Inseparable tandem: evolution chooses ATP and Ca²⁺ to control life, death and cellular signalling

Abstract

From the very dawn of biological evolution, ATP was selected as a multipurpose energy-storing molecule. Metabolism of ATP required intracellular free Ca²⁺ to be set at exceedingly low concentrations, which in turn provided the background for the role of Ca²⁺ as a universal signalling molecule. The early-eukaryote life forms also evolved functional compartmentalization and vesicle trafficking, which used Ca²⁺ as a universal signalling ion; similarly, Ca²⁺ is needed for regulation of ciliary and flagellar beat, amoeboid movement, intracellular transport, as well as of numerous metabolic processes. Thus, during evolution, exploitation of atmospheric oxygen and increasingly efficient ATP production via oxidative phosphorylation by bacterial endosymbionts were a first step for the emergence of complex eukaryotic cells. Simultaneously, Ca²⁺ started to be exploited for short-range signalling, despite restrictions by the preset phosphate-based energy metabolism, when both phosphates and Ca²⁺ interfere with each other because of the low solubility of calcium phosphates. The need to keep cytosolic Ca²⁺ low forced cells to restrict Ca²⁺ signals in space and time and to develop energetically favourable Ca²⁺ signalling and Ca²⁺ microdomains. These steps in tandem dominated further evolution. The ATP molecule (often released by Ca²⁺-regulated exocytosis) rapidly grew to be the universal chemical messenger for intercellular communication; ATP effects are mediated by an extended family of purinoceptors often linked to Ca²⁺ signalling. Similar to atmospheric oxygen, Ca²⁺ must have been reverted from a deleterious agent to a most useful (intra- and extracellular) signalling molecule. Invention of intracellular trafficking further increased the role for Ca²⁺ homeostasis that became critical for regulation of cell survival and cell death. Several mutually interdependent effects of Ca²⁺ and ATP have been exploited in evolution, thus turning an originally unholy alliance into a fascinating success story.

This article is part of the themed issue ‘Evolution brings Ca²⁺ and ATP together to control life and death’.

1. Introduction: the early pairing of ATP and Ca²⁺

The course of biological evolution was preordained from the very beginning, when nucleosides become elementary units of genetic code and ATP emerged as a universal energy substrate. These events defined the molecular canvass of living cell and shaped major signalling cascades; however diverse are the living forms, the basic property of life lies in tight control over its internal milieu that has to be compatible with phosphate-based energetics. Of course, our contemplation of the habitat at the dawn of biological evolution is purely speculative, and yet the survival and diversification of all life forms including the most early ones (be these common ancestral progenotes, bacteria or else unknown forms) resulted directly from their interaction with the immediate environment. In this interaction, cells used what was readily available to them; and by whatever chance, purines and inorganic phosphates were readily available, and so ATP become central for energy metabolism.

The pyrophosphate bonds, phosphorylated nucleosides and ATP most likely occurred during the prebiotic period, being fundamental for subsequent organic evolution and RNA/DNA formation [1,2]. The exact origin of organic phosphates remains enigmatic–the most exotic under consideration invokes an extraterrestrial source: phosphates have been discovered in meteorites [3]. Of course, even accepting the extraterrestriality of ATP, we still face the question of its genesis, however distal from our planet. Conceptually, nucleotide triphosphates were available for biological evolution from the very early stage because of their requirement for genetic code, first in a RNA world [4] and then for DNA and for translationally active molecules (tRNA, mRNA, etc.). Historically, the pairing of ATP and Ca²⁺ could reflect an idiosyncrasy of the primordial alkaline ocean that contained only a minute concentration of Ca²⁺; therefore, low Ca²⁺ concentration was natural for the cytosol of early organisms. Rapid decrease in ocean alkalinity coincided with a similarly rapid rise in seawater Ca²⁺ concentration ([Ca²⁺]), which arguably could represent the environmental switch that triggered the emergence of complex ion homeostasis, evolutionary diversity and even multicellularity [5]. It should be noted that this hypothesis is by no means accepted by all palaeobiologists, and many believe in a slightly acid primeval ocean containing high Ca²⁺ [6], so the truth is yet to be established.

The natural question is: why ATP, rather than other nucleotide triphosphates? Possible driving factors in eukaryote evolution may reflect the fact that ATP had been preset for energy conservation already in bacteria before Ca²⁺ has been established as a signalling molecule; that Ca²⁺ stimulates ATP synthesis by activating dehydrogenases of the tricarbonic acid (Krebs) cycle [7]; and that ATP prevails several fold over GTP, which is formed and used in only one step by transphosphorylation from ATP to GDP [8]. Acquisition of mitochondrial precursors is considered an early and critical step of eukaryote evolution [9,10]. Formation of an endomembrane system is attributed to the first eukaryotic common ancestor, which emerged in combination with formation of multiple paralogues of molecules dedicated to signalling and trafficking [11–13]. Production of ATP by mitochondria may have been a prerequisite because it greatly increased energy available for vesicle trafficking, be it for fuelling motor proteins for vesicle transport, for activation of the triple-A ATPase N-ethylmaleimide sensitive factor [14], a SNARE chaperone, or for increasingly sophisticated Ca²⁺ regulation. It is of note, however, that binding of ATP and GTP to F1-ATPase is similar [15]. All in all, there are only weak arguments for the selection of ATP, rather than other nucleotide triphosphates, for energy metabolism [16]. In contrast, the high free energy of GTP hydrolysis is proposed to be advantageous in allowing efficient transition from the active to the inactive state, as it occurs with GTPases [17] when they mediate specific vesicle interactions [18,19].

Be this all as it may, ATP remains quite a unique molecule, which arguably participates in more chemical reactions than any other compound on the Earth's surface, except water. At the same time, ATP metabolism is highly demanding, being operational only at very low levels of free Ca²⁺. This in turn stipulates that free Ca²⁺ inside the cells using ATP for energetics must be exceedingly low; and very low it is set indeed, being, as a rule, around 100 nM [20–22]. This low Ca²⁺ requirement shaped the biological evolution and linked ATP with ion composition of the intracellular milieu, which has minute amounts of Ca²⁺ and relatively low concentration of Na⁺; maintaining the ion composition of the cytosol, in turn, became the main energy expenditure of all living cells. Thus, ATP needs control over Ca²⁺ and control over Ca²⁺ needs ATP, hence making the two an inseparable tandem.

