Most intracellular Ca2+ signals result from opening of Ca2+ channels in the plasma membrane or endoplasmic reticulum (ER), and they are reversed by active transport across these membranes or by shuttling Ca2+ into mitochondria. Ca2+ channels in lysosomes contribute to endo-lysosomal trafficking and Ca2+ signalling, but the role of lysosomal Ca2+ uptake in Ca2+ signalling is unexplored. Inhibition of lysosomal Ca2+ uptake by dissipating the H+ gradient (using bafilomycin A1), perforating lysosomal membranes (using glycyl-L-phenylalanine 2-naphthylamide) or lysosome fusion (using vacuolin) increased the Ca2+ signals evoked by receptors that stimulate inositol 1,4,5-trisphosphate [Ins(1,4,5)P3] formation. Bafilomycin A1 amplified the Ca2+ signals evoked by photolysis of caged Ins(1,4,5)P3 or by inhibition of ER Ca2+ pumps, and it slowed recovery from them. Ca2+ signals evoked by store-operated Ca2+ entry were unaffected by bafilomycin A1. Video-imaging with total internal reflection fluorescence microscopy revealed that lysosomes were motile and remained intimately associated with the ER. Close association of lysosomes with the ER allows them selectively to accumulate Ca2+ released by Ins(1,4,5)P3 receptors.
Because Ca2+ can be rapidly moved across biological membranes to generate spatially organized increases in cytosolic free Ca2+ concentration ([Ca2+]i) (Smith and Parker, 2009), it is a versatile and ubiquitous intracellular messenger (Berridge et al., 2003; Rizzuto and Pozzan, 2006). Most Ca2+ signals are generated by regulated opening of Ca2+-permeable channels within either the plasma membrane (PM) or the membranes of intracellular organelles, most commonly the endoplasmic reticulum (ER). Within the latter, inositol 1,4,5-trisphosphate receptors [Ins(1,4,5)P3R] are almost invariably expressed (Foskett et al., 2007). In addition, the ER often expresses ryanodine receptors (RyR) and a variety of other Ca2+-permeable channels (Taylor and Dale, 2012). Regulated opening of these channels allows Ca2+ to flow rapidly into the cytosol. This, together with the high concentration of cytosolic Ca2+ buffers, allows an open Ca2+ channel to generate a local, but substantial, increase in [Ca2+]i that decays within about a hundred nanometres of the channel and persists only for as long as the channel is open (Shuai and Parker, 2005). These local microdomains of Ca2+ around the mouths of open channels are the fundamental units of Ca2+ signalling. They allow different channels to direct Ca2+ to different effector systems (Rizzuto and Pozzan, 2006). They allow Ca2+-mediated communication between Ca2+ channels to facilitate regenerative growth of Ca2+ signals by Ca2+-induced Ca2+ release, usually via Ins(1,4,5)P3R (Calcraft et al., 2009; Rahman et al., 2009; Smith and Parker, 2009) or RyR (Brailoiu et al., 2010; Cheng and Lederer, 2008). And where Ca2+ channels are close to another membrane, they allow Ca2+ to be selectively directed to proteins that remove Ca2+ from the cytosol. The Ca2+ pump of the ER, for example, can sequester Ca2+ entering via store-operated Ca2+ entry (SOCE) (Parekh and Putney, 2005); and mitochondria can selectively accumulate Ca2+ released by Ins(1,4,5)P3R (Csordás et al., 2010) or entering via SOCE (Parekh and Putney, 2005). These intimate relationships between Ca2+ channels and proteins that sequester, release or respond to Ca2+ endow the fundamental units of Ca2+ signalling, the local increase in [Ca2+]i evoked by opening of a single channel, with the versatility needed to regulate almost every cellular activity (Berridge et al., 2003).
The membranes that have attracted most attention in the context of Ca2+ signalling are the PM, the ER and the inner mitochondrial membrane. All three membranes express proteins that remove Ca2+ from the cytosol and so terminate Ca2+ signals. Mitochondrial Ca2+ uptake additionally links cytosolic Ca2+ signals to oxidative phosphorylation (Jouaville et al., 1999), mitochondrial motility (Yi et al., 2004) and cell death (Szalai et al., 1999). The ER and PM express a wide variety of Ca2+ channels that can generate cytosolic Ca2+ signals (Taylor and Dale, 2012; Taylor et al., 2009). The recent identification of regulated Ca2+ channels in lysosomes, notably the two-pore channels (TPC) gated by nicotinic acid dinucleotide phosphate (NAADP) (Calcraft et al., 2009) and various transient receptor potential (TRP) channels (Dong et al., 2010), has suggested that they too may contribute to cytosolic Ca2+ signalling. Lysosomes are the main degradative compartment of the cell, they contain substantial amounts of Ca2+ (Christensen et al., 2002; Morgan et al., 2011), membrane trafficking within the endo-lysosomal pathway is regulated by Ca2+ (Luzio et al., 2010), and dysfunctional lysosomes cause devastating lysosomal storage diseases (Vellodi, 2005), some of which are associated with defective lysosomal Ca2+ regulation (Lloyd-Evans and Platt, 2011; Morgan et al., 2011; Pereira et al., 2010). There is, therefore, a growing recognition that lysosomal Ca2+ channels contribute to the genesis of cytosolic Ca2+ signals and to regulation of endo-lysosomal membrane trafficking. However, the mechanism of Ca2+ uptake by mammalian lysosomes is not resolved (Lloyd-Evans and Platt, 2011) and nor has the contribution of lysosomal Ca2+ uptake to the shaping of Ca2+ signals evoked by conventional Ca2+ signalling pathways been established. Here, we show that intimate and dynamic associations between lysosomes and the ER allow lysosomes selectively to accumulate Ca2+ released from the ER. Lysosomes thereby shape the cytosolic Ca2+ signals evoked by activation of Ins(1,4,5)P3R without affecting those evoked by SOCE.
