Attenuation of calmodulin regulation evokes Ca²⁺ oscillations: evidence for the involvement of intracellular arachidonate-activated channels and connexons

Abstract

Intracellular Са²⁺ controls its own level by regulation of Ca²⁺ transport across the plasma and organellar membranes, often acting via calmodulin (CaM). Drugs antagonizing CaM action induce an increase in cytosolic Ca²⁺ concentration in different cells. We have found persistent Са²⁺ oscillations in cultured white adipocytes in response to calmidazolium (CMZ). They appeared at [CMZ] > 1 μM as repetitive sharp spikes mainly superimposed on a transient or elevated baseline. Similar oscillations were observed when we used trifluoperazine. Oscillations evoked by 5 μM CMZ resulted from the release of stored Ca²⁺ and were supported by Са²⁺ entry. Inhibition of store-operated channels by YM-58483 or 2-APB did not change the responses. Phospholipase A₂ inhibited by AACOCF₃ was responsible for initial Ca²⁺ mobilization, but not for subsequent oscillations, whereas inhibition of iPLA₂ by BEL had no effect. Phospholipase C was partially involved in both stages as revealed with U73122. Intracellular Са²⁺ stores engaged by CMZ were entirely dependent on thapsigargin. The oscillations existed in the presence of inhibitors of ryanodine or inositol 1,4,5-trisphosphate receptors, or antagonists of Ca²⁺ transport by lysosome-like acidic stores. Carbenoxolone or octanol, blockers of hemichannels (connexons), when applied for two hours, prevented oscillations but did not affect the initial Са²⁺ release. Incubation with La³⁺ for 2 or 24 h inhibited all responses to CMZ, retaining the thapsigargin-induced Ca²⁺ rise. These results suggest that Ca²⁺-CaM regulation suppresses La³⁺-sensitive channels in non-acidic organelles, of which arachidonate-activated channels initiate Ca²⁺ oscillations, and connexons are intimately implicated in their generation mechanism.

Introduction

Intracellular Ca²⁺ is a key messenger for various external signals that act on cell receptors and control a multitude of cell functions. The level of cytosolic Са²⁺ at rest and under stimulation is governed by autoregulatory effects on its channels, pumps and exchangers in the plasma membrane (PM) and membranes of intracellular organelles serving as Са²⁺ stores. In particular, direct Ca²⁺ binding to the subunits of inositol 1,4,5-trisphosphate receptor (IP₃R) and ryanodine receptor (RyR) of the endoplasmic reticulum (ER) opens these channels and causes calcium-induced calcium release (CICR) from the organelle filled by sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA). Many of Са²⁺ effects are mediated by ubiquitous protein calmodulin (CaM), which changes its conformation upon Ca²⁺ binding and activates or inhibits the targets including the Ca²⁺channels and PM Са²⁺-ATPase (PMCA). Such modulation by CaM can be indirect, proceeding through the CaM-dependent kinase or phosphatase, as well as through a range of intermediary effectors. Dysregulation of Ca²⁺ signaling is involved in cell death, altered proliferation and major diseases [1,2,3]. This complex Сa²⁺ signaling system typically responds to stimuli by oscillations. They are further decoded through CaM or other Ca²⁺ sensors that selectively activate various cellular processes, depending on both the frequency and the amplitude of the repetitive events. It is generally accepted that the mechanism of Са²⁺ oscillations is based on CICR [1, 4].

Calcium entry through PM necessary for persistent, long-lasting Сa²⁺ elevations and oscillations can be provided by different channels, among which store-operated Ca²⁺ (SOC) channels are widespread. Selective calcium release-activated-calcium (CRAC) channels of this type are formed by Orai1 proteins. Cationic TRPC channels mediate SOC entry (SOCE) and store-independent entry (non-SOCE) [5]. Under low agonist concentrations that cause oscillations, Са²⁺ entry may be determined by store-independent, arachidonate-regulated Ca²⁺ (ARC) channels of PM, which are also activated by leukotriene C₄ [5, 6]. ARC channels are heteromultimers of Orai1 and Orai3 subunits compared to homohexameric CRAC channels [5]. Connexin (Сх) hemichannels (connexons) were shown to be important for oscillations in some cells [7,8,9]. These hemichannels form intercellular gap junctions, while in an unopposed state they function as channels for ions and small molecules such as АТР and NAD⁺, and the released substances may activate cell receptors [10]. Connexons are also present in cytoplasmic structures, which perform their biosynthesis and degradation and communicate with PM by trafficking [10, 11]. It was suggested that vesicle-bound connexons might mediate NAD⁺ transport between subcellular compartments [12], but in general, the Ca²⁺-signaling role of intracellular connexons remains obscure. Pannexin proteins form channels (not gap junctions) that resemble connexons [10, 13]. Different PM components of Ca²⁺ transport can be transferred into the cell by endocytosis forming endosomes that are gradually acidified by the vacuolar-type H+-ATPase (V-ATPase) and eventually fuse with lysosomes. The proton gradient is used for accumulation of Са²⁺_, released thereafter by NAADP activating two-pore channels. The acidic stores can participate in CICR and formation of Са²⁺ signals [14].

Despite numerous data on CaM effects exerted on the participants of Ca²⁺ signaling, the total impact of CaM on Ca²⁺ dynamics, and especially its role in oscillations are not fully understood. Application of different СаМ antagonists increased cytosolic Ca²⁺ via stimulation of release and entry [15,16,17,18,19,20,21,22], suggesting the repressive Ca²⁺-CaM influence on the Ca²⁺ level at the resting state. The same approach indicated modulation by CaM of agonist-induced Ca²⁺ oscillations in hepatocytes [23] and megakaryocyte [24]. We had observed that calmidazolium (CMZ) itself induced prolonged Ca²⁺ oscillations in white adipocytes [25]. The imidazole derivative CMZ, or R24571, belongs to the compounds whose binding to the hydrophobic surfaces of СаМ prevents the interaction of CaM with the target proteins [26]. Adipocytes are prone to Ca²⁺ oscillations evoked by a number of agonists acting through IP₃R-dependent or RyR-dependent signaling pathways [25, 27, 28]. Apart from ER, these cells have endocytic structures and specialized vesicles providing recycling and storage of the glucose transporter GLUT4 [29]. Non-ER Ca²⁺ stores might be involved in the signaling network, since addition of exogenous NAADP to adipocytes resulted in Са²⁺ transients [30].

