Endosomes, lysosomes and lysosome-related organelles are emerging as important Ca2+ storage cellular compartments with a central role in intracellular Ca2+ signalling. Endocytosis at the plasma membrane forms endosomal vesicles which mature to late endosomes and culminate in lysosomal biogenesis. During this process, acquisition of different ion channels and transporters progressively changes the endolysosomal luminal ionic environment (e.g. pH and Ca2+) to regulate enzyme activities, membrane fusion/fission and organellar ion fluxes, and defects in these can result in disease. In the present review we focus on the physiology of the inter-related transport mechanisms of Ca2+ and H+ across endolysosomal membranes. In particular, we discuss the role of the Ca2+-mobilizing messenger NAADP (nicotinic acid adenine dinucleotide phosphate) as a major regulator of Ca2+ release from endolysosomes, and the recent discovery of an endolysosomal channel family, the TPCs (two-pore channels), as its principal intracellular targets. Recent molecular studies of endolysosomal Ca2+ physiology and its regulation by NAADP-gated TPCs are providing exciting new insights into the mechanisms of Ca2+-signal initiation that control a wide range of cellular processes and play a role in disease. These developments underscore a new central role for the endolysosomal system in cellular Ca2+ regulation and signalling.
- calcium store
- nicotinic acid adenine dinucleotide phosphate (NAADP)
- two-pore channel (TPC)
Although proton transport has been the most extensively studied ion movement across the membranes of the endolysosomal system, recent functional and biochemical studies have indicated a plethora of ion-transport mechanisms in endosomes and lysosomes which may play important roles in the functioning of these organelles.
The recognition that the organelles of the endolysosomal system contain appreciable amounts of Ca2+ [1,2] (Supplementary Table S1 at http://www.BiochemJ.org/bj/439/bj4390349add.htm), together with roles for this ion in regulating their fusion and trafficking, has prompted intense study of Ca2+-transport mechanisms in these organelles [3–9]. Interest in endolysosomes as Ca2+-storage organelles has intensified with the discovery that NAADP (nicotinic acid adenine dinucleotide phosphate), a Ca2+-mobilizing messenger first discovered to evoke Ca2+ signals in sea urchin eggs, does so in many cases by activating Ca2+-release mechanisms on acidic stores with characteristics of lysosomes .
In the present review, we highlight the role of the endolysosomal system as an acidic Ca2+-storage organelle, with a particular focus on recent evidence that it is the target of the Ca2+-mobilizing messenger NAADP which regulates Ca2+ release and ion fluxes across endolysosomal membranes, placing these organelles at centre stage for regulating intracellular Ca2+ signalling and homoeostasis in health and disease.
Discovery of the lysosome
The lysosome is a highly specialized acidic intracellular organelle that mediates a complex set of functions in eukaryotic cells . The concept of the lysosome can be traced to the pioneering studies by Metchnikoff (reviewed in ). He was the ‘father’ of innate immunity whose contribution was recognized by the 1908 Nobel Prize in Physiology or Medicine (along with Paul Ehrlich) for his seminal studies of phagocytosis . He found that particles fed to simple unicellular organisms and mammalian phagocytes were digested in an acidic compartment rich in digestive enzymes. In 1907, using simple pH-sensitive dyes, he reported that when peritoneal macrophages from the guinea pig were fed with bacteria, acidic reactions took place within phagosomes . It was many years later that the general relevance of this finding to non-phagocytic cells was appreciated and the ubiquitous catabolic organelle (the lysosome) proposed and identified.
In the 1950s Christian de Duve was studying the intracellular localization of acid phosphatase and identified a sedimentable particle surrounded by a continuous membrane that was enriched in hydrolases. Significantly, this particle was distinct from peroxisomes (which he also discovered) and mitochondria [14,15]. Electron microscopy studies revealed an organelle with highly heterogeneous electron-dense contents enclosed within a single limiting membrane that was termed at the time ‘pericanalicular dense bodies’ . De Duve proposed the term lysosome to be applied to this specialized organelle that was the cellular repository of mature acid hydrolases. Reviews by de Duve detail the experimental path that led to the discovery of the lysosome, for which he was awarded the Nobel Prize for Physiology or Medicine in 1974 along with Albert Claude and George Palade for their collective discoveries of the structural and functional organization of the cell [15,17]. We now know that the lysosome is the terminal organelle in the endocytic pathway and plays a complex role in multiple aspects of cellular homoeostasis [11,18].
The next major discovery (1963) was that genetic deficiencies in lysosomal enzymes were responsible for a group of inherited metabolic disorders termed ‘storage diseases’. The seminal work by Hers  revealed that the disease described in 1932 by Pompe (termed Pompe disease or generalized glycogenosis) was due to a defect in lysosomal α-glucosidase involved in lysosomal glycogen catabolism. During the process of autophagy, glycogen in the cytoplasm enters the lysosomal system when autophagosomes fuse with lysosomes and this enzyme is responsible for lysosomal glycogen catabolism. The concept of lysosomal storage diseases was born and over 50 disorders have now been identified, the majority of which are due to lysosomal enzyme deficiencies, but they can also be caused by defects in lysosomal membrane proteins . Indeed, the study of these diseases has contributed numerous insights into the fundamental functions of lysosomes and this knowledge is now being applied to the development of therapies for these devastating disorders.
Lysosomes and lysosome-related organelles
All eukaryotic cells contain lysosomes (termed vacuoles in plants and fungi), with the exception of some highly specialized cells such as mammalian erythrocytes, which lack multiple organelles including lysosomes. Lysosomes degrade exogenous and endogenous macromolecules derived from biosynthetic and endocytic pathways, and catabolize cytosolic components that are obtained from the autophagic pathway [11,18].
Following the discovery and definition of the lysosomal compartment subsequent studies identified several specialized organelles to be lysosome-related. This included melanosomes (in melanocytes), lytic granules (in natural killer and cytotoxic T lymphocytes of the immune system), the MHCII compartment (antigen-presenting cells), alpha granules of platelets and dense core granules from various haemopoietic cell lineages (e.g. basophils and neutrophils) [21,22]. In many of these cell types conventional lysosomes are present to mediate basic catabolic functions, whereas the lysosome-related organelles mediate specialist functions appropriate for the cell type. Lysosome-related organelles are believed to share common origins in terms of their evolution and biogenesis, although we still have an incomplete knowledge of how they are formed and how specialized proteins are sorted to them . The definition of these organelles is that they have a specialized function unrelated to macromolecule catabolism, share common hallmarks with lysosomes and differ from the endocine/exocrine granules, which lack lysosomal features . The other salient feature/definition of lysosome-related organelles is that they respond to physiological stimuli rather than functioning constitutively in the way that classical lysosomes do. Secretory lysosomes are another interesting subgroup of lysosome-related organelles, which are essentially classical lysosomes with additional proteins present such as Rab27a. Indeed, they combine many features of lysosomes with secretory granules . An interesting fact that has emerged is that a subgroup of lysosomes in non-secretory cells can fuse with the plasma membrane and mediate repair following plasma membrane damage .
Lysosomal integrity and enzyme targeting
The identification of the lysosome as a digestive compartment involved in macromolecular catabolism defined the first functional role of this compartment. One paradox was how the limiting membrane of the lysosome remains intact when in contact with multiple catabolic hydrolases. It became clear that the lysosomal membrane is protected by a complex and substantial glycocalyx on its luminal leaflet, due to the presence of heavily glycosylated integral membrane proteins. These are termed LAMPs (lysosomal-associated membrane proteins) and include LAMP-1, LAMP-2 and LAMP-3 (CD63). Evidence has emerged that processing of sialic acid residues on LAMP-1 regulates lysosomal exocytosis . The other membrane proteins of the lysosome are involved in transport of catabolites generated in the lysosome to feed into metabolic recycling and signalling pathways, or play roles in ion transport and pH maintenance.
Another question the identification of the lysosome raised was how were the lysosomal hydrolases that function in this compartment targeted to this organelle. The breakthrough came when the mannose 6-phosphate-targeting system was discovered, which most but not all soluble hydrolases utilize [26,27]. Alternative targeting systems include LIMP-2 and sortilin [27,28]. Membrane proteins of the lysosome are targeted via motifs in their cytoplasmic domains, which are typically dileucine- or tyrosine-based .
The operational definition of lysosomes that has emerged after decades of research is an organelle containing mature acidic hydrolases and LAMPs coupled with a lack of mannose 6-phosphate receptors. We are now in an era where understanding how the lysosome and lysosome-related organelles are regulated and interface with other cellular systems is a major focus of current research. Recent progress has been made in the area of Ca2+ regulation in the lysosome, informed in part by the study of lysosomal storage diseases, and this is the subject of the present review.
HOW DO ACIDIC STORES REGULATE pHL (luminal pH)?
Given that the luminal pH (pHL) and membrane potential are absolutely critical for endolysosomal Ca2+ homoeostasis, we shall first discuss what factors influence these fundamental properties.
Although the lumen of the major ER (endoplasmic reticulum) Ca2+ store is close to neutral pH [29–31], the lumen of the acidic organelle family is in the range 4.5–6.5 (Figure 1) [32,33]. How is this distinguishing acidity maintained? Multiple important physical factors impinge upon the steady-state pHL of small acidic vesicles that include, but are not restricted to, membrane capacitance (via its effect on Δψ), the luminal H+-buffering capacity (approximately 40–60 mM/pH [34,35]), the Donnan potential and ion-transport pathways . Focussing on the latter, acidification is the net result of the effect of multiple ion transporters (Figure 2) discussed below.
Naturally, the uptake of H+ into the lumen of the organelle requires energy either from ATP (or PPi) or by exploiting the gradients of other ions. The former represents families of H+ pumps, the latter families of H+ exchangers.
V-ATPase (vacuolar-type H+-ATPase)
Now understood and scrutinized at a detailed molecular level, the V-ATPase is the ubiquitous primary route that drives H+ into the lumen against the electrochemical gradient at the expense of ATP (with a stoichiometry of 2–4 H+ per ATP) [37–41]. In translocating H+, the pump is electrogenic and will therefore generate a lumen-positive membrane potential .
Weighing in at a hefty 910 kDa and comprised of 14 subunits, this impressive rotary machine is meta-organized into ATP-hydrolysing (V1) and the transmembrane H+-translocating (V0) domains, thereby resembling and functioning in much the same way as does its cousin the mitochondrial F1F0 ATP synthase [37,40]. Although virtually all cell types express isoforms of V-ATPase, the precise subunit composition is often tissue-specific and is a means of controlling subcellular targeting . For example, whereas the majority is found in the endolysosomal system, a small fraction is targeted to the Golgi and plasma membrane [37,40].
As befits such a key player, the V-ATPase is regulated in disparate ways including its subunit complement, subcellular location, reversible dissociation of its V1 and V0 domains (often as a function of metabolic status ), the efficiency of coupling to ATP hydrolysis, cytosolic pH , and regulatory proteins [38,42,44,45] including protein kinases PKA (protein kinase A), PKC (protein kinase C) and SOS (salt overly sensitive) [46–48].
Central to many studies investigating organellar pH (and Ca2+) have been the potent and specific V-ATPase inhibitors, the macrolides bafilomycin A1 and concanamycin , that bind to the V0 subunit in most species . Bafilomycin A1 is astonishingly selective (up to several million-fold) for the V-ATPase over other ATPases (e.g. F1F0, Na+/K+ and Ca2+) , but there are isolated reports of off-site actions .
H+-PPase and H+ P-ATPase
Some plants utilize a smaller P-type ATPase (P3A or the ‘autoinhibited H+-ATPase’) to regulate vacuole pH  (important for flower petal colour ). By contrast, the H+-PPases (or vacuolar proton pyrophosphatases V-PPases) pump protons at the expense of pyrophosphate hydrolysis. They are absent from mammalian cells and have been studied mainly in plants and lower organisms, where there are hundreds of homologous sequences available .
Complementing the H+ pumps are a retinue of H+ exchangers (antiporters) that modulate pHL by coupling H+ translocation to cation (or anion) movements; as such, they are normally expected to dissipate the organellar ΔpH and thereby contribute to setting the pHL point [55,56] (Figure 2).
As is typical for the acidic vesicle field, far more is known about the cation/H+ exchangers in plants and lower organisms, where detailed molecular analyses of a plethora of H+ antiporters for Na+, K+ and Ca2+ abound ([57–60] and references therein; Ca2+ antiporters are discussed below). By contrast, in the mammalian world our view of cation/H+ exchange is pretty much restricted to those electroneutral NHE (Na+/H+ exchanger) isoforms that are located principally on intracellular organelles (Table 1) including exocytotic granules, endosomes and the trans-Golgi network (NHE6–9) in contrast with the plasma membrane (NHE1–5) [55,56,63]. That said, vesicle trafficking events (exocytosis/endocytosis) do blur the delineation of the two groups, but importantly there are no (or few) NHEs in later acidic organelles.
