An Endosomal NAADP-Sensitive Two-Pore Ca2+ Channel Regulates ER-Endosome Membrane Contact Sites to Control Growth Factor Signaling

1How organelles communicate is a fundamental question that arises given the compartmentalized nature of eukaryotic cell function. Although vesicular traffic is an established means of information transfer, it is becoming clear that traffic also proceeds by non-vesicular means. In particular, membrane contact sites have emerged as potential platforms for both Ca2+ signaling and lipid transfer. Membrane contact sites are regions of close apposition between membranes that are stabilized by tethering complexes. The endoplasmic reticulum (ER) forms multiple classes of contacts with both the plasma membrane and organelles such as endosomes, lysosomes, and mitochondria. Endosome-ER contacts have been implicated in endosome positioning, dephosphorylation of internalized receptors, and components of the endosomal sorting complex required for transport (ESCRT) machinery, endosome fission, actin nucleation and retromer-dependent budding, and cholesterol transport. We have identified multiple populations of contact sites that form between the ER and different endocytic organelles, which include those dependent on VAPs. Notably, contact sites between the ER and EGF receptor-containing endosomes require annexin-A1 and its Ca2+-dependent binding partner S100A11, raising the possibility that Ca2+ fluxes may regulate contact.

2Ca2+ is a widespread signaling ion regulating a range of cellular processes including aspects of vesicle formation, fusion, and traffic. Ca2+ signals often invade the cell entirety (global) but they can also be spatially restricted (local), as exemplified by signals generated by the Ca2+-mobilizing messenger, nicotinic acid adenine dinucleotide phosphate (NAADP). NAADP is unusual in mediating Ca2+ release from the endo-lysosomal system, an acidic Ca2+ store filled by Ca2+/H+ exchange. It does so by activating two-pore channels (TPCs). Local NAADP-mediated Ca2+ release events from acidic organelles are amplified by Ca2+ channels on canonical Ca2+ stores of the ER to generate global signals. This occurs during signaling by external cues such as hormones and neurotransmitters. However, it is also evident that local TPC-mediated Ca2+ release events function in a constitutive manner. For instance, NAADP/TPC signaling regulates several membrane trafficking events, including retrograde traffic from endosomes to the Golgi and the trafficking of cholesterol, receptors, and viruses. This pathway also regulates endo-lysosomal morphology, likely through Ca2+-dependent vesicular fusion/fission events. However, what role TPCs play in non-vesicular trafficking is unexplored.

3Here, we reveal an essential requirement for NAADP and TPC1 in regulating membrane contact site formation between endosomes and the ER to control growth factor signaling.

4We examined the effect of inhibiting NAADP action on late endosome and lysosome morphology in primary human fibroblasts using four approaches.

5First, we tested NAADP antagonists. Figures 1A and 1B show the effect of an overnight treatment with Ned-19 on late endosome and lysosome morphology as assessed by immuno-fluorescence staining and confocal microscopy of the late endosome and lysosome marker LAMP1. Labeled structures were clustered in the perinuclear region and often appeared enlarged (Figure 1B; changes in staining intensity quantified in Figure 1H). Similar results were obtained with the recently described Ned-19 analog, Ned-K (Figure 1C) and upon shorter (2-hr) treatment with the antagonists (Figures S1A–S1C and S1I). Analysis of multiple individual labeled structures revealed an increase in the mean area (Table S1). LAMP1 protein levels were similar upon Ned-19 treatment (Figure S1J). We further examined the ultrastructure of the endo-lysosomal system by electron microscopy (EM). Consistent with results using light microscopy, late endosomes and electron-dense lysosomes were often clustered and more vacuolar in Ned-19-treated cells compared with controls (Figure 1D; quantified in Figure 1I). Immuno-EM confirmed that LAMP1 localizes to late endosome and lysosome clusters in Ned-19-treated cells (Figure S1K).

6Second, we pharmacologically targeted the TPC pore. Recent studies have shown that TPCs are inhibited by the plant alkaloid tetrandrine. Like NAADP antagonists, tetrandrine clustered late endosome and lysosomes and induced particularly pronounced vesicle enlargement (Figures 1E, 1J, S1D, and S1I; Table S1). TPCs have also emerged as targets for L-type Ca2+ channel blockers. Therefore, we examined the effect of isradipine and nifedipine (both dihydropyridines), diltiazem (a benzothiazapine), and verapamil (a phenylalkylamine). As shown in Figures S1E–S1I and Table S1, all three structurally distinct classes of inhibitors induced aggregation/vesicle enlargement following an acute treatment.