Dissipating the lysosomal H+ gradient enhances carbachol-evoked Ca2+ signals
Activation of endogenous muscarinic receptors in human embryonic kidney (HEK) cells with carbachol (CCh) stimulates phospholipase C (PLC), formation of Ins(1,4,5)P3, release of Ca2+ from intracellular stores, and stimulation of Ca2+ entry (Tovey et al., 2008). The initial release of Ca2+ from intracellular stores is entirely mediated by Ins(1,4,5)P3R and unaffected by inhibition of RyR or TPC (supplementary material Fig. S1).
Ca2+ accumulation by acidic organelles, including lysosomes, requires a transmembrane H+ gradient. It is probably mediated by a Ca2+/H+ exchanger or indirectly via a Ca2+/Na+ exchanger. The identities of these exchangers in mammalian cells are unknown (Klemper, 1985; Morgan et al., 2011). We used bafilomycin A1 to inhibit the vacuolar type H+-ATPase (V-ATPase) (Yoshimori et al., 1991), dissipate the H+ gradient and so prevent Ca2+ accumulation by acidic organelles (Fig. 1A). LysoTracker Red, a weak base that accumulates in acidic organelles (Bucci et al., 2000), substantially colocalised with a GFP-tagged lysosomal membrane protein (lysosomal-associated membrane protein 1; LAMP1–GFP) (Fukuda, 1991) transiently expressed in HEK cells (supplementary material Fig. S2). Bafilomycin A1 (1 µM, 1 h) almost abolished staining with LysoTracker Red without affecting the distribution of LAMP1–mCherry (supplementary material Fig. S3). This demonstrates that lysosomes are the major acidic organelles in HEK cells, and that bafilomycin A1 discharges their H+ gradient without affecting their distribution.
Pre-incubation of HEK cells with bafilomycin A1 caused the increase in [Ca2+]i evoked by a maximally effective concentration of CCh to increase by 1.7±0.1-fold without affecting the sensitivity to CCh (pEC50 = 4.3±0.2 and 4.5±0.2 for control and bafilomycin A1-treated cells, respectively; where pEC50 is the −log of the half-maximally effective concentration; Fig. 1A,B). Bafilomycin A1 also caused a small, but significant (P = 0.0035), increase in the [Ca2+]i of unstimulated cells from 30±2 nM to 46±2 nM (n = 5). This modest effect of bafilomycin A1 is probably due largely to basal activity of Ins(1,4,5)P3R since it was abolished by inhibition of PLC (supplementary material Fig. S1E). Analyses of single cells confirmed the results from cell populations. The number of cells responding to a maximal concentration of CCh was similar for control and bafilomycin A1-treated cells (98±1% and 96±3%, respectively) and the increase in [Ca2+]i was 1.9±0.3-fold greater after treatment with bafilomycin A1 (Fig. 1C; supplementary material Table S1).
Stimulation of HEK cells with a submaximal concentration of CCh (50 µM) evoked sustained Ca2+ oscillations, typical of the responses of many cells to physiological stimuli (supplementary material Fig. S4A,B). Bafilomycin A1 had no effect on the number of cells in which CCh evoked Ca2+ oscillations (38±9% and 50±6% in control and bafilomycin-treated cells, respectively), but it increased the amplitude of the Ca2+ spikes and reduced their frequency (supplementary material Fig. S4C,D).
The affinities of Ca2+ indicators and other Ca2+-binding sites for Ca2+ and of Ins(1,4,5)P3R for Ins(1,4,5)P3 are pH sensitive. We were therefore concerned that bafilomycin A1 might, via effects on cytosolic pH (pHi), affect the responses observed. Using SNARF-5F to measure pHi, we established that neither bafilomycin A1 nor any of the other inhibitors used affected pHi (Fig. 1D).
These results establish that the increase in [Ca2+]i evoked by CCh in HEK cells is potentiated by dissipating the lysosomal H+ gradient. Similar results were obtained with COS-7 cells, where activation of P2Y receptors by ATP evoked Ca2+ release from intracellular stores. Bafilomycin A1 caused the peak Ca2+ signal evoked by ATP to increase by 1.5±0.1 fold (Fig. 1E,F). However, in COS-7 cells, the exaggerated response to ATP occurred without an increase in basal [Ca2+]i (Fig. 1G).
Perforating lysosomal membranes potentiates the increase in [Ca2+]i evoked by CCh
Glycyl-L-phenylalanine 2-naphthylamide (GPN) allows selective disruption of lysosomes because its cleavage by the lysosomal enzyme, cathepsin C, causes osmotic swelling and thereby perforation of lysosomal membranes (Berg et al., 1994) (Fig. 2A). Treatment of HEK cells with GPN abolished staining with LysoTracker Red without affecting the distribution of LAMP1–mCherry (Fig. 2B) or pHi (Fig. 1D). GPN caused the peak Ca2+ signal evoked by a maximal concentration of CCh in Ca2+-free HEPES-buffered saline (HBS) to increase by 1.5±0.1-fold (Fig. 2C), without affecting the sensitivity to CCh (pEC50 = 4.68±0.03 and 4.64±0.06 for control and GPN-treated cells; Fig. 2D). GPN also caused a modest increase in basal [Ca2+]i (22±4 nM and 42±5 nM in control and GPN-treated cells, P = 0.025; Fig. 2C). To minimize any non-specific effects arising from release of lysosomal contents into the cytosol, we reduced the period of incubation with GPN from 30 to 10 min and included cathepsin inhibitor 1 (10 µM), which inhibits many lysosomal proteases, but not the cathepsin C that cleaves GPN (Demuth et al., 1996). The effects of GPN on CCh-evoked Ca2+ signals were unaffected by this revised protocol: the peak [Ca2+]i signal was increased by 1.6±0.2-fold, with no effect on the sensitivity to CCh (pEC50 = 4.5±0.1 and 4.3±0.2 in control cells and those treated with GPN and cathepsin inhibitor 1; Fig. 2E).
