Fine tuning of cytosolic Ca²⁺ oscillations

Introduction

Most of the time, the hormone-induced Ca²⁺ increases that activate a variety of essential intracellular processes take the form of Ca²⁺ oscillations. In non-excitable cells, these repetitive spikes mainly arise through the periodic exchange of Ca²⁺ between the endoplasmic reticulum (ER) and the cytosol, through the interplay between inositol 1,4,5-trisphosphate (InsP₃)-sensitive Ca²⁺ channels and SERCA pumps^1,2. This basic mechanism, summarized in Figure 1, has now been well characterized and accounts for the observed increase in the frequency of Ca²⁺ oscillations with increasing concentrations of InsP₃ accompanying the rise in stimulation. Such a process is referred to as frequency encoding. It was often hypothesized that oscillations provide a digital signal to downstream effectors that are in turn stimulated in an ON or OFF manner. Indeed, if a process is activated above a threshold Ca²⁺ concentration, oscillations allow Ca²⁺ to reach this threshold repetitively even if the average Ca²⁺ signal remains below the threshold^3,4.

Figure 1. Basic mechanism of cytosolic Ca²⁺ oscillations in non-excitable cells.

These oscillations are initiated by the stimulus-induced rise in inositol 1,4,5-trisphosphate (InsP₃) concentration and occur through a repetitive exchange of Ca²⁺ between the cytoplasm and the endoplasmic reticulum (ER).

Based on the observed frequency encoding of the extracellular signal, it was also postulated that the physiological response in the form of secretion, gene expression, proliferation, etc., would in turn be sensitive to the frequency of Ca²⁺ oscillations^5–8. Although intuitively attractive, such frequency sensitivity of the downstream targets of Ca²⁺ has not been well corroborated by data. Besides the beautiful example of Ca²⁺-dependent calmodulin kinase II regulation by high-frequency Ca²⁺ spikes⁹ or of selective gene expression in T-lymphocytes⁴, there are few clear examples of physiological responses to Ca²⁺ increases that are quantitatively controlled by the frequency of the Ca²⁺ spikes. This statement does not mean that frequency encoding does not occur or that the frequency of Ca²⁺ oscillations does not affect the extent of the Ca²⁺-mediated physiological response. Indeed, a higher frequency of oscillations implies a larger average Ca²⁺ level, which may be per se the reason for the larger response. However, modulating the amplitude of the oscillations, their baseline level, or the duration of the spikes also modifies the average level and hence the response. As another example, spikes preceded by an important pacemaker-like Ca²⁺ increase could activate slower downstream targets characterized by a low threshold of activation. In such cases, frequency cannot be considered as the key characteristic of the oscillatory pattern and the response is not simply frequency sensitive. However, in the numerous studies about Ca²⁺ oscillations, frequency is the most studied parameter and the most commonly related to the extent of Ca²⁺-mediated physiological responses.

In fact, the relative scarcity of phenomena that are purely controlled by the frequency of Ca²⁺ oscillations is not so surprising given that the period of Ca²⁺ oscillations can be subject to a significant level of randomness (Figure 2 and 8,10). In some instances, it has even been explicitly observed that the frequency does not by itself regulate the extent of the second-messenger-mediated response. This is the case, for example, for carbachol-induced salivary secretion by acinar cells¹¹. At mammalian fertilization, the total integrated Ca²⁺ signal input is the most relevant parameter ensuring completion of fertilization-associated events¹². Interestingly, frequency encoding is also not a universal feature of Ca²⁺ oscillations, as it was shown in some cases, such as in acetylcholine-stimulated pancreatic acinar cells¹³, methacholine-stimulated lacrimal cells¹⁴, fish hepatocytes¹⁵, or in cell lines expressing the metabotropic glutamate receptor 5¹⁶, that an increase in stimulation does not affect the frequency of the resulting Ca²⁺ oscillations. In these cases, of course, it cannot be expected that the frequency of Ca²⁺ oscillations would be the way by which cells encode the information related to the level of response that is precisely triggered by the stimulation.