Evolution perfected ion-controlling mechanisms and used ion gradients for many purposes, most notably for cellular signalling. Calcium as a signalling ion is employed by the most primitive bacteria [23,24] and by protozoa [25,26] before the system of Ca²⁺ signalling became highly elaborated in higher eukaryotes. Some bacteria already contain a Ca²⁺–ATPase/pump that is important for Ca²⁺ export and bacterial [27–29] and protozoan [30,31] survival. Most importantly, all Ca²⁺ signalling and homeostatic molecules are controlled by Ca²⁺ ions themselves, which imposes multiple feedbacks thus making the system highly versatile and adaptable [32]. Eukaryotic evolution proceeded along with a significant increase in the number of Ca²⁺-binding proteins [33,34] as well as in the number of signalling and metabolic processes that are regulated by Ca²⁺ [20,35]. Exocytotic SNARE proteins, Ca²⁺ signalling and directionality of organelle-specific GTPases must have been early evolutionary achievements maintained throughout eukaryotic kingdoms [11,36,37]. Figure 1 gives an overview of the widely branched regulation mechanisms and effects of Ca²⁺ that are exerted in cooperation with ATP as an energy source, from normal cell life to programmed cell death (apoptosis).

Figure 1.

Regulatory principles in Ca²⁺-based signalling and ATP-based energetics. As [Ca²⁺]_o is much higher than [Ca²⁺]_i, Ca²⁺ steadily leaks into the cell (not drawn), but this is counterbalanced by the Ca²⁺–ATPase/pump and by antiporters. Upon stimulation, Ca²⁺ can flow into the cell by influx channels of different types. Rapid downregulation is executed by Ca²⁺-binding proteins (CaBP) and by sequestration mainly into the endoplasmic reticulum (endowed with an organellar Ca²⁺-pump), but also into other organelles, such as cortical stores, the Golgi apparatus, different trafficking organelles and mitochondria. In their inner membrane, mitochondria have available a uniporter for Ca²⁺ uptake and a antiporter system for its rapid release. Ca²⁺-release channels, mainly of the type InsP₃R and RyR, are available in the different organelles. In the endophago/lysosomal trafficking system, such channels can mediate short-range Ca²⁺ signalling. Requirement of Ca²⁺ is also ascertained for stimulated exocytosis, whereby Ca²⁺ may originate from influx and/or release from cortical stores. Regulation of apoptosis (regulated cell death), among a plethora of molecules, involves release of Ca²⁺ from the endoplasmic reticulum and release of an ‘apoptosis-inducing factor’ as well as of cytochrome c from mitochondria. In summary, in the eukaryotic cell, there are many local signalling pathways, with many interactions, for a variety of subcellular processes.

2. Some evolutionary preconditions and basic considerations of having Ca²⁺ and ATP as crucial cell regulators

Signalling mediated by Ca²⁺ has developed early in evolution because of sheer omnipresence and abundance of this ion in the environment, whereas the ATP-centric metabolism made free Ca²⁺ toxic inside cells. Nonetheless, Ca²⁺ permeation through biomembranes, though variable and tightly controlled, is unavoidable. Like with oxygen which, when emerging in the atmosphere instigated the extinction of almost all living forms, evolution turned a disadvantage into an advantage by taming toxicity through spatial and temporal restriction [25,38,39]. Ionized Ca²⁺ has to be kept low in the cytosol to avoid interference with phosphate groups in free and bound form. On the other side, low Ca²⁺ concentration offers the advantage of economic handling. Assuming that Ca²⁺ was required to govern emerging intracellular trafficking, which may have developed only after full development of aerobic energy production with the consequence of high energy conservation by ATP, these two aspects represented a revolution in eukaryote evolution. Furthermore, considering the requirement of ATP for intracellular movement and of Ca²⁺ for membrane–vesicle interactions, the availability of both molecules in adequate concentrations was mutually interdependent from the very beginning of eukaryote evolution about 2.5 × 10⁹ years ago. As a rule, cytosolic concentrations are kept at millimolar range for ATP and at submicromolar range for Ca²⁺, the [Ca²⁺]_i being about 1000 times less than ambient concentration in the extracellular space [20]. From protists to humans, concentrations of total (free and bound) and free Ca²⁺ in the cytosol are very similar, being approximately 1 mM and approximately 0.1 µM, respectively. After stimulation, [Ca²⁺]_i usually rises by up to, or even above, 10 µM. In Ca²⁺-storing organelles, before release upon stimulation, both total [Ca] and ionized [Ca²⁺] are substantially higher compared with the cytosol, from protozoa [40] to mammals [41,42]. Sequestration into stores for trapping by binding to luminal proteins inactivates Ca²⁺, but also makes it available for rapid release upon stimulation.

3. Availability of Ca²⁺ for early evolution, and its properties predestining it for intracellular functions

The prevalence of divalent cations in the earth crust and in the ocean is: Ca²⁺ > Mg²⁺ > Ba²⁺ > Sr²⁺. While Ca²⁺ is mainly used for signalling, Mg²⁺ binds to many proteins and serves as a cofactor in biochemical reactions. For instance, together with GTP, Mg²⁺ activates the protein initiation factor or, as MgATP, frequently serves as a substrate. The preference of MgATP may reside in the highest binding constant for Mg²⁺ over other divalent cations [43,44]. In contrast to Mg²⁺, Ba²⁺ and Sr²⁺ drive several subcellular functions, similarly (albeit not always) to Ca²⁺ (e.g. exo–endocytosis in fish neurons [45] or ciliary movements in ciliates [46]), however their effects are, as a rule, weaker. Sometimes, however, Ba²⁺ can surpass the effects of Ca²⁺, for instance in chromaffin cells and in ciliated protozoa (the ‘barium dance’ [47]). It is plausible that Ba²⁺ and Sr²⁺ may have been overlooked by evolution because of their minor availability in nature. An alternative view considers different coordinated binding of Mg²⁺, Ca²⁺ and Ba²⁺ to peptide fragments, depending on polar aliphatic residues and ion radius [48], thus reflecting some gross dependency from the primary protein structure. Binding is much more restricted to Ca²⁺ when proteins possess specific Ca²⁺-binding motifs. Here specificity can be quite selective. For example, different paralogues and isoforms of the Ca²⁺-sensor synaptotagmin bind Ca²⁺ considerably better than Ba²⁺, Mg²⁺ or Sr²⁺ [49]. By these different aspects, during evolution, Ca²⁺ may have been selected as the main signalling molecule, while the other abundant divalent metal, Mg²⁺, was chosen for other functions.

