Introduction

The endoplasmic reticulum (ER), represented by a three-dimensional intracellular network of tubules and cisternae, serves as an integrating signalling organelle, which co-ordinates fast physiological Ca2+ signalling and long-lasting adaptive responses controlled by post-translational protein processing within the ER lumen [2, 4, 6, 9, 14]. Many ER-resident chaperones responsible for correct folding of proteins are regulated by the concentration of free Ca2+ within the ER lumen ([Ca2+]L), and therefore the latter bestows the link between fast physiological events and protein turnover [28, 51]. Fluctuations of [Ca2+]L are determined by the balance between Ca2+ release and Ca2+ uptake.

Two types of ligand-gated Ca2+ channels, the inositol-1,4,5-trisphosphate receptors (InsP3Rs) and ryanodine receptors (RyRs), provide the route for Ca2+ release and underlie the excitability of the ER membrane [3, 32, 36, 47]. Both types of Ca2+ release channels are abundantly expressed in nerve cells [50, 52], although their intracellular distribution shows considerable heterogeneity. That is, dendritic spines of Purkinje neurones are rich in InsP3Rs but are devoid of RyRs, although the latter are found in quantity in the dendritic shafts and in the cell body [27, 37]. In contrast, RyRs are predominant in dendrites of CA1 hippocampal neurones [38]; they are also preferentially expressed in axons and synaptic terminals of cerebellar basket neurones [25]. In agreement with such peculiar distribution, different types of Ca2+ release are activated upon physiological stimulation in distinct neuronal sub-compartments. Stimulation of synaptic inputs triggers InsP3-mediated Ca2+ release in the spines of Purkinje cells [13, 46], whereas Ca2+-induced Ca2+ release generated through RyRs plays an important role in postsynaptic Ca2+ signalling in hippocampus ([1, 12] but see [24]) and controls multivesicular neurotransmitter release in cerebellar synaptic terminals [25].

Therefore heterogeneous localization of Ca2+ release channels provides for a spatial control of Ca2+ signals, which is particularly important for highly polarized nerve cells. Yet, such a heterogeneity does not implicitly entail the existence of separate Ca2+ pools associated with different Ca2+ release mechanisms. This particular issue recently became a matter of controversy [5, 34].

Experiments on neuronal preparations have shown that depletion of RyR-sensitive Ca2+ store in Purkinje neurones completely abolished responses to photoreleased InsP3 suggesting that both receptors share the same interconnected Ca2+ pool [21]. Similar overlap between InsP3-sensitive and caffeine-sensitive Ca2+ pools was suggested for hippocampal [19], cerebellar granule [20, 39] and cultured myenteric [23] neurones. In contrast [Ca2+]i imaging in adrenal chromaffin cells revealed two distinct Ca2+ pools sensitive to caffeine and InsP3 respectively [8]. Finally the existence of separate Ca2+ pools in nerve cells was recently suggested by Blaustein and Golovina [5], who based their theory on a direct measurements of [Ca2+]L in astrocytes and atrial myocytes [15, 16, 17]. In the present paper we addressed the question of internal continuity of the ER Ca2+ pools in mammalian sensory neurones using direct monitoring of [Ca2+]L. Our evidence suggests that in this preparation the InsP3Rs and RyRs share the same functional Ca2+ pool.

Materials and methods

Real-time imaging of Ca2+ concentration in the store

Dorsal root ganglion neurones were enzymatically isolated from new-born (1–3 days old) Sprague-Dawley rats using a conventional treatment with 0.1% protease (type XIV) in HEPES-buffered MEM for 8 min at 37°C. Individual cells were separated mechanically and plated on poly-l-ornitine (1 mg/ml) and laminin (0.01 mg/ml) covered glass coverslips. Neurones were maintained in culture media (DMEM, supplemented with 10% horse serum, 50 U/ml penicillin/streptomycin mixture and 6 μg/ml insulin) at 37°C in an atmosphere of air supplemented with 5% CO2 for 1–2 days prior to the experiment. In the present study we investigated only large (proprioceptive) neurones with somas larger than 35 μm in diameter.

For [Ca2+]L recordings we have used Mag-Fura-2 (K D~50 μM) suitable for detecting high intraluminal [Ca2+] levels [7, 31, 42]. The neurones were incubated with 5 μM Mag-Fura-2 for 30 min at 37°C, and washed at 37°C for 1 h prior to the experiment (loading at 37°C promotes dye compartmentalization within the ER lumen [49]).