Functionally, NHE isoforms are thought to inhibit acidification. A ‘reverse mode’ of action for the internal NHE7–9 would extrude H+ and dissipate the pH gradient, at least experimentally [55,56]. This is energetically most feasible when NHEs transport K+ (as they can instead of Na+) considering the permissive luminal/cytosolic [K+] ratio (Supplementary Tables S1 and S2 at http://www.BiochemJ.org/bj/439/bj4390349add.htm).
Physiologically, organellar NHEs can be necessary for vacuole fusion , vesicular trafficking in development , metal/salt tolerance  and cytosolic pH regulation . Moreover, NHE8 regulates late-endosomal morphology and function .
Cl− conductance has long featured as a charge shunt for the V-ATPase (see below) and this was shown subsequently to be carried by the ClC (Cl− channel) family . However, it later emerged that whereas the plasma membrane ClCs are indeed Cl− channels (ClC-1, -2 and -K) those found on intracellular organelles are actually Cl− antiporters (coupled to H+) (ClC-3–7) (Table 1 and Figure 2) . Unlike NHEs, ClCs are present in all acidic compartments from recycling endosome to lysosome  and exchanger loss culminates in a variety of pathologies such as renal Dent's disease (ClC-5) [66,67], neurodegeneration (ClC-3, -6 and -7)  and osteopetrosis .
Membrane potential, counter ions and pHL
By definition, H+ pumps are electrogenic establishing a lumen-positive membrane potential (Δψ) and H+ gradient (ΔpH) and, eventually, the protonmotive force (a function of both) would oppose further acidification [32,33]. At chemical equilibrium, V-ATPase would theoretically lower pHL to ~2.6 [33,69,70], but this is not reached because the Δψ offers ‘more resistance’ to the H+ pumps than does ΔpH  requiring ~65% of the overall free energy . If the build-up of lumen-positive charge is a brake on acidification, then Δψ must be partially dissipated in order to unfetter the pump. How can this be achieved?
Charge neutralization (Figure 2) comes in the form of both anion co-transport and cation counter-transport. ClCs demand our attention as they are electrogenic (2Cl−:1H+) and voltage-gated carriers [66,71]. That is, their recorded currents reflect the net negative charge uptake which is thought to be a major counter ion for organellar acidification as knock-down (or mutation) of the ClC proteins testifies [66,71]. Indeed, the fact that they are Cl−/H+ exchangers is almost incidental: mutants that are pure Cl− channels are just as good at maintaining endolysosomal pHL [67,68]. The notion of other families of Cl− channels contributing to charge compensation has been quite polemic, with the CFTR (cystic fibrosis transmembrane conductance regulator) protein representing one hotly debated candidate , whereas the organellar CLICs (chloride intracellular channels) appear promising .
However, many contend that cation counter-transport is also fundamental for acidification [74,75], even in the absence of Cl− channels , and the endolysosomal lumen environment probably contain sufficient K+ to support this (Supplementary Tables S1 and S2). The relative contributions (and routes) of anion and cation fluxes are not fully understood (and not mutually exclusive), but, theoretically, cation fluxes dominate anion fluxes at least in newly formed vesicles . As others have astutely observed , the osmotic overhead that Cl− co-transport presents (net uptake of HCl drives water movement) might favour cation exchange.
Na+/K+-ATPase and Δψ
Finally, the organellar membrane potential can be shaped substantially by the Na+/K+-ATPase. The Na+/K+-ATPase is an electrogenic pump (3Na+:2K+) that is abundant in intracellular acidic vesicles (Table 1) where they profoundly affect pHL via Δψ. In the organellar membrane, they reinforce a lumen-positive membrane potential and therefore will be inhibitory to H+ pumping. This brake on acidification was first suggested in early (recycling) endosomes , but has now been woven into the fabric of pHL regulation across the endosomal spectrum  and the intricacies and permissive conditions (such as high luminal [K+]) modelled in detail [32,36]. Presumably, functional Na+/K+-ATPases are retained within the plasma membrane/endosomal axis as a result of endosome recycling.
What is the resting Δψ of acidic organelles?
For consistency, the plasma membrane and organellar membrane potentials would be best defined as Δψ=ψCyt−ψExo, where Exo refers to the exoplasmic compartment (the extracellular space or organelle lumen) and Cyt refers to the cytosol as proposed previously . Adopting this convention means that a cytosol-negative potential at the organelle membrane would also be returned as a negative value. Nonetheless, the (understandable) focus on the organelle lumen means that workers often prefer to emphasize the lumen-positive Δψ and consequently reverse the sign [79,80]. Except for the calculations below, we shall adopt the latter practice and refer to a lumen-positive potential as +XmV.
The membrane potential is comprised of two components, the ‘fixed charge’ Donnan potential (attributable to luminal anionic macromolecules) and the diffusion potential (governed by ionic gradients) . The Donnan potential is lumen-negative, spanning the −20 to −70 mV range [81,82], and not only provides a driving force for H+ entry into secretory granules [83–85] and lysosomes , but also drives Cl− clearance from the early endosomes . This Donnan negative charge is initially dominant in early endosomes, but is titrated as acidification proceeds . Similarly, neutralization of the Donnan potential by the entry of other cations (Na+, K+ and Ca2+) may also occur .
Although the Donnan equilibrium can be a sizeable potential, the diffusion potential is capable of dominating and reversing it (especially in the more acidic organelles). Accordingly, there are numerous studies that propose a lumen-positive resting membrane potential from the distribution of radioactive ions or fluorescent dyes in phagosomes , endosomes [34,74], lysosomes [88–90], secretory granules [83,91,92], yolk platelets  and vacuoles [75, 94]. Note that lumen-negative potentials are usually attributable to ‘non-physiological’ conditions (for example in the absence of counterions or ATP) in platelet alpha granules  and liver lysosomes .
Reports that actually put hard numbers on Δψ are less abundant and are confusingly scattered from +10 to +100 mV (Supplementary Table S2 at http://www.BiochemJ.org/bj/439/bj4390349add.htm), but cluster around lower potentials such as +30 mV. Recently lysosomal Δψ was determined in intact single cells as +19 mV using FRET (fluorescence resonance energy transfer)-based indicators ; realistically, a small Δψ is the more likely scenario when endolysosomes are highly permeable to counterions that dissipate charge . Only more in-depth surveys will reveal whether this data spread constitutes a biological variation (e.g. trends between organelles and/or cellular sources) or a methodological one.
The good news for Ca2+ signalling is that this range of Δψ is highly favourable for Ca2+ release from acidic vesicles (more so than the ER which has a smaller Δψ) and is far from the Ca2+ equilibrium potential of −72 to −110 mV (lumen-negative)2 where Ca2+ release could not occur.
How do different compartments maintain a characteristic pHL?
It should be self-evident that the characteristic pH set point (Table 1) is dictated primarily by the unique transporter protein profile of each organelle, although other factors such as the luminal H+-buffering capacity and organelle shape/size may also influence it to a degree  (but see ) and so the final pHL is determined chiefly by organelle-specific permutations of three factors: Δψ, ΔpH and the H+ pump/leak balance (see Table 1 for their differential distribution).
Given that H+ pumping is more sensitive to Δψ than to ΔpH [36,69], some suggest that their relative contribution along the endolysosomal pathway is reciprocal with endosomes being limited more by the Δψ, whereas lysosomes are limited by the ΔpH [36,90] (Figure 1). In theory, the Na+/K+-ATPase is the principal player that shapes Δψ to strongly inhibit acidification in the earlier compartments but not the later ones (Table 1). In contrast, the counterion conductances (promoting acidification) are ubiquitous throughout the endolysosomal system and so would not be expected to differentially set the Δψ. Unfortunately, as we have seen, reliable determinations of Δψ across the endolysosomal system are sorely lacking (and seldom within in the same cell type) so we do not know whether Δψ actually does vary across the endolysosomal system.
Pumps and leaks
For all the worthy discussion of the protonmotive force, the principal determinant of the pHL set-point is the H+ pump/leak balance. The current model states that the ratio of H+ pumps to leaks is highest in the most acidic organelles; that is, the less acidic organelles (e.g. endosomes) leak H+ more than do the ‘tighter’ acidic ones (e.g. lysosomes) [33,70].
Surprisingly, rigorous quantification of the H+ pump density throughout the acidic system has still not been conducted , but given the variations in V-ATPase assembly this is not a trivial undertaking. What is more certain is that the H+-leak rate is demonstrably different in different compartments as revealed by V-ATPase inhibitors3 and is typified by the leaky Golgi and tighter secretory granules . The sea urchin egg provides another more extreme example: the yolk platelets are acidic organelles believed to be endolysosomal in origin , but they are extremely ‘tight’ vesicles with low permeability to H+ (or indeed other ions) and application of bafilomycin A1 fails to change pHL at all unless combined with ionophores that enhance counterion and H+ leaks [7,93].
What pathway(s) underlies the H+ leak? Probably not the V-ATPase itself  and the involvement of NHEs is equivocal [55,63,97]. In the Golgi, it was concluded to be a hitherto uncharacterized H+ conductance .
In summary, H+ leakiness is the primary factor that determines the characteristic pHL set-point of each compartment. Nevertheless, Δψ also makes a substantial contribution to pHL regulation in earlier compartments and the concerted action of both brings the pHL to the appropriate level.
Ca2+ UPTAKE INTO ACIDIC Ca2+ STORES
For an organelle to be a long term Ca2+ store, mechanisms for Ca2+ uptake and its luminal sequestration must be present to maintain stored Ca2+ at a sufficiently high concentration. As is well known, Ca2+ filling in the neutral ER is effected primarily by members of the SERCA (sarco-endoplasmic reticulum Ca2+-ATPase) family, comprised of three pump isoforms (plus splice variants), each with its own unique tissue distribution . In stark contrast, the route(s) of Ca2+ uptake into acidic stores is not as well understood, at least in the animal kingdom.
Although the least acidic, the Golgi is the best studied Ca2+ store with a family of Ca2+ pumps called the SPCAs (secretory pathway Ca2+ ATPases) mediating Ca2+ uptake. SPCAs are P-type Ca2+ pumps sensitive to inhibition by vanadate, but relatively insensitive to the SERCA inhibitor thapsigargin (except at very high concentrations) . Mutations of the SPCA gene result in Hailey–Hailey disease of the skin . In the trans-Golgi network, the SPCA is reported to be the only Ca2+ pump required , although others have suggested that additional pumps are also present .
To the best of our knowledge, the SPCAs are unique to the Golgi meaning that other acidic compartments accumulate Ca2+ via different Ca2+ pumps and/or exchangers. Unsurprisingly, we understand less about acidic Ca2+-store refilling in mammals than we do in primitive life forms (yeast, plants and protists) where a higher affinity Ca2+ pump co-exists with a lower affinity (high capacity) Ca2+ exchanger (Figure 3).
Yeast, plant and protist vacuoles
In yeast and plants, the clearance of Ca2+ from the cytosol is dominated by the internal compartments such as the massive acidic vacuole (occupying ≤80% of the cell volume) which then becomes a major Ca2+-storing organelle . Thankfully, a wealth of genetic and functional data illuminate these more primitive organisms and we will briefly discuss the two transport pathways.
Vacuolar Ca2+-ATPases are P-type ATPases (i.e. their reaction cycle involves a phosphorylated enzyme intermediate ) which are, perhaps surprisingly for organellar pumps, related to mammalian PMCAs (plasma membrane Ca2+-ATPases) in terms of sequence motifs and topology, but which differ in their subcellular localization and finer properties . For instance, the affinity of the PMCA for Ca2+ is high (0.2–0.5 μM) , but the affinity of the vacuolar Ca2+-ATPases is somewhat lower (0.3–4.3 μM) [102,104].
Intriguingly, their regulation is very much species-specific. Plant Ca2+-ATPases emulate their mammalian cousins: PMCAs are activated by CaM (calmodulin) displacing an autoinhibitory domain  and a similar mechanism exists for the plant vacuolar enzymes [for which reason they are known as ACAs (autoinhibited Ca2+-ATPases)] . Consequently, CaM activation lowers the Km and increases the Vmax values and thus the pump responds efficiently and removes cytosolic Ca2+ .
