) complexes, like the AP-1A [19, 20].

The AP complexes mediate sorting of cargo proteins by binding to membrane lipids and target proteins, providing a scaffold for clathrin assembly on specialized vesicles [20-23]. As compared to the epithelial-specific adaptor AP-1B, which is involved in the basolateral sorting of proteins from recycling endosomes (e.g. the LDLR) in polarized cells [24], the ubiquitous AP-1A complex has been shown to specifically mediate trafficking between the TGN and endosomes [20, 25, 26].

In this study, we addressed the question of whether AP-1A is directly or indirectly implicated in the targeting of NPC1 or NPC2 to lysosomes.


NPC1 directly interacts with the clathrin adaptor AP-1

The AP complexes are assembled as heterotetramers composed of a small (σ), medium (μ) and two large (γ, α, δ, ϵ or ζ and β1–5) subunits for AP-1, -2, -3, -4 or -5, respectively [20, 26-28]. Using a yeast three-hybrid system, it was reported that the hemicomplexes γ1/σ1A, /σ1B or /σ1C (three isoforms) of AP-1 directly interact with the ‘acidic/di-leucine’ [D/E]XXXL[L/I] consensus motif present within cytoplasmic tails of various transmembrane proteins e.g. LIMP-II, HIV-1 Nef and tyrosinase [29, 30]. Although the multi-pass transmembrane glycoprotein NPC1 contains a typical and highly conserved acidic/di-leucine motif at the C-terminus of its 27-aa long cytoplasmic tail (ERERLLNF1278; Figure 1A), the characterization of its binding to AP complexes has not been reported. Accordingly, we generated four constructs with different NPC1 cytoplasmic tails: the full length (WT; aa 1252–1278); a truncated mutant lacking the acidic/di-leucine motif (TR8; ΔERERLLNF1278); a di-leucine motif mutant (AANF; positions 0 and +1 mutated to alanine) and an acidic residue mutant (ARER; glutamic acid at position −4 mutated to alanine). These were fused to the GAL4 DNA binding domain in the pBridge vector, which co-expresses or not the three different σ1 isoforms σ1A, σ1B and σ1C [30, 31], resulting in 16 different constructs. The latter were individually co-expressed in the yeast AH109 strain, with a pGADT7-AD vector containing γ1-adaptin fused to the GAL4 activation domain (γ1+). In this yeast three-hybrid system, the only co-transformants that would survive in restrictive media (−His) are those in which a productive trimeric complex is formed (γ1/σ1/NPC1), thereby activating transcription of the auxotrophic HIS3 gene. In this assay, only the cells expressing the WT full-length NPC1 survived, indicating that its cytoplasmic tail strongly interacts with the γ1/σ1 hemi-complexes (Figure 1B). Moreover, we show that all three σ1 isoforms (ubiquitously expressed A, tissue-specific B and C) exhibit strong complex formation, because they are insensitive to the addition of 2.5 mm 3-amino-1,2,4-triazole (3-AT), a competitive inhibitor of the HIS3-gene product (Figure 1B, middle panel). Finally, deletion of the last 8 aa (TR8), mutations of the di-leucine motif (AANF) or the acidic Glu residue (ARER) completely prevented binding of the cytoplasmic tail of NPC1 to the AP-1 hemicomplexes. We conclude that the integrity of the acidic/di-leucine motif is critical for NPC1 binding to AP-1.

Figure 1.

The acidic/di-leucine motif in the cytoplasmic tail NPC1 directly interacts with the hemicomplex γ1/σ1-adaptin of the clathrin adaptor AP-1. A) Sequence alignment of orthologous NPC1 cytoplasmic tails containing the AP complex recognition motif ([D/E]XXXL[L/I]). Critical residues of the consensus are denoted in bold and fore-grounded by gray boxes in the NPC1 sequence (acidic residue at position −4, di-leucine at 0, +1 and X is any aa). B) Saccharomyces cerevisiae tester strain AH109 were co-transformed with a pBridge vector encoding for both the GAL4 DNA-binding domain fused at MCS-I with NPC1 cytoplasmic tails (WT, GPSVNKAKSCATEERYKGTERERLLNF; TR8, GPSVNKAKSCATEERYKGT; AANF, GPSVNKAKSCATEERYKGTERERAANF; ARER, GPSVNKAKSCATEERYKGTARERLLNF) and in combination with different σ1-adaptin isoforms (σ1A, σ1B and σ1C) cloned into MCS-II for each of the constructs together with the pGADT7 AD vector (fusion of γ1-adaptin to the GAL4 activation domain; γ1 +). Plasmids without σ1 (Ø) served as negative control. Aliquots from liquid-cultures adjusted to OD600 0.05/mL and 50 μL were spotted onto selection media (−His). Plates were incubated at 30°C over 3 days in which direct interactions (NPC1/σ1/γ1) were evidenced by growth. Two concentrations of 3-Amino-1,2,4-triazole (3-AT; 0.2 and 2.5 mm; a competitive inhibitor of the HIS3-gene product) have been tested to titrate and evaluate binding avidities by growth inhibition. Yeast grown in media containing histidine (+His) was used as control (right panel).