Bafilomycin A1 is expected to inhibit acidification of most acidic organelles, but GPN selectively disrupts lysosomes (Berg et al., 1994). It is therefore significant that GPN abolishes staining by LysoTracker Red (Fig. 2B), confirming that lysosomes are the major acidic organelles in HEK cells. Furthermore, the effects of bafilomycin A1 (Fig. 1) and GPN (Fig. 2) on the Ca2+ signals evoked by CCh are similar (supplementary material Table S1), confirming that potentiation of Ca2+ signals by bafilomycin A1 is due to its effects on lysosomes.
Fusion of lysosomes increases the amplitude of CCh-evoked Ca2+ signals
Vacuolin, by an unknown mechanism, causes fusion of lysosomes (Fig. 3A) (Huynh and Andrews, 2005). In HEK cells transiently expressing LAMP1–mCherry, vacuolin caused lysosomes to fuse (Fig. 3B) without affecting pHi (Fig. 1D). Vacuolin caused a 1.4±0.1-fold increase in the peak [Ca2+]i evoked by CCh without affecting the sensitivity to CCh (pEC50 = 4.7±0.1 and 4.9±0.1 for control and vacuolin-treated cells; Fig. 3C,D). Single-cell analyses confirmed that vacuolin had no effect on the number of cells that responded to CCh (93±2% and 91±1% for control and vacuolin-treated cells), but increased the amplitude of the CCh-evoked Ca2+ signal by 2.0±0.2-fold (Fig. 3E). Vacuolin had no effect on basal [Ca2+]i in either single cells or cell populations (25±4 nM and 30±2 nM in control and vacuolin-treated cells, P = 0.31, supplementary material Table S1). The combined effects of maximal concentrations of bafilomycin A1 and vacuolin on the peak Ca2+ signals evoked by CCh were no different from the effects of either treatment alone (Fig. 3F). The results with vacuolin are important because vacuolin is not expected to affect lysosomal ion transport mechanisms, but instead to change the surface area to volume ratio of lysosomes and thereby to perturb interactions with other organelles.
Because some, though not all, treatments used to manipulate lysosomal activity caused modest increases in basal [Ca2+]i (Fig. 1E; supplementary material Table S1), we were concerned that potentiated responses to Ins(1,4,5)P3 might result from enhanced Ca2+ uptake by the ER. We therefore used a novel low-affinity Ca2+ indicator targeted to the ER, CatchER (Tang et al., 2011), to measure directly the free [Ca2+] within the ER of COS-7 cells. These are better suited than HEK cells for these studies because COS-7 cells are larger, flatter and their ER is more easily identified. The results demonstrate that under conditions where bafilomycin A1 increased the peak [Ca2+]i evoked by ATP, it had no effect on the free [Ca2+] within the ER (supplementary material Fig. S5). The larger increases in [Ca2+]i evoked by PLC-linked receptors in the presence of bafilomycin A1 are not therefore due to increased loading of the ER with Ca2+.
Three means of perturbing the behaviour of lysosomes, inhibition of the V-ATPase (Fig. 1), disruption of lysosomal membranes (Fig. 2) or modification of lysosomal morphology (Fig. 3), indistinguishably and non-additively (Fig. 3F) potentiate the Ca2+ signals evoked by CCh. Because CCh-evoked Ca2+ signals are entirely mediated by Ins(1,4,5)P3R (supplementary material Fig. S1), we examined the effect of bafilomycin A1 on the Ca2+ signals evoked by directly activating Ins(1,4,5)P3R.
Lysosomes accumulate Ca2+ released by direct activation of Ins(1,4,5)P3 receptors
Flash photolysis of caged Ins(1,4,5)P3 (ci-Ins(1,4,5)P3) non-disruptively loaded into cells allows Ins(1,4,5)P3 to be delivered directly to the cytosol (Dakin and Li, 2007). Photolysis of ci-Ins(1,4,5)P3 caused transient increases in [Ca2+]i, the amplitudes of which varied between cells (Fig. 4A), probably reflecting differences between cells in loading and/or de-esterification of ci-Ins(1,4,5)P3/PM. In a large sample (>500 cells), the peak response to photorelease of ci-Ins(1,4,5)P3 was 1.3±0.1-fold greater in bafilomycin A1-treated cells (Fig. 4A–C). This confirms that when Ins(1,4,5)P3R are directly activated by Ins(1,4,5)P3, Ca2+ signals are potentiated by inhibition of the lysosomal V-ATPase. The increase in [Ca2+]i evoked by photolysis of ci-Ins(1,4,5)P3 recovered with mono-exponential kinetics (Fig. 4B), but recovery was much slower (P = 0.029) in cells treated with bafilomycin A1 (half-time, t1/2 = 45.6±7.8 s) than in control cells (t1/2 = 18.9±1.7 s; Fig. 4D). Bafilomycin A1 also slowed recovery from CCh-evoked Ca2+ signals (supplementary material Fig. S6). The sustained effects of bafilomycin A1 on the Ca2+ signals evoked by CCh or photolysis of ci-Ins(1,4,5)P3 probably reflect cycling of Ca2+ through the ER fuelled by the activity of the Ca2+ pump of the ER (SR/ER Ca2+-ATPase; SERCA) and sustained activity of Ins(1,4,5)P3R. These results confirm that bafilomycin A1 potentiates Ins(1,4,5)P3- or CCh-evoked Ca2+ signals by inhibiting removal of Ca2+ from the cytosol through lysosomal uptake systems.
Bafilomycin A1 also potentiated the transient Ca2+ signals evoked by inhibiting the SERCA with thapsigargin in nominally Ca2+-free HBS (Fig. 5A,B). Similar results (1.9±0.2-fold increase in the peak Ca2+ signal, n = 3) were obtained when cyclopiazonic acid (CPA) was used to inhibit the SERCA (supplementary material Fig. S7). The route by which Ca2+ leaks from the ER is undefined, but it may include contributions from translocons (Lang et al., 2011) and Ins(1,4,5)P3R (supplementary material Fig. S1). Both channels have large Ca2+ conductances that might be expected to generate the large local increases in [Ca2+]i that we suggest are required for lysosomal Ca2+ uptake.
