Lipid Storage Disorders Block Lysosomal Trafficking By Inhibiting TRP Channel and Calcium Release

Abstract

Lysosomal lipid accumulation, defects in membrane trafficking, and altered Ca²⁺ homeostasis are common features in many lysosomal storage diseases. Mucolipin TRP channel 1 (TRPML1) is the principle Ca²⁺ channel in the lysosome. Here we show that TRPML1-mediated lysosomal Ca²⁺ release, measured using a genetically-encoded Ca²⁺ indicator (GCaMP3) attached directly to TRPML1 and elicited by a potent membrane-permeable synthetic agonist, is dramatically reduced in Niemann-Pick (NP) disease cells. Sphingomyelins (SMs) are plasma membrane lipids that undergo Sphingomyelinase (SMase)-mediated hydrolysis in the lysosomes of normal cells, but accumulate distinctively in NP cell lysosomes. Patch-clamp analyses revealed that TRPML1 channel activity is inhibited by SMs, but potentiated by SMases. In NP type C (NPC) cells, increasing TRPML1’s expression/activity was sufficient to correct the trafficking defects and reduce lysosome storage and cholesterol accumulation. We propose that abnormal accumulation of luminal lipids causes secondary lysosome storage by blocking TRPML1- and Ca²⁺-dependent lysosomal trafficking.

Introduction

The lysosome, which serves as the cell’s recycling center, can mediate the hydrolase-based degradation of a variety of biomaterials, including proteins, lipids, and membranes ¹. Most Lysosomal Storage Diseases (LSDs), of which more than 50 are known, are due to the defective activities of specific lysosomal enzymes, which result in lysosomal accumulation of specific substrates ². For instance, in Niemann-Pick (NP) type A and B diseases (NPA and NPB), sphingomyelins (SMs) accumulate in lysosomes due to insufficient activity of acid SMase (aSMase) (Supplementary Fig. S1) ³; similarly, insufficient aSMase activity and SM accumulation occur in Niemann-Pick type C (NPC), where the genetic defect occurs in components of an intracellular cholesterol/lipid trafficking pathway ^3,4. Interestingly, regardless of the nature of the primary undigested metabolites, common cellular defects are seen in a spectrum of LSDs ⁵. For example, defects in lipid trafficking (the exit of lipids from the late endosome and lysosome) and autophagome-lysosome fusion are commonly observed in almost all sphingolipidoses, including NPA, NPB, and NPC ^2,6. In addition, altered Ca²⁺ and Fe²⁺ homeostasis are also common factors in these LSDs ^2,5,7. Thus, it is likely that the common trafficking machinery is impaired in LSDs with unrelated origins or primary defects. One intriguing hypothesis is that the primary accumulated materials cause a sort of “lysosome stress,” which impairs common biochemical and/or signaling pathways in the lysosome, similar to the way unfolded proteins cause “ER stress” to trigger an “unfolded protein response” ^2,5.

Cells that lack mucolipin transient receptor potential (TRP) channel 1 (TRPML1) exhibit lysosomal defects similar to those seen in NP cells: defective autophagosome-lysosome fusion or lysosome reformation, abnormal lipid trafficking, and altered Ca²⁺ and Fe²⁺ homeostasis ^8-15. Furthermore, mutations in the human Trpml1 gene cause mucolipidosis type IV (ML4) LSD, which exhibits membrane trafficking defects and excessive lysosomal storage ^16-18. TRPML1 is a ubiquitously-expressed Fe²⁺ and Ca²⁺-dually permeable channel predominantly localized in late endosomes and lysosomes (LELs) ^11,13,19-21. TRPML1 is specifically activated by phosphatidylinositol 3,5-bisphosphate (PI(3,5)P₂), an LEL-localized low-abundance phosphoinositide ¹⁴. In addition, both TRPML1-lacking and PI(3,5)P₂-deficient cells exhibit defects in LEL-to-Golgi retrograde trafficking and autophagosome-lysosome fusion ^8,15,22,23, suggesting that the TRPML1-PI(3,5)P₂ system represents a common signaling pathway essential for late endocytic trafficking.

Because of the high degree of similarity in lysosomal defects, we hypothesize that TRPML1-PI(3,5)P₂ signaling is compromised in NPs and many other LSDs. In this study, by measuring lysosomal Ca²⁺ release using lysosome-targeted, genetically-encoded Ca²⁺ indicators, we found that TRPML1-mediated lysosomal Ca²⁺ release was compromised in NPC and NPA cells. Patch-clamp analyses demonstrated that TRPML1’s channel activity was inhibited by micromolar concentrations of SMs, but potentiated by SM-hydrolyzing enzyme SMases. In type C NP (NPC) cells, increasing TRPML1’s expression/activity was found to be capable of correcting the trafficking defects and decrease cholesterol accumulation. We conclude that SMs are negative regulators of TRPML1 channels under physiological conditions, and that abnormal accumulation of SMs in the lysosome lumen blocks Ca²⁺-dependent membrane trafficking by directly inhibiting TRPML1.

Results

Identification of a potent synthetic agonist for TRPML1

A recent high-throughput screen identified 15 small-molecule chemical compounds (SF compounds) that can activate recombinant surface-expressed TRPML3 ²¹. TRPML1 is primarily localized on the membranes of LELs ¹³. Using whole-endolysosome patch-clamp recordings ^11,14, we found that several SF compounds also weakly activated whole-endolysosome TRPML1-mediated currents (ITRPML1 or I_ML1, data not shown). In HEK293T cells transfected with surface-expressed mutant TRPML1 channels (TRPML1-L¹⁵L/AA-L⁵⁷⁷L/AA, abbreviated as ML1-4A ^9,21), one of the SF compounds, SF-51 21, induced small cytosolic [Ca²⁺] ([Ca²⁺]_cyt) increases (measured with Fura-2 ratios; Fig. 1a, b), and activated small but significant inwardly-rectifying whole-cell currents (I_ML1-4A; Fig. 1d).