Intracellular routing of NPC1 and NPC2 is dependent on AP-1A

It was reported that mouse embryonic fibroblasts (MEFs) derived from μ1A-deficient mice have a nonfunctional remaining trimeric AP-1A complex, which is unable to bind the TGN, resulting in an impaired sorting of cargo molecules such as gp48/MHC-I to endosomes and of MPRs to the TGN [32-34]. Consistent with the role of AP-1A, immunocytochemical analyses revealed that endogenous NPC1 was not detectable in LE/L [lysosome-associated membrane glycoprotein 1 (Lamp1)-positive vesicles; Lamp1 is sorted independently of AP-1A [32]] of μ1A-deficient MEFs (−μ1A; Figure 2A), compared to a cell line stably expressing recombinant μ1A (+μ1A; rescue of the μ1A−/− cell line and thus the AP-1A complex). Because the in house NPC1 polyclonal antibody used here was new, we validated its specificity in mouse and human fibroblasts (HF) by Western blot and immunocytochemistry (Figure S1). Moreover, pre-incubation of the NPC1 antibody with the immunizing peptide (Ctl versus +imm. peptide) readily displaced its LE/L (Lamp1-positive) labeling (+μ1A; Figure 2B). Despite the absence of NPC1 in LE/L of μ1A-deficient MEFs, the residual and diffused labeling of NPC1 was also reduced (Figure 2B). Compared to NPC1, immunocytochemistry did not reveal an evident quantitative mislocalization of NPC2 in LE/L of μ1A-deficient MEFs (−μ1A, Figure 2C).

Figure 2.

Lysosomal trafficking of NPC1 is impaired in μ1A-deficient cells. A–C) MEFs (+μ1A and –μ1A) were plated on 35 mm glass bottom dishes in complete media. Forty-eight hours later, the subcellular localization of NPC1 (A,B), NPC2 (C) and Lamp-1 (A–C) was analyzed by confocal microscopy. Inserts are numerical zoom of selected regions. B) Using the immunizing peptide (aa 1256–1274; NKAKSCATEERYKGTERER), we show that NPC1 labeling in LE/L of +μ1A cells (Ctl versus +imm. peptide, Lamp-1-positive; top panel) and the diffuse distribution in –μ1A is specific (Ctl versus +imm. peptide; bottom panel). Note that NPC1 was not present in lysosomes of μ1A-deficient cells as compared to +μ1A (emphasized by high magnification, Mag; right panels). Arrows point to NPC1 or NPC2 immunoreactivity. The results are representative of four independent experiments. Scale bar = 20 µm.

To support and extend the above observations, we performed subcellular fractionation of MEFs (+μ1A and –μ1A) followed by Western blotting. Using specific organelle markers, this technique allowed a good separation of endosomes/lysosomes (Endo/Lys), TGN and ER/mitochondria (Figure 3A). The absence of functional AP-1A did not result in any gross alteration in the density distribution of the various organelles, while Western blot analyses revealed that NPC1 and to a lesser extent, NPC2 are redistributed from lighter (Endo/Lys) to denser fractions (Figure 3B) in μ1A-deficient cells (−μ1A). These data agree with the absence of NPC1 in lysosomes and its wider subcellular distribution in cells lacking AP-1A (Figure 2B). However, subcellular fractionation showed that NPC2 is present at a reduced level in Endo/Lys fractions in addition to be more widely scattered in μ1A-deficient cells, a phenotype that was not evident from confocal microscopy. In order to further characterize the pronounced subcellular redistribution of NPC1 in the absence of functional AP-1A, both MEFs (+μ1A and –μ1A) were transiently transfected with either an ER-retained catalytically inactive site mutant of human PCSK9 [PCSK9-H226A-V5; [35]] or with the early endosomal-associated protein EGFP-Rab5 [36]. While NPC1 was mostly localized in LE/L in +μ1A MEFs (Figure 2A,B and 3C,D), the resulting distribution of NPC1 in punctate cytosolic vesicular structures in μ1A-deficient cells did not reveal co-localization with either the ER (PCSK9-H226A-V5) and/or early endosomes (EGFP-Rab5; Figure 3C,D; –μ1A) markers, a localization similar to that reported in these cells for γ-adaptin and clathrin heavy chain [32].