Binding to proteins is widely different for Ca²⁺ and Mg²⁺: relatively loose and reversible coordinative binding of Ca²⁺ to specific domains of dedicated Ca²⁺-binding proteins at low concentrations is an important factor. This enables low-capacity/high-affinity Ca²⁺-binding proteins for rapid and efficient signal transfer by conformational change. The archetypal and most widely distributed, multifunctional example is calmodulin [50], whose four EF-hand loops are hierarchically occupied, the lowest-affinity loop IV also binding Mg²⁺. Gradual conformational changes can thus serve for the transmission of a chemical signal into a mechanical one, be it at a subunit of plasmalemmal ion channels in protozoa [51] or intracellular Ca²⁺-release channels in mammals [52]. The Ca²⁺-sensor for membrane fusion, synaptotagmin, possesses distinct C2 domains for cooperative Ca²⁺-binding [53]. These C2 domains stick out from β-barrel structures and increase lipophilic properties upon Ca²⁺ binding. This mediates membrane binding and perturbation of adjacent membranes for Ca²⁺-dependent fusion [54].

An unusually high number of acidic amino acid residues in cytosolic Ca²⁺-binding proteins allows Ca²⁺ binding even at relatively high concentration. These high-capacity/low-affinity Ca²⁺-binding proteins, such as parvalbumin, serve as immobile Ca²⁺ buffers for rapid downregulation of cytosolic Ca²⁺ [55,56]. Centrin can operate not only by high-affinity binding sites belonging to the EF-hand motif, but also by its negative charges [57]. Centrin is the dominant cytosolic Ca²⁺-buffer in the cortex of ciliated protozoa [58]. In Ca²⁺ stores, high-capacity/low-affinity Ca²⁺-binding proteins are defined by their excess of acidic amino acids residues for Ca²⁺ storage and rapid release upon stimulation [59]. Calreticulin and calnexin in the endoplasmic reticulum are examples; they also exert a Ca²⁺-dependent chaperone function for folding newly synthesized proteins [60].

4. The specific case of magnesium

Magnesium is the fourth most abundant ion in eukaryotic cells, and, incidentally, it is the most abundant free divalent cation (total Mg²⁺ content of the body is approx. 24 g or approx. 1100 mmol, whereas ionized cytosolic [Mg²⁺] fluctuates between 0.25 and 1 mM [61,62]). This high cytosolic concentration could be another explanation for why evolution favoured Ca²⁺ over Mg²⁺ for signalling purposes. Well-defined domain structures evolved for Ca²⁺ binding, whereas Mg²⁺ buffers remain generally unknown; identifying Mg²⁺-binding domains is a tedious task and they generally are considered elusive. A similar conclusion arose from the bioinformatic analysis of Mg²⁺-binding proteins, the so-called magnesome, which includes rather diffuse Mg²⁺-binding motifs [63]. This reflects essential differences between the Mg²⁺ and the Ca²⁺ ion (and some other earth alkali metals, with the exception of Ba²⁺ and Sr²⁺ that closely resemble Ca²⁺). These differences [62,64] are in ionic radii (smaller for Mg²⁺), charge density, water shell binding (higher hydration for Mg²⁺), binding by inner and outer sphere coordination, etc. Coordination geometry is particularly important for binding to DNA, RNA and to various proteins and their substrates [65]. Several hundred enzymes are known to interact with Mg²⁺, mainly through using soluble MgATP, rather than ATP per se as a substrate. A large part of the cellular effects of Mg²⁺ are because of its abundance in ionized form, absence of toxicity (as opposed to Ca²⁺), less strict and less specific binding.

Data mining and molecular dynamics analyses have been applied to identify possible coordination structures based on crystal data for actin, myosin, DNA polymerase, RNA polymerase, DNA helicase and mitochondrial F₁-ATPase [66]. These analyses identified a broad range of molecules being affected by Mg²⁺ and revealed a diversity of both transitory and stable coordination arrangements between Mg²⁺ and ATP. In eubacteria, Mg²⁺ stabilizes the structure of 16S rRNA in a conformation favourable for binding to the ribosomal 30S subunit [67]. Similarly, the binding of transcriptional activator protein c of bacteriophage Mu to bacterial DNA is enhanced by Mg²⁺-induced change from a more β-sheath-like to a more α-helical conformation, as evidenced by circular dichroism analysis [68]. All these processes are not related to cell signalling.

However, indirectly, Mg²⁺ may contribute to intercellular signalling, because Mg²⁺ (and not Ca²⁺) controls interaction between the subunits of trimeric GTP-binding proteins (G-proteins) with GTP and GDP, respectively, thus determining their signal transduction activity [69]. As alluded to above, in the calmodulin molecule, which acts as an activator of many enzymes and ion channels, four Ca²⁺-binding EF-hand loops show hierarchical Ca²⁺ binding with different binding constants, which depend on [Mg²⁺]_i [70] and is different for each loop [71]. In addition, Ca²⁺ and Mg²⁺ often exert different, in part even antagonistic, effects on enzymes, such as protein kinases and protein phosphatases. For example, Ca²⁺ and Mg²⁺ differentially regulate cAMP-dependent protein kinase A [72]. Another example is associated with protein phosphatase type 2C, which is activated by physiological [Mg²⁺]_i, in contrast to Ca²⁺, which is inhibitory at micromolar concentrations [73].