The aim of the present study was to characterize the impact of CaM antagonists, attenuating CaM-mediated regulation, on cytosolic Ca²⁺ in adipocytes and to reveal the underlying mechanism with the use of inhibitors of potentially involved processes. We investigated whether this mechanism is based on IP₃R- or RyR-mediated Ca²⁺ release and tested the participation of other pathways able to mobilize Ca²⁺. The obtained data indicate stimulation of arachidonic acid (AA) production and Ca²⁺ release through AA-activated (probably, ARC) channels, with oscillations being driven by connexons. It is proposed that these channels, typical for PM, may be functional, at least under certain conditions, in organelles related to PM by trafficking, and the organellar connexons could generate oscillations due to their own CICR.

Materials and methods

Isolation of preadipocytes

All animal studies were approved by the Animal Ethic Committee of the Institute of Cell Biophysics, Russian Academy of Sciences, and are in accordance with the European Communities Council Directive (86/609/EEC). NMRI mice (aged 3–5 weeks) were decapitated after a brief (45–60 s) anesthesia with carbon dioxide before sacrifice. Mice were subjected to cervical dislocation and disinfected with 70% ethanol prior to dissection. All operations were performed in a sterile environment on ice. White adipose tissue was removed from the epididymal fat depot and placed in a Petri dish with cold DMEM-medium (Sigma, USA). Scissor-minced white adipose tissue was transferred into a tube containing sterile DMEM with 7 mg type II collagenase (Sigma-Aldrich, USA) and 4% bovine serum albumin (BSA, free from fatty acids, Sigma, USA). Then, the tissue was incubated for 18 min at 37 °C. To stop the collagenase enzymatic reaction, the tube was chilled on ice for 20 min with intermittent shaking followed by filtration through 250 µm filter and centrifugation at 1000 g for 10 min. The pellet was then resuspended in cold DMEM medium, filtered through 50 µm filters and centrifuged at 1000 g for 10 min. Finally, the pellet was resuspended in cultural medium containing: DMEM, 10% fetal bovine serum (FBS, Gibco, USA), 4 mM l-glutamine (Sigma, USA), 4 nM insulin (NovoNordisk, Denmark), 0.004% gentamicin and 25 µg/ml sodium ascorbate (Sigma, USA). The obtained suspension contained preadipocytes, since mature adipocytes carry vesicles of fat and do not precipitate under the given conditions.

Cultures of white adipocytes

100 µl droplets of culture medium containing 3 × 10⁴ preadipocytes were placed on round coverglasses (25 mm in diameter), which were then transferred into 35 mm Petri dishes. 6 h after adhesion of the cells to the glass, additional culture medium was added to the Petri dishes. On the third day, the medium in the dishes was replaced with a fresh portion of medium, which included 10 nM of cytosine arabinoside (Sigma, USA) to suppress proliferation of fibroblasts, and incubation in CO₂ atmosphere was continued for 8 h. After that the medium was replaced with fresh culture medium. On the ninth day of cultivation in the CO₂ incubator at 37 °C, the cells formed a monolayer and became differentiated.

Measurements of cytosolic calcium concentration

Measurements of cytosolic [Ca²⁺] were performed by fluorescent microscopy using fura-2/AM (Invitrogen, USA), a ratiometric fluorescent calcium indicator. Cells were loaded with the probe dissolved in Hanks balanced salt solution (HBSS), containing 10 mM HEPES (both from Gibco, USA), pH 7.4, at the final concentration of 5 µM at 37 °C for 40 min with subsequent 15 min washout. The coverslip containing the cells loaded with fura-2 was then mounted in the experimental chamber. During the experiment we used a perfusion system, which enables complete replacement of the cell bathing solution within 30 s. Calcium-free medium contained 0.5 mM EGTA. Axiovert 200M based imaging system (Carl Zeiss, Germany) equipped with a HBO100 mercury lamp, an AxioCam HSm CCD camera and a MAC5000 high-speed excitation filter wheel was used. Fura-2 fluorescence was excited at two wavelengths using band-pass filters BP 340/30 and BP 387/15; fluorescence was registered in the wavelength range of 465–555 nm. The excitation light intensity was lowered using 25 and 5% neutral density filters to prevent phototoxicity. Image frames were acquired at 3 s intervals with a Plan Neofluar 10×/0.3 objective.

Assessment of cell viability

The amount of viable and dead cells was estimated using staining with 1 µg/ml propidium iodide (PI, Invitrogen, USA). PI has red fluorescence and penetrates only into necrotic cells with damaged outer membranes and can also indicate abnormally opened Cx hemichannels. Cells were incubated with PI for 15 min and then washed 3 times with HBSS. To register the PI fluorescence, Filterset 45 (Carl Zeiss, Germany) was used.

Low temperature experiments

Probe loading (40 min) and washout (15 min) were performed at 37 °C. Then the chamber with cells (1 ml of medium) was placed into a refrigerator in an ice-cold container and chilled to 4 °C. Tubes containing reagents for additions and washing media were cooled simultaneously. The cells were loaded with PI on ice and then incubated in a refrigerator for 15 min. Before the experiment, the region of interest was rapidly photographed, and during the subsequent observations (up to 5 min) the temperature was maintained at about 4 °C.

Data analysis

The time lapse image sequences were analyzed using ImageJ 1.44 (NIH Image, Bethesda, MD, USA). Graphs were plotted using the OriginPro 8.0 software (Microcal Software Inc., Northampton, MA, USA). The statistical analysis was performed using the same software. The difference between the two sets of data was considered to be significant at P < 0.05 (lower levels are indicated) following the Student’s unpaired two-tailed t test. The results are presented as means ± SEM or as a representative calcium signal of the cells, where n stands for the number of cells possessing the given pattern of Са²⁺ behavior. A number of independent experiments are indicated (by default, data from a single experiment). N denotes the number of cells analyzed in the given experiment. Fractions of cells displaying any considerable Са²⁺ responses (RF), oscillations (OF), and a prolonged lag (taken as > 1 min) before the Ca²⁺ rise (LF) were calculated as their corresponding n/N, then used to determine the mean values. Oscillations were defined as Ca²⁺ changes with 3 or more peaks irrespective of their pattern. Cessation of oscillations was defined as a longer distance between the last peak and the end time compared to the maximum distance between the preceding peaks.