Figure 3.

Subcellular distribution of NPC1 and NPC2 in MEFs. A) MEFs (+μ1A and –μ1A) were washed, homogenized (20 mm Tris, 250 mm sucrose, 1 mm EDTA and protease inhibitors, pH 7.4) and overlaid on a 5–25% OptiPrep gradient and centrifuged at 200 000 × g for 3 h at 4°C. After centrifugation, 15 fractions (top to bottom) were collected and equal portion of each fraction was subjected to SDS-PAGE and Western blotting with specific organelle markers: from left (lighter) to right (denser), fractions are delineated as follows: Endo/Lys, endosomes/lysosomes; TGN, trans-Golgi network; ER/Mito, ER/mitochondria (B). As described above, NPC1 and NPC2 subcellular distribution was analyzed by Western blotting. C) MEFs (+μ1A and –μ1A) were transiently transfected with plasmids encoding either human V5-tagged PCSK9-H226A (ER-retained active site mutant) or with the chimera pEGFP-C1-Rab5 herein used as ER or early endosome (EE) markers, respectively. Twenty-four hours post-transfection, the localization of NPC1 together with subcellular markers was analyzed by confocal microscopy. Note that in μ1A-deficient MEFs, NPC1 immunoreactivity is widely diffuse in cytoplasmic vesicular-like structures that do not represent ER or EE. Arrows point to NPC1 immunoreactivity. The results are representative of two to three independent experiments. Scale bar = 20 µm.

We next wanted to determine if the absence of functional AP-1A could affect NPC1 and/or NPC2 protein levels or degradation. Quantitative PCR and Western blot analyses of immunoprecipitated NPC1 revealed that AP-1A did not affect its total protein levels (Figure S2A,B). Likewise, NPC1 was insensitive to proteasome inhibition by MG-132 in both MEFs (+μ1A and –μ1A; Figure S2B). Surprisingly, while mRNA levels of NPC2 were not significantly increased (Figure S2B), the absence of functional AP-1A complex enhanced its intracellular protein levels, but also resulted in its secretion into the media (−μ1A; Figure S2C). Thus, these data demonstrate that AP-1A is directly involved in the lysosomal trafficking of NPC1, and indirectly affects that of NPC2, most likely due to the rerouting of the MPRs [17, 32].

Lysosomal cholesterol efflux is impaired in μ1A-adaptin deficient cells

Because it was clearly demonstrated that both NPC1 and NPC2 cooperate within the LE/L [3, 37], we reasoned that cholesterol metabolism might be severely disturbed in the absence of AP-1A. By filipin staining, we first noticed that μ1A-deficient cells markedly accumulate intracellular cholesterol (Figure 4A) in LE/L (Figure 5B). In a situation of compromised cholesterol efflux from LE/L to the ER, the expression of cholesterogenic genes is expected to be activated by the transcription factor SREBP-2 cascade [4, 38]. In agreement, quantitative RT-PCR and Western blot analyses revealed that 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase (rate-limiting ER enzyme for cholesterol biosynthesis; HMGCR, Figure 4B,D) and LDLR (Figure 4C,E) are upregulated in the absence of functional AP-1A, suggesting a reduced flux of cholesterol toward the ER.

Figure 4.

The absence of AP-1A complex results in a typical Niemann-Pick type C phenotype. A) Forty-eight hours after plating, MEFs (+μ1A and –μ1A) were stained with filipin III. Scale bar = 20 µm. B–E) Relative mRNA and protein levels of cholesterogenic genes (HMGCR and LDLR) were analyzed by: (B,C) quantitative RT-PCR and (D,E) Western blotting. F–H) Prior to subcellular fractionation as defined in Figure 3, cells were pulse-labeled with 100 µg/mL of 3H-CL-LDL for 30 min and chased in unlabeled medium for (F) 0 min, (G) 30 min or (H) 60 min. Each subcellular fraction was subjected to Folch lipid extraction and separated on AgNO3 impregnated silica gel TLC plates with cholesterol standards. Cholesterol bands were scraped and scintillation counted. Data are the mean of two independent experiments for the 30 and 60 min time points and three independent experiments for 0 min time points; error bars represent the SEM. * = p < 0.02.

Figure 5.