5. Why ATP has been selected as an energy-storing nucleotide?

Interactions of Ca²⁺ and Mg²⁺ with ATP are different, with CaATP being much less soluble. As mentioned, Mg²⁺ binds to ATP to form a complex that serves as a substrate for many enzymes, such as ATPases including Ca²⁺–ATPases/pumps for extrusion or sequestration of Ca²⁺. Why among nucleotide triphosphates did ATP prevail over GTP, CTP, TTP, UTP, ITP? There may be evolutionary (‘historical’) reasons, e.g. because ATP had already been established in energy conservation. If so—why? Any other nucleotide triphosphate, if available in similar concentration, might do equally well because of similar free energy [74]. However, functions of the different nucleotide triphosphates may have been diversified already at the early evolutionary stages. As we see now, ATP has been selected preferably for energy conservation, whereas GTP is mainly used for signalling, e.g. for membrane interactions. This mainly concerns monomeric GTP-binding proteins (Rab-type GTPases) in the context of vesicle/membrane interaction [18,19,75], in addition to nucleo-cytoplasmic transport and microtubule dynamics. Only rarely is GTP used in bioenergetics, e.g. in only one step of the tricarbonic acid cycle. The reduction of nicotinamide dinucleotide (NAD) to NADH for subsequent exploitation by oxidative phosphorylation in the mitochondria greatly increases energy gain. Other nucleotide triphosphates are used for specific biosynthetic processes, such as N-glycosylation of proteins in the endoplasmic reticulum. Thus, ATP prevalence in energy metabolism may result from rather stochastic early evolutionary specialization, although the early pairing of ATP and Ca²⁺ could also play a role. Using different nucleotide triphosphates beyond ATP and GTP, including UTP and CTP, for different purposes may be advantageous for specific regulation of a range of processes.

In summary, it may be due to the evolutionary priority to have energetics and metabolism based on ATP, rather than on other nucleotide triphosphates. During evolution, ATP and Ca²⁺ became intimately connected in functional terms for many activation and deactivation processes (protein phosphorylation and dephosphorylation), whereas GTP has been assigned other functions.

6. Intracellular Ca²⁺ fluxes and Ca²⁺ regulation

Cellular Ca²⁺ fluxes are mediated by uniporters (influx- and release channels), primary active (Ca²⁺–ATPase/pump) and secondary active transport systems (e.g. H⁺/Ca²⁺ exchangers or related antiporters), all to be found in the plasmalemma and in the endomembranes [20,22,76,77]; intracellular Ca²⁺ dynamics is further regulated by Ca²⁺ buffers [56]. All these mechanisms were invented early in evolution, be it in ancestors of unikonts, such as choanoflagellates [78], or be it in bikonts, such as ciliates [38,40]. Cytosol contains both mobile and immobile Ca²⁺ buffers of different capacity and affinity. Mobile high-affinity Ca²⁺ buffers hinder Ca²⁺ diffusion, thus favouring localization of Ca²⁺ signalling events that often occur in a form of microdomains. In the endoplasmic reticulum, in contrast, low-affinity Ca²⁺ buffers allow diffusion and rapid equilibration of Ca²⁺ within the lumen through Ca²⁺ tunnels [79]. From immobile buffers, including high-capacity/low-affinity binding proteins, Ca²⁺ can slowly dissociate to be fed into active transport systems. For primary active transport, the ATP → ADP system is energetically roughly the same as it would be for pyrophosphatase activity The preference for the ATP → ADP system can be explained by the requirement of only one energy limiting step for ATP re-synthesis, rather than two for the pyrophosphatase system, whereas the same amount of free energy is available for both [80]. Thus, pyrophosphatase-driven Ca²⁺-sequestration is used only exceptionally, for example in acidocalcisomes of trypanosome parasites [81].

Furthermore, to keep energy expenditure low, Ca²⁺ influx and/or release from internal stores is frequently restricted to selective sites. The textbook example is the nerve terminal, where voltage-dependent Ca²⁺-influx channels are restricted to the active zone where transmitter is released by exocytosis [82,83]. In ciliated protozoa, this type of channels is restricted to cilia [84]. In the central nervous system of mammals and in ciliates, these Ca²⁺ channels are inactivated by formation of a Ca²⁺/calmodulin complex [46,85,86] using the very same Ca²⁺ they had conducted. This is another example of a principle conserved from ciliates to the human brain: it is generally appropriate to limit [Ca²⁺]_i increase to a small volume and short time, with values just supporting activation and, thus, greatly reducing ATP consumption for re-establishing [Ca²⁺]_i.

The principle of local restriction is also often applied to intracellular signalling. For intracellular Ca²⁺ release, the cell possesses different types of Ca²⁺-release channels activated by distinct compounds represented by (i) inositol 1,4,5-trisphosphate (InsP₃) formed from phosphatidyl inositol 4,5-bisphosphate (PInsP₂, a component of lipid membranes) and binding to InsP₃-receptor-type Ca²⁺-release channels (InsP₃R, [87]); (ii) by Ca²⁺ ions themselves, which activate ryanodine-receptors (RyR) evolutionarily related to the InsP₃-receptor at the protozoan level where intermediates of the two channel types are also found [26,38,39,88,89]; and (iii) by nicotinic acid adenosine dinucleotide phosphate (NAADP), which activates two-pore channels (TPCs) of acidic Ca²⁺ stores [90]. In metazoans, RyRs are activated by Ca²⁺ [91] and by cyclic adenosine diphosphoribose (cADPR) formed from the glycolytic H⁺ acceptor NAD [92]. Both NAD and nicotinamide adenine dinucleotide phosphate (NADP), but probably also PInsP₂, have been available in early eukaryote evolution—at least the key components [93] and enzymes [94] are found in protozoa. Both channel types, InsP₃R and RyR, are also predicted by database mining for choanoflagellates [78,95]. An intermediate form between InsP₃R and RyR was suggested from the nucleotide sequence in Purkinje cells [96]. Similar to Paramecium, the two types or Ca²⁺-release channels coexist in Ascidian egg cells [97]. In conclusion, there are diverse ways to release Ca²⁺ locally by different activators, only some being considered here. Considerable similarities between InsP₃Rs and RyRs from ciliated protozoa [88] to mammals [98] indicate that both may possibly have evolved from a common ancestral molecule, assembled from six transmembrane domains [38,89].

Another set of intracellular channels is represented by the TPCs that are members of the voltage-gated ion channel superfamily; the TPCs contain 12 transmembrane domains, being in essence a half of a voltage-gated channel [99–101]. These channels reside in acidic compartments (which form acidic Ca²⁺ stores), notably in the endo/lysosomal compartments and in the plant vacuoles [90,102], and are activated by NAADP, derived from the metabolite, NADP [90,99,103,104].