By contrast, yeast and slime mould Ca2+-ATPases lack the autoinhibitory domain [105,106] and so pump activity is controlled chronically at the transcriptional level (but still in a Ca2+-dependent fashion) . Finally, some pumps (such as the ACAs) may also be regulated by phosphorylation (e.g. PKC) in another parallel with PMCAs, although it is unclear whether this is by direct phosphorylation of the pump itself .
Unfortunately, the pharmacology of vacuolar Ca2+-ATPases is not well developed because they are relatively insensitive to the conventional SERCA inhibitors thapsigargin and cyclopiazonic acid. This is not so surprising for a PMCA-related protein that lacks the M3 helix Phe256 to which thapsigargin binds in SERCA .
Vacuolar CAXs (Ca2+/H+ exchangers)
Hijacking the potential energy of the vacuolar H+ gradient, CAXs transport substantial amounts of Ca2+ into the lumen. Compared with the pumps their Ca2+ affinity is dramatically lower (15–25 μM), but this is more than compensated for by their Vmax value which can be orders of magnitude greater than that of the Ca2+-ATPases . Such proteins are referred to overall as the CAX family and there are myriad members (six in Arabidopsis alone) which exchange with a stoichiometry of 1 Ca2+ to 2 or 3 H+ .
Like the Ca2+ pumps, these low-affinity exchangers are also tightly regulated, sometimes in a Ying-Yang relationship with the sister Ca2+-ATPases; for example, in yeast, whereas calcineurin up-regulates the Pmc1 Ca2+ pump, it inhibits the CAX (Vcx1), presumably to prevent Ca2+ overload of the vacuole (although how it actually inhibits is incompletely understood) . Interestingly, neither the pump nor the exchanger in yeast possess autoinhibitory domains.
By contrast, an autoinhibitory domain motif is present in many CAX proteins in plants, as it is in their ACA pumps . However, this CAX version of the domain rather discriminates against CaM (to which it is insensitive) in favour of other activating proteins such as novel CAX-interacting proteins, immunophilins and kinases .
Endolysosomal Ca2+ uptake
That the endolysosomes of higher organisms are gaining acceptance as Ca2+ stores bearing Ca2+ channels makes it all the more imperative to understand the mechanisms by which they fill, but unfortunately our understanding is woefully incomplete when compared with the vacuolar field. It has long been known that acidic organelles are packed with Ca2+ (Supplementary Tables S1 and S2), but how it actually gets there is more hazy.
What we do know is that Ca2+ filling of acidic stores is dependent upon the protonmotive force because manoeuvres that dissipate it usually empty the store(s) and/or prevent refilling: the abrogation of Ca2+ storage has been shown on numerous occasions by direct alkalinization of the lumen with membrane-permeant bases such as NH4Cl or with protonophores (nigericin or monensin) [10,94,107–111]. Alternatively, the H+ gradient can be indirectly collapsed by inhibiting the V-ATPase with bafilomycin A1 or concanamycin (thereby depending on a sufficient H+ leak [10,93]) and likewise this can prevent Ca2+ storage [1,2,10,112–118].
Δψ or ΔpH?
As discussed above, the protonmotive force is comprised of two components, Δψ and ΔpH, so are we right to assume that these agents primarily affect Ca2+ uptake via pH effects and ignore Δψ? Given that the V-ATPase is the principal generator of the lumen-positive Δψ, its inhibition results in vesicle depolarization [79,82–84,89,90,95] and similarly, simple H+-translocating ionophores [e.g. FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone)] also depolarize the organelles [83,89,92,95,119]. In stark contrast, electroneutral cation-exchanging protonophores (nigericin or monensin) or NH4Cl do not depolarize (and, if anything, hyperpolarize) [89,90,120,121]. In other words, these agents differentially affect Δψ (but commonly increase pHL) affirming that the ΔpH is, after all, the major component of the protonmotive force governing Ca2+ uptake. This also makes sense considering that Ca2+ uptake would otherwise be favoured upon vesicle depolarization by V-ATPase inhibitors or protonophores.
Irrespective of the organelle (acidic or neutral), inhibition of the Ca2+-uptake mechanism(s) in stores that are already replete with Ca2+ will only empty the store if Ca2+ leaks out (at steady-state, this Ca2+ leak is precisely balanced by the Ca2+ uptake). Hence in the ER, SERCA inhibition by thapsigargin unmasks the Ca2+ leak, evokes a ‘Ca2+ release’ and the stores empty. By the same token, dissipating ΔpH in acidic stores could inhibit Ca2+ uptake, but will only deplete them of Ca2+ if there is a sufficient Ca2+ leak. For many cell types, acidic stores are indeed ‘Ca2+ leaky’ and collapsing ΔpH causes a detectable Ca2+ release and/or inhibition of Ca2+ signals [2,112,114–117,122–128], but this need not be the case. In the sea urchin egg, bafilomycin A1 does not readily deplete acidic stores of Ca2+  (in part because they are not very H+ leaky , but their Ca2+ leak may also be small).
Endolysosomal Ca2+/H+ exchange
By analogy with the vacuolar systems described above, the effect of bafilomycin A1, nigericin etc. is explained most simply by the presence of a Ca2+/H+ exchanger (Figure 3), the reasoning being that if you abrogate the H+ gradient then the driving force for Ca2+ uptake is eliminated. However, this has almost become dogma when the supporting evidence is often circumstantial. For instance, addition of high concentrations of Ca2+ to the ‘cytosolic’ face of the acidic vesicle increase pHL, consistent with driving a Ca2+/H+ exchanger in reverse, as exemplified by sea urchin egg acidic vesicles [93,107], vacuoles , protist acidocalcisomes , synaptic vesicles , fungal spheroplasts  and endosomes . Considering there are other explanations for the Ca2+-induced alkalinization, e.g. H+-pump inhibition , it is reassuring when the converse is observed, i.e. a ΔpH-driven Ca2+ accumulation [94,130,132].
So, is the H+ gradient thermodynamically capable of generating the observed levels of luminal Ca2+ (free Ca2+ ~600 μM, total Ca2+ 4–14 mM; Supplementary Tables S1 and S2)? By analogy with plasma membrane exchanger thermodynamics , the equilibrium relationships between pHL, Δψ and luminal Ca2+ filling can be calculated for an acidic store exchanger (Figure 4). Electroneutral transport (1 Ca2+:2 H+) is, of course, unaffected by Δψ and a pHL of 5.0 could generate a 25000-fold Ca2+ gradient (2.5 mM luminal Ca2+). In contrast, the same ΔpH and an electrogenic 1 Ca2+:1 H+ could only generate a paltry ~50-fold gradient (assuming +30 mV, lumen-positive Δψ). The vacuolar (CAX) forms suitably utilize 2–3 H+ per Ca2+  and the endolysosomal system would require a similar stoichiometry on thermodynamic grounds alone.
Assuming that Ca2+/H+ exchange does occur across the acidic store membrane, what proteins might be responsible? It is frustrating that CAX proteins are absent from higher vertebrates  so they would have to belong to an entirely new family. Alternatively, Ca2+/H+ exchange may not be mediated by a single transporter, but rather by ‘nested’ exchangers acting in series and one plausible scenario is Na+/Ca2+ exchange coupled to Na+/H+ exchange (Figure 3).
For coupled transporters to work there are several prerequisites. First, the H+ gradient must drive Na+ uptake into the lumen (e.g. by a reverse-mode NHE) to establish a Na+ gradient that is sufficient, in turn, to drive Ca2+ uptake (e.g. by an Na+/Ca2+ exchanger; Figure 3). Indeed, Supplementary Tables S1 and S2 suggest that the [Na+] within acidic vesicles appropriately exceeds that of the cytosol (clearly, a Na+/K+-ATPase would also reinforce Na+ accumulation). In principle, K+ could substitute for Na+, but the luminal/cytosolic [K+] ratio appears less favourable for driving Ca2+ uptake (Supplementary Tables S1 and S2).
Secondly, Na+ (or K+) must leak promptly from the lumen when the V-ATPase is inhibited in order to dissipate the Na+ (K+) gradient and hence the driving force for Ca2+ uptake. The leak of monovalent cations is not particularly well studied, but reducing the ΔpH with FCCP or NH4Cl has been reported to lower lysosomal K+ [111,134] or Na+  hinting at a degree of leakiness consonant with the model.
Practically speaking, Na+/Ca2+ exchange has been implicated in secretory vesicles [135,136] and genomically speaking there is great scope for coupled transport. To compliment the NHEs, there are three families of CaCAs (Ca2+/cation antiporters) in higher organisms, namely NCX (Na+/Ca2+ exchangers), NCKXs (Na+/Ca2+-K+ exchangers)4 and CCX (Ca2+/cation exchangers) [137,138]. All are electrogenic [137,139] and some might be found on intracellular organelles (but reports are few) . Significantly, isoforms of NCKX have been suggested to contribute to Ca2+ filling of (secretory) chromaffin granules  and melanosomes  and, needless to say, endocytotic retrieval of plasma membrane forms of any of these proteins could recruit them for work in the endolysosomal system.
As tempting as the hypothesis is, there remains a problem with the coupled transport scheme as it stands. Its ΔpH-sensitive component is the NHE, but these are not expressed throughout the entire endolysosomal system (nor for that matter are the Na+/K+-ATPases) (Table 1). How then would later compartments like lysosomes fill with Ca2+ in a pH- and Na+-dependent manner? Clearly, these are issues that must be addressed in the future.
Endolysosomal Ca2+ pumps
Ever since SERCA inhibitors became widely available, they have been tested against organellar Ca2+ uptake mechanisms. However, vacuolar Ca2+ pumps show greater homology with the PMCA pumps that are insensitive to thapsigargin and cyclopiazonic acid, and there is no reason to suspect that an endolysosomal pump would be any different. The literature, however, is confusing because there are reports of thapsigargin-sensitive and -insensitive acidic Ca2+ stores. For example, a thapsigargin-insensitive Ca2+ uptake mechanism may exist in egg yolk platelets [10,142], slime mould vacuoles  and the acidic stores of HEK (human embryonic kidney)-293 cells , corneal endothelium , insect S2 cells  and pancreatic β cells [114,145–147]. By contrast, thapsigargin-sensitivity has been reported for the secretory granules of mouse pancreatic β cells , and acidic stores of MDCK (Madin–Darby canine kidney) cells  and lobster lysosomes (albeit with a low thapsigargin-sensitivity and Zn2+ as substrate) . Whether these are biological or methodological differences is presently unclear. Nonetheless, the sheer volume of reports does lend more support to a thapsigargin-insensitive route.
If stores are unaffected by thapsigargin, this could either mean that: (i) there are no Ca2+ pumps at all (just exchangers) or (ii) that there are thapsigargin-insensitive Ca2+ pumps. One way of distinguishing between these two possibilities is to test other Ca2+-ATPase inhibitors, and two informative ones have been vanadate and tBHQ (tert-butylhydroquinone).
Ca2+-ATPases are P-type pumps and as such are blocked by vanadate (resembling the γ phosphate of ATP at the active site)  (note that Na+/K+-ATPases are also inhibited by vanadate , but the V-ATPase is unaffected [89,119]). Importantly, vanadate also inhibits thapsigargin-insensitive Ca2+-ATPases, and so we can test for other pumps. Being membrane-impermeant, the use of vanadate is restricted to broken cell preparations, but nonetheless there are numerous reports of its inhibiting Ca2+ uptake into the slime mould vacuoles [104,143], synaptic vesicles , Golgi , protist acidocalcisomes , reticulocyte endosomes , and parotid or pancreatic acidic stores [155,156]. Using a slightly different readout, vanadate also affected the Ca2+-ATPase activity of rat liver lysosomes  and neurosecretory vesicles [119,158]. Taken together, these results are consistent with the presence of a P-type Ca2+-ATPase on some acidic Ca2+ stores.
Although the above is compelling evidence for Ca2+ pumps, not all preparations exhibit a vanadate sensitivity, and resistance to inhibition has been seen with fungal stores , egg yolk platelets , melanin granules  and neutrophil endosomes . Assuming that the non-trivial issues of vanadate polymerization and preparation  were adequately overcome, this implies that P-type Ca2+-ATPases are absent from particular organelles.
More recently, the use of another inhibitor, tBHQ, has led some to conclude that the Ca2+-ATPase on NAADP-sensitive acidic vesicles is a SERCA3-like pump [161,162]. According to this model tBHQ is selectively blocking the SERCA3 isoform, but not the SERCA2b on neutral stores (whereas the converse is true for thapsigargin which exhibits a modest selectivity for SERCA2b) . Indeed, the results are in accordance with the known pharmacology [164,165] and subcellular distribution of SERCA3 in platelets .