Mislocalization of NPC1 in μ1A-defficient cells is not due to cholesterol accumulation in lysosomes. A) Twenty-four hours after plating, +μ1A MEFs were incubated overnight with 2 µm of U18666A and (B) μ1A-deficient cells (−μ1A) incubated for 2 h with 10 mm MβCD. Intracellular cholesterol accumulation was revealed by filipin staining and subcellular localization of NPC1 and Lamp-1 was analyzed by confocal microscopy. Selected regions (dotted squares) are shown at higher magnification (Mag, right panels). Note the NPC1 labeling in ring-like structures in Lamp1-positive vesicles (see arrows). Arrows point to Lamp-1-positive vesicles. The results are representative of three independent experiments. Scale bar = 20 µm.

To validate this hypothesis, we decided to directly measure the extent of LDL-derived cholesterol efflux from LE/L in the absence or presence of AP-1A. Efficient clathrin-mediated endocytosis of the LDLR requires the AP-2 complex and the cytosolic adaptor ARH [39]. As expected from the exclusive role of AP-1A at the TGN and endosomes, μ1A-deficiency enhanced the receptor-mediated LDL endocytosis (Figure S3A), which correlates with the observed upregulation of LDLR levels (Figure 4C,E). From immunocytochemistry (Figure S3B) and Western blot analyses (Figure S3C,D), we also noticed that cellular incubations with alkalinizing agents (ammonium chloride; NH4Cl or Bafilomycin A1) further increased LDLR levels in both +μ1A and –μ1A MEFs. This indicates that LDLR trafficking and lysosomal turnover were not grossly affected by the absence of AP-1A.

To elucidate the major metabolic defect present in μ1A-deficient cells, we decided to study the intracellular fate of endocytosed LDL-derived cholesterol by pulse-chase analyses, followed by subcellular fractionation [8]. MEFs (+μ1A and –μ1A) were incubated with 3H-cholesteryl linoleate (3H-CL-LDL) and the fate of radiolabeled cholesterol derivatives were subsequently analyzed by thin-layer chromatography. Cells were pulsed for 30 min, chased in unlabeled media for 0, 30 or 60 min, cooled to 4°C, lysed and fractionated by density gradient centrifugation. Lipids were then extracted, separated by thin-layer chromatography, counted by scintillation and expressed as percentage of total cholesterol in each density fraction (Figure 4F–H). After the initial 30 min of pulse, the intracellular distribution of cholesterol was similar between the two cell lines (Figure 4F). Unfortunately, we were unable to measure appreciable changes in the rate of cholesterol re-esterification in the ER, most probably due to low levels of acyl-CoA:cholesterol acyltransferase (ACAT) in both cell lines (Figure S4). However, after 30 and 60 min of chase, the absence of AP-1A resulted in an accumulation of radiolabeled cholesterol in lighter density fractions mostly consisting of endosomal-like compartments (Endo/Lys; fractions 1–6, Figure 4G,H). In contrast, less cholesterol was found in heavier fractions, indicating a defect in redistribution of cholesterol from LE/L to denser cellular compartments (TGN, PM, ER/mitochondria; fractions 7–15). At 60 min of chase time, this difference is even more statistically significant in several of the higher density fractions (Figure 4H).

In summary, our data show that the absence of functional AP-1A causes a typical NPC-like phenotype, characterized by an intracellular accumulation of LDL-derived cholesterol and the loss of SREBP-2 negative feedback, both due to an impaired lysosomal efflux normally orchestrated by NPC1 and NPC2.

Cholesterol sequestration is not responsible for NPC1 mistargeting

We next decided to delineate the direct involvement of NPC1 trafficking defect with the NPC-like phenotype caused by the absence of AP-1A. It was shown that late endosomal cholesterol accumulation could affect the biological function of certain proteins by directly altering their intracellular targeting [40-42], which can be reversed by treatment with cyclodextrin (CD) [43]. Methyl-β- (MβCD) and hydroxypropyl-β-cyclodextrin (HPCD) are water-soluble molecules that have the ability to sequester cholesterol both in vitro and in vivo and have been shown to efficiently correct liver and neuronal defects in Npc1−/− mice and significantly increase life expectancy [44, 45]. Therefore, we first induced a NPC-like phenotype by treating +μ1A cells with U18666A, an intracellular cholesterol transport inhibitor [46, 47]. Overnight incubation with 2 µm U18666A caused a significant enlargement of LE/L compartments (Lamp1-positive vesicles) mainly generated by intracellular accumulation of cholesterol (filipin staining; Figure 5A). By immunocytochemistry, we observed that NPC1 LE/L localization and likely its trafficking were not overly disturbed following U18666A treatment (Figure 5A). Conversely, we decided to reverse the cholesterol accumulation by treating μ1A-deficient cells with 10 mm MβCD (Figure 5B). After 2 h incubation, we noticed that, without any obvious effect on global organelle morphology, MβCD normalized the size of LE/L (Lamp1-positive vesicles) by substantially reversing cholesterol accumulation, and effectively rescuing the NPC-like cholesterol accumulation to one comparable to +μ1A cells (Figure 5A; Ctl). However, cyclodextrin treatment could not correct the NPC1 mislocalization in LE/L (absence of NPC1 in Lamp1-positive vesicles; Figure 5B). Therefore, we conclude that lysosomal cholesterol accumulation was not responsible for the mislocalization of NPC1 in μ1A-deficient cells, but is rather a consequent phenotype.