Figure 2. Various characteristics of Ca²⁺ oscillations that participate in fine tuning.

Traces show typical curves of Fluo4 loaded HeLa cells challenged with either 2 µM histamine (upper trace) or 3 µM (lower trace). Calcium imaging was performed as described previously¹⁰. Fluorescence images were collected every 3 seconds by an EM-CCD camera (Hamamatsu), digitized, and integrated in real time by an image processor (Metafluor). Letters indicate characteristics of Ca²⁺ oscillations that, besides their frequency, can affect the cellular response to these repetitive Ca²⁺ increases (a: latency of the Ca²⁺ response to the stimulation, b: minimal Ca²⁺ level between the spikes or baseline Ca²⁺, c: duration, or half-width, of the spikes, and d: rate of decrease of the response or degree of sustainability).

Also, recent investigations tend to suggest that rather than the frequency alone, the detailed dynamic characteristics of the Ca²⁺ increase pattern play an important role in determining the extent of the cell response. As illustrated in Figure 2, in addition to frequency, Ca²⁺ oscillations can vary in the amplitude and the width of the spikes, the baseline Ca²⁺ level, and the degree of sustainability. We refer to modifications of one of these characteristics as fine tuning of Ca²⁺ oscillations to emphasize that they imply fine regulation of cytosolic Ca²⁺ that goes behind the mechanism schematized in Figure 1 accounting for the existence and the frequency of oscillations. Various observations corroborate that these properties are important determinants for the physiological response of the cell. For example, the CD147 factor promotes oncogenic activities and influences the progression of hepatocellular carcinoma by enhancing both the amplitude and the frequency of ER-dependent Ca²⁺ oscillations¹⁷. In intestinal stem cells, dietary and stress stimuli are integrated in such a way that frequent and robust Ca²⁺ oscillations are associated with a poised proliferative state, while smoother oscillations on a more elevated level accompany active proliferation¹⁸. Fine tuning of Ca²⁺ signals also plays a role in the differentiation of neuronal and muscle cells (see 19 for review). In astrocytes, knocking-down the Na⁺–Ca²⁺ exchanger (NCLX) that mediates Ca²⁺ release from mitochondria slightly affects cytosolic Ca²⁺ changes and, by doing so, significantly reduces Ca²⁺-dependent processes, such as glutamate release, wound closure, and proliferation²⁰. Cell survival, death, and adaptation are sensitive to changes in Ca²⁺ patterns due to the interplay between ER/cytoplasmic Ca²⁺ exchanges and mitochondria and lysosomes²¹. Shigella bacteria also fine tune the Ca²⁺ responses when they invade epithelial cells. While the wild-type strain induces rather smooth and low-amplitude Ca²⁺ variations in the cytoplasm of the host cell, a less-invasive mutant strain induces more robust Ca²⁺ responses which, paradoxically, are associated with a higher survival of the host cells during the first hours following invasion²².

On another level, the precise isoforms of InsP₃ receptors expressed by a given cell – which have been shown to substantially affect the shape of the Ca²⁺ oscillations^23–25 – are critical for cell death and survival decisions²⁶. Finally, bioinformatics analyses highlighted that in cancer cells and tissues, the main processes associated with Ca²⁺ dynamics that are perturbed are the mechanisms of store-operated calcium entry (SOCE) and of calcium reuptake into mitochondria²⁷. Both of these processes are related to the fine tuning of Ca²⁺ oscillations, as discussed below. Altogether, these observations call for a more detailed understanding of oscillation-associated Ca²⁺ dynamics. Understanding why Ca²⁺ oscillates and what regulates the frequency of oscillations is not sufficient to understand their physiological impact, but the duration and shape of the peaks, their sustainability, and the baseline Ca²⁺ level must be carefully taken into account. In the following sections, we elaborate on two key controllers of the InsP₃R-based Ca²⁺ oscillations, both related to Ca²⁺ stores other than the ER, namely the mitochondria and the extracellular medium. We briefly review and discuss some of the main recent observations about their interplay with the InsP₃-induced Ca²⁺ spikes.