Figure 1. Activation of TRPML1 by small molecule synthetic agonists.

(a) ML-SA1 induced increases in cytosolic [Ca²⁺], as measured by the Fura-2 ratio (F₃₄₀/F₃₈₀) in HEK293T cells expressing TRPML1-L¹⁵L/AA-L⁵⁷⁷L/AA (ML1-4A) at low pH (pH 4.6) external solution. SF-51 (10 μM) induced much smaller responses in the same cells. (b) SF-51 (30 μM) induced increases in cytosolic [Ca²⁺] in ML1-4A-expressing HEK293T cells at low pH external solution. (c) Chemical structures of SF-51 and ML-SA1. The difference between SF-51 and ML-SA1 is highlighted in red. (d) Activation of whole-cell currents by SF-51 and ML-SA1 in ML1-4A-expressing HEK293T cells. ML1-4A-mediated current (I_ML1-4A) was elicited by repeated voltage ramps (-140 to +140 mV; 400 ms) with a 4-s interval between ramps. Only a portion of the voltage protocol is shown; holding potential = 0 mV. (e) Activation of whole-endolysosome I_ML1 by ML-SA1 and PI(3,5)P₂. Enlarged vacuoles/endolysosomes were isolated from vacuolin-treated HEK293 cells stably-expressing ML1-4A. ML-SA1 and PI(3,5)P₂ were applied to the cytosolic sides of the isolated vacuoles. The pipette (luminal) solution was a standard external (Tyrode’s) solution adjusted to pH 4.6; the bath (internal/cytoplasmic) solution was a K⁺-based solution (140 mM K⁺-gluconate). Note that the inward current indicates cations flowing out of the endolysosome. (f) Synergistic activation of whole-endolysosome I_ML1 by ML-SA1 and PI(3,5)P₂ in human fibroblasts. I_ML1 was activated by PI(3,5)P₂ (1 μM) and ML-SA1 (10 μM); co-application of PI(3,5)P₂ and ML-SA1 further increased I_ML1. (g) ML-SA1 activated endogenous whole-endolysosome ML-like currents (I_ML-L) in CHO cells. (h) ML-SA1 activated whole-endolysosome I_ML-L in mouse primary macrophages. (i) ML-SA1 activated whole-endolysosome I_ML1 in WT (ML1^+/+), but not ML4 (ML1^−/−) human fibroblasts.

By conducting a Fura-2-based low-throughput-screen for more potent TRPML1 agonists, we identified a SF-51-related compound (see Fig. 1c for the chemical structures; Mucolipin Synthetic Agonist 1 or ML-SA1) that could induce significant [Ca²⁺]_cyt increases in HEK293 cells stably- or transiently-expressing ML1-4A (Fig. 1a). In electrophysiological assays, ML-SA1 robustly activated whole-cell I_ML1-4A (Fig. 1d) and whole-endolysosome I_ML1 (Fig. 1e). ML-SA1 also activated whole-cell ITRPML2 and ITRPML3, but not six other related channels (see Supplementary Fig. S2). ML-SA1 (10 μM) activation of whole-endolysosome I_ML1 was comparable to the effect of the endogenous TRPML agonist PI(3,5)P₂ (1 μM; Fig. 1e), and these agonists were synergistic with each other (Fig. 1f). ML-SA1 activated an endogenous whole-endolysosome TRPML-like current (I_ML-L) in all mammalian cell types that we investigated, including Chinese Hamster Ovary (CHO) (Fig. 1g), Cos-1, HEK293, skeletal muscle, pancreatic β, and macrophage cells (Fig. 1h). ML-SA1 activated whole-endolysosome I_ML-L in wild-type (WT; ML1^+/+), but not ML4 (ML1^−/−) human fibroblasts (Fig. 1i), suggesting that although ML-SA1 targets all three TRPMLs, the expression levels of TRPML2 and TRPML3 are very low, and TRPML1 is the predominant lysosomal TRPML channel in this cell type. Thus, we have identified a reasonably specific and potent agonist that can be a useful chemical tool for studying the functions of TRPMLs, especially TRPML1.

A lysosome-targeted genetically-encoded Ca²⁺ indicator

To measure lysosomal Ca²⁺ release in intact cells, we engineered GCaMP3, a single-wavelength genetically-encoded Ca²⁺ indicator 24, to the cytoplasmic N-terminus of TRPML1 (Fig. 2a). When transfected into HEK293T, CHO, or Cos-1 cell lines, GCaMP3-TRPML1 (GCaMP3-ML1) was mainly localized in the Lamp-1 and LysoTracker-positive compartments, i.e., LEL (Fig. 2b, c). To determine the specificity of the GCaMP3-ML1 fluorescence signal, we tested a variety of Ca²⁺ release reagents and found that GCaMP3 fluorescence (F₄₇₀) responded preferentially and reliably to lysosome-acting Ca²⁺-mobilization reagents (ΔF/F₀ =0.2-1) , including Glycyl-L-phenylalanine 2-naphthylamide (GPN, a Cathepsin C substrate that induces osmotic lysis of lysosomes; Fig. 2d) and bafilomycin A1 (Baf-A1, a V-ATPase inhibitor) 25. In a subset of cells, GCaMP3 also responded to large (ΔF/F₀> 2) Ca²⁺ increases triggered by thapsigargin (TG, targeting the ER Ca²⁺ store; Fig. 2d), which may presumably spread to close proximity of lysosomes. In the majority of the cells, however, TG failed to significantly increase GCaMP3 fluorescence (Fig. 2d), suggesting that small ER Ca²⁺ release was not detected by GCaMP3-ML1. Taken together, GCaMP3-ML1 readily and preferentially detects juxta-lysosomal Ca²⁺ increases.

Figure 2. A lysosome-targeted genetically-encoded Ca²⁺ indicator.