Ca2+ signals evoked by store-operated Ca2+ entry are unaffected by lysosomes
Our results so far demonstrate that Ca2+ released from the ER via Ins(1,4,5)P3R or leak pathways (after addition of CPA or thapsigargin) can be accumulated by lysosomes. We next assessed whether Ca2+ entering the cell across the plasma membrane via SOCE is also accumulated by lysosomes. Restoration of extracellular Ca2+ to cells treated with thapsigargin in nominally Ca2+-free HBS evoked sustained increases in [Ca2+]i (Fig. 5C). This is consistent with abundant evidence for SOCE in HEK cells (Parekh and Putney, 2005). Manipulating the extracellular [Ca2+] allowed the amplitude of the global increase in [Ca2+]i evoked by SOCE to match and exceed (Fig. 5D) the peak Ca2+ signals evoked by CCh in Ca2+-free HBS (Fig. 1B). Nevertheless, the Ca2+ signals evoked by SOCE were entirely insensitive to bafilomycin A1 (Fig. 5C,D). Similar results were obtained when CPA (100 µM, 15 min) was used to evoke SOCE: the peak Ca2+ signals evoked by addition of HBS containing 30 mM Ca2+ were 180±40 nM and 203±48 nM in control and bafilomycin A1-treated cells, respectively (supplementary material Fig. S7). A possible concern is that Ca2+ released from the ER after addition of thapsigargin or CPA is accumulated by lysosomes (Fig. 5A,B) and might thereby limit their capacity to accumulate further Ca2+ entering the cell via SOCE. To address this issue, we used a membrane-permeant low-affinity Ca2+ buffer, TPEN (N,N,N′,N′-tetrakis(2-pyridylmethyl)-1,2-ethylenediamide), to reduce the free [Ca2+] within the ER and so activate SOCE without causing Ca2+ release (Hofer et al., 1998) (supplementary material Fig. S8). Restoration of extracellular Ca2+ to HEK cells treated with TPEN (100 µM, 2 min) stimulated SOCE-mediated increases in [Ca2+]i that were insensitive to bafilomycin A1 (Fig. 5E). CCh-evoked Ca2+ release in these TPEN-treated cells was potentiated by bafilomycin A1 (Fig. 5F; supplementary material Fig. S8). There are two important conclusions. First, even when Ca2+ within the ER is heavily buffered by TPEN, responses to CCh are potentiated by bafilomycin A1. This reinforces our conclusion (supplementary material Fig. S5) that potentiation of Ins(1,4,5)P3-evoked Ca2+ signals by inhibitors of lysosomes is not due to enhanced Ca2+ loading of the ER. Second, under conditions where lysosomes can attenuate the increases in [Ca2+]i evoked by CCh, they have no effect on SOCE-evoked Ca2+ signals. We conclude that lysosomes accumulate Ca2+ released from the ER (Figs 1–Fig. 2,Fig. 3,4), but they do not accumulate Ca2+ entering the cell via SOCE (Fig. 5).
CCh increases the luminal pH of lysosomes
So far, our analyses of the contribution of lysosomes to shaping Ins(1,4,5)P3-evoked Ca2+ signals have relied on measurements of [Ca2+]i. Complementary analyses of free [Ca2+] within lysosomes ([Ca2+]ly) are more challenging. Indicators can be directed to the lysosome lumen by endocytosis, but their affinity for Ca2+ is reduced by the low lysosomal pH (∼pH 4) (Christensen et al., 2002; Lloyd-Evans et al., 2008). We assessed the effects of CCh on lysosomal pH (pHly) and [Ca2+]ly using pairs of endocytosed dextran-conjugated indicators: an inert marker (Texas Red, TR) and either Oregon Green (OG, to measure pH) or Oregon Green BAPTA (OGB, to measure [Ca2+]; Fig. 6).
In populations of HEK cells, CCh (1 mM) evoked a sustained increase in OG fluorescence, without affecting TR fluorescence (Fig. 6B,C). This demonstrates that CCh caused an increase in pHly. Parallel measurements of cytosolic pH (pHi) established that in both the presence and absence of extracellular Ca2+, CCh evoked a slow and very modest decrease in pHi (Fig. 6D). The effect was statistically significant (P = 0.009) only for cells stimulated with CCh for 5 minutes in the presence of extracellular Ca2+, when pHi fell from 7.42±0.02 to 7.34±0.01 (Fig. 6D). These results demonstrate that under conditions where pHi is either stable or very modestly reduced (Fig. 6D), CCh evokes an increase in pHly (Fig. 6B,C). The effect of CCh on pHly cannot therefore be mediated by changes in pHi. It is more likely to be caused by CCh-evoked Ca2+ signals and subsequent lysosomal Ca2+ uptake by Ca2+/H+ exchange. This interpretation is consistent with evidence that addition of Ca2+ to isolated acidic organelles increases luminal pH (Morgan et al., 2011).
It is more difficult to measure directly the free [Ca2+] within lysosomes because the Ca2+ affinity of OGB, like all other Ca2+ indicators, is pH sensitive. Nevertheless, in similar analyses using TR and OGB, CCh selectively increased the fluorescence of the lysosomal Ca2+ indicator (OGB; Fig. 6E–G). These direct measurements of pHly and indirect measurement of [Ca2+]ly concur with the analyses of [Ca2+]i (Figs 1–Fig. 2,Fig. 3,Fig. 4,5) by suggesting that lysosomes sequester Ca2+ released from intracellular stores.
Lysosomes are motile and associated with ER
We used total internal reflection fluorescence microscopy (TIRFM) to examine the distribution of ER and lysosomes in COS-7 cells by labelling ER with GFP–Ins(1,4,5)P3R1, GFP–Ins(1,4,5)P3R3 or GFP–ER, all of which colocalised with co-transfected SERCA–mCherry (supplementary material Fig. S9). Lysosomes, identified with either LysoTracker Red or LAMP1–mCherry, were closely associated with the ER (Fig. 7A): 79±3% (n = 7 cells) of lysosomes were close enough to the ER that their separation could not be resolved by TIRFM. As reported for other cells (Matteoni and Kreis, 1987), lysosomes were remarkably motile within COS-7 cells (supplementary material Movie 1).