Mitochondrial Ca²⁺ uptake and release

By stimulating the activity of key enzymes involved in mitochondrial ATP synthesis, Ca²⁺ entry into mitochondria stimulates metabolism, thereby coupling ATP synthesis with energy demand²⁸. That Ca²⁺ exchange between the cytosol and the mitochondria in turn affects InsP₃-induced cytosolic Ca²⁺ signals was put forward quite early^29,30, but this concept was somewhat put aside for a decade. The molecular identification of the mitochondrial Ca²⁺ uniporter (MCU), a voltage, cytosolic, and mitochondrial Ca²⁺-sensitive transporter^31–34, awakened interest in this question. Ca²⁺ entry into mitochondria through the MCU is a highly non-linear function of cytosolic Ca²⁺³³. The MCU is in fact the Ca²⁺ pore-forming component of the uniporter and is part of a large complex of proteins that are required for Ca²⁺ channel activity or to regulate it under various conditions. For example, MICU1 (mitochondrial Ca²⁺ uptake 1) limits mitochondrial Ca²⁺ influx at low cytosolic Ca²⁺ concentration and the interaction between the MCU and MICU1 requires the expression of another component, called EMRE for essential MCU regulator^34,35. Ca²⁺ efflux back into the cytoplasm occurs through a NCLX. As expected, modifying any of these pathways affects the frequency of the oscillations; interestingly, increasing the activity of the MCU can both increase and decrease the frequency of oscillations³⁶. In addition to its effect on the frequency of the oscillations, the MCU controls the width of the spikes and the sustainability of the oscillations, as knocking-down the MCU broadens Ca²⁺ oscillations and accelerates the rundown of the oscillations in rat basophilic leukemia (RBL)-1 cells (Figure 3). Such rundown suppresses gene expression in response to leukotriene receptor activation³⁷. Mitochondria also affect the rate of rise and fall of cytosolic Ca²⁺ and thus the half-width and duration of the spike. More specifically, mitochondria smooth out cytosolic Ca²⁺ changes mainly because they have a ~30 times larger Ca²⁺ buffering capacity than the cytoplasm³⁸. Also, because of their slow dynamics, mitochondria continue releasing Ca²⁺ between subsequent releases of Ca²⁺ from the ER, thus playing a key role in determining the baseline cytosolic Ca²⁺ level. Thus, mitochondrial Ca²⁺ handling through the MCU and the NCLX clearly fine tunes cytosolic Ca²⁺ oscillations.

Figure 3. The mitochondrial Ca²⁺ uniporter (MCU) participates in the fine tuning of Ca²⁺ oscillations.

A. Control recording of Ca²⁺ responses in rat basophilic leukemia (RBL)-1 cells stimulated by LTC4 in 2 mM external Ca²⁺. B. Same recording after MCU knockdown. Ca²⁺ entry in mitochondria via the MCU broadens cytosolic Ca²⁺ spikes and decreases their sustainability. From Samanta et al.³⁷. Shown are the ratios of fluorescence of Fura-2 loaded cells excited at 356 and 380 nm.

The kinetics of the MCU and the NCLX have been fairly well characterized, but much remains to be done to fully identify other fluxes. The permeability transition pore (PTP) in its low conduction mode participates in the Ca²⁺ exchange process in HeLa cells, as its inhibition by cyclosporine A affects Ca²⁺ oscillations^36,39. The functional role of LETM1-mediated Ca²⁺ transport also remains poorly understood. This EF-containing transmembrane protein has been functionally identified as a Ca²⁺/H⁺ exchanger of the inner mitochondrial membrane^40,41, although this remains controversial⁴². In electrically excitable cells such as cardiomyocytes and neurons, ryanodine receptors have been shown to transport Ca²⁺ into mitochondria^43,44. A rapid Ca²⁺ uptake mode (RaM) of poorly identified molecular nature has been reported in studies on isolated mitochondria from cardiac and liver cells^45,46. However, the implication of RyR and RaM in mitochondrial Ca²⁺ influx remains to be firmly established^47,48. Finally, in a more indirect manner, mitochondrial metabolism also affects cytosolic Ca²⁺ signals, mainly by modifying the mitochondrial voltage across the internal mitochondrial membrane, which greatly affects the activities of all of the above-mentioned fluxes²⁹. All of the above-cited phenomena are thus potentially implicated in the control of the detailed characteristics of Ca²⁺ oscillations. Their interplay with the activities of the MCU and the NCLX is regulated by an intricate and complex network of interactions implicating cytosolic and mitochondrial Ca²⁺ as well as mitochondrial voltage and numerous accessory proteins.