(a) GCaMP3-TRPML1 (GCaMP3-ML1) fusion strategy. GCaMP3 is fused to the N-terminus of TRPML1. (b) Co-localization of GCaMP3-ML1 fluorescence with Lamp-1 in HEK293T cells expressing both GCaMP3-ML1 and Lamp-1-mCherry. Scale bar = 5 μm. (c) Co-localization of GCaMP3-ML1 fluorescence with LysoTracker in GCaMP3-ML1-expressing HEK293T cells. Scale bar = 2 μm. (d) Preferential detection of Ca²⁺ release from lysosome stores by GCaMP3-ML1 in GCaMP3-ML1-transfected Cos-1 cells. ER and lysosome Ca²⁺ releases were triggered by thapsigargin (TG, 2 μM) and Glycyl-L-phenylalanine 2-naphthylamide (GPN, 200 μM), respectively, in low (nominally-free Ca²⁺ + 1 mM EGTA; free [Ca²⁺] estimated to be < 10 nM) external [Ca²⁺]. All cells responded to GPN; only 1 out 4 cells responded to TG.

In GCaMP3-ML1-transfected CHO (Fig. 3a, 3b), HEK293T, and Cos-1 cell lines, application of ML-A1 (20 μM) induced rapid increases in GCaMP3 fluorescence (ΔF/F₀ = 0.45 ± 0.05, n = 77 cells; Fig. 3a, k) in the presence of low external Ca²⁺ (nominally-free Ca²⁺ + 1 mM EGTA; free Ca²⁺ estimated to be < 10 nM). In contrast, no significant increase in GCaMP3 fluorescence (ΔF/F₀ = 0.03 ± 0.01, n = 32 cells; Fig. 3c, d, k) was observed in cells transfected with a plasmid carrying GCaMP3 fused to a non-conducting pore mutant of TRPML1 ^8,14 (GCaMP3-ML1-KK). Both GPN and Baf-A1 largely abolished ML-SA1-induced Ca²⁺ responses (Fig. 3e-k). In contrast, ML-SA1-induced Ca²⁺ response was mostly intact in TG-treated cells (Fig. 3k). Collectively, these results suggest that ML-SA1 activated endolysosomal TRPML1 to induce Ca²⁺ release from lysosomes. Thus we have established a robust Ca²⁺ imaging assay to measure TRPML1-mediated lysosomal Ca²⁺ release.

Figure 3. ML-SA1 induces TRPML1-mediated Ca²⁺ release from lysosomes.

(a) ML-SA1 (20 μM) induced rapid increases in GCaMP3 fluorescence (measured as change of GCaMP3 fluorescence ΔF over basal fluorescence F₀; ΔF/F₀) under low (< 10 nM) external Ca²⁺ in CHO cells transfected with GCaMP3-ML1. Subsequent application of GPN (200 μM) induced smaller responses. Maximal responses were induced by ionomycin (1 μM) application. (b) Representative micrographs from panel (a) to show the changes of GCaMP3 fluorescence upon bath application of ML-SA1 and ionomycin to GCaMP3-ML1-transfected CHO cells. Scale bar = 5 μm. (c) ML-SA1 failed to induce significant Ca²⁺ increases in CHO cells transfected with GCaMP3-ML1-KK (non-conducting pore mutation). (d) Basal GCaMP3 fluorescence (normalized to the maximal ionomycin-induced fluorescence) of GCaMP3-ML1-KK and GCaMP3-ML1. (e) GPN (200 μM) pretreatment abolished ML-SA1-induced responses in GCaMP3-ML1-expressing CHO cells. (f) Repetitive applications of ML-SA1 (20 μM) induced little or no responses in GCaMP3-ML1-expressing CHO cells. (g) Repetitive applications of GPN (200 μM) induced smaller GCaMP3 responses in GCaMP3-ML1-expressing CHO cells. (h) ML-SA1 failed to further increase GCaMP3 fluorescence in GCaMP3-ML1-expressing CHO cells in the presence of GPN (200 μM). (i) ML-SA1 induced small responses in GCaMP3-ML1-expressing CHO cells that had received an application of GPN. (j) Bafilomycin A1 (Baf-A1, 500 nM) pretreatment abolished ML-SA1-induced Ca²⁺ response in GCaMP3-ML1-expressing CHO cells. (k) ML-SA1-induced responses in CHO cells transfected with GCaMP3-ML1-KK (non-conducting pore mutation), and GCaMP3-ML1-transfected cells pretreated with Baf-A1, GPN, or TG. For panels d and k, the responses were averaged for 40-100 transfected cells from at least 3 independent experiments; data are presented as the mean ± SEM. Statistical comparisons were made using analysis of variance: * P <0.05.

Reduced TRPML1-mediated lysosomal Ca²⁺ release in NPC cells

Next we investigated whether TRPML1-mediated lysosomal Ca²⁺ release is altered in NP cells. We chose an easy-to-transfect NPC cellular model, i.e, NPC (NPC1^−/−) CHO cells, which exhibit all the hallmark features of NPC disease ^26,27. WT CHO cells transfected with GCaMP3-ML1 exhibited large TRPML1-mediated lysosomal Ca²⁺ release (Fig. 4a, l). In contrast, NPC CHO cells exhibited much smaller TRPML1-mediated Ca²⁺ release with a ~ 75% reduction compared to WT controls (Fig. 4b, l). GCaMP3-ML1 was still mainly co-localized with Lamp-1 in NPC cells (Fig. 4c). In addition, the expression levels of GCaMP3-ML1, as estimated from the maximal fluorescence intensity induced by ionomycin, were comparable for WT and NPC cells (Fig. 4d). A previous study reported that lysosomal Ca²⁺ stores are compromised in NPC cells ²⁵. However, we found that GPN-induced GCaMP3-ML1 increases were comparable for WT and NPC cells (Fig. 4e). Likewise, GPN-induced and NH₄Cl-induced Fura-2 responses were also comparable in non-transfected WT and NPC cells (Fig. 4f, g). Furthermore, TRPML1-mediated, but not GPN-induced lysosomal Ca²⁺ release, was also reduced in GCaMP3-transfected NPC human fibroblasts compared to WT cells (Fig. 4h). Collectively, these results suggest that TRPML1-mediated lysosomal Ca²⁺ release is selectively reduced in NPC cells.