In evolution, TPCs are present in monokonts, from choanoflagellates and the social amoeba Dictyostelium [105]) to mammals [100], and in bikonts, from protozoa to green plants [101,106]. In plants, TPCs can be activated by osmotic stress [107] as well as by Ca²⁺ ions (this Ca²⁺-dependent activation contributes to regulation of seed germination and stomatal movement [108]). Paramecium, which is another bikont, also seems to possesses TPCs [109]. In the parasitic monokont, Trypanosoma, NAADP does not trigger Ca²⁺ response in contrast to the bikont apicomplexan parasite Toxoplasma gondii, a relative of ciliates [110]. The T. gondii also possesses a plant-like vacuole and acidocalcisomes, which may bear TPC channels encoded by orthologous genes that are absent in Apicomplexa [106].

Different types of Ca²⁺-release channels frequently coexist within the membrane of Ca²⁺ storage organelles [111,112]. The RyRs and InsP₃Rs localized in the ER membrane may be functionally linked to TPCs in neighbouring acidic organelles [90,113–116]; Ca²⁺ release from TPCs can trigger Ca²⁺-induced Ca²⁺ release (CICR) [115]. From the fact that InsP₃Rs and RyRs can be activated by TPC activity, it has been speculatively concluded that TPCs are evolutionarily older than other Ca²⁺-release channels [90]. According to this hypothesis, channels that allegedly emerged later in evolution (i.e. InsP₃Rs and RyRs) may serve for signal amplification through consecutive channel activation [90,107,113–115]. In addition, NAADP can also activate Ca²⁺-influx through TPCs localized in the plasmalemma [115], which was confirmed in patch-clamp studies [90]. The TPCs were also proposed to be responsible for topologically restricted Ca²⁺ signals, which may locally ignite phosphoinositol turnover, store-operated Ca²⁺ entry (SOCE) activity and possibly also cAMP signalling [117]. From a more conventional perspective, Ca²⁺ signals arising from TPCs have been considered relevant for membrane fusions along the endolysosomal pathway [118]. For the time being, however, all these effects remain hypothetical. In summary, TPCs are universally distributed in eukaryotes, with the exception of some unicellular parasites. Otherwise, however, our knowledge about TPCs remains rather incomplete.

Transient receptor potential (TRP) channels owe their name to a phenotype of a Drosophila mutant whose electrical photoreceptor response is abnormally short (i.e. transient). Mammals possess 28 TRP channels classified into six subfamilies: TRPC (canonical), TRPV (vanilloid), TRPM (melastatin), TRPA (ankyrin), TRPML (mucolipin) and TRPP (polycystin) [119,120]. The TRP channels are widely distributed between different cell types and may be present in plasmalemma as well as in endomembranes (with only nucleus and mitochondrion being devoid of them). The TRP channels are strictly cationic with highly variable Ca²⁺ permeability; some TRP channels can conduct Mg²⁺ [119,120]. The intracellular TRP channels may mediate Ca²⁺ release from the stores; plasmalemmal TRP channels can be linked to intracellular Ca²⁺ release through a store-operated mechanism [120]. The TRP channels are broadly distributed throughout many species along the phylogenetic ladder. In protozoa, systematic analysis is scant and assignment to eukaryotic, specifically mammalian types is difficult. For the bikont Paramecium, originally no TRP channel genes had been found [106], but then three genes were retrieved [121]. Parasitic relatives of Paramecium, Apicomplexa, Toxoplasma gondii and probably also Plasmodium spp. all possess genes encoding TRP channels [106]. Parasitic flagellates Leishmania and Trypanosoma also possess TRP channel, but for all unicellular parasites, molecular characterization is yet to be done [110]. The green flagellate Chlamydomonas reinhardtii contains 19 genes encoding TRP channels [122], one of which shows considerable similarity to a human form and is localized probably in the flagellar membrane, contributing to sensory transduction [121]. Data mining revealed TRP channels encoding genes in Dictyostelium [121], where one of the members, PKD2, is localized in the cell surface and governs rheotaxis [123]. Choanoflagellates are endowed with TRP channels genes [124]; expanded molecular data mining revealed TRPA, TRPC, TRPM, TRPML and TRPV types [78]. Higher plants (viriplantae) apparently do not express TRP channels [122,125].

There is a crosstalk between intracellular Ca²⁺ stores, exemplified, for example, by variable coupling of endoplasmic reticulum and mitochondria [126,127], as well as between stores and plasmalemmal Ca²⁺ influx. In cardiomyocytes, Ca²⁺ entering the cytosol through voltage-gated channels directly activates RyRs, thus initiating CICR [128]. In skeletal muscle, activation of L-type plasmalemmal Ca²⁺ channels directly activates RyR type II, causing depolarization-induced Ca²⁺ release. Emptying the ER Ca²⁺ stores triggers plasmalemmal store-operated Ca²⁺-entry, SOCE [128], which ubiquitously occurs in non-excitable cells [129,130]. Activation of SOCE involves intraluminal Ca²⁺-sensor (Stim) and an influx channel (Orai and associated proteins or TRPC channels) in the plasmalemma [131,132]. Although operational SOCE is well characterized in the ciliate, Paramecium [40], Orai and Stim have not been detected (as yet), in contrast to informatics data obtained from choanoflagellates [78] that are placed at the basis of metazoans. However, several alternative modes of cell membrane–cortical store coupling are discussed, for instance by extended synaptotagmins with super-numerary C2 domains [133] which also occur in ciliates [38]. Altogether, SOCE appears to be an evolutionarily old mechanism.

The intracellular Ca²⁺ dynamics in the nerve terminal has to be exceptionally fast, in contrast to stimulus–secretion coupling in most other cells where no specific intimate structural and functional coupling exists and where the Ca²⁺ response, therefore, is less precise (an obvious example being the difference in vesicular secretion between neurons and neuroglia [134]). An exception is the ciliate, Paramecium, where a rapidly activated SOCE underlies the fastest known exocytosis of dense core-secretory vesicles important for predator defence [40,135]. The final target of Ca²⁺ during any type of secretion is the C2-domain-bearing protein synaptotagmin, and closely related Ca²⁺-sensors, such as extended synaptotagmin isoforms of which several (probably with different kinetics) are known [53].