Video-rate imaging of COS-7 cells co-transfected with GFP–ER and LAMP1–mCherry allowed tracking of individual lysosomes and showed that lysosomes maintain their association with the ER over prolonged periods (often for the entire 2-minute recording) as they move around the cell (Fig. 7B; supplementary material Movie 1). This movement often occurred along ER tubules, and occasionally lysosomes localised at the tips of ER tubules and moved concomitantly with ER tubule extension, further supporting a close spatial relationship between ER and lysosomes.
Lysosomes accumulate Ca2+ released from the ER
Disruption of lysosomal Ca2+ uptake mechanisms by preventing luminal acidification (Figs 1, 4) or by perforation of lysosomal membranes (Fig. 2), or changing the morphology of lysosomes and so their relationships with other organelles (Fig. 3) exaggerates the cytosolic Ca2+ signals evoked by Ins(1,4,5)P3. This occurs whether Ins(1,4,5)P3 is delivered by activation of endogenous receptors in the PM (Figs 1–f02,3) or by photolysis of ci-Ins(1,4,5)P3 (Fig. 4), and it occurs in both HEK and COS-7 cells. Despite the increased Ca2+ signals, there was no change in CCh sensitivity, suggesting that the treatments affected neither the intracellular concentrations of Ins(1,4,5)P3 nor the sensitivity of Ins(1,4,5)P3R. The decrease in frequency, but increase in amplitude, of CCh-evoked Ca2+ spikes after bafilomycin A1 treatment (supplementary material Fig. S4) is also inconsistent with a simple increase in either Ins(1,4,5)P3 concentration or Ins(1,4,5)P3R sensitivity because either would be expected to increase the frequency of Ca2+ oscillations. Furthermore, CCh-evoked Ca2+ release is accompanied by a sustained increase in lysosomal pH and perhaps lysosomal [Ca2+] (Fig. 6), consistent with sequestration of Ca2+ by a lysosomal Ca2+/H+ exchanger.
In HEK cells, inhibition of lysosomal Ca2+ accumulation with bafilomycin A1 or GPN was accompanied by a small (<20 nM) increase in basal [Ca2+]i. Nevertheless, we were concerned that these small increases in basal [Ca2+]i might contribute to the increased amplitude of the Ins(1,4,5)P3-evoked Ca2+ signals by enhancing loading of Ca2+ stores or by sensitization of Ins(1,4,5)P3R. The latter is unlikely because none of the treatments affected the sensitivity to CCh (pEC50; Fig. 1B; Fig. 2D,E; Fig 3D; and see above) and they potentiated the Ca2+ signals evoked by Ca2+ leaks from the ER (Fig. 5A; supplementary material Fig. S7A). Additional evidence establishes that enhanced Ca2+ loading of the ER is not the explanation. First, bafilomycin A1 enhanced the increase in [Ca2+]i evoked by ATP in COS-7 cells without affecting basal [Ca2+]i (Fig. 1E–G) and vacuolin mimicked the effects of bafilomycin A1 and GPN on CCh-evoked Ca2+ signals in HEK cells without affecting basal [Ca2+]i (supplementary material Table S1). Second, direct measurement of luminal [Ca2+] within the ER demonstrated that bafilomycin A1 did not affect Ca2+ uptake by the ER (supplementary material Fig. S5), and buffering ER Ca2+ with TPEN did not prevent bafilomycin A1 from potentiating CCh-evoked Ca2+ signals (Fig. 5F). Third, after CCh stimulation (supplementary material Fig. S6) or flash photolysis of ci-Ins(1,4,5)P3 (Fig. 4), [Ca2+]i recovered more slowly in bafilomycin A1-treated cells, confirming that bafilomycin A1 inhibits Ca2+ removal from the cytosol. Finally, CCh increased pHly and perhaps [Ca2+]ly in the absence of inhibitors (Fig. 6), suggesting that Ca2+ released via Ins(1,4,5)P3R is sequestered by lysosomes.
We conclude that the increased amplitude of the Ca2+ signals evoked by Ins(1,4,5)P3 after disruption of lysosomes is due to their diminished ability to sequester Ca2+ released from the ER. These observations are consistent with studies where thapsigargin evoked larger cytosolic Ca2+ signals (Pereira et al., 2010), and lysosomal Ca2+ uptake was diminished (Lloyd-Evans et al., 2008) in cells with lysosomal storage disorders.
ER and lysosomes are intimately associated
CCh evokes Ca2+ signals in HEK cells that are initiated by Ins(1,4,5)P3R and then sustained by SOCE. Although many lysosomes were present in the TIRFM field, suggesting that they lie within ∼100 nm of the PM, they did not sequester Ca2+ entering via SOCE even when the global increases in [Ca2+]i exceeded those evoked by Ins(1,4,5)P3. This demonstrates that lysosomes selectively sequester Ca2+ released from the ER. There are several implications of this conclusion. First, the global elevations in [Ca2+]i that occur as Ca2+ diffuses away from open channels are insufficient to allow significant stimulation of Ca2+ uptake by lysosomes. This suggests that the lysosomal uptake system responsible for these effects has a low affinity for Ca2+. The sparse studies of Ca2+ uptake by isolated lysosomes differ enormously in their estimates of the affinity (Km) for Ca2+: 108 nM and 5.7 mM for lysosomes isolated from neutrophils (Styrt et al., 1988) and fibroblasts (Lemons and Thoene, 1991), respectively. The significance of the latter study is unclear because the incubations excluded ATP. Second, the lysosomal Ca2+ uptake system responsible for the effects observed in our study is unlikely to be a Ca2+ pump because pumps are too slow (e.g. turnover number ∼10 s−1 for SERCA) (Lytton et al., 1992) to sequester Ca2+ effectively before it diffuses away from an open Ins(1,4,5)P3R that probably conducts ∼500,000 Ca2+ s−1 (Vais et al., 2010). Ca2+ exchangers, which have been proposed to mediate Ca2+ uptake by lysosomes (Lloyd-Evans and Platt, 2011) and are likely to have higher turnover numbers (up to 5000 s−1 for the NCX1 Na+/Ca2+ exchanger, for example) (Hilgemann et al., 1991), are better able rapidly to sequester Ca2+ released by Ins(1,4,5)P3R.