Figure 4. TRPML1-mediated lysosomal Ca²⁺ release, but not lysosome Ca²⁺ store, is reduced in NPC cells.

(a, b) ML-SA1-induced lysosomal Ca²⁺ release in GCaMP3-ML1-transfected WT (a) and NPC (b) CHO cells. (c) Co-localization of GCaMP3-ML1 fluorescence with Lamp-1 in NPC CHO cells doubly-transfected with GCaMP3-ML1 and Lamp-1-mCherry. Scale bar = 5 μm. (d) Comparable expression levels of GCaMP3-ML1 in WT and NPC CHO cells. Expression level of GCaMP3-ML1 was estimated by the maximal GCaMP3 fluorescence signal induced by ionomycin (1 μM). (e) Comparable GPN-induced GCaMP3 responses in WT, NPC, or U18666A-treated GCaMP3-ML1-transfected CHO cells. (f) NH₄Cl (10 mM) and GPN (200 μM) induced comparable levels of Ca²⁺ increases (measured with Fura-2 ratios) in WT, NPC, and U18666A-treated CHO cells. (g) NH₄Cl (10 mM) and GPN (200 μM) induced comparable levels of Fura-2 Ca²⁺ responses in WT and NPC human fibroblasts. (h) TRPML1-mediated, but not GPN-induced lysosomal Ca²⁺ release, was reduced in GCaMP3-ML1-transfected NPC human fibroblasts compared to WT cells. (i) WT CHO cells were treated with U18666A (2 μg/ml) for 16 h, and then subjected to filipin staining. Scale bar = 10 μm. (j) Sphingomyelins were stained with Lysenin (0.1 μg/ml) in WT CHO cells, WT CHO cells treated with U18666A (2 μg /ml) for 16 h, and NPC CHO cells. Scale bar = 10 μm. (k) Ca²⁺ responses in GCaMP3-ML1-transfected WT CHO cells treated with U18666A (2 μg /ml) for 16-20 h. (l) ML-SA1-induced peak GCaMP3 responses were reduced in GCaMP3-ML1-transfected NPC or U18666A-treated WT CHO cells. (m, n) Endogenous TRPML1-mediated lysosomal Ca²⁺ release (measured with Fura-2 ratios) in WT (m) and NPC (n) CHO cells. (o) Average endogenous ML-SA1-induced lysosomal Ca²⁺ responses in WT and NPC CHO cells, and WT, NPC1^−/−, and ML1^−/− mouse macrophages. For panels d, e, f, g, h, k, l, and o, the results were averaged for 40-100 cells from at least 3 independent experiments; data are presented as the mean ± SEM. Statistical comparisons were made using analysis of variance: * P <0.05.

NPC-like cellular phenotypes can be pharmacologically induced with a small-molecule drug U18666A ²⁸. After incubating GCaMP3-ML1-transfected WT CHO cells with U18666A (2 μg/ml) for 16-20 h, cells exhibited significant increases in cholesterol level as detected by filipin staining ⁴ (Fig. 4i), and sphingomyelin (SM) level as detected by lysenin staining (Fig. 4j). U18666A-treated cells exhibited a ~ 70% reduction in TRPML1-mediated lysosomal Ca²⁺ release (Fig. 4k, l). Conversely, drug treatment failed to result in noticeable changes in GPN-induced GCaMP3 responses (Fig. 4e) or Fura-2 Ca²⁺ responses (Fig. 4f). Taken together, these results suggest a specific reduction of TRPML1-mediated lysosomal Ca²⁺ release in pharmacologically-induced NPC cells.

We next investigated the effect of NPC on endogenous Ca²⁺ release using Fura-2 Ca²⁺ imaging. In WT CHO cells, small but significant Ca²⁺ increases were seen upon application of 100 μM ML-SA1 in low external Ca²⁺ (ΔFura-2 = 0.08 ± 0.01, n = 89; Fig. 4m, o). In contrast, in NPC CHO cells, ML-SA1 failed to induce significant Ca²⁺ responses (ΔFura-2 = 0.01 ± 0.002, n = 61; Fig. 4n, o). Cyclodextrin, a polysaccharide that reportedly alleviates NPC cellular phenotypes 4, was found to partially restore ML-SA1-induced lysosomal Ca²⁺ release (Supplementary Fig. S3). Finally, NPC1^−/− mouse macrophages, one of the major cell types affected in NPC ³, also exhibited a significant reduction in ML-SA1-induced lysosomal Ca²⁺ release compared to WT cells (Fig. 4o & Supplementary Fig. S4). The ML-SA1-induced lysosomal Ca²⁺ responses are most likely mediated by TRPML1, as demonstrated by the lack of response in ML1^−/− macrophages (Fig. 4o & Supplementary Fig. S4). Collectively, the reduction in TRPML1-mediated lysosomal Ca²⁺ release may be a general feature of most NPC cell types.

Reduced TRPML1-mediated currents in NPC cell lysosomes

We next investigated the effect of NPC on whole-endolysosome I_ML1 in human fibroblasts. Compared to WT fibroblasts (Fig. 5a, c), much smaller ML-SA1-activated I_ML1 was seen in NPC fibroblasts at three different concentrations of ML-SA1 (Fig. 5b, c). On average, whole-endolysosome I_ML1 activated by 25 μM ML-SA1 was 75 ± 25 (n = 8) pA/pF for NPC cells, a ~ 70% reduction from the values observed for WT cells (442 ± 150 pA/pF; n = 6). Notably, the mRNA expression levels of TRPML1-3 were not decreased in NPC cells (Fig. 5d). These results suggest that a reduction of whole-endolysosome I_ML1 accounts for the reduced TRPML1-mediated Ca²⁺ release observed in NPC cells.