Results

Са²⁺ responses to CaM antagonists CMZ and TFP

Addition of CMZ to white adipocytes caused dose-dependent changes in the cytosolic Са²⁺ concentration. Quantitative characteristics of these responses are presented in Fig. 1a. At 1 µM CMZ, no appreciable responses were detected among 90 cells in three independent experiments, and the next concentration of 2.5 µM was half-effective for the response fraction (RF). This implies the existence of a steep threshold between 1 and 2.5 µM CMZ. The responses consisted mainly of oscillations (defined as having not less than three peaks) represented by the oscillation fraction (OF), reaching maximum at 5 µM CMZ. The fraction of the remaining patterns, pulses or transients with 1–2 peaks, when expressed as a percentage to the responded cells, was the lowest (5.4%) at 5 µM CMZ and increased to 46.2% at 10 µM CMZ, thus replacing the oscillatory mode. Responses of all types could have a lag phase before a rapid Ca²⁺ rise. The relative number of responses with lag > 1 min (lag fraction, LF) was rather constant, and the ratio to the responded cells was 29–39%. For the detailed analysis presented below, 5 µM concentration of CMZ was chosen, at which the oscillatory mode was most expressed.

Fig. 1

Dose–response relationship and oscillations of cytosolic Ca²⁺ in single adipocytes at 5 µM CMZ. a RF, OF and LF are fractions (relative to N) of cells displaying, correspondingly, any responses, oscillations, and time lag > 1 min before a response. Data of 7 (for 5 µM CMZ), 4 (10 µM) or 3 (1 and 2.5 µM) independent experiments with N ≥ 25 cells incubated for 5.5–16.5 min with CMZ are included. *P < 0.05, **P < 0.01 vs 2.5 µM CMZ. b–e represent same experiments as in (a) for 5 μM CMZ, N = 28–69, and 361 cells in total. RF = 57.1 ± 3.0%, OF = 54.1 ± 4.3%, LF = 16.5 ± 5.2%, and 7.2 ± 1.5% oscillations stopped. b Immediate Ca²⁺ rise followed by oscillations with a gradual decrease in the Ca²⁺ level. c Lag phase is present, with no decline in the Ca²⁺ level. d Baseline oscillations without the lag. e Complex oscillations with a high-frequency component and the initial lag. Mode (d) was minor (≤ 17.6% of oscillations, mean 6.2 ± 2.3%), and mode (e) occurred occasionally

Figure 1b–e shows time courses of responses, mostly possessing the shape of a rapid Са²⁺ rise with subsequent sustained oscillations on the elevated level that could slowly drift, still staying above the resting state (Fig. 1b, c). Only ≤ 12% of oscillations in each of seven experiments ceased during exposure times up to 16.5 min. The oscillations were mainly represented by sharp spikes with a sub-minute period, ranging from several seconds to a few minutes in some cells. During the baseline oscillations (Fig. 1d), the Ca²⁺ concentration between all spikes resumed the pre-stimulation level. High-frequency oscillations also occurred, which could be imposed upon increases of low frequency in the form of rapid bursts (Fig. 1e). The level of Ca²⁺ during the time lag did no change for some period right after CMZ application, and then an accelerating or abrupt Ca²⁺ elevation took place as shown in Fig. 1c, e. These response patterns suggest that the mechanism of CMZ-induced Ca²⁺ rises can possess specific traits of its manifestation in particular adipocytes. Analogous patterns of drifting oscillations were previously found in the same cell cultures stimulated by acetylcholine and phenylephrine [25, 27].

In view of possible non-specific effects of CMZ not mediated by CaM, we also used another antagonist of CaM, trifluoperazine (TFP), belonging to the phenothiazine family [26]. TFP applied in the same concentration range as CMZ, led to robust oscillations (Fig. 2) closely resembling the CMZ-evoked ones in respect to their notched form, amplitude, frequency and duration. At 5 µM TFP (Fig. 2a), oscillations occurred in the vast majority (92%) of responding cells. As a result of higher TFP concentration 10 µM, the oscillatory responses decreased to 59.1%, and subsequent application of 50 µM TFP produced only a transient (Fig. 2b). Taking into account essential similarities between Ca²⁺ responses to CMZ and TFP, we further used CMZ.

Fig. 2

Ca²⁺ responses to three doses of TFP. a Typical oscillations at 5 µM TFP, n = 23, RF = 65.8%, N = 38. b Oscillations at 10 µM TFP (n = 13, RF = 53.7%) followed by a transient at 50 µM TFP, N = 41

Role of Ca²⁺ entry and SOC channels

The medium with normal Са²⁺ concentration was replaced for nominally Ca²⁺-free with addition of EGTA as Ca²⁺ chelator to define the role of the total Са²⁺ entry via calcium-conducting channels of all types. After 30-min preincubation in this medium, cells responded to CMZ with RF, OF and LF indices, not significantly different from those (Fig. 1) in the normal medium. The main pattern was a transient with superimposed oscillations declined close to the initial Ca²⁺ level (Fig. 3a). Most oscillations terminated before the end time in contrast (P < 0.05) to persistent oscillations in Fig. 1, where only 7.2% vanished. Subsequent inhibition of the sarco/endoplasmic reticulum Ca²⁺-ATPase (SERCA) by thapsigargin (TG), that empties the TG-sensitive stores, elicited the response of a similar height as that of CMZ-induced rises (Fig. 3a), not differing from that without CMZ (Fig. 3b) or from that without the preceding Ca²⁺ withdrawal (see the legend to Fig. 7d). This indicates no substantial depletion of the overall TG-sensitive stores during the 30-min absence of Ca²⁺ entry, or during the period of CMZ action. Besides, since CMZ did not lower the rate of Ca²⁺ recovery driven by PMCA (Fig. 3a), its activation by Ca²⁺-CaM [1] did not account for the increased net Ca²⁺ influx in the normal medium. Thus, oscillations in the CMZ-stimulated adipocytes are produced by the periodic Са²⁺ release from an intracellular store, whereas Са²⁺ entry enables sustained high-amplitude oscillations.