The cytoplasmic portion of dye was removed either via intracellular dialysis through the patch pipette (as described in our previous paper [43]) or by permeabilization of the plasmalemma with saponin [41, 42]. In the latter case the cellular membrane was permeabilized by brief (7–10 s) application of saponin (0.001%) in "intracellular" solution. The permeabilization technique was used exclusively for probing the neurones for InsP3-induced Ca2+ release as InsP3 does not penetrate through intact plasmalemma.

Calibration of Mag-Fura-2 signals

The [Ca2+]L values were calculated using the 340/380 nm ratio with the equation [Ca2+]L=K*(R−R min)/(R max−R). R min, R max and K* were determined using exposure of Mag-Fura-2/AM loaded intact neurones or saponin-permeabilized neurones to 20 μM ionomycin and four calibrating solutions with [Ca2+]<10 nM (10 mM EGTA); 100 μM; 400 μM and 10 mM; solutions were prepared as described previously [43]. The calibration procedure on permeabilized cells consistently yielded higher (~40%) values for K*, most likely due to a higher Mag-Fura-2 K D within the ER lumen. As we assumed this being more accurate, the values from the latter procedure were used throughout. Values of R min, R max and K* were 0.3, 1.9 and 287 μM respectively.

Real-time video-imaging

Fluorescence images were captured using an Olympus IX70 inverted microscope (40× UV objective) equipped with a charge-coupled device (CCD) cooled intensified camera (Pentamax Gene IV, Roper Scientific, UK). The specimen was alternately illuminated at 340, 380 and 488 nm by a monochromator (Polychrom IV, TILL Photonics, Germany) at a cycle frequency 0.5–5 Hz. Control over the experiment, image storage and off-line analysis was performed by use of MetaFluor/MetaMorph software (Universal Imaging Corporation, USA) running on a Windows 98 workstation.

Electrophysiology and solution exchange

Whole-cell recordings were made by using EPC-9 amplifier run by the PC-based PULS software (both from HEKA, Germany). The pipette resistance was 3–5 MΩ. All solutions were applied using a fast local superfusion technique [53] which ensured complete exchange of the milieu surrounding the cell within 100 ms.

Solutions and reagents

The extracellular bathing solution contained (in mM): NaCl 135, KCl 3, CaCl2 2, glucose 20, HEPES/NaOH 20, pH 7.4. The Ca2+-free solution contained 5 mM EGTA with no CaCl2 added. The "intracellular" solution used in permeabilization experiments contained (in mM): KCl 140, Na2ATP 3, MgCl2 2, CaCl2 0.4, BAPTA 5, HEPES/KOH 20, pH 7.2, free Ca2+ concentration ~70 nM). The intra-pipette solution used for intracellular dialysis contained (in mM): CsCl2 122, TEA-Cl 20; Na2ATP 3, HEPES/CsOH 10, EGTA 0.1, pH 7.3. All reagents were purchased from Sigma (Dorset, UK), and fluorescent probes were obtained from Molecular Probes (Ore., USA).

Results

Sensory neurones co-express functional InsP3Rs and RyRs

We monitored intraluminal Ca2+ dynamics in single sensory neurones using low-affinity Ca2+ probe Mag-Fura-2 compartmentalized within the ER lumen. The cytosolic portion of the dye was removed either by intracellular dialysis under whole-cell patch-clamp configuration [43], or by permeabilization of the plasmalemma by brief application of saponin [41, 42]. We used the latter technique to permit direct activation of InsP3Rs by InsP3, which otherwise cannot penetrate through the cell membrane.

The resting [Ca2+]L determined with both techniques varied between 100 and 400 μM. Since [Ca2+]L is an important determinant of the velocity and magnitude of Ca2+ release [43] we restricted our analysis to neurones with [Ca2+]L higher than ~300 μM.

Brief extracellular applications of both InsP3 (3–10 μM; 10 s) and caffeine (20 mM, 5 s) to permeabilized neurones triggered transient fall in [Ca2+]L, which recovered to the pre-stimulated level after washout (Fig. 1). We found these responses in all neurones subjected to such an application protocol (n=12). The InsP3-induced [Ca2+]L decrease was substantially slower as compared to that induced by caffeine: on average maximal velocity of [Ca2+]L decrease induced by InsP3 was about three times lower as compared with [Ca2+]L decrease induced by caffeine (12±6 μM/s versus 41±10 μM/s, respectively; n=12, see also Fig. 3). Quite obviously the depletion of ER store by InsP3 required much more time as compared with caffeine (see for example Fig. 5). Such a difference may result from either lower density of InsP3Rs or from denial of Ca2+-induced potentiation of IICR in heavily buffered "intracellular" solution in permeabilized experiments.