The third implication is the need for selective association of lysosomes and ER to allow lysosomes, some of which are no more than ∼100 nm from the PM (Fig. 7; supplementary material Movie 1), to ignore Ca2+ signals generated by SOCE, but accumulate Ca2+ released by Ins(1,4,5)P3R. Local increases in [Ca2+]i near Ins(1,4,5)P3R with their large unitary currents are likely to be much greater (perhaps 50 µM) (Vais et al., 2010) than those near SOCE channels where unitary Ca2+ currents are ∼100-times smaller (Zweifach and Lewis, 1993). Ins(1,4,5)P3R are probably better able than SOCE to generate high local [Ca2+]i and so allow Ca2+ sequestration by low-affinity transport pathways. However, these high [Ca2+]i dissipate steeply within ∼100 nm of the open channel (Shuai and Parker, 2005). Selective sequestration of Ca2+ by lysosomes probably requires that Ins(1,4,5)P3R are within ∼100 nm of lysosomes. Our analyses are consistent with such an intimate association. Lysosomes and ER are closely associated and maintain their association during organelle movements (Fig. 7; supplementary material Movie 1).
ER–organelle junctions: a recurring theme in Ca2+ signalling
We have shown that dynamic lysosomes and ER are intimately associated (Fig. 7; supplementary material Movie 1), allowing lysosomes selectively to accumulate Ca2+ released from the ER. The reciprocal relationship is also important because in many cells NAADP-evoked Ca2+ release from lysosomes triggers Ca2+ release from the ER by Ca2+-induced Ca2+ release via Ins(1,4,5)P3R (Calcraft et al., 2009) or RyR (Brailoiu et al., 2010; Cancela et al., 1999; Kinnear et al., 2008; Lee et al., 1997). Lysosome–ER junctions are reminiscent of those between mitochondria and ER (Fig. 8). The latter are maintained by specific tethering proteins between dynamic organelles (Csordás et al., 2006) and they allow mitochondria, despite the low affinity of their Ca2+ uptake pathway, to sequester Ca2+ released by Ins(1,4,5)P3R (Rizzuto et al., 2009). Mitochondrion–ER junctions allow bi-directional interactions between organelles: mitochondria can relieve Ca2+ inhibition of Ins(1,4,5)P3R by rapidly sequestering Ca2+ (Olson et al., 2010), while Ca2+ provided by ER Ca2+ channels allows mitochondrial Ca2+ uptake to regulate oxidative phosphorylation (Jouaville et al., 1999), apoptosis (Szalai et al., 1999) and mitochondrial motility (Yi et al., 2004). We suggest a similar bi-directional interplay between ER and lysosomes (Fig. 8). Ca2+ release via TPC2 can trigger Ca2+ release via RyR or Ins(1,4,5)P3R in the ER (Brailoiu et al., 2010; Calcraft et al., 2009; Kinnear et al., 2008), and Ca2+ release via Ins(1,4,5)P3R is selectively accumulated by lysosomes. The latter may regulate the behaviour of lysosomes by increasing lysosomal pH; by priming TPC2, which appears to be stimulated by luminal Ca2+ (Pitt et al., 2010), to respond to NAADP; and it may regulate endolysosomal trafficking (Luzio et al., 2010). Conversely, lysosomal Ca2+ sequestration shapes cytosolic Ca2+ signals, including the frequency and amplitude of the Ca2+ oscillations typically evoked by physiological stimuli, and by analogy with mitochondria (Olson et al., 2010) lysosomes may modulate feedback regulation of Ins(1,4,5)P3R gating (Fig. 8).
Materials and Methods
Culture media, Lipofectamine 2000, LysoTracker Red DND-99, SNARF-5F/AM, fluo-4/AM, fura-2/AM, dextran-conjugated Texas Red (Mr, 70,000), dextran-conjugated Oregon Green BAPTA 488 (Mr, 10,000), dextran-conjugated Oregon Green 488 (Mr, 10,000) and the Ca2+ standard solutions used to calibrate fura-2 fluorescence signals to [Ca2+]i were from Invitrogen (Paisley, UK). Cell culture plastics and 96-well assay plates were from Greiner (Stonehouse, UK). Imaging dishes (35-mm diameter with a 7-mm No. 0 glass insert) were from MatTek Corporation (Ashland, USA) or PAA Laboratories (Yeovil, UK). U73122 (1-[6-[[(17β)-3-methoxyestra-1,3,5(10)-trien-17-yl]amino]hexyl]-1H-pyrrole-2,5-dione) and CPA were from Tocris (Bristol, UK). Bafilomycin A1 was from AG Scientific (California, USA) or Fluorochem (Hadfield, UK). Glycyl-L-phenylalanine 2-naphthylamide (GPN) was from Bachem (St. Helens, UK). Cathepsin inhibitor 1 was from Calbiochem (Nottingham, UK). BAPTA was from Molekula (Dorset, UK). CCh, ATP, DMSO, foetal bovine serum, poly-L-lysine and vacuolin were from Sigma-Aldrich (Poole, UK). Ionomycin was from MerkEurolab (Nottingham, UK). Thapsigargin was from Alomone Labs (Jerusalem, Israel). Caged cell-permeant Ins(1,4,5)P3 (ci-Ins(1,4,5)P3/PM) was from SiChem (Bremen, Germany). NED 19 and TPEN were from Enzo Life Sciences (Exeter, UK). Ryanodine was from Ascent Scientific (Bristol, UK).