Figure 5. TRPML-mediated currents are reduced in the lysosomes of NPC cells.

(a, b) Concentration-dependent activation of whole-endolysosome I_ML-L in WT (a) and NPC (b) human fibroblasts. ML-SA1 (3, 10, 25 μM) was applied to the cytosolic sides of the vacuoles isolated from human fibroblasts. (c) Average current densities (pA/pF) of I_ML-L were reduced in NPC fibroblasts compared to WT cells. The current density was calculated from the current size (pA) normalized with the capacitance of the vacuole (pF). (d) RT-PCR analysis of TRPML1-3 mRNA expression in WT and NPC human fibroblasts, and NPC1 ^−/− mouse macrophages. Expression levels were normalized to that of GAPDH (human fibroblast) and L32 (mouse macrophage). For panels c and d, data are presented as the mean ± SEM; the n numbers are in parentheses. Statistical comparisons were made using analysis of variance: * P <0.05.

Inhibition of TRPML1 by SMs

The reduced TRPML1-mediated lysosomal Ca²⁺ release and I_ML1 in the presence of normal TRPML1 expression in NPC cells suggest tonic inhibition of TRPML1 by cytoplasmic and/or luminal factors. One possibility is that lipids accumulated in NPC lysosomes, such as cholesterol, SM, sphingosine (Sph), and glycosphingolipids (GSLs) ^4,29, might directly inhibit TRPML1. To test this possibility, we performed whole-cell recordings in ML1-4A-expressing HEK293T cells. Under this configuration, the extracellular side is analogous to the luminal side (see Supplementary Fig. S1), making it possible to study the effects of luminal lipids on I_ML1-4A. SMs, but not cholesterol or GSLs (data not shown), acutely inhibited SF-51- (Fig. 6a, d) or ML-SA1-activated I_ML1-4A. In contrast, SMs failed to inhibit I_ML1- Va , a gain-of-function mutation that locks TRPML1 in the open state ¹³ (Fig. 6b, d). SMs can be hydrolyzed into ceramide and phosphocholine via SMase (see Supplementary Fig. S1). No significant inhibition was seen with ceramide (Fig. 6c, d) or phosphocholine (Fig. 6d). Sphingosine, a hydrolyzed product of ceramide, potentiated I_ML1-4A (Fig. 6e).

Figure 6. Regulation of TRPML1 by sphingomyelins and sphingomyelinases.

(a) SF-51-activated whole-cell I_ML1-4A was inhibited by sphingomyelins (SMs; 20 μM) at low extracellular pH in ML1-4A-expressing HEK293T cells. (b) Insensitivity of whole-cell I_ML1-Va to SMs (60 μM). (c) Lack of effect of ceramide (Cer, 20 μM) on whole-cell I_ML1-4A. (d) Summary of the effects of various sphingolipids on whole-cell I_ML1-4A in ML1-4A-expressing HEK293T cells. PC: phosphocholine. (e) SF-51-activated whole-cell I_ML1-4A was potentiated by sphingosine (Sph; 5 μM) at low external pH in ML1-4A-expressing HEK293T cells. (f) The effects of SMase (50 ng/ml) treatment on I_ML1-4A at low external pH. I_ML1-4A was elicited by repeated voltage ramps (− 140 to + 140 mV; 400 ms) with a 4-s interval between ramps. Current amplitudes at -140 mV were used to plot the time dependence. (g) Representative traces of I_ML1-4A before (black) and after (dark yellow) SMase treatment at two different time points, as shown in f. (h, i) SMase (50 ng/ml) treatment alone had no (6 out of 7 cells; see h for an example) or small (1 out of 7 cells; i) activation effect on basal whole-cell I_ML1-4A in ML1-4A-expressing HEK293T cells. (j) The effects of SMase (50 ng/ml) treatment on SF-51-activated I_ML1-4A at neutral external pH (pH 7.4). (k) The potentiation effect of SMase was more dramatic at pH 4.6, but abolished by heat inactivation of the enzyme activity. HI-SMase: heat-inactivated SMase. For panels d and k, data are presented as the mean ± SEM; the n numbers are in parentheses. Statistical comparisons were made using analysis of variance: * P <0.05.

The robust inhibition of I_ML1-4A by SMs, but not ceramide or phosphocholine, suggests that the activity of SMase may regulate I_ML1-4A. Indeed, bath application of SMases, but not heat-inactivated enzymes, significantly potentiated SF-51-activated I_ML1-4A (Fig. 6f-k), with the potentiation being more dramatic at acidic extracellular (luminal) pH than at neutral pH (Fig. 6k).

SM accumulation reduces TRPML1-mediated Ca²⁺ release

The results presented above suggest that SM-mediated inhibition underlies NPC-mediated reduction in TRPML1-mediated lysosomal Ca²⁺ release. Consistent with this, in fibroblasts derived from NPA patients, which completely lack the activity of acid SMase ³, TRPML1-mediated , but not GPN-induced lysosomal Ca²⁺ release (Fig. 7a-c), was significantly reduced. To further test this hypothesis, we induced lysosomal SM accumulation by incubating cells with desipramine and fluoxetine for 2-4 hrs, which cause proteolytic degradation of acid SMase ^30,31. Compared to non-treated cells, TRPML1-mediated lysosomal Ca²⁺ release was dramatically reduced in desipramine/fluoxetine -treated CHO cells, and the reduction was abolished in the presence of a protease inhibitor leupeptin ^30,31 (Fig. 7d-g). Taken together, these results suggest that SM inhibition of TRPML1 underlies NPA or NPC-mediated attenuation of TRPML1-mediated lysosomal Ca²⁺ release.

Figure 7. SM accumulation in the lysosome reduces TRPML1-mediated Ca²⁺ release.