Fig. 3

Ca²⁺ responses upon elimination of Ca²⁺ entry and inhibition of SOC channels. a Мedium without Ca²⁺ addition but containing 0.5 mM EGTA was applied 30 min before the zero time point. The trace represents n = 111 from 4 experiments with N = 35–69, RF = 80.4 ± 11.6%, OF = 59.9 ± 17.9%, LF = 6.0 ± 2.8% (P > 0.05 for each), where 59.6 ± 28.9% oscillations stopped before the addition of TG (P < 0.05), compared to Fig. 1. Responses to 10 μM TG averaged in each experiment had the mean amplitude 0.0612 ± 0.0087 fluorescence ratio units and the maximal slope of decline (2.98 ± 0.95)·10^− 4 units/s, close (P > 0.05) to the corresponding values in (b). b The same zero-Ca²⁺ preincubation as in (a), but without CMZ, 5 experiments with N = 32–43. The amplitude and the slope of responses to 10 μM TG were 0.0792 ± 0.0129 and (3.26 ± 0.60)·10^− 4, respectively, and 10 µM ionomycin gave the amplitude of 0.0237 ± 0.0055. c Oscillations continue after application of 1 µM YM-58483 as indicated, n = 40, N = 37, 46 and 47. Only 9.0 ± 1.8% of oscillations stopped before the end time (P > 0.05 vs Fig. 1). d Preincubation with YM-58483 during 2 h has no effect; oscillations occurred in n = 78 cells, N = 31, 45 and 56. RF = 78.8 ± 12.8%, OF = 56.4 ± 13.6%, LF = 25.2 ± 7.0%, and 18.4 ± 3.3% of oscillations ceased, the values differed insignificantly (P > 0.05) from the control in Fig. 1

The presence of activatable SOC channels in adipocytes [25] together with the lack of store depletion raise the question whether these channels are involved in Са²⁺ responses to CMZ. It was shown that the SOC channel inhibitor, YM-58483 (also known as BTP2), had a submicromolar half-maximum concentration, with its potency rising by more than an order for hours of preincubation [31]. In our experiments, 1 µM YM-58483 added during CMZ-induced oscillations did not cause a clear-cut immediate effect in respect to the slow component, frequency or amplitude, and oscillations largely persisted (Fig. 3c). Ca²⁺ oscillations were also unaffected by short preincubation with 2-APB (see below), that inhibits [32] the SOC channels. Hence, these PM channels rapidly accessible to the two inhibitors are not indispensable to CMZ-induced oscillations. The possibility of delayed effects was explored by prolonged incubation of cells with YM-58483 during 2 h. It was found that this YM-58483 action did not affect (P > 0.05) the response indices (Fig. 3d). These results suggest that the SOC channels are not involved in CMZ action in our experiments.

Contribution of PLA₂ and PLC

CMZ was shown to stimulate Ca²⁺ fluxes into the cytosol through activation of PLA₂ [18,19,20, 25], with requirement for functional Са²⁺-independent phospholipase А₂ (iPLA₂) inhibited by CaM [19]. Considering this, we examined the mechanism of CMZ action using PLA₂ inhibitor arachidonyl trifluoromethyl ketone (AACOCF₃) also affecting iPLA₂ [33] and selective inhibitor of iPLA₂, bromoenolactone (BEL). In white adipocytes, adipose-specific PLA₂ (AdPLA) is abundantly expressed, that is sensitive to AACOCF₃ and not to BEL, with much lower expression of iPLA₂ and other PLA₂s [34, 35]. When 15 µM AACOCF₃ was applied in the course of CMZ-induced response, all oscillations (n = 45, N = 31, 43 and 59) continued, 8.5 ± 1.0% of them ceasing before the end time (P > 0.05 vs the data in Fig. 1), so no effect of AACOCF₃ was seen. Preincubation with AACOCF₃ and the following administration of CMZ showed that all responding cells without exception had a protracted lag of more than 1 min before the first Ca²⁺ spike, reaching the maximum of 10.5 min. No considerable Ca²⁺ rise was obtained during this period indicating that PLA₂ inhibition completely abolished the early Ca²⁺ release typical of control responses. Subsequent Ca²⁺ changes were oscillatory, with normal percentage and pattern (Fig. 4a). This suggests that PLA₂ products are responsible for the early events of Ca²⁺ mobilization and launching of oscillations, but are not involved in the mechanism of oscillations. When BEL was added 5.5 min prior to incubation with CMZ, no change (P > 0.05) in RF, OF and LF was found (3 experiments, N = 30–52) compared to Fig. 1. Therefore, the form iPLA₂ does not significantly contribute to CMZ action.

Fig. 4

Effects of PLA₂ (a) and PLC (b) inhibition. a Cells were treated with 15 μM AACOCF₃ before CMZ addition, N = 33, 49 and 50. All responses had the initial lag > 1 min (n = 57), and one cell with the maximum lag of 10.5 min had no oscillations. RF = 43.8 ± 6.0% and OF = 43.2 ± 6.6% are not significantly different (P > 0.05) from the control values in Fig. 1. b Transient (n = 71) and oscillations (n = 32), whose indicated fractions are relative to all 103 responding cells in 3 experiments with N = 54, 59 and 90. U73122 (10 µM) was added 310 s before time zero. RF = 47.6 ± 9.9%, OF = 17.6 ± 6.7%, LF = 39.57 ± 13.01%, where OF and LF/RF = 79.0 ± 10.0% are significantly different (P < 0.05) from the control in Fig. 1

As found previously, CMZ-evoked Ca²⁺ release was prevented with phospholipase C (PLC) inhibition by U73122 [18, 20], or the release was PLC-independent [22]. U73122 also inhibited the non-SOCE pathway [36]. In our experiments, cell incubation with U73122 at the concentration of 10 μM sufficient for the inhibition [18, 20, 36] did not suppress Ca²⁺ responses in general, but modified their prevalent pattern (Fig. 4b). To be exact, oscillations were reduced (OF decreased threefold, P < 0.05) being replaced by transients, and lag > 1 min before the response, when estimated as LF/RF, increased (P < 0.05) to 79% and became predominant. Therefore, PLC is not a primary target for CMZ, although PLC activity contributes to both early Ca²⁺ mobilization and subsequent oscillations.

TG-sensitive stores, RyR and IP₃R, acidic stores

То assess significance of the TG-sensitive Ca²⁺ pool in generation of the CMZ-evoked responses, the pool was emptied in Ca²⁺-free medium prior to CMZ application. It can be seen from Fig. 5a that TG induces a transient with no additional Ca²⁺ release by CMZ, thus indicating that CMZ effect requires the TG-sensitive Са²⁺ pool to be filled. Addition of Ca²⁺ ionophore ionomycin at the end of the control experiments without CMZ resulted in additional Ca²⁺ release making 30% of the TG-produced amplitude (see Fig. 3b). This demonstrates the existence of а considerable TG-independent pool, which, however, is not engaged by CMZ, at least after emptying the TG-dependent pool. Thus, Ca²⁺ rises and oscillations in adipocytes under CMZ action are totally dependent on organelles accumulating Са²⁺ via SERCA.