Plasmids encoding LAMP1–mCherry and LAMP1–GFP were made by transferring the LAMP1 fragment from LAMP1–tdTomato (a gift from T. Carter, MRC National Institute for Medical Research, London, UK) (Babich et al., 2008) into pmCherry N1 or pAcGFP-N1 (Clontech, Mountain View, USA) respectively, using EcoRI/BamHI restriction sites. A plasmid encoding the ER marker, GFP–ER, in which GFP is attached to sequences that allow insertion into the cytosolic leaflet of the ER membrane, was a gift from V. J. Allan (University of Manchester, UK) (Woźniak et al., 2009). A plasmid encoding SERCA type-1 tagged at its C-terminus with mCherry was prepared by insertion of mCherry from pmCherry-N1 into SERCA-GFP (a gift from J. M. East, University of Southampton, UK) (Newton et al., 2003) using AgeI/NotI restriction sites. Plasmids encoding rat types 1 and 3 Ins(1,4,5)P3R tagged at their N-termini with GFP have been described previously (Pantazaka and Taylor, 2011). The coding sequences of all plasmids were verified. A plasmid encoding CatchER, an EGFP-based Ca2+ indicator targeted to the ER lumen, was generously provided by J. J. Yang (Georgia State University, USA) (Tang et al., 2011).
Cell culture and transfection
HEK 293 cells were cultured at 37°C in Dulbecco's modified Eagle's medium/Ham's F12 with GlutaMAX and foetal bovine serum (10%) in a humidified atmosphere containing 95% air and 5% CO2. Medium was replaced every third day, and cells were passaged when they reached ∼80% confluence. For experiments with cell populations, cells were seeded (∼2×104 cells/well) into 96-well plates. For single-cell analyses or TIRFM, cells were seeded (1.2×105 cells/well) onto either 22-mm round glass coverslips or glass-bottomed culture dishes pre-coated with 0.01% (w/v) poly-L-lysine. Cells were grown for a further 2–3 days before experiments or transfection. The latter used Lipofectamine 2000 according to the manufacturer's instructions with 1 µg DNA/well for cells in glass-bottomed culture dishes.
COS-7 cells were cultured at 37°C in RPMI 1640 with foetal bovine serum (10%) in a humidified atmosphere containing 95% air and 5% CO2 (Pantazaka and Taylor, 2011). For TIRFM experiments, COS-7 cells were transiently transfected with plasmids using Neon nucleofection (Invitrogen) with 5 µg DNA/100 µl cells (105 cells). Cells were then plated onto poly-L-lysine-coated glass-bottomed culture dishes and used 1–2 days after transfection.
Measurements of [Ca2+]i
Measurements of [Ca2+]i in HEK cells were performed at 20°C in HBS (Tovey et al., 2008). HBS had the following composition: 135 mM NaCl, 5.9 mM KCl, 1.2 mM MgCl2, 1.5 mM CaCl2, 11.5 mM glucose, 11.6 mM HEPES, pH 7.3. Ca2+ was omitted from nominally Ca2+-free HBS, and replaced by BAPTA (10 mM) in Ca2+-free HBS. For cell populations, confluent cells in 96-well plates were loaded with fluo-4/AM (2 µM, 1 h, 20°C), washed and incubated for a further 1 h in HBS. Fluorescence, from which [Ca2+]i was determined, was then measured using a fluorescence plate reader (FlexStation 3, MDS Analytical Devices, Wokingham, UK). Fluo-4 fluorescence was calibrated to [Ca2+]i using the equation: [Ca2+]i = KD (F−F min)/(F max−F), using a KD for Ca2+ of 345 nM. Fluorescence from Ca2+-saturated (Fmax) and Ca2+ free (Fmin) fluo-4 was measured in adjacent wells by addition of Triton X-100 (0.1%) with CaCl2 (10 mM) or BAPTA (10 mM) (Tovey et al., 2008).
For single-cell imaging, confluent cultures of HEK or COS-7 cells on 22-mm round, poly-L-lysine-coated glass coverslips were loaded with fura-2/AM (2 µM, 1 h, 20°C), washed and incubated for a further 1 h in HBS. Fluorescence, detected at >510 nm after alternating excitation at 340 and 380 nm, was detected using an Olympus IX71 inverted fluorescence microscope with a Luca electron multiplying charge-coupled device (EMCCD) camera (Andor Technology, Belfast, UK) with a 40×1.35 NA objective and analysed using MetaFluor software (MDS Analytical Devices, Wokingham, UK). After correction for background fluorescence, fluorescence ratios (F340/F380) were calibrated to [Ca2+]i using Ca2+ standard solutions (Tovey et al., 2008).
Measurement of lysosomal pH and [Ca2+]
Almost confluent cultures of HEK cells grown on poly-L-lysine-coated, glass-bottomed culture dishes, were incubated in culture medium with dextran-conjugated Texas Red (TR; 0.1 mg/ml) together with a dextran-conjugate (0.1 mg/ml) of either Oregon Green (OG) or Oregon Green BAPTA (OGB) for 12 h at 37°C, and then for 4 h without the indicators. Cells were then washed with HBS and fluorescence was recorded using an Olympus IX81 microscope with a 40×/1.35 NA objective. Cells were illuminated with a mercury xenon lamp using alternating filter sets: U-MNIBA (Olympus; excitation 470–495 nm, emission 510–550 nm) and LF561A (Semrock; excitation 550–570 nm, emission 580–630 nm) for OG/OGB and TR, respectively. Images were captured at 2-second intervals using an EMCCD camera (Andor iXon 897) and analysed using Cell∧R software (Olympus, Milton Keynes, UK). All records were corrected for background fluorescence determined under identical conditions from cells that had not been loaded with indicators. Fluorescence changes from defined regions of interest (ROI) were then expressed as F/Fo, where Fo and F denote the average fluorescence within the ROI at the start of the experiment (Fo) and at each time point (F).