The scheme is intriguing, so is this a platelet peculiarity or a universal model? In support of universality: (i) SERCA3 does exhibit a low (or undetectable) sensitivity to thapsigargin and a low Ca2+ affinity  (compare with plants and yeast); (ii) tBHQ does preferentially inhibit acidic store Ca2+-ATPases in other systems ; or (iii) of the canonical SERCAs, SERCA3 exhibits a 6–30-fold greater sensitivity to vanadate (IC50 ~10 μM) , which is in reasonable agreement with the vanadate sensitivity of some Ca2+-ATPases on other stores (IC50 1–30 μM) [104,143,152,153]. On the negative side: (i) other acidic store Ca2+-ATPases are substantially less sensitive to vanadate (IC50 200–320 μM) [119,157]; and (ii) it is difficult to envisage this being a universal model for Ca2+ uptake into endolysosomal compartments when SERCA3 expression is, as far as we know, restricted to very few tissues , even accounting for the six splice variants of SERCA3a–f . For the moment we should perhaps be cautious about extrapolating the platelet work to other cell types at least until the SERCA3 expression pattern is better defined.
In conclusion, there is compelling evidence for Ca2+ pumps on many types of acidic Ca2+ store, although their molecular nature remains uncertain (with the possible exception of platelet alpha granules). However, the fact that other acidic Ca2+ stores are apparently resistant to Ca2+ inhibition suggests that pumps are absent from certain preparations and highlights the differences between acidic vesicle classes.
Ca2+-ATPases and ΔpH
An often overlooked fact is that for P-type Ca2+-ATPases, ATP hydrolysis not only drives the translocation of Ca2+ across a membrane (against its electrochemical gradient), but also transports H+ in the opposite direction [103,169]. The ramifications of this obligate exchange mechanism should be obvious. First, whether the pump is electrogenic or not will depend on the stoichiometry (a matter of debate, some suggest it is electroneutral [103,169]); secondly, and perhaps more important for this discussion, the ΔpH will affect Ca2+-ATPase activity. At the plasma membrane or neutral ER, the ΔpH is negligible and can ostensibly be ignored by resident pumps, but a Ca2+-ATPase on an acidic vesicle is a very different matter. Could the potential energy of the H+ gradient help drive Ca2+ uptake as has been suggested in lower organisms [104,170]?
Similar to analyses in yeast , one can calculate5 that the Ca2+ electrochemical driving force (ΔG) across a lysosome might be expected to be of the order of −28 to −42 kJ/mol. In comparison, the ΔG of ATP hydrolysis in vivo is of the order of −52 to −64 kJ/mol . A vacuolar/PMCA-type Ca2+ pump displays a stoichiometry of 1 ATP:1 Ca2+ , so ATP hydrolysis itself should be more than sufficient to translocate Ca2+. However, the SERCA family translocate 2 Ca2+ per cycle  which doubles the energy requirement (−56 to −84 kJ/mol). In other words, this is close to (or exceeds) the energy from a single ATP molecule which may not be competent to fuel the pumping of 2 Ca2+ into the lumen of lysosomes (depending on the membrane potential), and therefore the H+ gradient could provide the energetic shortfall6. That is, the intrinsic H+-exchange activity of SERCA is not just an epiphenomenon, but is probably an essential requirement for Ca2+ pumping into acidic compartments, and this may be pertinent to the platelet SERCA3 story.
This is not mere nit-picking because it has a bearing on interpreting the effect of altering ΔpH on Ca2+ signals. Classically, dissipating the pH gradient with protonophore or bafilomycin A1 is expected to inhibit Ca2+/H+ exchange, but what would be the effect on a Ca2+ pump in light of the above? On the one hand, pump activity might be severely inhibited by eliminating the H+ gradient (the thermodynamic ‘top up’). On the other hand, the store becomes more ER-like (little Δψ) which presents a lower energy obstacle for Ca2+ pumping7.
The upshot of collapsing the H+ gradient may be that the (formerly) ‘acidic’ store may not empty (or might even increase their Ca2+ filling) and the involvement of an acidic Ca2+ store may be erroneously dismissed. For example, it is one inventive explanation for an apparently bafilomycin A1-insensitive NAADP response (e.g. ). Indeed, these considerations remind us that the relationship between luminal [H+] and [Ca2+] need not be the reciprocal one that it is often suggested (i.e. the more acidic, the greater the Ca2+ filling), and this complexity is borne out by looking at the endolysosomal system as a whole (Figure 4).
One or two uptake pathways?
For a given endolysosomal organelle, does Ca2+/H+ exchange and a Ca2+ pump exist simultaneously, as is the case for plants and yeast vacuoles? In these organisms, the diversity of the Ca2+ affinity (and capacity) of the two pathways is commensurate with their different roles (e.g. store priming compared with cytosol clearance), and this would equally make sense in other organisms. Indeed, multiple Ca2+ uptake pathways do seem to co-exist in secretory granules  and pancreatic acidic stores , but we will have to wait and see if this is ubiquitous.
Endocytosis and Ca2+ filling
One final point is that processes such as endocytosis and phagocytosis may also contribute to Ca2+ filling of acidic vesicles. Having engulfed extracellular fluid, the earliest phago- or endo-somal structures would contain millimolar [Ca2+]. Nonetheless, the luminal free [Ca2+] is lowered with some rapidity (in minutes) to the micromolar range [172–174]. Assuming that the Ca2+-buffering capacity of the lumen will, if anything, be decreased as these vesicles acidify, this most probably reflects a real decrease in total Ca2+ due to extrusion from the vesicle. Will a modest 30–300-fold gradient still be sufficient for endosomes to act as Ca2+ stores? Circumstantial evidence suggests that they can. They've been implicated in endocytosis in pancreatic Ca2+ signalling  and TPCs (two-pore channels) (see below) are indeed found on endosomes [123,127,128], but, again, this awaits empirical determination.
All this serves to illustrate that the Ca2+-filling pathways of acidic stores are still ill-defined in systems other than plants and yeast. There is patchy (but consistent) evidence for Ca2+/H+ exchangers and/or Ca2+-ATPases, and thermodynamics impose constraints on their stoichiometry and modes of function.
Luminal Ca2+ buffers
Once Ca2+ is translocated into any Ca2+ store, only a small fraction of the total Ca2+ is free, the remainder is bound (to help reduce the Ca2+ electrochemical gradient and thereby facilitate Ca2+ storage). Within the ER, Ca2+ binds to the ER-selective proteins calsequestrin, calreticulin and calnexin [176–178], which possess highly acidic domains that bind Ca2+ with an appropriately low affinity yet high capacity (~500 μM and 20–50 mol/mol respectively [176,177]), ideal properties to reversibly sequester 50–90% of the ER Ca2+ [179,180].
Acidic Ca2+ stores emulate this precept and chelate substantial quantities of Ca2+, as a comparison of Supplementary Tables S1 and S2 attests (note the orders of magnitude difference in the total and free [Ca2+], up to 99.9% buffered ). What chemical groups could bind Ca2+ in such an acidic environment? Intuitively, one would expect that Ca2+ chelation might occur through anionic groups such as phosphate, sulfate and carboxylate (recall that S and P are abundant in acidic vesicles; Supplementary Table S1). The side-chain carboxylate on proteins (i.e. aspartate/glutamate) has a pKa value of ~4, sulfate conjugates <1, and polyprotic phosphate extends over 2–12. Therefore, despite the aggressively low pHL of 4.5–5.0, these groups will all be anionic (to some extent) and available to form complexes with Ca2+ ions (albeit with a lower affinity than at neutrality owing to H+ competition).
What molecules provide these groups in situ is more open to debate with a paucity of information available, but work has implicated several classes of molecules, in increasing size: (i) small organic acids such as oxalic acid that are abundant and Ca2+-binding in vacuoles (to the point of forming crystals) [102,105]; (ii) polyanionic matrixes that come in several forms, including polyphosphates (acidocalcisomes , vacuoles , lysosome-related organelles ), and glycosaminoglycans such as heparin ; and (iii) proteins, either binding Ca2+ through specific metal-binding motifs (e.g. melanin , toposome ) or non-specifically via extensive decoration by glycosylation (e.g. mucin ). It is worth recalling that it is these larger molecular mass molecules that contribute to the Donnan potential mentioned above, as well as to the pHL buffering capacity [34,35].
Ca2+ RELEASE MECHANISMS IN THE ENDOLYSOSOMAL SYSTEM
The study of Ca2+ release from acidic stores has been galvanized by the finding that in many cell types NAADP appears to trigger directly Ca2+ release from acidic organelles. We will initially describe the studies leading to NAADP as a Ca2+-mobilizing messenger, and how the dissection of its mechanism of action has led to the discovery of a novel class of endolysosomal channels in animal cells, the TPCs, whose functional characterization is beginning to enhance our understanding of Ca2+-release mechanisms from the endolysosomal system.
NAADP-evoked Ca2+ release
In a landmark paper, Lee and colleagues  discovered that two pyridine nucleotide metabolites released Ca2+ from homogenates prepared from sea urchin eggs. The sea urchin egg had been one of the earliest systems in which the then newly discovered Ca2+-mobilizing messenger IP3 (inositol 1,4,5-trisphosphate) had been shown to be active in releasing Ca2+ from intracellular stores . Previously, it had been shown that such egg homogenates displayed a robust response to IP3 which releases Ca2+ from non-mitochondrial stores in this preparation . Cell fractionation showed that the IP3-sensitive stores correlated with the distribution of ER markers. An important finding was that following IP3-evoked Ca2+ release, homogenates remained refractory to a further challenge with IP3 for some hours. Pyridine nucleotides were known to undergo dramatic changes at fertilization in the sea urchin egg , prompting Lee and colleagues to investigate their possible role in regulating vesicular Ca2+ transport processes in their newly defined cell-free system . In egg homogenates rendered refractory to IP3, the pyridine nucleotides NAD+ and NADP+ both caused Ca2+ release, but apparently by distinct mechanisms which varied in several important respects. First, whereas NADP+ gave a rapid response, NAD+ did so but after a latency of several seconds. Secondly, although, like IP3, NAD+ and NADP+ showed the property of homologous desensitization, they did not affect the ability of each other to evoke a response. Thirdly, homogenate fractionation indicated that NAD+ and NADP+ released Ca2+ from different populations of vesicles.
Further analyses of these three key observations led to several important discoveries in Ca2+ signalling. First, was the chemical identification of the Ca2+-mobilizing molecules themselves. The effect of NAD+ was ascribed to a novel cyclic metabolite cADPR (cyclic adenosine diphosphate ribose) . The conversion of NAD+ into cADPR by enzymes termed ADP-ribosyl cyclases was shown to account for the latency between NAD+ addition and the initiation of Ca2+ release . The rapid Ca2+-mobilizing effects of NADP+ was due to the contamination of commercial sources of the compound by the related analogue NAADP . The homologous self-desensitization of the three Ca2+-release mechanisms activated by IP3, cADPR and NAADP suggested that they activated distinct Ca2+-release mechanisms. The rapid kinetics of the responses to each Ca2+-mobilizing agent was indicative of activation of three separate Ca2+-release channels. In the late 1980s, two related families of Ca2+-release channel from the sarco-endoplasmic reticulum had been identified, IP3Rs (IP3 receptors) activated by IP3  and RyRs (ryanodine receptors)  which could be opened by increases in cytoplasmic Ca2+, so called CICR (Ca2+-induced Ca2+ release) [195,196]. An important finding was that cADPR triggered Ca2+ release through the activation of RyRs , and with the cognate receptor for IP3 identified , this left the question of the nature and identity of the NAADP-gated Ca2+ release channel. Pharmacological studies showed that NAADP-evoked Ca2+ release was not affected by blockers of IP3Rs or RyRs, but sensitive to VGCC (voltage-gated Ca2+ channel) blockers .
The other major finding was that NAADP released Ca2+ from a population of vesicles distinct from those sensitive to IP3 and cADPR. The depletion of endoplasmic stores by the SERCA pump inhibitor thapsigargin abolished Ca2+ release by both IP3 or cADPR, but the Ca2+-mobilizing effect of NAADP still persisted , even in intact eggs . In addition, it was found in stratified sea urchin eggs, that NAADP mobilized Ca2+ from an area of the egg which was distinct from that sensitive to IP3 or cADPR . The identification of the principal NAADP-sensitive store followed in an important study in 2002, where NAADP was demonstrated to release Ca2+ from organelles identified as reserve granules, acidic lysosomal-related organelles . This study for the first time linked NAADP-evoked Ca2+ release with acidic Ca2+ stores which has now been strengthened by studies in cell types from a variety of organisms including many from vertebrate systems . In addition, NAADP may also trigger Ca2+ influx at the plasma membrane, but it is not clear whether this is a direct effect on plasma membrane channels or a consequence of Ca2+ release from internal stores.