To our knowledge, this study provides the first evidence that a fully functional AP-1A complex is important to maintain intracellular cholesterol homeostasis by directly targeting NPC1 and NPC2 to LE/L compartments. More precisely, we showed that deficiency of the μ1A-adaptin subunit results in a typical NPC-like phenotype, characterized by an intracellular accumulation of cholesterol (Figure 4A) and increased expression of SREBP-2 target genes (Figure 4B–E), mainly due to a defect in LDL-derived cholesterol efflux from LE/L (Figure 4F–H). Our cumulative data indicated that such impairment in cholesterol trafficking is a consequence of the mislocalization of NPC1 into LE/L, in addition to aberrant secretion of NPC2 (Figures 2, 3 and S2), likely due to rerouting of MPRs [11, 17, 32].

Yeast three-hybrid experiments clearly indicated that the acidic/di-leucine motif in the NPC1 cytoplasmic tail (aa 1271–1278, ERERLLNF) directly interacts with the γ1/σ1 hemicomplex (constitutive components of AP-1A hetero-tetramer; Figure 1), consistent with its essential role in lysosomal targeting [19, 48]. In an earlier study, it has been proposed that the adaptor-protein complex AP-3 could be implicated in the sorting of NPC proteins [49]. It is important to mention that AP-1A and AP-3A mediate diverse steps in lysosomal targeting [34] and that AP-3 behaves differently in yeast versus mammalian cells [50]. In contrast to AP-1A-deficient cells (this work), AP-3-deficient MEFs showed a decrease, rather than an increase, in cholesterol content. This suggests that AP-1A and AP-3 may have distinct effects on cholesterol metabolism and/or possibly NPC trafficking. Therefore, it would be interesting to study a possible complementary effect of AP-3 on AP-1A using doubly deficient cells. AP-1A-dependence does not exclude that NPC1 is recognized by other AP-complexes mediating sorting in the endosomal pathways. However, all the others fulfill tissue- and/or subcellular-specific functions. In contrast, NPC1/NPC2 functions are required in all cell types and sorting by AP-1A ensures proper targeting independent of other pathways. It might very well be that tissue-dependent specializations of endo-lysosomal pathways also lead to slight modifications of the NPC1/NPC2's trafficking pathways and/or kinetics. This would have to be tested in the respective tissues. AP-1A is ubiquitously expressed and its deficiency is embryonic lethal in contrast to AP-3, -4 and -5. Thus AP-1A is definitely the critical sorting factor in all tissues.

AP-1A directly mediates the vesicular sorting of cargo molecules in a bi-directional fashion, between the TGN and endosomes [25, 32, 34, 51, 52]. Although it was shown that cholesterol sequentially binds NPC2 and then NPC1 [3], and that the latter directly promote lysosomal cholesterol efflux [5], the exact mechanism remains to be determined. Urano et al. recently demonstrated that newly liberated cholesterol from endocytosed LDL particles in LE/L transit through the TGN before reaching the ER and/or the PM [8, 9]. Moreover, it has been shown that while NPC1 is a resident lysosomal protein, it is also found to be transiently associated with the TGN [22]. One possible explanation for this retrograde transport is that once bound to NPC1, cholesterol could be transported directly from endosomal compartments to the TGN by AP-1A concomitant to the anterograde route deduced from this work (schematized in Figure 6).

Figure 6.