Measurement of luminal free [Ca2+] in the ER
The luminal free [Ca2+] of the ER was measured using an ER-targeted Ca2+-sensor CatchER (Tang et al., 2011). COS-7 cells in 35-mm imaging dishes were transfected with the CatchER coding sequence in pcDNA3.1 (2 µg/well) using Lipofectamine 2000 (1∶1, DNA/Lipofectamine). After 48 h, cells were imaged using an Olympus IX81 TIRF microscope with 60× or 150× TIRFM objectives. In time-lapse experiments, images were captured using wide-field illumination (488 nm, 200-ms exposure time) with a 60× TIRF objective and an Andor iXon 897 EMCCD camera. Control cells were used to correct images for photobleaching (∼15% over 1 h).
Flash photolysis of caged Ins(1,4,5)P3
HEK cells grown on poly-L-lysine-coated, glass-bottomed culture dishes were loaded with ci-Ins(1,4,5)P3/PM (1 µM, 45 min) (Dakin and Li, 2007), and then with fluo-4/AM (2 µM) and ci-Ins(1,4,5)P3/PM (1 µM) for a further 45 min. After washing and incubation in HBS for a further 45 min, cells were imaged (20°C) using an Olympus IX81 microscope equipped with a 40×/1.35 NA objective. Cells were illuminated with a 488-nm diode-based solid-state laser (Olympus Digital Laser Systems) and emitted fluorescence (500–550 nm) was captured with an EMCCD camera (Andor iXon 897). After 60 s, ci-Ins(1,4,5)P3 was uncaged by exposing an entire field to three pulses of UV light (<345 nm, ∼1 ms, 3000 µF, 300 V, ∼170 J) delivered within ∼6 s using a JML-C2 xenon flash-lamp (Rapp OptoElectronic GmbH, Hamburg, Germany). Images were acquired at 1-second intervals with an Andor iXon 897 camera (512×512 pixels) and analysed using Cell∧R software (Olympus, Milton Keynes, UK). For these measurements of [Ca2+]i using a non-ratiometric indicator (fluo-4), responses are reported as F/Fo, where Fo is the average fluorescence intensity recorded from a ROI immediately before flash photolysis, and F is the fluorescence intensity from the same region after the flash.
Measurement of cytosolic pH
Cytosolic pH (pHi) was measured using the fluorescent pH-sensitive indicator, SNARF-5F (Liu et al., 2001). Cells in 96-well plates were incubated with SNARF-5F/AM (2 µM, 30 min), washed with HBS, and fluorescence (excitation at 561 nm, and emission at 580 and 640 nm) was measured using a FlexStation fluorescence plate reader. Fluorescence ratios (F580/F640) were calibrated to pHi using standard pH solutions (Owen, 1992).
Total internal reflection fluorescence microscopy
For TIRFM, cells on poly-L-lysine-coated, glass-bottomed culture dishes were imaged using an Olympus IX81 microscope with 150×/1.45 NA or 60×/1.45 NA TIRF objectives. For staining of lysosomes, cells were incubated with LysoTracker Red DND-99 (50 nM, 1 h, 20°C). Cells were illuminated with 488 nm (for GFP) or 561 nm (for mCherry or LysoTracker Red) diode-based solid-state lasers, and images were acquired with an Andor iXon 897 EMCCD camera. With the filters used there was no significant crosstalk between green and red channels. Images were processed using Cell∧R software (Olympus, Milton Keynes, UK).
Concentration-effect relationships for each experiment were individually fitted to Hill equations using non-linear curve-fitting (GraphPad Prism, version 5) and the results obtained from each (pEC50, Hill coefficient h, maximal response) were pooled for statistical analysis and presentation.
Rates of recovery of [Ca2+]i after flash photolysis of ci-Ins(1,4,5)P3 were determined by fitting mono-exponential decay equations (GraphPad Prism, version 5) to the averaged responses of all cells within the field exposed to the UV-flash (30–70 cells/field). At least three such fields were analysed for each coverslip and the half-times (t1/2) were pooled to produce a single value for each coverslip. Rates of recovery of [Ca2+]i after stimulation of cell populations with CCh were determined by fitting bi-exponential decay equations (GraphPad Prism, version 5). Each determination comprises the average response from at least three wells in a single experiment.
For quantitative analyses of the colocalisation of two fluorophores, Cell∧R software was used to correct for background fluorescence using an area outside the cell and then to define ROI (∼200–400 µm2) within the peripheral cytoplasm that excluded the nucleus and perinuclear area. For each ROI examined (Fig. 7; supplementary material Figs S2, S9), there was a statistically significant (P<0.05) colocalisation of fluorophores defined using the Colocalisation Analysis/Colocalisation Test plugin (ImageJ). This applies the Costes randomization method with 100 iterations and ignores pixels in which there is no fluorescence from either fluorophore (Costes et al., 2004). Colocalisation was then quantified using the Colocalisation Analysis/Colocalisation Threshold plugin (ImageJ). This was applied to threshold images and then calculate Pearson's correlation coefficient (Rcoloc), ignoring pixels with intensities below threshold. Rcoloc = Σ(Ri−Rm)(Gi−Gm)/√Σ(Ri−Rm)2Σ(Gi−Gm)2, where Gi and Ri are the intensities of individual green and red pixels respectively, and Gm and Rm are the mean intensities of green and red pixels. Rcoloc = 1 denotes perfect colocalisation.
Student's t-test or 1-way ANOVA was used for statistical analyses with P<0.05 considered significant. All statistical analyses were performed on raw data, although for clarity of presentation some results are shown normalized and with statistical significance shown for the underlying raw data (Fig. 1C,F; Fig. 3E; Fig. 4C).
We thank Viki Allan (Manchester University, UK), Tom Carter (MRC NIMR, London, UK), Malcolm East (Southampton University, UK) and Jenny Yang (Georgia State University, USA) for generous gifts of plasmids.
This work was supported by the Wellcome Trust [grant number 085295 to C.W.T.]; an equipment grant from the Isaac Newton Trust, Cambridge [to C.W.T.]; a Meres senior research associateship from St John's College, Cambridge [to D.L.P.]; and studentships from the Caixa Galicia Foundation and Obra Social La Caixa, Spain [to C.I.L.S.]. Deposited in PMC for release after 6 months.