Fig. 1A, B.
figure 1

InsP3Rs and RyRs coexist in DRG neurones. A The endoplasmic reticulum lumen [Ca2+] ([Ca2+]L) recording (top trace) and its first derivative (bottom trace) obtained from permeabilized DRG neurone pre-loaded with Mag-Fura-2/AM. The neurone was challenged by 10 μM InsP3 and 20 mM caffeine as indicated on the graph. B The [Ca2+]L recordings from another permeabilized DRG neurone alternately treated with 10 μM inositol-1,4,5-trisphosphate (InsP3) and 20 mM caffeine as indicated on the graph

Full size image

Both agents, caffeine and InsP3, triggered [Ca2+]L decrease in a dose-dependent fashion (Fig. 2). The responses to InsP3 saturated at 10 μM, i.e. at a rather high concentration, which might reflect either hampered diffusion of InsP3 towards ER membrane even in permeabilized preparation, or generally lower sensitivity of neuronal InsP3Rs to InsP3 [22]. Responses to caffeine saturated at concentrations higher than 10 mM. Therefore, these experiments demonstrate that ER membrane in DRG neurones possesses both InsP3-induced and Ca2+-induced Ca2+-release mechanisms (IICR and CICR, respectively).

Fig. 2A–D.
figure 2

Dose-dependence of InsP3-induced and caffeine-induced Ca2+ release in DRG neurones. A The changes in [Ca2+]L measured from permeabilized DRG neurone in response to applications of increasing concentrations of InsP3. B The concentration dependence of InsP3-induced decrease in [Ca2+]L. The data are mean±SD from 5 cells. C The changes in [Ca2+]L measured from dialysed DRG neurone in response to applications of increasing concentrations of caffeine. D The concentration dependence of caffeine-induced decrease in [Ca2+]L. The data are mean±SD from 5 cells

Full size image

Inhibition of SERCA pumps and direct activation of RyR deplete the Ca2+ store

The ability of the ER Ca2+ store to generate Ca2+ signals is regulated by intraluminal free Ca2+ concentration, so that store depletion prevents development of Ca2+ release. We have already demonstrated that the store replenishment following CICR (induced either by caffeine or by Ca2+ entry) is determined by thapsigargin (TG)-sensitive sarco(endoplasmic)reticulum Ca2+-ATPase (SERCA) pumps [43]. Here we investigated the mechanisms of store depletion in greater detail. Conceptually the stores can be depleted either by inhibition of Ca2+ uptake, which leaves the resting Ca2+ leakage unopposed, or by stimulation of Ca2+ release channels. In DRG neurones the inhibition of SERCA pumps by both TG (5 μM; n=14) and cyclopiazonic acid (CPA, 50 μM; n=11) triggered a progressive decrease in [Ca2+]L which stabilized after complete depletion of the store (Fig. 3A, B). Alternatively the stores can be depleted by activation of the Ca2+-release route. For this purpose we incubated the neurones with a low (0.5 μM) concentration of ryanodine. At this concentration the latter is known to promote opening of RyRs either by direct interaction with gating mechanisms [40] or by a dramatic (~1,000-fold) increase in RyRs sensitivity to cytosolic Ca2+ [26]. As shown in Fig. 3C, application of 0.5 μM of ryanodine effectively decreased [Ca2+]L, thus indicating the depletion of the ER store. The maximal velocities of [Ca2+]L decrease initiated by various pharmacological agents (caffeine, InsP3, ryanodine, TG and CPA) are compared in Fig. 3D.

Fig. 3A–D.
figure 3

Activation of RyRs and inhibition of SERCA pumps deplete the Ca2+ store. A–C Examples of [Ca2+]L recordings from Mag-Fura-2 pre-loaded dialysed DRG neurones in response to extracellular application of thapsigargin (TG), cyclopiazonic acid (CPA) and ryanodine (rya). D Average values (mean±SD, n is indicated on the graph) of maximal rate of [Ca2+]L decrease induced by 10 μM InsP3, 20 mM caffeine, 0.5 μM ryanodine, 5 μM TG and 50 μM CPA

Full size image

RyRs and InsP3Rs share the common pool sensitive to TG and CPA

The experiments described above have evidently demonstrated that the ER in DRG neurones is endowed with functional InsP3Rs, RyRs as well as with TG and CPA-sensitive SERCA pumps. These data enabled us to address the central question of this study, i.e. whether all these mechanisms operate within a single Ca2+ pool or the ER Ca2+ store is represented by several independent compartments. For this purpose we utilized the ability of the agents described above to deplete the ER store. After Ca2+ pool was depleted by one of these agents, we tested the ability of others to initiate a further decrease in [Ca2+]L. The appearance of such a decrease would indicate the coexistence of separate Ca2+ pools.