(a,b) ML-SA1-induced lysosomal Ca²⁺ release was reduced in GCaMP3-ML1-transfected NPA fibroblasts (b) compared to WT fibroblasts (a). (c) Average ML-SA1 (20 μM) and GPN (200 μM)-induced responses in GCaMP3-ML1-transfected WT and NPA fibroblasts. (d) Desipramine (25 μM) treatment for 2-4 hrs reduced TRPML1-mediated lysosomal Ca²⁺ release responses (20 μM ML-SA1) in GCaMP3-ML1-transfected CHO cells. (e) Pretreatment of a protease inhibitor Leupeptin (25 μM) abolished the effects of Desipramine in ML-SA1-induced GCaMP3 responses. (f) Summary of the effects of Desipramine (25 μM) and Fluoxetine (25 μM) treatment for 2-4 hrs on TRPML1-mediated lysosomal Ca²⁺ release responses (20 μM ML-SA1) in GCaMP3-ML1-transfected CHO cells. (g) Desipramine and fluoxetine treatment for 2-4 hrs reduced endogenous lysosomal Ca²⁺ release responses (measured with Fura-2 ratios; 100 μM ML-SA1) in CHO cells. For panels c, f, and g, the results were averaged for 40-100 cells from at least 3 independent experiments; data are presented as the mean ± SEM. Statistical comparisons were made using analysis of variance: * P <0.05.

TRPML1 regulates lysosomal lipid and cholesterol trafficking

Human NPC, NPA, and NPB fibroblasts exhibit defects in LEL-to-Golgi trafficking of lipids and proteins ^6,27. Consistent with this, BODIPY-LacCer, a fluorescent analogue of lactosylceramide that is localized in the Golgi apparatus of WT cells after a pulse-chase period ³², was predominantly localized in the LEL-like compartments of NPC1^−/− mouse macrophages (Fig. 8a). Because of the established roles of PI(3,5)P₂/TRPML1/Ca²⁺ in LEL-to-Golgi lipid trafficking ¹⁵, it is possible that the trafficking defects seen in NP cells are due to the aforementioned attenuated TRPML1-mediated lysosomal Ca²⁺ release. If so, one would expect that boosting TRPML1 activity may correct the trafficking defects. Indeed, more than 60% of ML-SA1-treated NPC macrophages exhibited a Golgi-like pattern of LacCer staining, suggestive of a significant restoration in LEL-to-Golgi transport (Fig. 8a). In NPC1^−/− mouse macrophages and TRPML1-transfected NPC CHO cells, the puncta number of LacCer staining that reflects the function of LEL-to-Golgi transport, was significantly reduced in the presence of ML-SA1 (Fig. 8b & Supplementary Fig. S5). Collectively, these results suggest that increasing TRPML1’s channel activity may restore the trafficking defects in NPC cells, and that TRPML1 deregulation is a pathogenic mechanism underlying lipid trafficking defects in NP cells.

Figure 8. ML-SA1 and TRPML1 overexpression rescue the trafficking defects and reduce cholesterol accumulation in NPC cells.

(a) A fluorescent analogue of lactosylceramide, BODIPY-LacCer, was mainly localized in the Golgi-like structures of WT macrophages after a pulse-chase for 1 h, but accumulated in the LEL-like vesicles of NPC macrophages. ML-SA1 (10 μM) treatment of NPC1^−/− macrophages for 12 h resulted in a primary Golgi-like localization for LacCer. Scale bar = 10 μm. (b) ML-SA1 reduced the number of LacCer puncta in NPC1^−/− macrophages. (c, d) ML-SA1 reduced unesterified cholesterol (filipin) staining in ML1, but not ML1-KK-expressing U18666A-treated WT CHO cells. Differential interference contrast (DIC) images are shown for comparison. Scale bar = 10 μm. (e) Unesterified cholesterol (filipin) staining in NPC CHO cells was partially reduced in cells expressing DsRed-aSMase. (f) Average filipin staining (intensity normalized to non-transfected control cells) in DsRed-aSMase-transfected cells. (g) Unesterified cholesterol (filipin) staining in NPC CHO cells was partially reduced in cells expressing ML1-Va-EGFP (ML1^Va). (h) Average filipin staining (intensity normalized to non-transfected control cells) in ML1^Va-transfected cells. Scale bar = 10 μm for panels a, c, e, and g. For panels b, d, f, and h, the results were averaged for multiple randomly-taken micrographs from at least 3 independent experiments. Data are presented as the mean ± SEM. Statistical comparisons were made using analysis of variance: * P <0.05.

NPC cells accumulate cholesterol, as demonstrated by elevated filipin staining ²⁵. However, in U18666A-treated CHO cells, ML-SA1 induced a significant reduction in filipin staining in TRPML1, but not TRPML1-KK-transfected cells (Fig. 8c, d). Likewise, in TRPML1-transfected NPC CHO cells, a significant reduction of filipin staining was also seen in the presence of ML-SA1 (Supplementary Fig. S5). Reduction of filipin staining was also seen in aSMase- (Fig. 8e, f) and TRPML1^Va-transfected NPC CHO cells (Fig. 8g, h). These results suggest that increased TRPML1 activity may reduce cholesterol accumulation in NPC cells.

Discussion

We used a lysosome-targeted GCaMP3-based Ca²⁺ imaging method to demonstrate that TRPML1-mediated lysosomal Ca²⁺ release, but not the lysosomal Ca²⁺ store as has been reported previously ²⁵, is selectively compromised in NPC and NPA cells. SM, a sphingolipid that is known to accumulate in all three types of NP cells ⁴, inhibits TRPML1. Pharmacologically-induced SM accumulation significantly reduces TRPML1-mediated lysosomal Ca²⁺ release. Conversely, increasing TRPML1’s expression/activity is sufficient to correct the trafficking defects and reduce cholesterol accumulation in NPC cells. Collectively, our work has provided a feedback mechanism for luminal lipids to control lipid efflux from the lysosome, as well as a pathogenic cause for the secondary lysosome storage present in many LSDs.