The trigger hypothesis: juxta-organellar Ca2+ signalling
In many cells, NAADP evokes sizeable Ca2+ signals . In a minority that includes sea urchin eggs, this may be explained by a preponderance of acidic Ca2+ stores [7,10], but in most cells lysosomes and lysosomal-related organelles constitute a much smaller cellular volume than the ER, and consequently mobilization of Ca2+ stores in lysosomes might be expected to produce only small and localized cytoplasmic Ca2+ signals . One explanation of this apparent paradox has been advanced in the formulation of the ‘trigger hypothesis’ to explain Ca2+ mobilization by NAADP .
Early analysis of NAADP-induced Ca2+ release from homogenates of sea urchin egg [142,205] or brain , indicated that the NAADP-sensitive Ca2+ release mechanism was not modulated by cytosolic Ca2+ (or Ca2+ surrogate ions). Nevertheless, NAADP can initiate regenerative global Ca2+ signals (waves and oscillations) and this is because ‘trigger’ Ca2+ provided by NAADP is subsequently amplified by recruitment of IP3Rs and RyRs that exhibit the characteristic property of CICR [199,207]. This ‘channel chatter’  may occur by two modes: most obviously by the trigger Ca2+ stimulating ER channels via the CICR, but also by trigger Ca2+ being sequestered into and ‘priming’ the ER (which sensitizes ER channels from the luminal face) (Figure 5).
Such communication between Ca2+-storing organelles demands close appositions as typified by the interactions between SR (sarcoplasmic reticulum)/ER and mitochondria  that is cemented physically by mitofusins acting to tether the organelles together . Is there an analogous juxtaposition of acidic organelles and ER? Lysosome/SR junctions have certainly been observed in vascular smooth muscle cells which are the site of initiation of NAADP and agonist-evoked Ca2+ signals [117,211] and sea urchin egg yolk platelets are cradled by proximate ER [7,212]. Whether this is ubiquitous and requires specialized tethering proteins remain to be shown.
Modulation of pHL by NAADP
Given that NAADP appears to mobilize Ca2+ from acidic stores in sea urchin eggs, the effect of NAADP on organellar luminal pH was investigated in egg homogenates and intact sea urchin eggs [93,213]. Employing acridine orange or Lysosensor dyes as pH indicators, NAADP was found to evoke an alkalinization of acidic stores [93,213]. This effect was highly specific in that neither IP3 or cADPR affected pHL, and the effects of NAADP were blocked by prior desensitization of the NAADP-sensitive Ca2+ release mechanism . This is not peculiar to eggs because NAADP-dependent alkalinization has been recapitulated in pancreatic acinar cells . The mechanism by which NAADP modulates pHL is unclear (see  for discussion), but it has been proposed to be secondary to Ca2+ release from acidic stores. Given that pHL is important for endolysosomal function, including membrane fusion and luminal enzyme activities , and NAADP receptor channel gating (see below) it may be of major physiological importance.
The search for the NAADP receptor
Since the discovery of the Ca2+-mobilizing properties of NAADP, there has been intense effort to identify the mechanism by which this molecule evokes Ca2+ signals in cells. A number of candidate ion channels have been implicated , for which there are various degrees of evidence, both for and against.
Although there is substantial evidence that RyRs are the principal effectors of cADPR-induced Ca2+ release from the ER, a number of studies have also implicated RyRs in NAADP-evoked Ca2+ release [116,216]. In many cells, NAADP-evoked Ca2+ release is certainly sensitive to inhibition by RyR blockers, and in many cases this may be a manifestation of the trigger hypothesis outlined above, where NAADP recruits ER stores via CICR [117,211]. However, in a number of studies, NAADP has been proposed to directly activate RyRs on the ER [217,218], their major site of subcellular localization, or RyRs on acidic stores, for which there is some evidence .
Extensive studies by Dammermann and Guse  in Jurkat cells have characterized a robust Ca2+-mobilizing response to microinjection of NAADP, which has the characteristic bell-shaped concentration–response curve observed widely in mammalian cells. However, this response is blocked by RyR inhibition, RNAi (RNA interference)-based knockdown of RyR expression, and is apparently insensitive to agents that interfere with Ca2+ storage by acidic organelles . Similarly, in pancreatic acinar cells, NAADP releases Ca2+ from the nuclear envelope, a contiguous Ca2+ store with the ER, which again is blocked by RyR inhibitors, but not by agents affecting acidic store Ca2+ storage .
Direct evidence for NAADP regulation of RyRs has come from studies on purified RyRs incorporated into lipid bilayers, whereby NAADP at nanomolar concentrations was found to activate the skeletal muscle RyR1 isoform with conductances typical for authentic RyRs . Such conclusions are based on the purity of the incorporated protein fractions. However, other bilayer studies of purified RyRs have failed to show significant activation by NAADP [219,220], and heterologous expression of RyRs in HEK-293 cells enhances cADPR, but not NAADP-evoked Ca2+ release . Furthermore, NAADP mobilizes Ca2+ in cells lacking RyRs, but requiring TPC expression [127,221].
TRPML1 (mucolipin 1)
Genetic analysis of the lysosomal storage disease MLIV (mucolipidosis type IV) uniquely identified mutations in a gene encoding a putative ion channel , whereas other storage diseases are due to defects in enzymes or transporters. The protein encoded by this gene, TRPML1, has homologies with ion channels of the Trp (transient receptor potential) family . Two homologues mucolipin-2 and mucolipin-3 were subsequently identified . TRPML1 was found as expected to localize to lysosomes, and lysosome-targeting motifs were identified [225,226]. TRPML1 is a cation channel with evidence that it is permeant to H+, Ca2+ and Fe2+, among others [227–229]. As the first ion channel definitively localized to lysosomes, it was an obvious candidate for mediating NAADP-evoked Ca2+ release from acidic stores. The first study of TRPML1 to examine the possible link with NAADP failed to find any modulation of the channel by NAADP, and neither did overexpression enhance [32P]NAADP binding to membranes from overexpressing cells . However, two subsequent reports from Li and colleagues based on single channel recordings implicated TRPML1 as targets for NAADP [230,231]. Nonetheless, a recent study involving both overexpression of all three isoforms of mucolipin as well as RNAi, failed to confirm these findings  and ruled out direct gating of mucolipins by NAADP. Therefore it may be highly relevant that TPCs co-immunoprecipitate with TRPML1  and may explain why purified TRPML1 complexes generate convincing NAADP responses in bilayers [218,219]. Instead, TRPML1 may be modulated physiologically by PIP2 [PtdIns(3,5)P2] . In conclusion, on balance, the case for TRPML1 as being an obligatory component of the NAADP-gated Ca2+-release channel is not supported. As an endolysosomal ion channel, TRPML1 could nonetheless influence luminal [Ca2+], pHL or Δψ to indirectly affect TPCs (see the Acidic Ca2+ Store Pathology section for further discussion).
TRPM2, a member of the melastatin family of Trp channels, is a channel with enzyme activity termed a chanzyme . It is regulated by ADP-ribose and the channel expresses ADP pyrophosphatase activity associated with its Nudix hydrolase domain . Although it is expressed primarily at the plasma membrane, a recent report has shown localization to lysosomes . Furthermore, the channel is sensitive to NAADP. However, the concentrations of NAADP required to modulate channel gating are in the high micromolar concentration range, very different from the nanomolar concentrations of this most potent of Ca2+ messengers that have routinely been found to be effective in most cells .
Whereas both metabotropic (P2Y) and ionotropic (P2X) families of purinoceptors are well-studied cell-surface receptors for ATP and other nucleotides, P2X4 has been found to be expressed in lysosomes and targeting motifs identified . Topologically, nucleotide-binding sites would be predicted to be luminal rather than cytoplasmic, requiring intra-organellar synthesis of ligands or transport into the lysosome. However, there are no reports that NAADP can modulate this channel, although high concentrations of NAADP may interact with P2Y receptors at the plasma-membrane-activating phospholipase C-linked signalling pathways.
A recent addition to the network of ion channels regulating lysosomal function is a small family of proteins termed TPCs encoded by TPCN genes (Figure 6) which have been the subject of recent reviews [3,5,8,9,239,240]. The founding member of this family TPC1 was cloned in 2000 by screening a rat kidney cDNA library for sequences based on known voltage-gated channels, the superfamily of which TPC is a member . α (pore-forming) subunits of voltage-gated sodium channels contain four homologous domain repeats of 6TMD (transmembrane domain) architecture, whereas α subunits of K+ channels contain only one domain and tetrameric assemblies are required to form functional channels . In contrast, a full sequence of a protein was identified with only two domains of 6TMD and thus named TPC1. These channels thus appear to represent an evolutionary intermediate form between putative voltage-gated Ca2+ 6TMD channel from the bacterium Bacillus halodurans , and 24TMD four-domain channels typified by mammalian voltage-gated Ca2+ or Na+ channels, generated by gene duplication during evolution. In the S4 domain (TMD1), repeats of charged amino acids are partially conserved with those in the voltage-sensor region of mammalian voltage-gated channels, suggesting that the TPC1 protein may possess a degree of voltage-sensitivity.
The initial description of the TPC1 protein was confined to mammalian tissue distribution by Northern blot analysis, and mRNA transcripts were found to be distributed across most rat tissues, with particularly high signals in kidney and liver . Immunostaining of kidney tubules with a polyclonal antibody raised against the TPC1 recombinant protein showed that the protein showed a polarized expression in renal tubules with more prominent staining in the apical domain . However, expression of rat TPC1 cRNA in Xenopus oocytes failed to generate any voltage-dependent plasma membrane currents significantly different from background . In this study, the authors noted that a distantly related protein with approximately 25% homology, had been deposited in the Arabidopsis database, and a study subsequently followed describing an initial characterization of AtTPC1 (Arabidopsis thaliana TPC1) .
AtTPC1 was found to be expressed in most Arabidopsis tissues and initially it was thought to be a plasma membrane channel. Employing aequorin as a Ca2+ probe, AtTPC1 expression was found, in part, to mediate Ca2+ signals elicited by osmotic shock with sucrose solutions . Modulation of membrane potential by genetic manipulation of H+-transport proteins was also found to affect the response. Finally, in yeast the expression of AtTPC1 was found to rescue Ca2+ transport processes in mutant strains. The authors concluded that AtTPC1 is a voltage-sensitive Ca2+-permeable transporter. In an important study, Sanders and colleagues showed that AtTPC1 was a vacuolar channel, and that it mediates the well-characterized SV (slow vacuolar) current important in stomatal movements and germination . In contrast with animal TPCs, the plant TPC1 proteins have two cytoplasmic EF-hands in the loop between the two 6TMDs (Figure 6) which may be more physiologically relevant to its regulation than its voltage-sensitivity conferred by charged residues in the S4 segments. This domain allows the channel to function as a CICR effector, which may be its major role in plants.
For 8 years, the plant TPC1 was the only TPC protein to be studied electrophysiologically, revealing an additional regulation by luminal Ca2+ (increases in Ca2+ decrease the channel open probability) an effect that is reduced in the D454N (fou) mutation in a putative luminal loop (Figure 6) of the TPC1 channel . Thus plant and animal TPCs, although only weakly homologous in sequence, exhibit similarities and differences in structure and function . Whether plant TPC1 channels are also gated by NAADP has not been demonstrated, but in the only report of NAADP-evoked Ca2+ release in plants to date, microsomal rather than vacuolar fractions appeared to comprise the major NAADP-sensitive Ca2+ stores .
Identification of animal TPCs as NAADP-gated channels
In another landmark study, major candidates for NAADP-gated Ca2+-release channels were identified as TPCs . The strength of the candidature of TPCs as possible mediators of NAADP-induced Ca2+ release was based on two properties of these channels. The first was their localization to the endolysosomal system, indeed inspection of the human TPC2 sequence revealed a putative dileucine lysosomal-targeting motif. The second was their homology with voltage-gated cation channels, a possibility given that dihydropyridines and other cation channel blockers inhibit NAADP-stimulated Ca2+ release . Subsequently, other groups confirmed that TPCs could be gated by NAADP [123,144].