Proposed model for NPC1 and NPC2 intracellular trafficking to lysosomes. Cells acquire mainly cholesterol from circulating LDL particles via LDLR-mediated endocytosis. The receptor is internalized in clathrin-coated vesicles (CCVs) with the help of AP complex AP-2 and the cytosolic adaptor ARH [39]. Within the endosomal compartments (LE/L), LDL particles are released and apolipoprotein B100 is catabolized into amino acids and cholesterol esters hydrolyzed by the lysosomal acid lipase (AL) [63, 64]. Cholesterol (yellow circles) is then sequentially transferred from NPC2 to NPC1 protein and effluxed to other cellular compartments such as the TGN, plasma membrane, mitochondria and ER where it exerts a negative feedback on the SREBP machinery [3, 4, 8, 65]. Inasmuch as the presence of both NPC2 and NPC1 within the LE/L compartments is critical, they depend on the proper coordination of trafficking components. It was shown that the soluble NPC2 protein is targeted to the LE/L via MPRs [11, 17], whereas NPC1 traffics directly by the AP-1A complex (this work). It was recently demonstrated that the retrograde vesicular trafficking is an important route by which a portion of LDL-derived cholesterol travels through the cell [8]. Thus, it is likely that it could directly involve the transport of NPC1 mediated by the adaptor AP-1A (dotted line).

Niemann-Pick type C disease is a lethal, autosomal recessive, neurovisceral disorder that is associated with gene mutations at the NPC1 [∼95%; [10]] or NPC2 [∼5%; [11]] loci. To date, no human mutation has been reported for the acidic/di-leucine motif in NPC1 cytoplasmic tail that should affect its binding to AP-1A [53]. On the other hand, targeted disruption of either μ1A or γ-adaptin led to embryonic lethality in mice [32, 54] and to impaired zebrafish development due to μ1-deficiency [55]. While NPC1 is not essential during development in mice [56] and human [10], it was recently demonstrated that NPC1-dependent cholesterol availability is crucial in zebrafish [57]. Thus, the requirement for NPC1 may contribute to the developmental defects of μ1A and γ-adaptin knockout mice and μ1-deficient zebrafish.

Lysosomal storage diseases are human genetic disorders, typically caused by deficiency of a single lysosomal protein, which confers severe health consequences and often early death [58]. In some cases the specific molecular lesion cannot be readily identified. This work shows that lacking functional AP-1A does not perturb the intrinsic activities of NPC1 and NPC2, but instead affects their targeting to the LE/L compartments. Absence of functional AP-1A could also play a role in other lysosomal lipid storage disorders. For example, it has been shown that the lysosomal integral membrane protein type 2 (LIMP-II) binds to the γ1/σ1 hemi-complex (AP-1) [30]. This represents a MPR-independent route by which the lysosomal hydrolase β-glucocerebrosidase, implicated in Gaucher disease [59], traffics to lysosomes. The data presented in this work should lead to future studies designed to better define the targeting routes of enzymes causing lysosomal storage diseases.

Materials and Methods

Chemicals and plasmids

Low-density lipoprotein coupled with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocya-nine perchlorate (DiI-LDL) was from Biomedical technologies. Filipin III, U18666A, methyl-β-cyclodextrin (MβCD), ammonium chloride (NH4Cl) and Bafilomycin A1 were purchased from Sigma-Aldrich and TO-PRO3 Iodide from Molecular Probes. The proteasome inhibitor MG-132 was obtained from EMD Millipore. pCMV-HA-tagged-Ubiquitin (HA-Ub) plasmid was kindly provided by Dr Meloche (Université de Montréal) and pEGFP-C1-Rab5 by Dr Cohen (Clinical Research Institute of Montreal). Catalytically inactive V5-tagged human proprotein convertase subtilisin/kexin 9 (PCSK9-H226A-V5) was subcloned into pIRES2-EGFP vector [35].

Yeast three-hybrid assay

Human NPC1 cytoplasmic tails were subcloned as EcoRI/BamHI fragments into MCS-I of pBridge vector (GAL4-BD, Clontech Laboratories) using human pSG5-NPC1 plasmid generously provided by Dr Ioannou [60]. σ1-adaptins (σ1A, σ1B and σ1C) were cloned into MCS-II in each of the constructs. Plasmids without σ1 served as control. γ1-adaptin was cloned as a EcoRI/BamHI fragment into pGADT7 (GAL4-AD, Clontech Laboratories). AH109 tester strain (Clontech laboratories) was transformed with corresponding NPC1 cytoplasmic tail/σ together with the γ1-adaptin construct. Aliquots from liquid-cultures adjusted to OD600 0.05/mL and 50 μL were spotted onto selection media. Plates were incubated at 30°C over 3 days in selected media.

