Endo-lysosomal TRP mucolipin-1 channels trigger global ER Ca²⁺ release and Ca²⁺ influx

The remarkable versatility of Ca²⁺ as an intracellular messenger is underpinned by the spatiotemporal organisation of Ca²⁺ signals (Berridge et al., 2003). Targeted delivery of Ca²⁺ to a particular subcellular locale is one way in which Ca²⁺ can selectively influence a particular cell outcome (e.g. secretion). Equally, global Ca²⁺ signals, which are often oscillatory, can also be selective through information encoded in oscillation frequency. Interplay between Ca²⁺ release from intracellular Ca²⁺ stores and Ca²⁺ influx from the extracellular space ensures tight control of Ca²⁺ levels and their downstream targets (Clapham, 2007). Although we have gained much mechanistic insight into the mobilisation of ER Ca²⁺ stores and the subsequent entry of Ca²⁺ across the plasma membrane that ensues (Prakriya and Lewis, 2015), we know relatively little about Ca²⁺ handling by acidic Ca²⁺ stores such as lysosomes (Patel and Muallem, 2011).

Lysosomes are best known for their degradative role but they maintain an intraluminal Ca²⁺ concentration (∼500 µM) similar to the ER (Christensen et al., 2002; Lloyd-Evans et al., 2008). They are likely filled by Ca²⁺-H⁺ exchange (Melchionda et al., 2016) and express members of the transient receptor potential mucolipin (TRPML) and two-pore channel (TPC) families to effect Ca²⁺ release (Grimm et al., 2012; Kiselyov et al., 2012; Patel, 2015; Waller-Evans and Lloyd-Evans, 2015). Three TRPML isoforms are present in humans (García-Añoveros and Wiwatpanit, 2014). TRPML1 (encoded by MCOLN1) is ubiquitously expressed and targets to lysosomes through di-leucine motifs (Vergarajauregui and Puertollano, 2006). It is activated by the endo-lysosomal phosphoinositide phosphatidylinositol 3,5-bisphosphate [PI(3,5)P₂] (Dong et al., 2010) and is a non-selective cation channel permeable to a number of ions including Ca²⁺ and Fe²⁺ (Dong et al., 2008). Importantly, mutation of TRPML1 results in the lysosomal storage disorder mucolipidosis type IV (Bargal et al., 2000). This neurodegenerative disorder is characterised by build-up of various lipid species in lysosomes, suggesting a role for TRPML1 in lipid trafficking. It has long been recognised that localised Ca²⁺ release from the endo-lysosomal system can influence vesicular fusion and fission events (Pryor et al., 2000). Specifically, Ca²⁺ release through TRPMLs likely underlies functional roles in vesicular trafficking as well as lysosomal exocytosis (Samie et al., 2013), autophagy (Medina et al., 2015) and regulated secretion (Park et al., 2016). Notably, recently described TRPML agonists have proven beneficial in correcting trafficking defects not only in mucolipidosis type IV (Chen et al., 2014) but also in Niemann–Pick type C disease (Shen et al., 2012) and HIV-associated dementia (Bae et al., 2014). Defective TRPML activity has also been implicated in Alzheimer's disease (Lee et al., 2015). Despite their demonstrable (patho)physiological relevance, TRPML-mediated Ca²⁺ signals are difficult to resolve due to their localised nature. To overcome this, a number of recent studies have used genetically encoded Ca²⁺ indicators such as GCaMP-3 or GECO fused directly to TRPML1 to record Ca²⁺ signals in response to agonist activation (Bae et al., 2014; Cao et al., 2015; Medina et al., 2015; Shen et al., 2012). These signals are suggested to reflect local peri-lysosomal Ca²⁺ release. But it is not clear how the relatively high-affinity Ca²⁺ indicators used distinguish between local and global Ca²⁺ signals.