7Third, we buffered Ca2+ levels. Treatment of cells with the cell-permeable form of the Ca2+ chelator, 1,2-bis(o-aminophenoxy)ethane-N,N,N′,N′-tetraacetic acid (BAPTA), also altered the appearance of late endosome and lysosomes (Figure 1F). In contrast, treatment with the slower Ca2+ chelator EGTA did not (Figure 1G). This differential sensitivity to chelators summarized in Figure 1J suggests that morphology of the endo-lysosomal system is regulated by local Ca2+ fluxes.

8Fourth, to directly probe the role of TPCs in endo-lysosomal morphology, we examined the effect of TPC knockdown. Treatment of fibroblasts with small interfering RNAs (siRNAs) targeting two independent sequences in both TPC1 or TPC2 reduced transcript levels >50% (Figure S2A). Knockdown of TPC1 protein was confirmed by western blotting (Figures S2B and S2C). As shown in Figures 2A–2C, confocal microscopy of TPC1-silenced cells revealed marked changes in LAMP1 staining (quantified in Figure 2F and Table S1), similar to chemical blockade of NAADP signaling (Figure 1). TPC2 silencing, however, had little effect (Figures 2D and 2E). In accord, late endosome and lysosomes appeared more clustered and less distinct upon TPC1, but not TPC2, silencing at the ultrastructural level (Figures 2G–2K, quantified in Figure 2L). To assess specificity of our molecular manipulations, we performed rescue experiments with a siRNA-resistant TPC1 construct. Expression of this construct in TPC1-depleted cells partially reversed clustering of late endosome and lysosomes as assessed by both LAMP1 immunocytochemistry (Figures 3A and 3B) and correlative light and electron microscopy (CLEM) (Figures 3C and 3D), thereby attesting to specificity. Late endosome and lysosomal morphology was also unchanged by Ned-19 in cells where both TPC1 and TPC2 had been silenced (Figures S2D–S2F), further attesting to specificity. Late endosome and lysosomal form is thus specifically determined by TPCs in an isoform-selective manner.

9Taken together, we identify an unexpected role for NAADP, a target channel, and associated local Ca2+ fluxes in maintaining late endosome and lysosomal morphology.

10The distribution of TPC1 within the endocytic pathway is unclear and likely more diffuse than that of TPC2, which is expressed predominantly on lysosomes. Immuno-EM revealed localization of GFP-tagged or untagged TPC1 to the limiting membrane of multi-vesicular endosomes, rather than electron-dense lysosomes, with some additional ER localization that might be related to ectopic expression (Figures 4A and S3A–S3C). Parallel studies with TPC2, however, were ambiguous because expression of TPC2 resulted in a proliferation of endocytic organelles with disorganized membranous content, often aggregated in clusters (Figure 4A). TPC2-mediated disruption of late endosome-lysosomal morphology is consistent with our previous analysis. Intriguingly, we found that TPC1 was often found at membrane contact sites between endosomes and the ER (Figures 4B and S3A–S3C). Quantitative analysis of TPC1-positive endosomes showed that there was a ∼5-fold increase in the number of gold particles/unit endosomal membrane in contact with the ER compared to regions not associated with the ER (Figure 4B). TPC1 thus emerges as a component of ER-endosome contacts.

11The presence of TPC1 at ER-endosome contact sites raised the possibility that local Ca2+ signals deriving from the endosome may regulate contact with the ER. We therefore examined the effect of NAADP blockade on ER-endosome contact site formation (Figure 4C). As shown in Figure 4D, Ned-19 reduced the percentage of endosomes with an ER contact. Similar results were obtained in HeLa cells where Ned-19 reduced ER-endosome contact site formation in a concentration-dependent manner (Figure S3D). The effects of Ned-19 were recapitulated by knocking down TPC1, whereas depletion of TPC2 had little effect on ER-endosome contacts in fibroblasts (Figure 4D). Chemical and molecular inhibition of NAADP signaling on contact site formation thus mirrors the effect on gross late endosome and lysosomal morphology (Figures 1 and 2).

12The ER forms contacts with lysosomes that are biochemically distinct from those with endosomes. To assess specificity of our manipulations, we examined the effect of NAADP blockade on ER-lysosome contact site formation (Figure 4C). Ned-19 had little effect on the percentage of lysosomes with an ER contact in both fibroblasts (Figure 4E) and HeLa cells (Figure S3D). Silencing of TPC1 was also largely without effect, whereas silencing of TPC2 reduced ER-lysosome contact sites (Figure 4E).

13In summary, these data uncover a highly isoform- and compartment-specific role for NAADP signaling in the formation of membrane contacts between endosomes and the ER.