Cell culture and transfection

MEFs (μ1A-deficient, –μ1A) were established from day 11.5 μ1A−/− embryos by continuously passaging the cells. A cell line stably expressing recombinant μ1A (resulting in a rescued AP-1A complex) was derived from μ1A-deficient MEFs and was used as a control [+μ1A; [32]]. Before their use in experiments, MEFs were grown to 95% confluence in DMEM (Gibco) supplemented with 10% FBS (Wisent) at least 15 days after thawing and were passaged <10 times. For drug treatments, 24 h after plating, MEFs were incubated overnight with 2 µm of U18666A [46] or 2 h with 10 mm of MβCD [44]. Intracellular alkalinization was achieved after overnight incubation with either 10 mm of NH4Cl or 20 nm of Bafilomycin A1. For the qualitative assessment of free cholesterol, cells were incubated with 50 µg/mL of filipin III diluted in PBS for 1 h at room temperature (RT) or overnight at 4°C and analyzed by confocal microscopy. Low-density lipoprotein receptor (LDLR) endocytic activities was evaluated by incubating the cells for 4 h with 5 µg/mL of 1,1′-dioctadecyl-3,3,3′,3′-tetramethyl-indocarbocyanine perchlorate (DiI-LDL) and analyzed by immunocytochemistry. MEFs were transfected with selected plasmids using Lipofectamine 2000 (Invitrogen) according the manufacturer's instructions.

Subcellular fractionation

MEFs (−μ1A and +μ1A) were grown to 95% confluence in DMEM supplemented with 10% FBS, washed, harvested in cold homogenization buffer (20 mm Tris, 250 mm sucrose, 1 mm EDTA and protease inhibitors, pH 7.4) and homogenized in a hand held stainless steel homogenizer on ice. The post-nuclear supernatant of the cell homogenate was overlaid on a 5–25% OptiPrep gradient medium (Axis-Shield) and centrifuged at 200 000 × g for 3 h at 4°C. Fifteen equal fractions were collected, starting from the top of the tube, of which an equal portion of each fraction was either subjected to Western blot analysis or had lipids extracted by the Folch method and subjected to TLC and scintillation counting.

3H-cholesteryl linoleate LDL pulse-chase

As described previously (8), cells were pulse-labeled with 100 µg/mL of 3H-cholesteryl linoleate labeled human LDL for 30 min and chased in unlabeled media for various times. A control well, containing medium but no cells, was also pulse-labeled for 30 min and chased. At the end of the chase period, the cells were subjected to subcellular fractionation or harvested in 0.2 M NaOH, neutralized and subjected to Folch lipid extraction. The lipids were spotted on AgNO3 impregnated silica gel TLC plates and run in a solvent system of 2:1 benzene:hexanes with cholesterol, cholesteryl linoleate and cholesteryl oleate standards. The lipids were visualized with 2,7-dicholorofluorescine under a UV light and the respective cholesterol, cholesteryl linoleate and cholesteryl oleate bands were scraped off the TLC plate and scintillation counted.


Forty-eight hours after plating, cells were washed three times with PBS, fixed with 4% paraformaldehyde for 15 min at RT or with methanol (−20°C/5 min, for NPC2), permeabilized with 0.1% Triton X-100/PBS or with 0.5% saponin (for NPC1, Lamp1 or V5-tag) for 10 min and incubated with 150 mm glycine to stabilize the aldehydes. The cells were then incubated for 30 min with 1% BSA (Fraction V, Sigma-Aldrich) containing 0.1% Triton X-100, followed by overnight incubation at 4°C with selected antibodies: rabbit anti-NPC1 (1:250; Figure S1), rabbit anti-NPC2 (1:25; Santa Cruz, sc-33776), rat anti-Lamp1 (1:100; 1D4B developed by August, J.T. was obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA 52242) or goat anti-mouse LDLR (1:100; R&D Systems, AF2255). Afterward, the cells were incubated for 60 min with corresponding Alexa Fluor-conjugated secondary antibodies (Molecular Probes) and mounted in 90% glycerol containing 5% 1,4-diazabicyclo[2.2.2]octane (DABCO, Sigma). Nuclei were stained with TO-PRO3 Iodide (1/1000, Molecular Probes). Immunofluorescence analyses were performed with a Zeiss LSM 710 or Olympus Fluoview FV10i confocal microscope. The specificity of NPC1 staining was confirmed by pre-adsorption of the antibody with the immunizing peptide (human NPC1 aa 1256–1274; NKAKSCATEERYKGTERER; Invitrogen). The peptide was resuspended at a final concentration of 1 mg/mL in 0.05% acetic acid of which 2 µg was mixed with 3 µg/mL of the affinity-purified NPC1 antibody and rotated overnight at 4°C in a final volume of 1 mL of PBS containing 1% BSA. Then, the complex was added on cells and processed for immunocytochemistry has described above.