SMs are present on the outer leaflet of the plasma membrane 33, suggesting that SM inhibition of TRPML1 might serve as a physiological mechanism to prevent this lysosome channel from being active at the wrong subcellular location, i.e., the plasma membrane. For normal cells, SMs in the lysosome would be rapidly degraded by acid SMase activity ³³. In NP cells, however, SMs accumulate in the lumen and limited lysosomal membranes (Supplementary Fig. S1). Although the primary storage is likely to be cholesterol, NPC cells are found to have a much lower level of aSMaseactivity, and a restoration of aSMase activity could revert cholesterol accumulation ³⁴. It is not known whether the level of SM is perturbed in other sphingodoses. However, rather than absolute SM levels, the relative localization (plasma membrane versus lysosome) is the factor critical for lysosomal Ca²⁺ release. In addition, other offending metabolites might also inhibit TRPML1, and/or PI(3,5)P₂-metabolizing enzymes. Thus, impaired Ca²⁺ release might be a common pathogenic mechanism for many LSDs. Although TRPML1 is the dominant lysosomal TRPML channel in most cell types, in other cell types where TRPML2 and TRPML3 are more abundantly expressed ¹³, our proposed SM-mediated inhibitory mechanism may also be applied to TRPML2/3, causing lysosomal trafficking defects. Finally, SM and cholesterol accumulation might also contribute to NP phenotypes via mechanisms unrelated to lysosomal trafficking and TRPMLs ³.

LEL-to-Golgi trafficking defects are seen in PI(3,5)P -deficient 22 2 cells , ML4 fibroblasts ^15,35, and NP fibroblasts, as well as fibroblasts from several other sphingolipid LSDs ^6,36. Our results suggest that reduced TRPML1-mediated lysosomal Ca²⁺ release directly causes trafficking defects. We previously reported that impaired Fe²⁺ homeostasis facilitates the formation of autofluorescent lipofuscin in ML4 cells 11. Interestingly, lipofuscin formation is also increased in NPC cells (Supplementary Fig. S6). For most LSDs, impaired lysosomal degradation due to a lack of hydrolytic enzymes leads to trafficking defects and lysosome storage. However, lysosome storage can in turn affect lysosomal degradation and membrane trafficking, which results in a positive feedback loop and a vicious cycle ². Our work suggests a novel concept that lysosome enhancement by manipulating TRPML1 to speed membrane trafficking may break the vicious cycle, providing a novel therapeutic approach for NPs and many other LSDs.

Methods

Molecular biology and biochemistry

The GCaMP3-ML1 construct was made by inserting the full length GCaMP3 sequence ²⁴ between the HindIII and BamHI sites of a pcDNA6 plasmid that contains the mouse TRPML1 cDNA at the XhoI site. TRPML1-L¹⁵L/AA-L577L/AA (abbreviated as ML1-4A) and the TRPML1 non-conducting pore mutant (D471K/D472K; abbreviated ML1-KK) were constructed using a site-directed mutagenesis kit (Qiagen). All constructs were confirmed by sequencing, and protein expression was verified by western blot. CHO, HEK293T, Cos-1, and human fibroblast cells were transiently transfected with GCaMP3-ML1 for electrophysiology, Ca²⁺ imaging, and confocal imaging.

PCR primers used for real-time qPCR analysis of TRPML expression in human fibroblasts and mouse macrophages are as follows:

Human TRPML1: F, CGGATGACACCTTCGCAGCCTAC; R, ACGCATACCGGCCCAGTGACAC

Human TRPML2: F, GGTCACCACACAGCTTGTTCGT; R, TGATGATACTGATTAATAGCA

Human TRPML3: F, CCAGAAATTGAAACTGAGTGTT; R, ATGTTATAGTCAGAGTAAAGTC

Mouse TRPML1: F, AAACACCCCAGTGTCTCCAG; R, GAATGACACCGACCCAGACT

ML4 and NPC cell lines

NPC (CT43; NPC1^−/−) and WT control (RA25) CHO cells were a gift from Dr. Chang TY (Dartmouth Medical School). Human skin fibroblast cells from an ML4 patient (ML1^−/−, clone GM02048), an NPC patient (NPC1^−/−, clone GM03123), an NPA patient (clone GM00112), and non-disease control cells (ML1^+/+/ NPC1^−/−, clone GM00969) were obtained from the Coriell Institute for Medical Research (NJ, U.S.A). Human fibroblasts were transiently transfected by electroporation (260 V, 950 μF) with 100 μg of GCaMP3-ML1 plasmid, and then used for Ca²⁺ imaging experiments 24 h after electroporation.

Mouse lines

NPC1 KO mice (BALB/cNctr-Npc1m1N/J) and WT littermates were ordered from the Jackson Laboratories. TRPML1 KO mice were kindly provided by Dr. Susan Slaugenhaupt (Harvard Medical School) and Dr. Jim Pickel (NIH) ³⁷. Animals were used under approved animal protocols and University of Michigan Institutional Animal Care Guidelines.

Preparation and culture of mouse macrophages

Bone marrow cells from femurs and tibias were harvested and cultured in macrophage differentiation medium (RPMI-1640 medium with 10% fetal bovine serum (FBS) and 100 unit/ml recombinant colony stimulating factor from PeproTech, Rocky Hill, NJ). After 7 d in culture at 37 °C with 5% CO₂, the adherent cells (> 95% are expected to be macrophages) were harvested for assays ³⁸.