Fig. 5

CMZ-induced Ca²⁺ rises are prevented by discharge of the TG-dependent stores, and oscillations do not require RyR, IP₃R or acidic Ca²⁺ stores. a Records for 5 of 68 cells responding to 10 µM TG without Ca²⁺ increase by CMZ, N = 69. The medium was Ca²⁺-free from the moment of TG application. No influence of CMZ was observed in a similar experiment with the Ca²⁺-containing medium (n = N = 48). bn = 21, N = 53, RF = 43.4%, 100 µM ryanodine (Rya). сn = 42, N = 68, RF = 63.2%, 400 nM XeC. d 100 µM 2-АРВ, N = 38, 46 and 58, RF = 92.5 ± 5.1, OF = 39.6 ± 6.0, with P < 0.01 and P > 0.05, correspondingly, vs the data in Fig. 1. Experiments with the same 2-АРВ addition prior to 10 µM phenylephrine gave RF = 0.011 ± 0.011 (N = 48, 56 and 58), P < 0.05 compared to RF = 0.772 ± 0.122 without 2-APB (N = 36, 56 and 64). e 300 nM Baf A1, n = 18, N = 52, RF = 44.2%

The most recognized routes of stored Ca²⁺ mobilization are IP₃Rs and RyRs of ER [1, 2]. In white adipocytes, RyRs can be activated by an agonist, with their participation being revealed by ryanodine application [27]. Preincubation with this drug at the concentration of 100 µM that locks the channel in a closed state [37] did not eliminate the CMZ-induced oscillations (Fig. 5b). Oscillations also retained (Fig. 5c) in the presence of xestospongin C (XeC), inhibiting IP₃R with IC₅₀ = 358 nM [38]. Simultaneous action of ryanodine and XeC did not lead to any attenuation of oscillations (n = 45, N = 68, 500 nM XeC, 100 µM ryanodine) as in the cases of individual treatment. Multiple Са²⁺ spikes were retained with this mixture in Ca²⁺-free medium (two experiments with 400 or 500 nM XeC, n ≥ 24), suggesting no role of IP₃Rs and RyRs in short-lived oscillations that arise without Ca²⁺ entry.

At the same time, 400–500 nM XeC applied here may be insufficient for a strong IP₃R suppression, whereas higher doses can inhibit SERCA [39]. For this reason, we complementarily used 2-APB known to inhibit IP₃R [32]. Its efficiency was proved by incubation of cells with 2-APB prior to phenylephrine, an agonist acting through IP₃ production. Such preincubation caused a drastic drop in cell responsiveness to phenylephrine, but did not inhibit CMZ-induced responses: RF even increased due to transients, while OF was unchanged (Fig. 5d). Since 100 μM 2-APB inhibits SOCE [32], our data with the use of 2-APB also indicate that this Ca²⁺ entry pathway is not activated. Remarkably, ARC channels are unaffected by 100 μM 2-APB [40], and their participation, therefore, remains possible.

As to acidic stores, it was previously shown, that 30 s preincubation of white adipocytes with V-ATPase inhibitor, bafilomycin A1 (Baf A1) at 300 nM concentration, eliminated Са²⁺ increase caused by exogenous NAADP [30]. As seen from Fig. 5e, preincubation with the same concentration of Baf A1 for 60 s did not prevent oscillations evoked by CMZ. Oscillations were also resistant to Baf A1 when its concentration was increased to 1 µM (n = 40, N = 63), or preincubation time was extended to 375 s (n = 49, N = 75). Cell-permeable antagonist of NAADP, NED-19, added 12 min before CMZ at 100 µM concentration that inhibits calcium spiking in pancreatic β-cells [41], had no effect either (n = 27, N = 47).

In adipocytes, calcium signals can be created via activation of CaM-dependent NO-synthase (eNOS), followed by activation of ADP-ribosyl cyclase [28] synthesizing cADPR, NAADP and some other Ca²⁺ active metabolites [12]. To exclude periodic modulation of Ca²⁺ channels by this system, we used the eNOS inhibitor, L-NAME. This inhibitor at 1 mM concentration if applied 6 min before CMZ in Ca²⁺-free medium together with 100 µM ryanodine and 400 nM XeC, did not influence the CMZ-induced oscillations (n = 34, N = 66). Therefore, the oscillations are generated by some kind of intracellular channel that is distinct from RyR, IP₃R and acidic store channels, and is independent of NO.

Connexin and pannexin channels

Connexin and pannexin channels are regulated by several factors including Ca²⁺, with possible involvement of СаМ [10, 13], so they could participate in Ca²⁺ responses to CMZ. We assessed the role of these channels with carbenoxolone (CBX) and octanol. Non-selective inhibitor CBX affects the channels that possess Cx26, Cx30, Cx32, Cx43, Cx46 and Pannexin1 [13]. The effect of СВХ was time dependent. After a short preincubation with СВХ, cells responded to CMZ by usual oscillations (Fig. 6a). Their premature cessation, characteristic of Ca²⁺ entry abolishment, did not happen, that is the Са²⁺ entry component dependent on PM hemichannels was not considerable. CBX with Ca²⁺-free medium allowed for CMZ-induced oscillations (see the legend to Fig. 6a), showing explicitly that their intracellular source was not inhibited by the short-term action of CBX. Longer preincubation (35 min) restrained oscillations in the following way: most cells displayed a single initial pulse (Fig. 6b, black trace), and the others demonstrated the same pulse, which, after a certain period of silence, was accompanied with oscillations, mainly with a low amplitude (Fig. 6b, grey trace). Two-hour preincubation with СВХ (Fig. 6с) did not significantly lower cell responsiveness, while oscillations were strongly inhibited (P < 0.0001). The influence of CBX was not caused by the inhibition of Ca²⁺ entry supporting oscillations, since there were no oscillations just after the first peak, in contrast to their existence in the Ca²⁺-free medium. To verify and specify these results, we also applied octanol that inhibits Cx38, Cx43, and Cx50, with no reported action on pannexins [13]. Two-hour preincubation with octanol gave the same results as with CBX. Namely, responsiveness to CMZ was retained suggesting the fullness of the CMZ-sensitive stores, and the responses had the form of transients without oscillations (Fig. 6d). Subsequent addition of TG typically increased the level of Ca²⁺ (N = 35 and N = 41), revealing non-empty stores. We supposed, therefore, that the suppressive effect of prolonged preincubation with CBX or octanol was due to the postponed inhibition of connexons in the intracellular structures. Remaining CBX-independent initial spikes are evidently of another origin related to activation of phospholipases (see above).