14We have shown previously that ER-endosome contacts enable the interaction between endocytosed EGF receptor (EGFR) tyrosine kinase and the protein tyrosine phosphatase, PTP1B, on the ER. This contact allows receptor de-phosphorylation—a determinant of signaling by EGF. We reasoned that disrupting NAADP signaling would enhance EGF signaling due to compromised contact at the ER-endosome interface. To test this, we examined the effect of Ned-19 on the phosphorylation state of EGFR. Acute stimulation with EGF induced a transient rise in EGFR tyrosine phosphorylation that was significantly enhanced and prolonged by Ned-19 treatment (Figure 5A, quantified in Figures 5B, S4A, and S4B). For these experiments, we used HeLa cells due to their high EGFR levels, but similar results were obtained in fibroblasts (Figures S4B–S4D).

15To examine downstream functional consequences of disrupting contacts, we took two approaches.

16First, we measured ERK activity because activated EGFRs are classically coupled to the MAP kinase pathway. Consistent with the prolonged EGFR activation observed in cells treated with Ned-19, tyrosine phosphorylation of ERK1/2 was significantly increased and extended upon NAADP inhibition (Figures 5C and 5D).

17Second, we measured cytosolic Ca2+ levels. Activated EGFRs also couple to phospholipase C-gamma (PLCγ), which generates inositol 1,4,5-trisphosphate, resulting in Ca2+ release from ER Ca2+ stores. EGF evoked readily measurable Ca2+ signals that were exaggerated and more sustained in cells treated with Ned-19 for 2 hr or overnight (Figures 5E–5H). Total levels of the EGFR were unchanged (Figures S4E–S4G), indicating that enhanced signaling effects are not attributable to increased expression of EGFR. These data are again consistent with prolonged activation of EGFR due to perturbed contact at NAADP-sensitive endosome-ER contact sites.

18Taken together, these data reveal a role for NAADP in regulating both EGFR activity and downstream signaling by MAP kinase and phospholipase C.

19Membrane contact sites between endosomes and the ER are gaining much attention as novel coordinators of cell function. Contact with the ER is regulated by cholesterol and increases as endocytic vesicles mature. Previous studies have shown marked upregulation of contacts in response to expression of STARD3/STARD3NL, ORP1L, and protrudin, but whether these proteins are necessary for contact site formation is less clear. Here, we use EM, which, relative to light microscopy, is better suited for resolving inter-organelle junctions to provide direct evidence that TPC1 is a contact site component (Figure 4). Importantly, inhibiting TPC1 activity using chemical and molecular means significantly decreased ER contact site formation with late endosomes (Figure 4). This was associated with disruptions in late endosome and lysosome morphology (Figures 1 and 2), although whether TPC1 and LAMP1 co-localize to junction-forming organelles remains to be established. Consistent with a role for ER-endosome contact sites in endosomal positioning, we observed a more perinuclear population of LAMP1-positive late endosome and lysosomes when contact was reduced by inhibition of NAADP or TPC1 (Figures 1 and 2). We thus provide key evidence implicating endogenous NAADP signaling in regulating ER-endosome contact and the subcellular distribution of endocytic organelles.

20Localized Ca2+ release from endocytic organelles is a driver of endosome-lysosome fusion. In accord, TPCs associate with the fusion apparatus and have been implicated in a number of classical organellar vesicular trafficking events. Our identification of ER-endosome contact sites dependent on annexin A1 and NAADP (this study) establishes a paradigm whereby localized Ca2+ release from endocytic organelles might regulate non-vesicular traffic. Of relevance are recent studies showing that Ca2+ regulates the formation of contacts between the ER and the plasma membrane. Notably, the measured affinity for Ca2+ is in the low micromolar range, suggesting that large global signals, such as those evoked during Ca2+ influx, regulate these junctions. However, because of the restricted volume at contacts, it is possible that even modest, constitutive fluxes in unstimulated cells, as alluded to here, could achieve the necessary Ca2+ concentrations to modulate Ca2+-dependent contacts. The effects of interfering with NAADP/TPC signaling on contact sites correlated well with effects on gross late endosome and lysosomal morphology. However, possible endosome-lysosome fusion defects might also contribute to morphological changes.

21Activated EGF receptors undergo internalization onto endosomes where they continue to signal in their phosphorylated form until they are dephosphorylated by PTP1B on the ER. Disrupting ER-endosome contacts by inhibiting NAADP prolongs EGFR phosphorylation (Figure 5). These data support our findings that EGFR dephosphorylation occurs at endosome-ER contact sites populated by PTP1B and the Ca2+-binding protein annexin A1. Importantly, we also report that disruption of NAADP-dependent contacts substantially enhances downstream signaling by EGF through both ERK and PLCγ (Figure 5). Whether EGF receptors are coupled to NAADP production similar to vascular endothelial growth factor (VEGF) receptors is not known at present. Rather our data identify NAADP as a negative regulator of EGF action through local signaling at the endosome-ER interface (Figure S4H). Many external stimuli elevate NAADP levels, raising the possibility that second messenger and mitogenic signaling crosstalk may occur through modulation of contact site strength.