In addition to supporting local Ca²⁺ fluxes required for endocytic ‘well-being’ (Hockey et al., 2015), lysosomes can also initiate global Ca²⁺ signals during signalling. This process has been established in the context of the actions of the messenger nicotinic acid adenine dinucleotide phosphate (NAADP) (Galione, 2015). According to the trigger hypothesis, increases in NAADP levels activate endo-lysosomal Ca²⁺ channels, generating local Ca²⁺ signals that sensitise neighbouring Ca²⁺ release channels on the ER. Activation of the latter (on the more substantial store) results in global Ca²⁺ release. Indeed, in fibroblasts, direct mobilisation of lysosomal Ca²⁺ stores through osmotic permeabilisation is sufficient to trigger ER-dependent complex Ca²⁺ signals (Kilpatrick et al., 2013). Physiologically, it is the TPCs that have emerged as the target channels for NAADP (Brailoiu et al., 2009; Calcraft et al., 2009; Patel, 2015) although a role for TRPMLs has also been advanced (Zhang et al., 2011, but see Yamaguchi et al., 2011). Coupling between TPCs and ER Ca²⁺ channels likely occurs at membrane contact sites between lysosomes and the ER (Kilpatrick et al., 2013; Patel and Brailoiu, 2012; Penny et al., 2014). Although TRPMLs can evoke global Ca²⁺ entry, at least when mutated as exemplified by the Va mutation in TRPML3 (encoded by MCOLN3) (Grimm et al., 2007; Kim et al., 2007), whether they support lysosome–ER cross talk is not known.

Here, we take advantage of TRPML agonists ML-SA1 (Shen et al., 2012) and MK6-83 (Chen et al., 2014) to probe the Ca²⁺ permeability of TRPMLs in an intact cell setting. We show that activation of TRPML1 unexpectedly evokes global Ca²⁺ signals. Mechanistically, we dissect these signals into release and influx components implicating both ER and extracellular Ca²⁺, in addition to lysosomal Ca²⁺, for their genesis. Our data suggest that the action of TRPML1 is not limited to local lysosomal Ca²⁺ signalling.

Fusion proteins comprising TRPML1 and either GCaMP-3 (Shen et al., 2012) or GECO (Cao et al., 2015) have been used to resolve presumed local Ca²⁺ fluxes from lysosomes upon TRPML1 activation. But as shown in Fig. 1A and Movie 1, Ca²⁺ responses to ML-SA1 (20 µM) were readily resolvable in a proportion of Hela cells loaded with the fluorescent Ca²⁺ indicator Fura-2. This organic dye distributes throughout the cytosol and thus records bulk changes in cytosolic Ca²⁺. ML-SA1 responses were concentration dependent (Fig. 1A). Similar results were obtained using primary cultured human fibroblasts. As in HeLa cells, ML-SA1 evoked strong Ca²⁺ responses in Fura-2-loaded cells in a concentration-dependent manner (Fig. 1B; Movie 2). The amplitude of the Ca²⁺ signals and the proportion of responsive cells are quantified in Fig. 1C,D. Fibroblasts were more responsive than HeLa cells, although we noted considerable variability between experiments, particularly for fibroblasts (Fig. 1D). To probe specificity, we examined the effects of the new TRPML antagonists ML-SI1 and ML-SI3 (Samie et al., 2013). ML-SI1 could not be used because it evoked Ca²⁺ signals (Fig. S1A). ML-SI3, however, inhibited ML-SA1-evoked Ca²⁺ signals (Fig. 1E,F). Collectively, these data suggest that, contrary to the prevailing view, activation of endogenous TRPMLs is capable of evoking global Ca²⁺ signals.

To investigate the mechanism underlying agonist-evoked Ca²⁺ signals, we examined the effects of expressing active TRPML1 and inactive TRPML1 mutated within the pore region (TRPML1^D471K) (Yamaguchi et al., 2011). Confocal analyses of HeLa cells expressing TRPML1 and TRPML1^D471K revealed a comparable punctate distribution consistent with localisation to lysosomes (Fig. 2A). Accordingly, there was marked colocalisation of both proteins with lysotracker (a fluorescent acidotrope) and LAMP1 (a lysosomal marker) (Fig. 2B–E).

In control experiments, ML-SA1 evoked Ca²⁺ signals in cells expressing LAMP1 (Fig. 3A). The responses were similar to those in un-transfected cells (Fig. 1). In cells expressing TRPML1, however, there was a large potentiation of the Ca²⁺ signal (Fig. 3A) such that essentially all of the cells responded (Fig. 3B,C). The potentiating effect of TRPML1 was specific because it was not observed with TRPML1^D471K (Fig. 3A–C). Similar results were obtained in fibroblasts (Fig. S1C,D) although this analysis was more limited due to the difficulty in transfecting these cells. Notably, ML-SA1-evoked Ca²⁺ signals were inhibited in cells expressing TRPML1^D471K relative to cells expressing LAMP1 (Fig. 3A–C). The inhibitory effect of TRPML1^D471K is likely due to dominant-negative activity resulting from oligomerisation with endogenous TRPML1 (Venkatachalam et al., 2006).