Store-operated Ca²⁺ entry

Cytosolic Ca²⁺ oscillations are sustained by SOCE from the extracellular medium⁴⁹. This mechanism involves the stromal interaction molecule (STIM) and the Orai protein⁵⁰. The transmembrane ER protein STIM is sensitive to Ca²⁺ changes in the ER through an EF-hand facing the lumen of the store. Decrease in luminal Ca²⁺ below ~200 μM (for the STIM1 isoform) leads to STIM aggregation, followed by migration to ER–plasma membrane junctions. Here, STIM oligomers can bind and activate Orai, a four-transmembrane-domain plasma-membrane-spanning protein, thus forming a channel (known as CRAC for Ca²⁺-release-activated Ca²⁺ channel) allowing Ca²⁺ to enter down the chemical gradient. Another STIM isoform, STIM2, has a lower affinity for ER Ca²⁺, which allows for activation of Ca²⁺ entry at moderate ER depletion, although at a reduced rate⁵¹. Mammalian cells have genes for the three homologs Orai1, Orai2, and Orai3, and it is thought that Orai2 and/or Orai3 act as compensative types for the lack of Orai1. Orai channels are made of multiple subunits, and CRAC channel gating by STIM is best described by a Monod-Wyman-Changeux scheme in which tetramers of Orai have four STIM binding sites^50,52.

Although the mechanism just described has most of the time been investigated in conditions when the Ca²⁺ pools are emptied artificially, studies performed in a variety of cell types demonstrate that STIM expression is essential for an ensemble of physiological processes⁵³. To quote here just one recent example, in airway smooth muscle, altered expression and function of STIM/Orai proteins have been linked to pathologies including restenosis, hypertension, and atopic asthma⁵⁴.

The STIM-Orai pathway for Ca²⁺ entry displays a hysteretic behavior: STIM-Orai association and dissociation do not occur at similar ER Ca²⁺ concentrations⁵⁵. Although the origin and the physiological significance of this unusual behavior remains unknown, it might be related to the inactivation of SOCE-mediated Ca²⁺ entry by cytosolic Ca²⁺ itself, a process that has long been thought to be mediated by calmodulin⁵⁶ but was recently suggested to be due to a calmodulin-independent conformational change within the pore allowed by two specific Orai residues, Y80 and W76^57,58. This Ca²⁺-induced inactivation (CDI) of SOCE allows for a modulation of Ca²⁺ entry depending on the level of cytosolic Ca²⁺, thus shaping the oscillations.

The key effect of STIM and Orai on the oscillatory Ca²⁺ pattern and its downstream targets are also much documented. Through the specific ER Ca²⁺ sensor STIM2 that has a high K_M for Ca²⁺, SOCE determines the basal level of Ca²⁺ in HeLa cells⁵¹. More generally, SOCE-mediated Ca²⁺ entry has a significant effect on Ca²⁺ oscillations, as it can in turn affect all Ca²⁺ exchanges between the cytoplasm and the internal stores. A less straightforward but highly interesting effect was uncovered in RBL-2H3 cells. Because the activity of the plasma membrane phosphatidylinositol 4-phosphate 5 kinase that replenishes the PIP₂ pool is Ca²⁺ sensitive, SOCE is necessary to avoid the rundown of the oscillations. Indeed, in the absence of SOCE, cysteinyl leukotriene type I receptor activation leads to the exhaustion of the PIP₂ pool and hence to the disappearance of InsP₃-induced Ca²⁺ release from internal stores⁵⁹.