Elaborated Ca²⁺ signalling is also required for intracellular trafficking from organelle to organelle, which can be concluded from the distribution of synaptotagmins [136,137]. The signal may come from the very same trafficking organelles, as they contain Ca²⁺-release channels of different types, including InsP₃Rs and RyRs. This is true from protozoa [38,88] to mammals [98], even though other types of Ca²⁺-release channels can also be involved (not considered here). Organelles of the endophagocytotic/lysosomal pathway contain Ca²⁺ at different concentrations [138]; release of Ca²⁺ drives local vesicle–vesicle interactions and fusion. Requirement for local availability of Ca²⁺ has been shown for instance for endosomes [139], endoplasmic reticulum and Golgi vesicles [140,141], as well as for lysosome–phagosome fusion [142]. Besides influx channels, endocrine cells also possess InsP₃Rs in their secretory vesicles [143]. Again, limiting the signal to the organelles involved is a prerequisite for avoiding toxicity and to keep energy cost low.

Which Ca²⁺-release channel is the oldest? Different types of Ca²⁺-release channels, i.e. InsP₃R, RyR, TPCs and TRPCs are all present at the lowest evolutionary stage when monokonts separated from bikonts, which happened approximately 1.6 × 10⁹ years ago [144]. This makes it difficult to decide upon the nature of ancestral Ca²⁺-release channel, although low selectivity of TRPCs [120] may argue in their favour. The mechanosensor TRPY1 of yeast may be such a prototype [145] (of note however, yeasts emerged by secondary reduction). Channels with high selectivity for Ca²⁺, such as InsP₃Rs and RyRs, require Ca²⁺ stores endowed with uptake mechanisms for specific sequestration of Ca²⁺; therefore, high selectivity of Ca²⁺-release channels reflects availability of Ca²⁺ in the store lumen (for TPCs, see above). Whichever is the oldest, Ca²⁺-release channels appear as inevitable prerequisite for ‘constructing’ large eukaryotic cells with abundant vesicle trafficking.

7. Molecular targets of Ca²⁺ in cells

Salient features of targets for Ca²⁺ signals are disparate in different cell types. For membrane–vesicle interaction, various SNARE proteins and synaptotagmin-type Ca²⁺-sensors are present in spatially separated cellular compartments [146,147], in contrast to non-stimulated (constitutive) exocytosis that may not depend on Ca²⁺ signals, if one makes painstaking scrutiny [148]. Endocytosis via clathrin-coated pits or uncoated vesicles is synchronized with exocytosis by the pulse of Ca²⁺ required for both processes in mammalian [149,150] as well as in protozoan cells [151]. Endocytosis requires dephosphorylation of the large GTPase dynamin, by the Ca²⁺/calmodulin-activated phosphatase 2B/calcineurin [152]—an enzyme maintained from protozoa [153] to human, where it is essential for many signalling processes, including long-term potentiation and immune defence [154–156]. Calcineurin is a true multipurpose enzyme [157]. Nevertheless, in evolution, only one of the two calcineurin subunits is maintained in plants, where it serves for coping with ionic stress [125]. Unfortunately, no data are available for characaean algae placed at the roots of higher plant evolution. The situation with trafficking vesicles deep inside a cell is different because it requires locally restricted and finely tuned Ca²⁺ signals. The availability of synaptotagmin at the surface of vesicles of the endophagolysosomal vesicles [136,137] and of Ca²⁺ in their lumen allows for local signalling and fusion. Conceptually, Ca²⁺ ions have to be provided ‘at the place’, because diffusion from the far remote influx channels at the cell surface would be too inefficient, as Ca²⁺ would be trapped on its way through the cell by Ca²⁺-binding proteins.

The Ca²⁺/CaM-activated protein kinase (‘CaM kinase’) is important for neurotransmission, because it regulates, by phosphorylation of synapsin, the release of vesicles from the actin web, thus facilitating their access to the cell membrane [158]. During evolution, CaM kinase has been maintained only in monokonts from myxamoebae Dictyostelium to man [159]. This is different for bikonts, from ciliates up to viridiplantae, where CaM kinase proper does not exist. Here, Ca²⁺-dependent protein kinases are kinases with integrated CaM-like motifs [125,160]. The situation is again different for cyclases for cyclic nucleotide formation. These are Ca²⁺-dependent enzymes from ciliates [161,162] to humans [163]. Thus, via cyclases, cyclic nucleotide-activated protein kinases at the end of a signalling cascade also depend (indirectly) on Ca²⁺.

The multiple ways in which Ca²⁺-dependent cell motility is regulated are well characterized. The Ca²⁺-sensitive molecules involve severing proteins (gelsolin), actin bundling proteins (caldesmon), troponin, etc. In muscle cells, Ca²⁺ binds to troponin C; the resulting conformational change of the troponin–tropomyosin complex allows the interaction of the myosin head with the actin filaments, which causes cell contraction by tilting of the myosin [164]. During stimulus–contraction coupling, this movement from molecular to macroscopic scale is fuelled by ATP hydrolysis. Ca²⁺ exerts pleiotropic effects on ciliary activity, especially in ciliated protozoa. The availability of a minimum of cytosolic ionized Ca²⁺ is required for ciliary/flagellar activity [165], which can be further modulated by Ca²⁺ [166]. In ciliates, normal ciliary beat depends on Ca²⁺ [165], as does any change in beat activity. For instance, ciliary beat is reversed when intraciliary [Ca²⁺] increases by activation of voltage-gated Ca²⁺ channels, as described for ciliated protozoa [166], and in ctenophores (comb jellies). In both systems, these channels are restricted to the cilia [84,167]. In ciliates, Ca²⁺ mediates dynein subunit phosphorylation by Ca²⁺ followed by calmodulin [168]; this type of regulation appears to be absent in cilia at a higher evolutionary level. In ciliates, the Ca²⁺-regulatory mechanism involves activation of a cGMP-activated protein kinase [169] formed by a Ca²⁺-dependent cyclase [170]. It also should be mentioned that many basic metabolic processes, such as glycogen turnover, are regulated by Ca²⁺. In metazoans, the same is true of intercellular communication through gap junctions (which close when [Ca²⁺] rises in the cytosol [171]). This may contribute to epigenetic effects achieved during evolution by non-structural RNA species, which can be transferred from one cell to another, thus harmonizing tissues.