22In summary, our findings provide new information on the molecular makeup, regulation, and function of ER-endosome contact sites.

23Primary cultured human fibroblasts and HeLa cells were maintained in 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 by scraping (fibroblasts) or with trypsin (HeLa cells) and plated onto coverslips (for immunocytochemistry, Ca2+ imaging, and electron microscopy) or directly onto tissue culture plates/flasks (for western blotting) before experimentation.

24For chemical treatments, drugs were dissolved in DMSO or H2O, diluted into culture medium, and then sterile filtered. The NAADP antagonists, trans-Ned-19 and Ned-K were synthesized as described by and, respectively. Both were kind gifts from Raj Gossain, A. Ganesan, and Sean M. Davidson. BAPTA-AM was from Biovision. EGTA-AM was from AnaSpec. Tetrandrine was from Santa Cruz. Isradipine, nifedipine, verapamil, and diltiazem were from Sigma.

25Knockdown was performed by transfecting cells with control siRNA duplex (Allstars Negative Control siRNA; QIAGEN) and duplexes targeting human TPC1 (#1, AGCUGUAUUUCAUCAUGAAtt; #2, GGCUACUAUUAUCUCAAUAtt) and TPC2 (#1, GGUGGGACCUCUGCAUUGAtt; #2, CGGUAUUACUCGAACGUAAUtt) (#1, QIAGEN; #2, Ambion) using Lipofectamine RNAiMAX as described previously. For overexpression, cells were transfected with untagged TPC1 or C-terminally GFP-tagged human TPC1 and TPC2 as described by, or a rescue construct in which the target sequence of siRNA#1 in TPC1-GFP was mutated to TGAACTCTACTTTATA using the Q5 site-directed mutagenesis kit (NEB) with forward (actttataATGAACCTGCTTCTGGCT) and reverse (agagttcaATGGAGAGGTACACGATG) primers.

26Stimulation with EGF (100 ng/mL; Sigma) was performed following serum starvation for 1 hr in serum-free DMEM.

27LAMP1 labeling and confocal microscopy were performed as described by. Briefly, all images were acquired under identical confocal settings and mean fluorescence intensity per cell quantified to allow comparison between controls and the various treatments. Automated size analysis of labeled structures was performed using the “Analyze Particle” function in Fiji from binary images created by local thresholding using the Bernsen algorithm (5-pixel radius with the default contrast threshold) and Watershed segmentation.

28EM was performed essentially as described. Briefly, serum-starved cells were stimulated with EGF and BSA-gold. After fixation in paraformaldehyde (PFA)/glutaraldehyde, cells were post-fixed in osmium tetroxide/potassium ferricyanide and embedded. Clustering was quantified by calculating the area (using Fiji) of three or more late endosomes/lysosomes in close apposition relative to the cytoplasmic area (excluding the nucleus). For pre-embedding labeling, cells were fixed in PFA, permeabilized with digitonin, and incubated with primary and nanogold-secondary antibodies prior to fixation for EM. ER-endosome contact sites in random sections were defined as regions where apposing membranes were <30 nm apart, with no minimum length. For correlative light and electron microscopy (CLEM), cells were plated on gridded dishes, fixed in 4% PFA, and imaged by light microscopy (Nikon Ti-E), prior to preparation for conventional EM as above.

29Cytosolic Ca2+ concentration was measured essentially as described. In brief, cells were loaded with the fluorescent Ca2+ indicator Fura-2 in HEPES-buffered saline. Dual excitation time-lapse fluorescence imaging was performed using a CCD camera. Data are presented as fluorescence ratios upon excitation at 340 and 380 nm.

30Quantitative PCR and western blotting were performed exactly as described by. The primary antibodies used for western blotting were anti-LAMP1 (mouse; Santa Cruz; 1/500; 1 hr room temperature [RT]), anti-TPC1 (rabbit; Abcam; 1/200; 1 hr RT), anti-actin (goat; Invitrogen; 1/500; 1 hr RT), anti-phosphotyrosine 1068 EGFR (rabbit; Cell Signaling; 1/2,000; 1 hr RT), anti-EGFR (sheep; Fitzgerald; 1/2000; 1 hr RT), anti-phosphotyrosine 204 ERK1/2 (mouse; Santa Cruz; 1/1,000), and anti-ERK1/2 (rabbit; Cell Signaling; 1/1,000).

31B.S.K. performed the cell culture, knockdowns, and Ca2+ imaging. B.S.K. and E.R.E. performed the western blotting. E.R.E. performed the electron microscopy. L.N.H. and B.S.K. performed the drug treatments, immunocytochemistry, and confocal microscopy. B.S.K. and E.Y. performed the qPCR and LAMP1 area analysis. C.E.F. and S.P. conceived and jointly supervised the study. S.P. and B.S.K. wrote the paper with input from all authors.