Fig. 6

Responses to CMZ after preincubation of adipocytes with CBX and octanol. Cells were exposed to 100 µM CBX or 1 mM octanol as indicated (a), or starting from 35 min (b) and 2 h (c, d) before time zero. an = 12 (1 stopped), N = 47, RF = 34.0%. In another experiment with Ca²⁺-free medium applied together with CBX, and CMZ added 156 s later, oscillations occurred with n = 30 (10 stopped), N = 64, RF = 50.0%. bn = 60 (black trace), n = 39 (gray trace), percents at the traces relative to 100 responding cells, one of which demonstrated delayed oscillations without the initial pulse. N = 110, RF = 90.9%. c Example typical of 3 experiments, n = 11, 15 and 16, N = 31, 39 and 39. Oscillations remained in 7 cells, including 5 cells with small amplitudes. RF = 44.3 ± 4.7% (P > 0.05) and OF = 6.0 ± 3.7% (P < 0.0001), where P is determined relative to the control in Fig. 1. dN = 35, 41 and 43, RF = 61.8 ± 17.7, OF = 6.2 ± 2.7 (P < 0.0001 vs Fig. 1)

Possible ATP secretion through PM hemichannels causing the stimulation of P2Y₂ receptors and Ca²⁺ increase in adipocytes [42] prompted us to test the contribution of this pathway. Cell treatment with 5 U/mL hexokinase before CMZ had no effect on both the initial Ca²⁺ release and subsequent oscillations (N = 63), implicating that such ATP secretion plays no role in the observed responses.

Effect of La³⁺

Lanthanides La³⁺ and Gd³⁺ block channels for Са²⁺ entry into the cell, affecting SOC and, at higher concentrations, non-SOC channels [5, 32, 36]. In particular, it was demonstrated that ARC channels [40, 43] and Сх43 hemichannels [44] can be inhibited by these ions, which gave us grounds to expect an action covering both AACOCF₃ and CBX/octanol effects. La³⁺does not permeate the cells and blocks Са²⁺ entry by acting extracellularly [45], so we supposed that La³⁺ could be used to reveal the role of the PM Ca²⁺ channels located in endocytic vesicles and in organelles to which these La³⁺-containing vesicles are fused. Taking into account the characteristic time of the endocytic transport that can reach 1–2 h [46] and the obtained development of CBX inhibition on the same time scale, preincubation of adipocytes with La³⁺ was performed for 30 min, two and 24 h.

Incubation with 1 mM La³⁺ for 30 min resulted in partial deterioration of oscillations, as exemplified by a complex transient in Fig. 7a. Although responsiveness decreased insignificantly, oscillations were sensitive to La³⁺ inasmuch as the fraction of oscillations in responses decreased by 1/3 with P < 0.001. Only about 1/4 of the remaining oscillations were rather regular and prolonged. Subsequent addition of TG to the Ca²⁺-free medium evoked Ca²⁺ rise in all tested cells (N = 42 and 45) with an amplitude comparable to the foregoing peaks, similar to those shown in Fig. 3a. Prolongation of the preincubation time to 2 h led to strong inhibition of Ca²⁺ responses (Fig. 7b). Response and oscillation indices decreased by 74.3% (Р<0.01) and 86.9% (Р<0.0001), respectively. This effect was not a result of depletion of Ca²⁺ stores, since subsequent TG addition together with Ca²⁺-free medium again produced usual response amplitudes (not shown).

Fig. 7

Treatment of adipocytes with La³⁺ for increased time periods suppresses Са²⁺ responses to CMZ without reducing Са²⁺ stores. Preincubation with 1 mM La³⁺ for 30 min (a), 2 h (b) and 24 h (c and d). 10 µM TG and 1 µM ionomycin (Iono) were used. P values in (a, b) are determined relative to the data in Fig. 1. a A representative response deviating from typical oscillations. In three experiments, n = 64 (any responses), N = 42, 45 and 59, RF = 44.5 ± 6.5% (P > 0.05), OF = 28.3 ± 6.8% (Р<0.005), OF/RF = 63.8 ± 2.4 (P < 0.001). b Traces of 25 out of the 59 non-responding cells (for clarity, the remaining traces are not presented) and only one response in the experiment with N = 60. In all 4 experiments, N = 23, 29, 33 and 60, RF = 14.7 ± 7.5 (Р<0.01) and OF = 7.1 ± 4.0 (Р<0.0001). c Time course of the mean Ca²⁺ level, based on 3 experiments with N = 36, 37 and 39, where only 2 cells responded to CMZ, both with oscillations. d Resting Ca²⁺ level (Rest), RF for CMZ (CMZ effect), and the amplitude of the TG-evoked responses (TG effect) for experiments of panel (c) in comparison with control without La³⁺. *P < 0.05. Values are normalized to the mean in the control (N = 31, 40, 42), for which RF = 41.8 ± 7.2%, the fura-2 ratio at rest is 0.473 ± 0.013, and the amplitude measured from the averaged (including oscillations) level is 0.0577 ± 0.0113, the same (P > 0.05) as the amplitude in Fig. 3a

To estimate the maximal La³⁺ effect and to exclude emptying of stored Ca²⁺ more reliably, the culture was incubated with La³⁺ for 24 h. The preincubation did not change vitality of cells and their morphology. Like with a shorter 2-h preincubation (Fig. 7b), the La³⁺ presence during 24 h inhibited responses to CMZ almost completely, leaving them in only 2 of 112 analyzed cells (Fig. 7c). Importantly, subsequent responses to TG and ionomycin, applied in Ca²⁺-free medium, retained, demonstrating the fullness of the Ca²⁺ stores. Statistical comparison showed that Ca²⁺ responsiveness to CMZ dropped at P < 0.05, while the TG-induced response amplitude even increased, although insignificantly (Fig. 7d). Thus, the lack of signals after the long-lasting La³⁺ presence is not a consequence of reduced internal stores determining the Ca²⁺ release by CMZ. From Ca²⁺ decline in Fig. 7c, it сan be deduced that PMCA was not blocked by La³⁺. Note that 24-h exposition of adipocytes to La³⁺ slightly decreased (P < 0.05) the fluorescence intensity of unstimulated cells (Fig. 7d). Consequently, La³⁺ able to change the fura-2 fluorescence excitation spectrum similar to Ca²⁺ [45] did not permeate into the cytosol and did not accumulate in it during this period, thus indicating the organellar localization of La³⁺ and its targets. Despite the possible interaction of La³⁺ with multivalent anions in the solution [32], its amount was de facto enough to inhibit Ca²⁺ rises.