Western blot analysis

Cells were washed three times in PBS and lysed in complete radio-immune precipitation assay (RIPA) buffer [50 mm Tris/HCl, pH 8.0, 1% (v/v) Nonidet P40, 0.5% sodium deoxycholate, 150 mm NaCl and 0.1% (v/v) SDS] supplemented with 1× Complete Protease Inhibitor Mixture (Roche Applied Science). Proteins were separated by 10% SDS-polyacrylamide gel electrophoresis and blotted on HyBond nitrocellulose membranes (GE Healthcare), which were blocked for 1 h in Tris-Buffered Saline-Tween 20 (TBS-T, 50 mm Tris–HCl, pH 7.5, 150 mm NaCl, 0.1% Tween 20) containing 5% nonfat dry milk. Membranes were then incubated overnight in 1% nonfat milk with the respective antibodies goat anti-mouse LDLR (1:1000; R&D Systems), rabbit anti-NPC1 (1:2500) or anti-β-actin (1:5000; Sigma-Aldrich, A2066). Monoclonal antibody against hydroxy-3-methyl-glutaryl coenzyme A reductase (HMGCR) was obtained from supernatants of a hybridoma cell line (IgG-A9, CRL-1811) from American Type Culture Collection [61]. For total NPC1 protein levels detection, confluent 60 mm2 MEFs were lysed in 1 mL RIPA buffer and immunoprecipitated with NPC1 antibody (1:250) together with 50 μL of protein A/G PLUS-Agarose (Santa Cruz) and rotated overnight at 4°C. Following incubation, samples were centrifuged at for 1 min at 5000 × g and were washed six times with 1 mL RIPA buffer. Pellets were resuspended in 75 μL of 2× Laemmli sample buffer and subjected to Western blot analysis. For NPC2, media were aspirated and replaced with 1 mL of fresh media containing 1% FBS, for 3 h, and then collected and spun down to remove debris. The media loaded on the gel were normalized according to cellular protein concentration. The cells were washed with PBS and then lysed with 10% SDS supplemented with 1× protease inhibitor cocktail (Sigma-Aldrich). Cell lysates were homogenized 30 times with a 27 1/2 G syringe, and then spun down to remove debris. Five microliter aliquots of the supernatant were taken to measure the protein concentration in the sample by Lowry assay [62]. Dithiothreitol (DTT) was immediately added to a final concentration of 100 mm. After measuring protein concentration, samples were prepared in sample buffer that contained DTT and 20% SDS, such that the final concentration of SDS was ∼8%. Samples were run on 6% SDS-PAGE gel for NPC1, and 15% gel for NPC2. The membranes were blocked in skim milk or goat serum and then blotted with either anti-rabbit NPC1 (1:2000; 1% skim milk in TBS-T) or anti-rabbit NPC2 (1:1000; 1% goat serum, graciously provided by Dr Peter Lobel, Robert Wood Johnson Medical Center). All antibodies used for subcellular fractionation analysis are as described previously [8]. Appropriate horseradish peroxidase-conjugated secondary antibodies (1:10 000, Sigma) were used for detection with enhanced chemiluminescence using the ECL plus kit (GE Healthcare).

Quantitative real-time PCR

Forty-eight hours after plating, total RNA was isolated from confluent cells using TRIzol Reagent and cDNA was prepared using the SuperScript II Reverse transcriptase according the manufacturer's instructions (Invitrogen). Quantitative real-time PCR was performed with the MX3000p real-time thermal cycler (Agilent) using the PerfeCTa SYBR Green SuperMix, UNG, Low ROX (Quanta Biosciences). For each gene of interest, dissociation curves and agarose gel electrophoresis was performed to ensure unique PCR product. Arbitrary unit was determined using the ribosomal S16 (Rps16) as a normalizer. Oligonucleotides sequences are provided in Table S1.


Many thanks to all members of the Seidah Laboratory for helpful discussions and to Brigitte Mary for secretarial assistance. We are also grateful to Dr Ioannou (Mount Sinai School of Medicine, New York), Josee Hamelin (Clinical Research Institute of Montreal), Maya Mamarbachi (Montreal Heart Institute) and Sarah Zafar (University of Göttingen) for providing reagents and for excellent technical assistance. S. P. was supported by a F. Banting and C. Best – doctoral Canadian Institutes of Health Research (CIHR) fellowship and S. R. M. by the National Institute of General Medical Sciences – Award #T32GM008704. G. M. is a Research Scholar funded by the Fonds de la Recherche en Santé du Québec (FRSQ). This research was supported by CIHR grants MOP-102741 and CTP-82946 and a Canada Chair #216684 (N. G. S.) and by the Montreal Heart Institute Foundation (G. M.). P. S. was supported by the DFG Schu 802/3-1 & 2 and T. Y. C. by the NIH grant HL036709.