We also tested the effect of the structurally distinct mucolipin analogue MK6-83. Like ML-SA1, MK6-83 evoked large Ca²⁺ signals in cells expressing TRPML1 but not TRPML1^D471K (Fig. 3D,E). Ca²⁺ signals evoked by both ML-SA1 and MK6-83 in TRPML1-expressing cells were blocked by ML-SI3 (Fig. 3F–I). These molecular and chemical analyses establish the requirement for TRPML1 in global Ca²⁺ signalling.

To define the Ca²⁺ sources underlying TRPML1-dependent Ca²⁺ signals, we first established the relative contribution of store release and Ca²⁺ entry to the evoked responses. As shown in Fig. 4A,B, removal of extracellular Ca²⁺ reduced the peak response upon ML-SA1 stimulation and eliminated the sustained phase. ML-SA1-evoked Ca²⁺ signals in Ca²⁺-free medium were TRPML1 dependent, as judged by the lack of ML-SA1-evoked Ca²⁺ signals in TRPML1^D471K-expressing cells and upon treatment with ML-SI3 (Fig. 4C). These data suggest that ML-SA1 evokes both Ca²⁺ release and Ca²⁺ influx.

We used the lysosomotropic agent Gly-Phe β-naphthylamide (GPN) to disrupt lysosomes. This dipeptide causes osmotic destabilisation of lysosomes and other cathepsin-C-positive compartments (Jadot et al., 1984). GPN treatment resulted in a rapid loss of Lysotracker^® red fluorescence consistent with its action on lysosomes (Fig. S1E,F) and inhibited the responses to ML-SA1 indicating a requirement for lysosomes in ML-SA1-evoked Ca²⁺ release (Fig. 4D). These data are consistent with the lysosomal localisation of TRPMLs (Fig. 2). However, despite the complete compromise of lysosomes upon GPN treatment, the resulting Ca²⁺ signals were modest relative to those evoked by ML-SA1 (Fig. 4D). This suggested that ML-SA1 did not exclusively release Ca²⁺ from the lysosomes.

We considered the possibility that activation of TRPMLs might be followed by Ca²⁺ release from the ER in a manner similar to activation of TPCs by NAADP (Cancela et al., 1999; Patel et al., 2010). To test this, we examined the effects of ML-SA1 after depleting ER Ca²⁺ stores with thapsigargin. As shown in Fig. 4D, thapsigargin treatment blocked the responses to ML-SA1 indicating a clear requirement for ER Ca²⁺ stores in ML-SA1 action. To further probe the role of the ER, we used 2-APB to block inositol 1,4,5-trisphosphate (IP₃) receptors. As shown in Fig. 4D, 2-APB (100 µM) inhibited ML-SA1-evoked responses. A summary of these data is provided in Fig. 4E. We conclude that activation of TRPMLs on lysosomes triggers Ca²⁺ release from the ER.

Finally, we examined the nature of the Ca²⁺ entry pathway evoked by TRPML1 activation. Because store-operated Ca²⁺ entry is a ubiquitous pathway underlying Ca²⁺ influx in non-excitable cells, we tested the effects of the Ca²⁺ entry inhibitor BTP2. BTP2 (20 µM) completely prevented Ca²⁺ signals evoked by Ca²⁺ add-back following depletion of ER Ca²⁺ stores with thapsigargin (Fig. S2A,B). However, BTP2 did not affect the amplitude of Ca²⁺ signals evoked by ML-SA1 (Fig. 5A,B) although it did slow the rate of rise (Fig. S2C). These data suggest that store-operated Ca²⁺ entry does not play a major role in Ca²⁺ signals evoked by TRPML1 activation.