It was found that endocytosis is negligible below 10 °C [47]. Given this, low temperature should block La³⁺ delivery to intracellular structures and restore Ca²⁺ responses to CMZ. Storage of adipocytes at 4 °C for 1 h (not more) preserved cell vitality, and RF = 90.2% and OF = 60.3% were obtained for La³⁺-treated cells (the mean of two experiments with N = 30 and N = 47). These values are 2 and 2.1 times higher than the corresponding values for the cells incubated with La³⁺ at normal temperature during even a shorter period of time (30 min, Fig. 7a). We consider this increase of responsiveness as a support of the proposal that CMZ-induced responses depend on organelle-located PM calcium channels accessible to the inhibitor by endocytosis.

Discussion

Our experiments revealed that CMZ interfering with CaM-dependent regulation of Ca²⁺ fluxes initiates persistent oscillations of the cytosolic Ca²⁺ concentration by activating both Ca²⁺ release and entry. The oscillations existed in a rather wide range of 2.5–10 µM CMZ, with no responses in any cell at 1 µM and with reduction of this mode at 10 µM. The same threshold action of low CMZ concentrations on the Ca²⁺ level was observed in other cell types, with similar [17, 18, 20] or higher sensitivity [22] to CMZ. Decaying oscillations in response to 10 µM CMZ were previously obtained in HeLa cells, but only in Ca²⁺-free medium, leading to the conclusion that Ca²⁺ entry did not support, but rather compromised oscillations [20]. Our results with and without Ca²⁺ in the medium have shown that Ca²⁺ entry positively influences the intracellularly generated oscillations by maintaining them.

Phospholipases PLA₂ and PLC, both capable to supply AA had different effects on responses to CMZ. The early Ca²⁺ release was provided mainly by PLA₂, but the subsequent oscillations were partially dependent on PLC and not on PLA₂. The ability of exogenous AA to evoke Ca²⁺ release [20], non-SOCE activation [20, 36, 48] and Ca²⁺ response in adipocytes [25], as well as CMZ action not requiring AA-derived metabolites [20], suggests that effects of phospholipases in our data implicated direct activation of ARC channels by AA. Nevertheless, the action on the same channels through LTC₄ [5] is not excluded. This activation is in line with the observed here attenuation of responses to CMZ after incubation with La³⁺ inhibiting ARC [43] among other channels. Lowering the AA level by PLA₂ inhibition resulted in oscillations emerging with a delay, which could reflect weak AA-induced Ca²⁺ release and slow (within minutes) development of the CMZ effect on connexons responsible for oscillatory Са²⁺ release. The influence of PLC on Ca²⁺ responses was not related to the IP₃-mediated Ca²⁺ release, because no IP₃R impact was detected, and early Ca²⁺ rise strongly depended on PLA₂ functioning. This suggests the involvement of the second PLC product, diacylglycerol (DAG). Upon DAG accumulation, AA can be produced by DAG lipase, resulting in activation of non-SOCE [36]. Given that PLC shortened the initial lag in adipocytes, although less efficiently than PLA₂, it is possible that CMZ released Ca²⁺ partially due to the AA formation by DAG lipase. The subsequent oscillations seem to depend on DAG-derived AA, with the much less contribution of AA made at this phase by PLA₂. The absence of BEL effect, consistent with insignificant expression of iPLA₂ in adipocytes [35], implies that CMZ could activate AdPLA, conceivably by antagonizing its inhibition by CaM.

No influence of SOC channels on Ca²⁺ responses to 5 µM CMZ was found here using YM-58483 or 2-APB, so the entire entry was represented by non-SOCE. This fact is likely a consequence of insignificant decrease in stored Ca²⁺ content indicated by subsequent full-size TG responses. Moreover, our data agree with vast predominance of non-SOCE at [CMZ] ≥ 5 µM [17, 19, 20, 48] and SOCE attenuation by CMZ [15], probably caused by accumulation of AA that activates non-SOCE and inhibits SOCE [48]. Since there are evidences for YM-58483 [49] and 2-APB [50] action on several TRP family channels, our negative results with these inhibitors argue against the contribution of this channel type to Ca²⁺ entry and oscillations initiated by CMZ.

Non-involvement of IP3R or RyR in our results complies with Ca²⁺ release by CMZ at inhibited PLC [22] and atypical CICR not requiring these channels [51] found in living cells. In search for an alternative source of the CMZ-induced oscillations, our key finding was that sufficiently prolonged incubation of the culture with La³⁺, inhibiting Ca²⁺ entry channels, prevented Ca²⁺ responses, including Ca²⁺ release. We interpret this deferred action as a non-permeable inhibitor reaching its organelle-located targets through endocytosis. The subcellular presence of PM channels typically serving for Ca²⁺ entry is exemplified by Orai1, whose constitutive recycling between PM and endosomes provides a considerable percentage of total Orai1 in these organelles [52]. According to our results, neither PM-located, nor organellar Orai1-containing SOC channels play a role in the CMZ-induced responses, as seen from the incubation with YM-58483 up to 2 h. This drug does not permeate through the cell membrane during 24-h preincubation [31], but it could be uptaken and passed to organelles like La³⁺. Therefore, La³⁺-sensitive organellar channels distinct from CRAC channels were involved, which could be ARC channels and connexons. It is important that even 24-h incubation with La³⁺ did not empty the intracellular Ca²⁺ pools, thus rejecting the unresponsiveness of adipocytes as a simple consequence of this pool loss. This resistance to depletion may stem from La³⁺-inhibition of virtually all Ca²⁺ entry channels, including those in organelles, together with lowering the PMCA activity at the normal resting Ca²⁺ level and below it.