Intimate links between ATP and Ca²⁺ are observed in mitochondria [172,173]. Mitochondrial Ca²⁺ is connected with ATP production through stimulation of some mitochondrial dehydrogenases [7], of oxidative phosphorylation [174] and of ATP synthase [175]. This requires rapid crosstalk between the cytosol and the organelle. The key player is the mitochondrial calcium uniporter (MCU) for rapid Ca²⁺ influx [176,177]; the MCU and its regulators are present at the protozoan level. An MCU-encoding gene is found in Trypanosoma [31]. Paramecium also possesses MCU and associated MIUC protein, as well as the MCU regulator, EMRE [176,178]. Accordingly, in Paramecium, during synchronous exocytosis, Ca²⁺ rapidly (on a subsecond time scale) flushes into mitochondria; most of the Ca²⁺, however, is released back into the cytosol within seconds, with only a minor fraction remaining in the organelle [109,179]. Mitochondrial exchangers, H⁺/Ca²⁺ and Na⁺/Ca²⁺ (NCLX) are known in higher eukaryotes [172,173], but they have not been identified in protozoa.

8. Ca²⁺ as a regulator of programmed cell death

In mammalian cells, apoptosis is triggered by release of Ca²⁺ from the endoplasmic reticulum, influx of Ca²⁺ into mitochondria [180] and release of cytochrome c, parallelled by a rise in [Ca²⁺]_i, and further Ca²⁺ entry into mitochondria. A protease (caspase) cascade is activated and DNA is cut to nucleosome-size fragments (or multiples, called DNA ladder). Breakdown of subcellular structures, autophagy and phagocytosis of cell fragments [181] are additional events. Altogether, there are multiple effects of Ca²⁺ during apoptosis in mammalian cells [182], and mitochondria, before degeneration, play a key role in this process [183]. Toxic levels of [Ca²⁺] and failure of energy supply in conjunction with cytochrome c release [183], functional and structural disturbance of mitochondria and phagocytosis are final events.

Some molecules pertinent to apoptosis can be encountered already in unikont protists, such as choanoflagellates [184]. In bikont protists, such as ciliates, the macronucleus (equivalent to the soma, as opposed to the generative micronucleus), rather than the whole cell, is degraded in a process called ‘nuclear death’ [185]. This includes DNA fragmentation and autophagy of mitochondria, although the role for ATP and Ca²⁺ remain unsettled. The apoptotic stratagem is only slightly more complex in lower metazoans [186], including secondary reduction of the process in the nematode C. elegans [184]. Altogether, many important details of apoptosis at the roots of unikonts and bikonts are not yet well understood, as summarized recently [187]. It appears that these crucial aspects of apoptosis, including mitochondrial function and involvement of Ca²⁺, have only successively evolved in the course of metazoan evolution.

9. Intercellular signalling: purinergic transmission links to Ca²⁺ signals

The concentration gradient for ATP (aimed extracellularly) is likely the highest existing in living cells. Indeed, intracellular concentration of ATP is set at approximately 1–5–10 mM, whereas the extracellular concentration does not exceed 1–10 nM creating thus a difference of approximately 1 million times. An extremely low concentration of extracellular ATP results from constant operation of ectonucleotidases [188], which degrade ATP to ADP, AMP and adenosine. This ATP degrading system has ancient evolutionary roots, being present in bacteria, in early eukaryotes and in protists [188–190]. The steep concentration gradient favours ATP exit upon opening of transmembrane pores and following plasmalemmal damage; it should not be surprising therefore, that evolution rapidly began to use ATP as an intercellular signalling molecule. Already some prokaryotes are known to release ATP, which could be used for intercellular communications [191], whereas the plasmalemmal channels-mediated ATP release is described in polyphenic fungi, type Candida albicans [192].

Biological effects of extracellular ATP are many, and they are widespread throughout virtually all life forms [190]. In prokaryotes, exposure to ATP affects bacterial growth and development, sporulation and germination, changes ion fluxes and alters gene expression [193–196]. In protozoans, extracellular ATP controls multiple responses; in social amoeba, Dictyostelium discoideum, ATP inhibits amoeboid movement, and triggers depolarization and Ca²⁺ influx; in the ciliates, Paramecium and Tetrahymena, ATP induces avoidance reactions through altering the rate of cilia beating and swimming [47]; in Trypanosoma cruzi, ATP induces parasitosis [193]. Purinergic signalling operates in all kingdoms of life, in plants, in fungi and in animals [190,197,198]. Chemical signalling by ATP, adenine nucleotides and adenosine is present in various forms virtually in all organisms and contributes to a wide variety of functions, from food detection (purinoceptors are an important part of food search behaviour in crustaceans such as lobsters and in insects such as mosquitoes [199,200]) to neurotransmission [201]. ATP-mediated transmission is operative in most (if not in all) organs and systems in vertebrates [197].

At the molecular level, purinergic chemical transmission is mediated by several systems responsible for transmitter release, and by specific plasmalemmal purinoceptors. Purinergic receptors are represented by: (i) ionotropic P2X receptors that are ATP-gated cationic plasmalemmal channels [202,203] and (ii) by metabotropic, G-protein-coupled seven-transmembrane receptors classified as P1 adenosine receptors, P2Y nucleotide (ATP, ADP, UTP, UDP and UDP-sugar) receptors and P0 adenine receptors [204,205]. The ATP-gated ion channels are similar in their architecture to P2X receptors, and had appeared already in protozoa; hitherto, P2X receptors have been identified and characterized in D. discoideum, in the marine green alga Ostreococcus tauri, in the choanoflagellate Monosiga brevicollis, in trematode Schistosoma mansoni and in a tardigrade Hypsibius dujardini [206–210]. The P2X receptors retain their structure throughout evolution, being composed of three subunits with each subunit containing two transmembrane domains.