Mammalian whole-cell electrophysiology

HEK293T, Cos-1, and CHO cells were used for the heterologous expression experiments and were transfected using Lipofectamine 2000 (Invitrogen). The pipette solution contained 147 mM Cs, 120 mM methane-sulfonate, 4 mM NaCl, 10 mM EGTA, 2 mM Na₂-ATP, 2 mM MgCl₂, and 20 mM HEPES (pH 7.2; free [Ca²⁺]_i < 10 nM). The standard extracellular bath solution (modified Tyrode’s solution) contained 153 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 20 mM HEPES, and 10 mM glucose (pH 7.4). The “Low pH Tyrode” solution contained 150 mM Na-Gluconate, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM glucose, 10 mM HEPES, and 10 mM MES (pH 4.6). All bath solutions were applied via a perfusion system to achieve a complete solution exchange within a few seconds. Data were collected using an Axopatch 2A patch clamp amplifier, Digidata 1440, and pClamp 10.0 software (Axon Instruments). Whole-cell currents were digitized at 10 kHz and filtered at 2 kHz. All experiments were conducted at room temperature (21-23 °C), and all recordings were analyzed with pClamp 10.0, and Origin 8.0 (OriginLab, Northampton, MA).

Endolysosomal electrophysiology

Endolysosomal electrophysiology was performed in isolated endolysosomes using a modified patch-clamp method ^11,14. Briefly, cells were treated with 1 μM vacuolin-1 for 2-5 h to increase the size of endosomes and lysosomes. Whole-endolysosome recordings were performed on isolated enlarged LELs. The bath (internal/cytoplasmic) solution contained 140 mM K-Gluconate, 4 mM NaCl, 1 mM EGTA, 2 mM Na₂-ATP, 2 mM MgCl₂, 0.39 mM CaCl₂, 0.2 mM GTP, and 10 mM HEPES (pH adjusted with KOH to 7.2; free [Ca²⁺]_i approximately 100 nM based on the Maxchelator software (http://maxchelator.stanford.edu/)). The pipette (luminal) solution consisted of a “Low pH Tyrode’s solution” with 145 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM HEPES, 10 mM MES, and 10 mM glucose (pH 4.6).

Fura-2 Ca²⁺ imaging

Cells were loaded with 5 μM Fura-2 AM in the culture medium at 37 °C for 60 min. Fluorescence was recorded at different excitation wavelengths using an EasyRatioPro system (PTI). Fura-2 ratios (F₃₄₀/F₃₈₀) were used to monitor changes in intracellular [Ca²⁺] upon stimulation. GPN (Glycyl-L-phenylalanine 2-naphthylamide; 200-400 μM, a lysosome-disrupting agent) and Bafilomycin A1 (500 nM, a V-ATPase inhibitor) were used as positive controls to induce Ca²⁺ release from lysosome and acidic stores, respectively ³⁹. Ionomycin (1 μM) was added at the conclusion of all experiments to induce a maximal response for comparison. For endogenous lysosomal Ca²⁺ release measurement, Ca²⁺ imaging was carried out within 0.5-3 h after plating and when cells still exhibited a round morphology.

GCaMP3 Ca²⁺ imaging

18-24 h after transfection with GCaMP3-ML1, CHO or human fibroblast cells were trypsinized and plated onto glass coverslips. Most experiments were carried out within 0.5-3 h after plating when cells still exhibited a round morphology. The fluorescence intensity at 470 nm (F₄₇₀) was monitored using the EasyRatioPro system. Lysosomal Ca²⁺ release was measured under a “low” external Ca²⁺ solution, which contained 145mM NaCl, 5 mM KCl, 3 mM MgCl₂, 10 mM Glucose, 1mM EGTA, and 20 mM HEPES (pH7.4). Ca²⁺ concentration in the nominally-free Ca²⁺ solution is estimated to be 1-10 μM. With 1 mM EGTA, the free Ca²⁺ concentration is estimated to be < 10 nM based on the Maxchelator software (http://maxchelator.stanford.edu/).

LacCer-BODIPY trafficking

Cells for the Lactosylceramide (LacCer)-BODIPY uptake experiment were grown on glass coverslips and incubated with 5 μM BSA-complexed LacCer in “Tyrode” solution for 45 min at room temperature. Cells were then washed twice with PBS and incubated with normal DMEM/F12 medium containing 10% FBS at 37 °C for 1h. For the correction experiments, NPC1^−/− macrophages were pre-treated with ML-SA1 (10 μM) for 12 hrs, and chased in the presence of ML-SA1 (10 μM). Images were taken using a Leica (TCS SP5) confocal microscope.

Filipin Staining

Cells were fixed in 3% paraformaldehyde for 1 h, washed twice with PBS, and then incubated with 1.5 mg/ml glycine in PBS for 10 min. Cells were then stained for 2 h with 0.05 mg/ml filipin ²⁵ in PBS supplemented with 10% FBS. All procedures were conducted at room temperature. Images were taken using a fluorescence microscope with a UV filter. The lysosome-like storage organelle filipin intensity was calculated based on low- and high-threshold intensities in the image ³⁴.

Lysenin Staining

Cells were fixed with 4% PFA for 20 min, blocked with 1% BSA in PBST for 2h, and then incubated with lysenin (0.1 μg/ml, Peptides International) for 20 h at 4 °C. The cells were then incubated with anti-lysenin antibody (Peptides International) for 1h, followed by Alexa Fluor 488 anti-rabbit (Invitrogen) for 45 min. Images were taken using a Leica confocal microscope.

Reagents

Sphingomyelins (Sigma) were dissolved in “Tyrode” or “Low pH Tyrode” solutions by sonication. Beta-cyclodextrin, filipin, and fluoxetine were obtained from Sigma; desipramine was purchased from Santa Cruz; C6-ceramide was obtained from Matreya, LLC; sphingomyelinase was obtained from MP Biomedicals; SF-51 and ML-SA1 were purchased from Princeton BioMolecular Research Inc; PI(3,5)P₂ was purchased from Echelon; U18666A was purchased from Enzo; LacCer-BODIPY was purchased from Invitrogen, and lysenin and anti-lysenin antibody were purchased from Peptides International.

Data analysis

Data are presented as the mean ± standard error of the mean (SEM). Statistical comparisons were made using analysis of variance (ANOVA). A P value < 0.05 was considered statistically significant.