The above data raised the possibility that TRPML1 might mediate Ca²⁺ entry more directly. To test this, we examined the effects of ML-SA1 on Fe²⁺ entry. Fe²⁺ permeates TRPML1 and, like several other metal ions, quenches Fura-2 fluorescence (Dong et al., 2008). Fura-2-loaded HeLa cells were therefore challenged with FeCl₂ (1 mM) in nominally Ca²⁺-free medium after stimulation with ML-SA1. FeCl₂ induced a significant quench of Fura-2 fluorescence in cells expressing TRPML1 (Fig. 5C). This effect was blocked by ML-SI3 (Fig. 5C,D) and was not observed in cells expressing TRPML1^D471K (Fig. 5G,H). Similar results were obtained upon stimulation with MK6-83 (Fig. 5E–H). Challenge with FeCl₃ (1 mM) after stimulation with ML-SA1 was without effect (Fig. 5I,J) attesting to specificity. BTP2 slowed Fe²⁺ entry (Fig. S2D,E), however, add-back of Fe²⁺ following depletion of Ca²⁺ stores with thapsigargin did not result in a significant quench (Fig. S2F,G).

Collectively, these data provide evidence that TRPML1 supports divalent cation entry across the plasma membrane.

In this study, we have shown that activation of TRPMLs by two structurally distinct agonists, in both a human cell line and primary cultures, evokes global Ca²⁺ signals. These data indicate that TRPMLs do not solely mediate local Ca²⁺ release events from endo-lysosomal compartments. Indeed, we provide evidence that activation of TRPMLs can evoke both Ca²⁺ release and Ca²⁺ entry.

We have dissected global Ca²⁺ release signals into both lysosomal and ER components. There is much evidence supporting the transmission of Ca²⁺ signals from endo-lysosomes to the ER but to date such ‘chatter’ has been ascribed exclusively to the actions of NAADP and TPCs (Patel and Brailoiu, 2012). Our study extends such coupling to TRPMLs, raising the possibility that other Ca²⁺-permeable endo-lysosomal channels such as P2X4 receptors and TRPM2 might also functionally couple to ER Ca²⁺ release channels (Patel and Cai, 2015).

Inhibition of TRPML1 by plasma membrane PI(4,5)P₂ is thought to ensure channel activation only in endo-lysosomal compartments (Zhang et al., 2012). In accordance with this idea, basal (unstimulated) whole-cell currents are not detectable in TRPML1-expressing cells. SF-51 (Grimm et al., 2010), the precursor to ML-SA1, however, has been reported to mediate a TRPML1 plasma membrane current (Zhang et al., 2012) consistent with our Ca²⁺ and Fe²⁺ entry measurements. Divalent cation entry might therefore result from targeting of a small fraction of TRPML1 to the plasma membrane. Although blockade of store-operated channels reduced TRPML1-dependent Ca²⁺ and Fe²⁺ entry (albeit modestly), the amplitude of the Ca²⁺ signal upon maximal Ca²⁺ ER store depletion was substantially smaller than that evoked by agonist activation. Indeed, store-operated Fe²⁺ entry was not demonstrable under our conditions. These data raise the possibility that BTP2 might directly inhibit TRPML1 similar to its effects on TRP-canonical (TRPC) channels (He et al., 2005).

The multi-compartment actions of TRPML1 offer fresh perspective on interpreting Ca²⁺ signals recorded using Ca²⁺ reporters fused to TRPML1 (Bae et al., 2014; Cao et al., 2015; Medina et al., 2015; Shen et al., 2012). We suggest that such signals might additionally reflect Ca²⁺ release from the ER and Ca²⁺ entry. Indeed, defects in Ca²⁺ signals reported using such indicators in scenarios of TRPML1 dysfunction are perplexing given that expression of TRPML1 is able to rescue phenotypic defects such as lysosomal storage (Cao et al., 2015; Shen et al., 2012). Use of inert lysosomal proteins for targeting (McCue et al., 2013) and/or low-affinity Ca²⁺ indicators, to better insulate against bulk cytosolic Ca²⁺ changes, might aid in isolating local lysosomal Ca²⁺ release events. Finally, how TRPMLs are endogenously activated remains to be established, but PI(3,5)P₂ is a clear candidate raising the possibility that changes in its levels in response to cues such as growth factors might regulate global Ca²⁺ dynamics.