Purinergic metabotropic receptors generally emerged at later evolutionary stages. Although the rather unique cAMP metabotropic receptors (defined as CAR1–4) are present in D. discoideum [211], the proper metabotropic purinoceptors appeared later. The most evolutionarily ancient adenosine receptors were identified in the sea anemone Nematostella vectensis, and they are abundant in insects, in bivalves (mussels) and in echinoderms [212]. The P2Y metabotropic receptors seem to emerge in vertebrates: the earliest homologues have been found in sharks and skate, Raja erinacea [213]. The evolutionary roots of P0 adenine receptors are unknown; these have been cloned only from mouse and hamster [205].

Release of ATP seems to be a common feature of life forms, and it has been identified in the earliest life forms: ATP is secreted by bacteria, by protozoa, by yeasts and plants, and by many types of cells in all multicellular animals [193]. The pathways for ATP release are represented by (i) transmembrane diffusion via ATP-permeable channels; (ii) active transport; and (iii) exocytosis [214,215]. Arguably, diffusional ATP release that uses plasmalemmal ion (usually anion) channels or unpaired hemichannels of connexin and innexin family, or pannexins (which though related to connexins/innexins do not form gap junctional channels and hence cannot be properly called hemichannels; there are indications that pannexins act as anion channels [216]) is the most ancient and general mechanism; indeed, diffusional ATP release is found throughout phylogeny and is present in many types of cells in mammals, from kidney and osteoclasts to neurons and neuroglia [134,217–220]. Active plasmalemmal transport of ATP is associated with ABC transporters that are expressed already in protists [221]. Finally, exocytotic ATP release that uses vesicles bearing specific nucleotide transports (VNUT/SLC17A9) allows spatially restricted and tightly regulated ATP release characteristic for nervous system, the VNUT-bearing vesicles are found in invertebrates and characterized in the nematode C. elegans, in sea anemone, in sea urchin and in some insects [222,223]; they are present in vertebrates, where they are expressed in various tissues including the brain [215].

In its physiological action, ATP transmission is tightly connected with Ca²⁺ signalling. Indeed, generation of Ca²⁺ signals seems to be the general consequence of activation of purinoceptors. All P2X receptors are permeable to Ca²⁺: the P_Ca/P_monovalent is approximately 5 for P2X₁ receptors, approximately 2.5 for P2X₂ receptors, approximately 1.5 for P2X₃ receptors, approximately 4.2 for P2X₄ receptors and approximately 1.5 for P2X₅ receptors; the heteromeric receptors similarly had an appreciable Ca²⁺ permeability. The P2X₇ receptors' permeability for Ca²⁺ may vary depending on the state of the pore: dilation of the latter increases P_Ca/P_monovalent [193,224–226]. Fractional Ca²⁺ currents through P2X receptors measured directly varied between 3 and 15 depending on channel composition (with rat P2X₁ and human P2X₄ receptors having maximal fractional Ca²⁺ currents at 12.4% and 15%, respectively [227]). This is remarkable, because Ca²⁺ fluxes through P2X receptors are fully comparable to Ca²⁺ current through NMDA receptors (fractional Ca²⁺ permeability approx. 14% [227]) and hence P2X-mediated Ca²⁺ signals may seriously contribute to postsynaptic plasticity. Activation of ionotropic purinergic receptors triggers Ca²⁺ signals in many types of cells [228–232]. Incidentally, P2X receptors may also serve as intracellular Ca²⁺-release channels, as has been shown in D. discoideum [233], further corroborating an intimate relation between ATP and Ca²⁺ from the very early evolution of cellular signalling. The P2X receptors seem to operate in the membranes of other organelles, including mitochondria, lysosomes and nuclei [234], although whether activation of these intracellular receptors is linked to Ca²⁺ movements remains to be found.

Metabotropic purinoceptors are distributed even more abundantly than P2X channels; the P2Y receptors and A1 adenosine receptors are expressed virtually in every cell type in mammals [197,235]. In many cells, these receptors are coupled, via several G proteins, to phospholipase C that produces InsP₃ and initiates Ca²⁺ release from the endoplasmic reticulum. The ATP-induced Ca²⁺ signalling mediated through the metabotropic route seems to be almost universal in non-excitable cells, being present in immune cells [236], in neuroglia [237], in kidney [238], in cells of urinary tract [239], in skin, bone and cartilages [240], in endocrine glands [241], etc. The P2Y receptors and their downstream Ca²⁺ signalling can be activated not only by ATP, but also by other nucleotides including ADP, UTP, UDP and UDP-glucose [242,243]. The adenosine-mediated signalling and A1 receptors are similarly widely distributed, being abundant in the nervous system, in the heart, in the lungs, in the blood vessels and so on [244–246] and in many cases, these receptors are coupled with cellular Ca²⁺ signalling. Adenosine is the product of ATP catabolism and is also released (through equilibrating transports) when cells experience high ATP breakdown (for example under stress or in pathology), thus directly coupling ATP biochemistry with ATP/Ca²⁺ intercellular signalling.

10. Conclusion

During evolution, inevitable leakage of Ca²⁺ into the eukaryotic cell was not only a problem—because of toxicity even in moderate concentrations—but also a chance. A compromise has been developed between ATP-based energetics and very low solubility of CaATP. However, considerable amounts of ATP have to be invested to keep [Ca²⁺] low in the cell, thus enabling Ca²⁺ to serve as a low cost second messenger. As a result, ATP is used by the cell preferably as an energy-storing molecule, whereas Ca²⁺ mainly serves for intracellular signalling. Intracellular Ca²⁺ signals can be generated by Ca²⁺ influx, by Ca²⁺ release from stores or by a combination of both. The termination of signal is achieved by binding to cytosolic buffers, sequestration and extrusion. There are many intracellular targets for Ca²⁺, the number of Ca²⁺-binding proteins increasing considerably during evolution. Another ancient strategy of the eukaryotic cell is to exploit the unavoidable leakage of nucleotide triphosphates from the cell as an extracellular messenger by purinergic receptors. Incidentally, this mirrors the stratagem of how the cell exploits leakage of Ca²⁺, but in opposite direction. Again, Ca²⁺ signals triggered by activation of purinoceptors provide for intracellular signalling. Considering the legion of cellular activities that depend on, or are modulated by Ca²⁺, this is an essential signal for life and death of a cell.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.