The homology of TPCs with voltage-gated cation channels has led to the prediction of their transmembrane arrangement (Figure 6). Each of the two repeated 6TMD regions contain a putative pore-forming region between the putative S5 and S6 membrane-spanning sequences, and this prediction is consistent with topology studies mapping trypsin and antibody access to fluorescent protein tags at different sites along TPCs .
Like other lysosomal proteins, TPCs are glycosylated [128,144,250], probably luminally (Figure 6), affording them protection in a highly acidic environment (see above) and giving rise to different TPC electrophoretic mobilities (apparent molecular masses ranging from 80–100 kDa depending on the degree of glycosylation). Such TPC glycosylation may well regulate their sensitivity to activation by NAADP as judged by mutation of glycosylation sites near the putative pore .
Given their domain structure, TPCs might be predicted to form dimeric functional channels. Indeed, co-immunoprecipitation studies have indicated that TPC2 homodimerization may occur , but how the TPC family assembles must be examined in greater detail.
Localization of TPCs
There are three isoforms of TPCs expressed in most animals, an exception being the primate and rodent genomes which either completely lack TPC3 or contain a pseudogene . The three TPCs are equally distant from each other (see below), and from plant TPC1, with approximately 30% conserved amino acid identity in the conserved transmembrane regions . Heterologous expression of human and mouse TPC1 and TPC2, and chicken TPC3 in HEK-293 cells showed that although all three isoforms localize to components of the endolysosomal system, there are differences . TPC2 appears to localize predominantly to lysosomes and late endosomes, whereas TPC1 is more endosomal. TPC3 may be largely in recycling endosomes (Figure 7). A polyclonal antibody raised against the HsTPC2 (human TPC2) sequence, showed that endogenous TPC2 protein co-localizes with the lysosomal marker LAMP2 in HEK-293 cells. There have been no reports of endogenous localization of TPCs to the ER or plasma membrane. Both the N-terminal sequence and certain leucine residues appear to be important for HsTPC2 targeting to lysosomes, since deletion or changes to these residues results in their appearance at the plasma membrane .
TPCs and NAADP-evoked Ca2+ release
Several vertebrate and invertebrate TPCs have been cloned and their Ca2+-releasing properties and sensitivity to NAADP has been examined by heterologous expression in cell lines. Three broadly consistent reports appeared in 2009, each supporting the notion that TPCs are mediators of Ca2+ release from endolysosomal stores. The first report was a comprehensive study which encompassed all three vertebrate TPC isoforms . Introduction of NAADP or caged NAADP into control HEK-293 cells through a patch pipette elicited only a small Ca2+ response. However, in cells overexpressing lysosomal HsTPC2 channels, NAADP or photolysis of caged NAADP, induced a large biphasic Ca2+ release . The first phase was proposed to be due to direct NAADP-evoked Ca2+ release from acidic stores since it was blocked by bafilomycin. Following this, a second larger Ca2+ release was found to be heparin-sensitive and probably due to amplification through the recruitment of IP3Rs on ER stores, and nicely consistent with the trigger hypothesis for NAADP action.
Conversely, NAADP responses were abolished in cells treated with siRNA (small interfering RNA) to TPC2, and also in pancreatic β cells derived from TPC2-knockout mice . Thus it was proposed that TPCs are likely to be excellent candidates for the elusive NAADP-gated Ca2+-release channels on acidic stores. In a subsequent independent study, expression of TPC2 was similarly found to evoke Ca2+ release from lysosomal Ca2+ stores, although under the conditions of this study, amplification of the Ca2+ response by recruitment of ER Ca2+ stores was not apparent .
Subsequently, Ca2+ responses to microinjected NAADP in the breast cancer SKB3 cell line were shown to be affected by altering TPC expression. TPC1 shRNA (short hairpin RNA) reduced both the endogenous expression and Ca2+ responses, whereas TPC1 overexpression enhanced the sensitivity of NAADP-induced Ca2+ release . Since bafilomycin and ryanodine both abolished NAADP responses, the authors invoked an NAADP trigger hypothesis implicating RyRs on the ER. In summary, these three studies support the idea that TPCs are likely to be the targets for NAADP action at acidic stores, but differences in details, especially with regard to juxta-organellar coupling between different classes of Ca2+-release channel, are likely to be due to cell-specific differences, and variations in the levels of heterologous expression.
In view of this, it was thus important to assess properties of endogenous TPC proteins with regard to their interactions with NAADP. Because of the prior extensive studies in the sea urchin egg, TPCs were immunopurified from solubilized membrane fractions from these cells. Remarkably, [32P]NAADP binding with these immunoprecipitates exhibited all of the hallmark features of binding to intact membranes. Binding was high affinity (Kd ~1 nM), showed a high degree of selectivity over NADP+, and was irreversible in the presence of high concentrations of K+ ions . In terms of Ca2+ signals, pre-treatment with subthreshold concentrations of NAADP inactivated NAADP-induced Ca2+ release as occurs with bona fide sea urchin egg NAADP receptors ; moreover, heterologous expression of SpTPC (sea urchin TPC) 1 and SpTPC2 in HEK-293 cells enhanced Ca2+ release in response to patch-applied NAADP or the membrane-permeant NAADP-AM (NAADP-acetoxymethyl ester). In contrast, overexpression of SpTPC3, failed to enhance NAADP responses, and acted in a dominant-negative fashion, and abolished the enhanced response to NAADP in cells co-expressed with TPC2 . However, in a separate study of SpTPCs, heterologous expression in SKB3 cells demonstrated that all three SpTPCs could mediate NAADP-evoked Ca2+ release .
The localization of TPCs to the endolysosomal system presents difficulties in their study by electrophysiological techniques to examine their channel properties. However, three complimentary approaches have now been reported, all examining HsTPC2 channel properties, all with various degrees of strengths and weaknesses (Figure 9). However, useful parameters from these comparative approaches have been obtained, giving deeper insights into the nature of these proteins and their regulation.
The first is patching of purified lysosomes [80,252]. Using the planar patch-clamp technique (Figure 9), currents were recorded from single lysosomes swollen with the small molecule, vacuolin, prepared from cells overexpressing TPC2 (or wild-type controls). This approach has the advantage of examining the properties of channels in their native membrane, although it is not clear whether vacuolin affects channel activity or whether other lysosomal channels (or other organelles) confound the preparation. Only in the TPC2-overexpressing cells were NAADP-evoked whole lysosomal currents detected, and the current was abolished by mutations in the proposed pore loop 1 (S5–S6) (N257A) suggesting a role in channel gating . The current was selective for Ca2+ over K+ by a factor of >1000, but mutations in E643A in the pore loop of the second putative S5–S6 domain changed the ion selectivity in favour of monovalent cations . The currents were modulated by intraluminal pH, with currents only observed at acidic pH .
In a traditional approach much used to study IP3 and RyRs, immunopurified TPC2 was incorporated into artificial planar lipid bilayer membranes (Figure 9). In contrast with whole organelle electrophysiology, this approach may give an indication of TPC properties uncontaminated by the presence of other channels that exist in cellular membranes (depending on the purity). This study gave the first insight into single-channel characteristics of TPC2 channels, and details of the molecular mechanisms by which NAADP may evoke Ca2+ release by TPC2 activation . The single-channel conductance was approximately 300 pS for K+, but only 15 pS for Ca2+ . NAADP increased Po (mean open probability), and was dramatically sensitized by increasing luminal Ca2+ concentrations (EC50 for NAADP of 5 nM at 200 μM Ca2+, in the range reported for lysosomal Ca2+ content [1,2]).
Channel activity was reported at both acidic and neutral pH, but at acidic pH, the characteristic bell-shaped concentration–response curve was observed with high concentrations (millimolar) of NAADP failing to open the channels, in line with mammalian cell NAADP-induced Ca2+ release . In addition, pH appears to regulate the reversibility of NAADP effects on channel activity. Alkalinization of the lumen, as shown to be induced by NAADP, would promote an increase in channel Po, but also the irreversibility of NAADP effects on the channel. This latter effect could conceivably be related to the profound desensitization properties of NAADP-sensitive mechanisms widely observed. Closure of channels would allow both the re-acidification, which would promote NAADP dissociation ready for a future round of activation by NAADP, and refilling of Ca2+ stores would sensitize the channel to another round of NAADP activation (Figure 8). Finally, validation of TPC2 as a target for the selective inhibitor of NAADP-evoked Ca2+ release Ned-19, was achieved by showing that it blocked the channels and their activation by NAADP, although interesting at low concentrations it may activate the channels . This study highlighted the complex interactions between cytoplasmic NAADP, and pHL and luminal Ca2+ concentrations in dictating the activity of these channels and presumably Ca2+ release from lysosomes.
In a third approach, TPC2 was mutated and the N-terminus deleted resulting in its mistargeting to the plasma membrane, where it is more amenable to study by patch-clamp electrode techniques (Figure 9) . Although this approach may raise a few concerns about changes in the channel structure or non-native location, it does provide elegant support for the trigger hypothesis (Figure 5). In this setup, the plasma membrane form is uncoupled from the CICR via ER Ca2+ channels and it conducts currents across the plasma membrane that are appropriately sensitive to NAADP and Ned-19 . Single-channel currents (Cs+ as a carrier) were ~130 pS and, again, mutations in putative pore-forming loops abolished this current .
In all three studies, current–voltage relationships were virtually ohmic (perhaps a slight inward rectification) suggesting that the channel is not voltage-gated in spite of modest conservation of charged amino acid residues in the putative S4 domains. That the membrane potential across lysosomes (Supplementary Table S2) is much more modest than that across the plasma membrane also mitigates against a major role for voltage in this channel's regulation . The major conclusions from these studies are that the TPC2 protein functions as a cation-selective channel, is permeant to Ca2+, modulated by cytoplasmic NAADP at physiologically relevant concentrations, and is also highly dependent on luminal Ca2+ and pH, the latter consistent with its lysosomal localization. The possible lack of sensitivity to cytoplasmic Ca2+ is in marked contrast with IP3Rs and RyRs, and consistent with its role as a trigger of local rather than global Ca2+ release.
PHYSIOLOGICAL ROLES OF Ca2+ RELEASE FROM ENDOLYSOSOMAL Ca2+ STORES
Although there had been sporadic reports of agonist-evoked Ca2+ release from acidic stores, it has been NAADP that has galvanized this field over the last decade. The identification of acidic Ca2+ stores and TPCs as NAADP targets has provided chemical and molecular tools by which to study the physiological roles of Ca2+ release from the endolysosomal system. This new component of the Ca2+ toolkit provides new ways of controlling how and where Ca2+ signals are generated.
Agonist-mediated NAADP signalling
A growing number of receptors have now been implicated as coupled to NAADP-mediated Ca2+ signalling . Until now, four approaches of various directness have been employed to implicate NAADP in receptor-mediated Ca2+-signalling pathways. These include inhibition of agonist-evoked Ca2+ signals by self-inactivation of NAADP , use of the selective membrane- permeant NAADP antagonist Ned-19 [253,254], measurement of cellular NAADP levels following agonist stimulation  and more indirectly by ablation of Ca2+ storage by acidic stores with, for example, GPN (glycyl-L- phenylalanine-naphthylamide) or bafilomycin. The pathways leading to NAADP synthesis need clarification, but several reports in different cell types have implicated CD38 [118,214,256,257], an ADP-ribosyl cyclase enzyme shown previously to catalyse the synthesis of NAADP from NADP in vitro.
The characteristic ‘Ca2+ signature’ response evoked by a given cell-surface receptor may, in part, be the product of ‘mixing and matching’ the three Ca2+-mobilizing messengers and cognate Ca2+ stores [114,258]. This then allows Ca2+ signals to be fine-tuned and in turn specify how the cell will respond. Thus receptors previously ascribed to couple to IP3 production may also link to NAADP-dependent pathways . In view of the trigger hypothesis, NAADP may act to provide co-agonist Ca2+ at the juxta-organellar junctions to sensitize IP3Rs when cellular IP3 levels are also elevated. Similar arguments can be advanced for NAADP/cADPR combinations which would recruit RyRs, as found for endothelin-1-evoked Ca2+ signals in rat pulmonary arteriolar smooth muscle cells .
Alteration in Ca2+ release from lysosomes may be important in several lysosomal disorders as discussed below, but also in other pathological processes, which may result in alterations in agonist-mediated Ca2+ signalling. Enlargement of lysosomes by treatment with cathepsin inhibitors, modulates the responses to both NAADP and glutamate, and such aberrant Ca2+ signalling may be of relevance in the pathophysiology of neurodegeneration . In an accumulating body of work examining aberrant Ca2+ signalling in pancreatic acinar cells linked to premature intracellular trypsin activation as occurs in acute pancreatitis, Petersen et al.  have found that bile acids, and alcohol and metabolites may disproportionately evoke Ca2+ release from acidic stores over the ER.