HeLa cells and primary cultured human skin fibroblasts from healthy individuals (Kilpatrick et al., 2013) were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum, 100 units/ml penicillin and 100 µg/ml streptomycin (all from Invitrogen) at 37°C in a humidified atmosphere with 5% CO₂. Cells were passaged using trypsin (HeLa) or scraping (fibroblasts) and plated onto 13-mm glass coverslips prior to microscopy. For HeLa cells, coverslips were coated with poly-L-lysine (Sigma). Cells were transfected with plasmids using Lipofectamine 2000 according to the manufacturer's instructions. Plasmids encoding the following proteins were described previously: C-terminally GFP-tagged LAMP1 (Falcón-Pérez et al., 2005), C-terminally mRFP-tagged LAMP1 (Sherer et al., 2003) and N-terminally GFP-tagged wild-type or D471K human TRPML1 (Yamaguchi et al., 2011).

All live imaging experiments were performed in HEPES-buffered saline (HBS) comprising (in mM): 1.25 KH₂PO₄, 2 CaCl₂, 2 MgSO₄, 3 KCl, 156 NaCl, 10 glucose, and 10 HEPES (pH 7.4; all from Sigma). For measurement of cytosolic Ca²⁺ concentration and Fe²⁺ or Fe³⁺ entry, cells were incubated with Fura-2 AM (2.5 µM) and 0.005% v/v pluronic acid (from Invitrogen) for 1 h. For measurement of lysosome distribution, cells were incubated with Lysotracker^® Red (100 nM) (Invitrogen) for 15 min. After labelling, cells were washed with HBS and mounted in a 1-ml imaging chamber (Biosciences Tools) prior to microscopy. For measurement of colocalisation, cells were co-transfected with GFP–TRPML1 or GFP–TRPML1^D471K together with LAMP1–mRFP and fixed with 4% paraformaldehyde prior to confocal microscopy. For visualisation of nuclei, cells were incubated with 1 µg/ml DAPI for 5 min.

Epifluorescence images were captured every 3 s with a cooled coupled device camera (TILL photonics) attached to an Olympus IX71 inverted fluorescence microscope fitted with a 20× objective, and a monochromator light source. Fura-2 was excited at 340, 360 or 380 nm, and Lysotracker^® Red at 560 nm. Emitted fluorescence was captured using a 440-nm long-pass filter (Fura-2) and 590-nm filter (Lysotracker^® Red). TRPML1-expressing cells were identified by monitoring fluorescence of GFP (excitation 488 nm, emission 505 nm). Cells were stimulated with ML-SA1, GW405833 hydrochloride (Sigma; also known as ML-SI1, Haoxing Xu, Department of Molecular, Cellular, and Developmental Biology, University of Michigan, USA, personal communication), ML-SI3 (a kind gift from Haoxing Xu), MK6-83 synthesised as described previously (Chen et al., 2014), GPN (Santa Cruz Biotechnology), BTP2 (Sigma), thapsigargin (Merck), 2-aminoethoxydiphenyl borate (2-APB, Sigma). Where indicated, Ca²⁺ was omitted from the HBS (nominally Ca²⁺-free) or replaced with 1 mM EGTA (Ca²⁺-free). For Fe²⁺ or Fe³⁺ quench experiments, cells were stimulated using freshly prepared 100 mM stock solutions of either FeCl₂ or FeCl₃ (Sigma). The maximal fluorescence ratio change (Ca²⁺) or the fractional quench at 360 nm excitation (Fe²⁺ or Fe³⁺) were quantified on an individual cell basis and averaged for all cells in a given field of view. Experiments were repeated at least three times and all individual data points plotted. The total number of cells analysed is stated in the legends.

Confocal images were captured using an LSM510 confocal scanner (Zeiss) attached to a Zeiss Axiovert 200M inverted microscope fitted with a 63× Plan Apochromat water-immersion objective. DAPI, GFP and Lysotracker^® Red or mRFP were excited at 364 nm, 488 nm and 543 nm, and emitted fluorescence was captured using 385–470 nm, 505–550 nm and 560–615 nm band-pass filters, respectively. Pearson's correlation coefficients were calculated from z-stacks (1-µm intervals, 8–12 slices) using the ImageJ plugin Coloc2. The fraction of colocalised vesicles was calculated from the middle slice or slices using the ImageJ plugin SQUASSH.

We thank Shmuel Muallem and Haoxing Xu for provision of reagents, and Manuela Melchionda, Anthony Morgan, Christopher Penny and Jonathan Soboloff for useful discussion.