In RBL cells displaying Ca²⁺ oscillating, gene expression is entirely driven by SOCE and proceeds as an all-or-nothing process in individual cells⁶⁰. During maturation of mouse oocytes, STIM1 and Orai1 control the basal Ca²⁺ level and the whole Ca²⁺ homeostasis, thus controlling meiosis resumption⁶¹. At fertilization of pig eggs, overexpression of STIM1 and Orai1 substantially decreases the number of Ca²⁺ spikes induced by sperm binding (Figure 4). Moreover, these spikes are broader and their frequency is reduced as compared to control eggs⁶². This observation contrasts with the observed decreased frequency of fertilization-induced Ca²⁺ oscillations in hamster eggs when decreasing external Ca²⁺ concentration⁶³. It shows that the control of Ca²⁺ signaling by SOCE cannot be directly assimilated to the control of Ca²⁺ signals by the extracellular Ca²⁺ concentration. Interestingly, if SOCE is inhibited, fertilization is also impaired, as oscillations last for about 1 hour instead of at least 2 hours. In mice, cytoplasmic Ca²⁺ levels are elevated for ~50% of the time in STIM1+Orai1-overexpressing oocytes in the first 2 hours after fertilization, as compared to only less than 20% of the time in control oocytes. Despite this larger Ca²⁺ signal, most of the STIM1/Orai1-overexpressing oocytes do not reach the two-cell stage⁶⁴. However, female mice lacking Orai1 are fertile⁶⁵, while male mice are sterile due to severe defects in spermatogenesis⁶⁶.

Figure 4. Effect of store-operated calcium entry (SOCE) activity on fertilization-induced Ca²⁺ oscillations in pig eggs.

A. Co-overexpression of Orai1 and stromal interaction molecule 1 (Stim1) leads to broader spikes with reduced frequency and that stop prematurely. B. Control situation. Reproduced with permission from Chunmin Wang, Lu Zhang, Laurie A. Jaeger, and Zoltan Machaty. Store-Operated Ca2+ Entry Sustains the Fertilization Ca²⁺ Signal in Pig Eggs. Biol Reprod 2015; 93(1):25. DOI:10.1095/ biolreprod.114.126151⁶².

Conclusion

It is by now clear that in many cases the existence of Ca²⁺ oscillations does not provide an ON/OFF signal for the Ca²⁺-mediated response to the stimulus nor is the extent of the response only determined by the frequency of the oscillations. How the exact shape of this Ca²⁺ signal is controlled, i.e. what we refer to here as its fine tuning, can alter the response qualitatively and quantitatively. Ca²⁺ exchanges with mitochondria and SOCE play an important role in fine tuning cytosolic Ca²⁺ oscillations. Interestingly, there is some dynamic interplay between these two Ca²⁺ sources as, in mesothelial cells in the absence of external Ca²⁺, mitochondrial Ca²⁺ takes over to provide a Ca²⁺ influx pathway during oscillations⁶⁷. Moreover, other organelles such as the Golgi^68,69 or the acidic organelles^21,70,71 are also involved. Genetic regulations further complicate the intricate network of Ca²⁺ fluxes: in lymphocytes, Ca²⁺-dependent activation of CREB controls the level of expression of the MCU, which explains why the expression of this uniporter is modified in the absence of InsP₃ receptors or of the STIM/Orai machinery⁷². Much remains to be done to understand how diverse factors interact to control the detailed pattern of Ca²⁺ oscillations and how this pattern can in some cases significantly affect the physiological response. Integration of the Ca²⁺ signal over long periods of time may explain why small changes in the pattern of the Ca²⁺ spikes become significant in some cases. By such integration, the extent of activation of the downstream targets of Ca²⁺ is modified by apparently minor changes in the Ca²⁺ oscillatory pattern, which are less visible than its frequency. Spatial aspects most certainly also play an important role in this respect, as the Ca²⁺-sensitive targets are far from being homogeneously distributed within the cell⁷³. Finally, the kinetics and thresholds for Ca²⁺ activation of these targets are expected to be at least as important, as in other signaling cascades playing a key role in the storage of information⁷⁴.