Regulation of plasma membrane excitability by NAADP
It has long been known that Ca2+ release from intracellular stores by NAADP may modulate plasma membrane channels and membrane excitability. NAADP was found to depolarize the plasma membrane of invertebrate eggs and evoke Ca2+ influx [260,261], and may be important in mediating the fast block to polyspermy at fertilization, a property not shared by either IP3 or cADPR . NAADP activates membrane Ca2+-activated currents in pancreatic acinar cells , important for fluid secretion, whereas in excitable cells it can trigger changes in membrane potential [115,127]. In many cases, these effects are secondary to Ca2+ release from acidic stores which may or may not involve amplification by ER Ca2+-release mechanisms, although direct action at plasma membrane channels has also been proposed . Importantly, it was found that NAADP activates plasma membrane currents in mouse pancreatic β cells, effects that are abolished in cells from TPC2-knockout mice . Thus Ca2+ release from acidic stores proximal to the plasma membrane may be important determinants of plasma membrane excitability, and since such TPC-expressing organelles are dynamic , they can be targeted to different subcellular sites to initiate various Ca2+-dependent cellular responses.
The localization of TPCs to the endolysosomal system, and in particular TPC2 to lysosomes, has prompted an investigation of the possible role for these channels in endolysosomal trafficking and lysosome biogenesis. Ca2+ release and luminal Ca2+ in endolysosomal vesicles has been proposed to play a key role in vesicular fusion mechanisms in the endolysosomal system [6,263–265]. TPCs thus are an attractive mediator for local Ca2+ signals regulating both luminal and juxta-vesicular Ca2+ levels in the endolysosomal system. Indeed, overexpression of TPC1 and TPC2 cause enlarged lysosomal structures to form, consistent with enhanced endolysosomal Ca2+ release promoting fusion, and also changing endocytosis and trafficking of material from the plasma membrane . Importantly, these effects were rescued by the NAADP antagonist Ned-19 fully implicating the NAADP/TPC axis in these processes . NAADP-mediated Ca2+ signalling may be affected in pathological processes affecting lysosomes, the most prominent being lysosomal storage diseases. These are discussed below.
TPC-dependent Ca2+ signalling
With the identification of TPCs as targets for NAADP, cells from TPC-knockout mice and treatment with TPC siRNAs, have offered a new approach to identify stimuli and physiological processes coupled to NAADP/TPCs.
In mouse detrusor smooth muscle, muscarinic receptor activation induces contraction by releasing Ca2+ from both acidic and SR stores . In permeabilized detrusor muscle strips, NAADP evoked contraction with the characteristic bell-shaped curve, but had no effect in muscle derived from TPC2-knockout mice . Consequently, muscarinic receptor activation contracted muscle from TPC2-knockout mice exclusively through the SR and confirms acetylcholine signals through NAADP/TPCs .
In a recent study, histamine-evoked Ca2+ release via the H1 receptor in human endothelial cells was suggested to be linked to NAADP signalling by pharmacological studies, and accordingly, the secretion of von Willebrand's Factor was significantly reduced by siRNA constructs for both TPC1 and TPC2 .
With the realization that there are multiple messengers for Ca2+ mobilization, an obvious question is whether activation of distinct mechanisms control specific cellular responses. One fundamental cellular process regulated by Ca2+, cell differentiation, has been selectively linked to Ca2+ release by NAADP from acidic stores. In skeletal muscle, differentiation was affected by Ned-19, bafilomycin or siRNA to TPC2 (but was less sensitive to blocking IP3Rs or RyRs) . Similarly, liposomal delivery of NAADP (but not IP3 or cADPR) induced neurite extension . These studies underscore the role of Ca2+ release from acidic stores to mediate specific cellular responses that cannot be substituted by ER-mediated Ca2+ release.
Finally, in a genome-wide association study it was reported that SNPs (single nucleotide polymorphisms) in the hsTPC2 gene were associated with pigmentation . Since melanosomes are lysosome-related organelles, changes in the ion-transport processes across the membrane of these melanin-storing organelles might impair pigment storage giving rise to associations with particular hair and skin pigmentation.
Why different TPCs?
Obvious questions are why are there different TPCs and in what ways are they similar and different? First, the primary sequences of the three isoforms are quite different from one another (within a given species only ~20% identical, ~35% similar), but each orthologue is quite well conserved across species (40–90% identical, 55–93% similar) [123,127,128]. This hints at specific conserved functions for each isoform. Accordingly, the subcellular distribution throughout the endolysosomal system is exquisitely isoform specific (Figure 7), with TPC1 especially exhibiting a far broader, more nebulous, distribution in the endosomal compartments, whereas TPC2 is restricted to late endosomes/lysosomes (Figure 7). The paucity of reliable antibodies means that the subcellular localization of endogenous TPCs has only occasionally been investigated with TPC1 found apically in rat kidney tubules  and TPC2 in lysosomes in HEK-293 cells , whereas TPC3 is localized in cortical puncta in sea urchin eggs (agreeing with heterologous expression of SpTPC1–3 in echinoderm oocytes) .
As to tissue distribution, Northern blot analyses revealed that both TPC1  and TPC2  appear to be expressed widely, the corollary being that multiple TPCs may be co-expressed within individual cells [123,128,266]. Although early days, it seems that TPC1 dominates TPC2 expression, at least in human endothelial cells (real-time PCR analysis revealed a transcript ratio of 9:1 ), skeletal muscle , SKBR3 and PC12 cells, and sea urchin eggs . To add to the complexity, their weakly overlapping subcellular distribution may well allow some heterodimerization in selected subcompartments (compare with TPC2 homodimerization ). This could potentially generate an even broader spectrum of TPC assemblies and this crucially awaits experimental clarification.
Properties of TPC isoforms
Despite their divergence, the NAADP-binding properties of individual TPCs seem well conserved, at least for endogenous SpTPC1 and SpTPC3 immunoprecipitates  and for heterologously expressed HsTPC2 . Translating this into Ca2+ release, there are manifest differences in the pattern mediated by TPC isoforms. Compared with TPC2, TPC1 is a less-efficient ‘trigger’ and couples weakly to ER Ca2+ release in some cell types. Activation of heterologously expressed human TPC1 in HEK-293 cells evoked only a local non-propagating Ca2+ response  and similarly, the sea urchin orthologue SpTPC1 recruited IP3Rs, but was delayed ~100 s compared with SpTPC2 . A working hypothesis is that endosomal TPC1 is not so closely apposed to the ER as is the late endosome/lysosome TPC2 and therefore the kinetics of Ca2+ accumulation at the juxta-organellar interface is suboptimal with TPC1.
In contrast, expression of HsTPC1  or SpTPCs  in a breast cancer cell line did not reveal any obvious ER-coupling deficiencies when microinjected with NAADP and all gave robust Ca2+ responses. Why this seems to differ with the other studies is unclear, but may be due to experimental differences. SKBR3 cells may inherently couple more efficiently (due to a different cell architecture or different amplifiers, RyRs in this case) or the rapid NAADP delivery by injection might be too fast to reveal coupling delays (the previous studies used slower cytosolic dialysis).
The story for SpTPC3, however, is less clear, with reports of overexpression of this isoform inhibiting  or potentiating  NAADP-evoked Ca2+ signals. How can there be such differences and how could TPC3 be inhibiting at all? The differences may not be so surprising when considering that the sea urchin genome is highly prone to polymorphisms  and so the respective sequences need to be scrutinized for potentially informative mutations that alter function. How some TPC3 sequences act ostensibly as ‘dominant–negatives’ (with either exogenous or endogenous TPC1/2) is less clear. If pHL is not affected by TPC3 , then it is possible that TPC3/TPC2 heterodimerize to form dysfunctional channels (e.g. if SpTPC3 mediated the characteristic self-inactivation of sea urchin egg NAADP receptors at sub-threshold [NAADP]). Further work is required to examine the relevance of these observations to other TPC3 orthologues and polymorphisms.
TPC isoform physiology
Finally, what physiological processes require TPCs? All three sea urchin egg isoforms are cortical in eggs/oocytes  and, circumstantially, in the right place to deliver Ca2+ for exocytosis of the fertilization envelope and for the pHLASH (a term coined to describe the selective alkalinization of cortical acidic vesicles) . In mammalian cells, although TPC2 may be the minority isoform, its abundance belies its importance, and knockout of TPC2 in the mouse model abolishes NAADP-induced responses such as Ca2+-dependent plasma membrane currents in mouse pancreatic β cells  and bladder muscle contraction . Similarly, in studies of skeletal muscle differentiation, TPC2 siRNA was more effective than TPC1 siRNA in inhibiting differentiation of myoblasts . Although of unknown physiological significance, heterologously expressed TPC2 forms complex with TRPML1 (and TRPML3 to a lesser extent), whereas TPC1 does not .
However, TPC1 shRNA profoundly reduced NAADP-induced Ca2+ release in SKBR3 cells, in spite of the additional presence of TPC2 . Moreover, overexpression of SpTPC1 or SpTPC2 had similar effects upon endolysosomal trafficking .
On balance, the pre-eminence of TPC2 thus far might simply be a function of coincidence (the few systems studied are more TPC2-dependent), a global disruption of heterodimerization, or a real possibility that TPC1 is less crucial (or efficient) for Ca2+ release. Only more studies will illuminate this most important of issues.
ACIDIC Ca2+ STORE PATHOLOGY
Potential role in infection?
Many intracellular pathogens enter macrophages through phagocytosis and need to avoid fusion with lysosomes if they are to survive and replicate. Although the intracellular habitat is potentially hostile (the acidic and hydrolase-rich lysosomes will rapidly degrade bacteria) significant survival advantages are associated with the intracellular environment, including reduced exposure to immune surveillance and a rich supply of nutrients provided by the host cell.
We now know of many human pathogens that have evolved mechanisms to manipulate the endocytic pathway to their advantage, including escaping from the phagosome into the cytosol (e.g. Listeria and Shigella species), manipulating the phagosome membrane to hide the nature of the compartment to prevent fusion with lysosomes (e.g. Legionella) and blocking phagosome–lysosome fusion (e.g. pathogenic Mycobateria). These mechanisms have been reviewed recently in .
However, in light of our new knowledge about acidic store Ca2+ regulation, we speculate that micro-organisms may have evolved mechanisms to subvert these processes for their own ends. This would potentially prevent phagosome–lysosome fusion and would appear to be an excellent evolutionary target and a future direction for research.
Lysosomal storage diseases
Diseases that result from defects in any aspect of lysosomal homoeostasis are termed lysosomal storage disorders. Impaired lysosomal function leads to the accumulation (‘storage’) of undegraded macromolecules in the late endocytic/lysosomal system . Currently approximately 50 disorders are known and the majority are inherited as autosomal recessive traits . They occur at a collective frequency of 1:5000 live births and are the most frequent cause of neurodegeneration in infants and children . Recent studies [resulting from the surveillance program to monitor vCJD (variant Creutzfeldt–Jakob disease) cases] have also highlighted that lysosomal disorders are a common cause of progressive intellectual and neurological deterioration in children in the U.K. .
Lysosomal disorders were identified clinically over a century ago with the biochemical nature of storage elucidated from the 1960s onwards . The advent of molecular biology led to the identification of the causative gene defects. What has emerged is that the majority of these disorders result from defects in lysosomal hydrolase function in line with Hers original prediction following his seminal studies on the glycogen storage disorder Pompe disease .
However, not all lysosomal disorders result from enzyme deficiencies and a significant subset are caused by mutations in lysosomal membrane proteins . We currently know of two lysosomal storage diseases that involve defective acidic store Ca2+ regulation [NPC (Niemann–Pick type C) and MLIV] and interestingly these diseases have opposing defects. NPC is characterised by reduced NAADP-induced Ca2+ release leading to impaired late endosome/lysosome fusion, whereas in MLIV there is increased Ca2+ release leading to enhanced endocytic pathway fusion. The current status of our understanding of the mechanism leading to dysregulated acidic store Ca2+ in these two diseases is discussed below.
NPC presents as a progressive neurodegenerative disease and patients typically die in childhood/adolescence , although more chronic forms exist and usually present in young adults. Cerebellar atrophy and a profound yet selective loss of Purkinje neurons is characteristic, which leads to ataxia . Additional symptoms include learning difficulties, psychiatric symptoms, epilepsy, speech loss, vertical gaze palsy and respiratory dysfunction . Neurofibrillar tangles similar to those found in Alzheimer's disease are present and dementia is a characteristic symptom, suggesting potential pathophysiological convergence between these two disorders .