Intracellular connexons seem to be most closely related to the mechanism of oscillations, since their inhibition by prolonged incubation with CBX or octanol effectively abolished oscillations leaving the initial transient, and inhibition by La³⁺ suppressed all responses, including oscillations. Similar to our data, 30-min preincubation with CBX suppressed bradykinin-induced oscillation, but left the initial peak in MDCK cells expressing Cx32 and Cx43 [8]. It was concluded that hemichannels contribute to the IP₃R-based oscillations [8], and involvement of Cx43 hemichannels was specified [9]. Cx43 channels are common targets for CBX and octanol [13], and they are also inhibited by La³⁺ [44], so the three lines of evidence converge on the Cx43 isoform to be responsible for Ca²⁺ oscillations studied here. These connexons could transport an intracellular messenger triggering oscillations through another Ca²⁺ channel, but our data ruled out participation of evident candidates, such as NAADP, acidic stores or IP₃R and RyR. On the other hand, the bell-shaped Са²⁺ influence on the Сх43 hemichannel opening was shown based on ATP release, and such Ca²⁺ effect was also observed at the single channel level in the patch-clamp experiments [10]. We propose, therefore, a more straightforward possibility of CICR arising from Ca²⁺ release through organellar connexons and activation of just these channels. The bell-shaped Са²⁺ dependence of Cx43 connexons can be sufficient for the steady state instability and appearance of oscillations, as shown for IP₃R [53] with the same dependence character. CMZ presumably initiates oscillations by evoking the production of AA that releases Ca²⁺ through AA-activated channels causing CICR through intracellular connexons, and AA may also promote [10] the opening of connexons. The lag preceding oscillations may reflect the time required for the subthreshold accumulation of activating Ca²⁺, AA or AA-derived intermediates. The rapid cessation of Ca²⁺ oscillations after blocking the hemichannels [7,8,9], that did not occur in our results, is consistent with a minor role of PM hemichannels in the cultured adipocytes in respect to CMZ-evoked Ca²⁺ changes. The processes able to account for the data of the present work are summarized in Fig. 8.

Fig. 8

Scheme of CaM-mediated regulation of Ca²⁺ fluxes in adipocytes inferred from the obtained results. One generalized compartment with SERCA and both Ca²⁺-releasing channels is provisionally shown, but these channels may also be located in different organelles. Dashed arrows with sharp and T-shaped ends stand for activation and inhibition, which can proceed via intermediate steps. PIP₂: phosphatidylinositol 4,5-bisphosphate, PL: phospholipid, LPL: lysophospholipid. All other abbreviations and explanations are given in the text

As for organellar identity of Ca²⁺ pools that generate oscillations under CMZ action, several versions can be considered. The dependence of responses on SERCA pumping points to the SERCA-possessing organelles, in particular ER, as the main source of CMZ-releasable Ca²⁺. Meanwhile, the effect of TG can also be indirect, in the sense that a massive TG-induced Ca²⁺ release could merely discharge other organelles not equipped with SERCA, and therefore, prevent subsequent Ca²⁺ rises by CMZ. A candidate for the generator role is the Golgi apparatus that contains SERCA in the cis and medial sections [54], and also incorporates oligomerized connexins derived (as monomers in the case of Cx43) from ER [11]. Trans-Golgi and secretory vesicles, accumulating Ca²⁺ via the secretory pathway Ca²⁺-ATPase, can possess RyRs allowing for CICR [54], so these stores may represent the case of indirect TG dependence. Trafficking of the PM-derived endosomes could explain the long delay in the effect of Ca²⁺ entry inhibitors as the time needed to deliver the extracellular solution through the vesicles to the intracellular structures capable of oscillations. In this process, the endosomal cargo can reach the trans-Golgi network owing to bidirectional vesicle exchange [55] and then pass to other compartments. Endosomes themselves contain internalized Cx [11], and probably, ARC channels, although their Ca²⁺ is lost within minutes to few µM, in parallel with acidification [56]. Specialized GLUT4 storage vesicles in adipocytes may contain V-ATPase, and therefore, belong to acidic compartments [57], apparently not involved in the present study. Nevertheless, Ca²⁺ transport in these vesicles should be characterized in more detail, and their role in the responses is still possible.

CaM antagonists have some non-specific effects not mediated by CaM, of which direct inhibition of SERCA [58] might be the most relevant to our data. However, disparate IC₅₀ values for CMZ and TFP were reported [58] in contrast to that similar oscillations were observed here at the same concentrations of either CMZ or TFP. Besides, SERCA was not inhibited by CMZ judging from the conserved Ca²⁺ stores after CMZ action.

Targeted inhibition of CaM has shown that this protein is essential for cell viability and proliferation. Accordingly, CaM antagonists arrest proliferation and induce apoptosis of tumor cells, which makes promising their use for therapeutic purpose. Anti-CaM drugs could also exert protective effect on non-transformed cells, such as neurones, cardiomyocytes, β-cells, and cells of the immune system, in cases when CaM has proapoptotic function [3]. Since Ca²⁺ overload can cause cell death by various ways [1, 3], Ca²⁺ behavior under such treatments becomes an important factor. We showed that, starting at a critical dose, CaM antagonist can destabilize Ca²⁺ homeostasis and produce oscillations. This regime is not necessarily deleterious for cells, because the Ca²⁺ rises are of short duration. Cytotoxicity could be related rather to large Ca²⁺ transients, caused by the abrupt drug action on the cell, as well as to strong permanent Ca²⁺ elevations because of the excessive stimulation of PLA₂, PLC and downstream ARC and Cx channels. Therefore, our results can be useful in developing strategies for selection and administration of anti-CaM drugs, possibly in combination with agents reducing AA impact, to control viability and correct abnormal functions of adipocytes and other cells.

In conclusion, CaM mediates regulation of the Ca²⁺ level in white adipocytes, and alleviation of this regulation leads to Ca²⁺ transients and sustained oscillations. The underlying mechanism implicates control of AA-activated and connexin channels by phospholipases A₂ and C via their products (AA, DAG) distinct from IP₃. Organellar connexons appear to play an essential role in the oscillations and may generate them due to own CICR. Operation of this phosholipase-triggered mechanism may be not limited by the conditions of weakened CaM regulation, and its involvement in agonist-induced Ca²⁺ signaling is plausible.