NPC disease is one of the most complex lysosomal disorders with multiple lipid species stored including cholesterol, sphingomyelin, sphingosine and glycosphingoipids . A clue to the cellular pathogenesis of this disorder is that it has a unique late endosome/lysosome fusion defect, suggestive of altered Ca2+ homoeostasis. The storage bodies in NPC visualized by electron microscopy have a heterogeneous morphology, presumably reflecting the diverse range of lipids stored (Figure 10). This is in contrast with the much more homogeneous inclusion bodies observed in diseases such as Sandhoff disease where a single major ganglioside species is stored as the result of a lysosomal hydrolase defect (β-hexosaminidase; Figure 10).
NPC is unusual in that it is caused by mutations in either of two independent genes termed NPC1 and NPC2 , responsible for 95% and 5% of clinical cases respectively . Although the NPC1 and NPC2 genes were cloned over a decade ago, the precise functions of the proteins they encode and how they interact still remains controversial .
NPC1 is a 13TMD protein of the limiting membrane of late endosomes/lysosomes, whereas NPC2 is a soluble cholesterol-binding protein (previously known as HE1 due to its high levels in human epididymal fluid) . There are currently two main theories in the field as to the function of the NPC pathway: (i) that it transports cholesterol out of late endosomes/lysosomes [281,282] or (ii) it transports non-sterol cargo(s) out of the acidic compartment and is potentially regulated by cholesterol [2,283–285].
In support of the second hypothesis, we identified NPC1 as a candidate lysosomal sphingosine transporter . When NPC1 is inactivated the first metabolite to accumulate is sphingosine  and exogenous addition of sphingosine to healthy cells induces NPC cellular phenotypes . Sphingosine storage also has a profound effect on cells as it causes a dramatic reduction in luminal acidic store Ca2+ content . Indeed, the Ca2+ levels in the acidic compartment are reduced to 30% of the wild-type, which translates into reduced Ca2+ release in response to NAADP  (Figure 10). A drop in the luminal [Ca2+] not only reduces the electrochemical driving force for Ca2+ release, but also may lower the NAADP sensitivity of TPCs . This defect in luminal Ca2+ content in NPC was not due to altered acidic store pH, but due to impaired store filling .
It is currently not known if sphingosine storage directly or indirectly affects the enigmatic protein(s) involved in acidic store Ca2+ filling, but the central role for defective lysosomal Ca2+ regulation in the pathogenesis of this disease came from two lines of experimental evidence. First, NPC disease phenotypes could be induced in healthy cells by chelating luminal lysosomal Ca2+, and secondly, NPC cells could be corrected by elevating cytosolic Ca2+ using the weak SERCA inhibitor curcumin to compensate for lack of adequate Ca2+ release from acidic stores . Indeed, curcumin treatment of a mouse model of NPC1 also resulted in clinical benefit suggesting that mild elevation of cytosolic Ca2+ may be a novel therapeutic approach to treat NPC disease in patients . We believe that the low luminal lysosomal Ca2+ levels in NPC disease explains the block in late endosome–lysosome fusion in this disorder. Fusion and vesicular release is a Ca2+-dependent process  and the Ca2+ that facilitates these processes is derived from the lysosomal compartment itself . The failure to release sufficient Ca2+ in this disease therefore leads to a block in the normal trafficking and fusion essential for the correct functioning of the endosomal/lysosomal system, causing the secondary storage of cholesterol, glycosphingolipids and sphingomyelin .
Just as NPC is a disease that involves an endolysosomal Ca2+ defect so too does MLIV, but via a different mechanism with very different consequences. MLIV is a rare neurodegenerative lysosomal storage disease caused by mutations in the Mcoln1 gene encoding TRPML1 [223,286–288]. MLIV patients have progressive early developmental decline, increased blood gastrin levels and neuronal loss followed by a period of stabilization that lasts several decades .
TRPML1 was recently uncovered as a major gene product involved in a network of lysosomal genes believed to regulate lysosomal generation and function  and MLIV cells accumulate a unique combination of sphingolipids, phospholipids, glycosaminoglycans and autofluorescent lipofuscin within the endolysosomal system [291–294]. This storage profile probably reflects alterations in endocytosis and lipid trafficking [225,295].
It has been suggested that retarded recycling of lipids out of the endocytic system results from either enhanced late endosome–lysosome fusion or defective fission [225,295–297]. Since ionic changes (pH and Ca2+) are known to modulate fission and fusion, are either affected in MLIV? It is germane that TRPML1 is a lysosomal transmembrane protein that is an inwardly rectifying channel permeable to multiple ions including Ca2+, Fe2+, Na+, K+ and H+ [229,298].
Several groups have monitored lysosomal pH in MLIV cells, but there is no consensus as to whether there is a decrease, increase or no change when compared with normal cells [225,299,300]. However, lysosomal enzyme activity in situ is unaltered  suggesting no major alteration in pH. Further evidence against a change in pH in MLIV lysosomes is that treatment of MLIV cells with low concentrations of bafilomycin does not improve lysosomal storage in MLIV fibroblasts .
One possible explanation for these disparate pH measurements is that they are measuring pH in different organelles. In control cells a hybrid compartment  is transient and undetectable using these pH measurement techniques so the reading reflects true lysosomes. However, in MLIV cells the hybrid organelle persists and its higher pH skews the ‘lysosomal’ pH values [6,222]. As a consequence it is not possible to conclude whether pH alterations play a central role in MLIV. Because of the homology between TRPML1 and other TRP family members , an alternative hypothesis to altered pH is altered Ca2+ homoeostasis and enhanced fusion would be consistent with enhanced Ca2+ release from acidic stores [227,228,264,302,303].
As discussed above TRPML1 is unlikely to be an NAADP receptor. Indeed, NAADP-mediated Ca2+ release is not reduced in TRPML1-null MLIV patient fibroblasts, but is in fact elevated thereby potentially explaining the aberrant fusion observed in MLIV (K. Peterneva, K. Rietdorf, A. M. Lewis, A. Galione, G. C. Churchill, F. M. Platt and E. Lloyd-Evans, unpublished work; Figure 10). Indeed, in terms of both Ca2+ homoeostasis and defective endocytic pathway fusion, MLIV is the opposite of NPC. The exact function of TRPML1 remains unknown, but recent studies have highlighted some interesting modulators of its function, including PIP2 and the Ca2+-sensitive ALG-2 (apoptosis-linked gene-2) [233,304].
It may be highly relevant that a related mucolipin family member (TRPML3) was recently suggested to function as an early endocytic Ca2+-leak channel modulated by pH ; interestingly, TRPML3 impairment likewise resulted in enhanced luminal Ca2+ levels and enhanced endosomal fusion . Similarly, TRPML1 may also be a pH-modulated Ca2+ channel , so whether it functions in a similar manner to TRPML3, but at a later stage in the endocytic pathway, remains to be determined.
The property of high endolysosomal [H+] important for the well-established degradative function has been known for more than a century, but endolysosomal cation and anion channels/transporters have only emerged more recently as fundamental regulators of organelle physiology. There is now overwhelming evidence that the endolysosomal system also has an appreciable Ca2+ storage capacity, with a reported luminal Ca2+ concentration not dissimilar from that of the ER in the case of lysosomes. Consequently, altered ion transport defects are emerging as major contributors to certain pathologies within the lysosomal storage disease family.
In the present review we have focused on mechanisms of Ca2+ uptake and release from endolysosomal stores, and in particular the NAADP/TPC pathway as a mediator of a plethora of physiological signals including fertilization, secretion, membrane trafficking, contractility and differentiation. Moreover, the properties of TPCs may make them unique integrators of converging signals from NAADP, luminal Ca2+ and pHL. This convergence may then disseminate signals either locally (e.g. endolysosomal trafficking) or globally (e.g. plasma membrane excitability or regenerative Ca2+ signals by organellar cross-talk).
However, key questions still remain. Although NAADP is now established as a Ca2+-mobilizing messenger from the endolysosomal system, we still do not know if there is a broad sensitivity to this messenger or whether there is a subset of organelles that are specialized in mediating NAADP-evoked Ca2+ release. We know little about the molecular mechanism mediating Ca2+ uptake and storage, and the study of the different TPC isoforms is still in its infancy, but already hints at distinct roles. More fundamentally, it is still unclear whether NAADP binds directly to TPC proteins themselves or to accessory proteins within the complex and we still do not understand the mechanism of NAADP self-inactivation of the channel. Finally, aberrant NAADP/TPC signalling and endolysosomal Ca2+ homoeostasis may not only illuminate pathology, but also give clues as to which cellular processes are under the control of endolysosomal Ca2+ to offer unique drug targets.
We thank Professor Sergio Grinstein (University of Toronto, Toronto, Canada), Dr Jon Pittman (University of Manchester, Manchester, U.K.), and Professor George Ratcliffe and Dr Margarida Ruas (University of Oxford, Oxford, U.K.) for helpful comments. We also thank Wendy Tynan for preparing the micrographs.
↵2 Calculated using ECa=(RT/2F)·ln([Ca2+]o/[Ca2+]i), where R is the gas constant, T is temperature and F is the Faraday constant. The free luminal ER [Ca2+]o=30–600 μM and cytosolic [Ca2+]i=0.1 μM. At 20°C, RT/2F=12.63 mV.
↵3 By definition, the steady-state pHL reflects an exact balance between the rate of H+ uptake (pumps) and H+ loss (leaks) and therefore, the rate of H+ leak is unmasked when the V-ATPase is inhibited by bafilomycin A1 or concanamycin
↵4 Also referred to as K+-dependent Na+/Ca2+ exchange since the counter-flow of Na+ and K+ down their concentration gradients drives Ca2+ extrusion in a stoichiometry of 4 Na+:(1 Ca2++1 K+) compared with NCX stoichiometry of 3 Na+:1 Ca2+.
↵5 From the Nernst equation ΔGCa=2FΔψ + RTln(Cacyt/CaLum) where T=310 K (37°C), the cytosolic [Ca2+] (Cacyt)=0.1 μM, luminal [Ca2+] (CaLum)=600 μM (Supplementary Table S2). When Δψ=−0.03 V (cytosol-negative, lumen-positive), ΔG=−28 kJ/mol, when Δψ=−0.1 V, ΔG=−42 kJ/mol.
↵6 The protonmotive force=17 kJ/mol if ΔG=2.3RTΔpH – FΔψ, and assuming pHL=4.8, cytosolic pH=7.2 and Δψ=−0.03 V (cytosol-negative, lumen-positive) at 310 K.
↵7 Note that SERCA on neutral stores would not have to contend with a substantial Δψ and simply removing this component reduces the ΔGCa overhead to −44 kJ/mol for 2 Ca2+ which is amply covered by ATP hydrolysis.
Abbreviations: ACA, autoinhibited Ca2+-ATPase; AtTPC1, Arabidopsis thaliana TPC1; cADPR, cyclic adenosine diphosphate ribose; CaM, calmodulin; CAX, Ca2+/H+ exchanger; CICR, Ca2+-induced Ca2+ release; ClC, Cl− channel; ER, endoplasmic reticulum; FCCP, carbonyl cyanide p-trifluoromethoxyphenylhydrazone; HEK, human embryonic kidney; IP3, inositol 1,4,5-trisphosphate; IP3R, IP3 receptor; LAMP, lysosomal-associated membrane protein; MLIV, mucolipidosis type IV; NAADP, nicotinic acid adenine dinucleotide phosphate; NAADP-AM, NAADP-acetoxymethyl ester; NCKX, Na+/Ca2+-K+ exchanger; NCX, Na+/Ca2+ exchanger; NHE, Na+/H+ exchanger; NPC, Niemann–Pick type C; pHL, luminal pH; PKC, protein kinase C; PIP2, PI(3,5)P2; PMCA, plasma membrane Ca2+-ATPase; Po, mean open probability; RNAi, RNA interference; RyR, ryanodine receptor; SERCA, sarco-endoplasmic reticulum Ca2+-ATPase; shRNA, short hairpin RNA; siRNA, small interfering RNA; SPCA, secretory pathway Ca2+-ATPase; SR, sarcoplasmic reticulum; tBHQ, tert-butylhydroquinone; TMD, transmembrane domain; TPC, two-pore channel; HsTPC2, human TPC2; Trp, transient receptor potential; TRPML1, mucolipin 1; SpTPC, sea urchin TPC; V-ATPase, vacuolar-type H+-ATPase
- © The Authors Journal compilation © 2011 Biochemical Society