First we tested the ability of TG and CPA to deplete the caffeine-sensitive Ca2+ pool. As shown in Fig. 4A, dialysed neurones were initially challenged with caffeine to probe for the existence of a caffeine-sensitive pool. Subsequently the neurones were incubated with 5 μM TG which resulted in a decrease in [Ca2+]L. After complete depletion of the pool (as was judged by stabilization of [Ca2+]L in the presence of TG) the cells were challenged with 20 mM caffeine and 50 μM CPA. In all nine neurones exposed to such a protocol neither caffeine nor CPA were able to affect [Ca2+]L after the stores were depleted by TG. Similar results were obtained when Ca2+ pool was initially depleted by 50 μM CPA: both caffeine and TG applied in the presence of CPA failed to affect [Ca2+]L (Fig. 4B, n=8). Likewise, depletion of stores with 0.5 μM ryanodine rendered caffeine, CPA and TG, applied in the presence of ryanodine, totally ineffective (n=7; Fig. 4C, D).

Fig. 4A–D.
figure 4

Ryanodine- and caffeine-releasable Ca2+ pool is sensitive to sarco(endoplasmic)reticulum Ca2+-ATPase (SERCA) inhibition by TG and CPA. All traces represent [Ca2+]L recordings from Mag-Fura-2 pre-loaded dialysed DRG neurones. A [Ca2+]L recording in response to applications of 20 mM caffeine, TG and CPA. Depletion of Ca2+ store by TG completely eliminates responses to caffeine and CPA. B Similarly to A depletion of Ca2+ store by CPA prevents [Ca2+]L responses to 20 mM caffeine and to TG. C, D Activation of RyRs by 0.5 μM ryanodine depletes Ca2+ store and abolishes [Ca2+]L responses to CPA, TG and 20 mM caffeine

Full size image

After completion of these experiments we switched to permeabilized neurones, thus gaining the possibility to test for InsP3-induced Ca2+ release. The cells were initially incubated with 10 μM InsP3, which resulted in a drop in [Ca2+]L. The [Ca2+]L stabilized at a steady-state level after complete depletion of the InsP3-sensitive pool. Application of caffeine performed at this moment failed to further affect [Ca2+]L (Fig. 5A, n=7). Vice versa, when the stores were depleted in the presence of 20 mM of caffeine, application of 10 μM InsP3 did not induce any changes in [Ca2+]L (Fig. 5B, n=6). Despite the clarity of the traces resulting from the protocols described above, doubt remained, as caffeine is known as an effective inhibitor of InsP3Rs [11, 33, 54]. Therefore, we looked for alternative means to specifically activate RyRs, and used 0.5 μM ryanodine for this purpose. Once more, after the Ca2+ pool was depleted by incubation with 10 μM of InsP3, ryanodine was unable to induce any further fluctuations in [Ca2+]L (Fig. 5C; n=5). Likewise, when the permeabilized neurone was treated with 0.5 μM ryanodine to achieve full exhaustion of the Ca2+ pool, neither InsP3 nor caffeine were able to activate any additional Ca2+ release (Fig. 5D, n=5).

Fig. 5A–D.
figure 5

InsP3Rs and RyRs share a common Ca2+ pool in sensory neurones. All traces represent [Ca2+]L recordings from Mag-Fura-2 pre-loaded permeabilized DRG neurones. A Depletion of Ca2+ store by incubation with 10 μM InsP3 completely abolishes [Ca2+]L response to 20 mM caffeine. B Similarly, depletion of the Ca2+ store in continuous presence of caffeine eliminates the InsP3-induced Ca2+ release. C Depletion of Ca2+ store by incubation with 10 μM InsP3 completely abolishes [Ca2+]L response to ryanodine. D When the Ca2+ store was depleted by 0.5 μM ryanodine neither InsP3 nor caffeine (20 mM) were able to produce further Ca2+ release

Full size image

It has to be noted that SERCA inhibitors, caffeine and InsP3 all decreased [Ca2+]L to the same, relatively high residual level. As we have demonstrated before [43] application of ionomycin in Ca2+-free extracellular solution following TG and/or CPA treatment reduced [Ca2+]L to zero (R min). This high residual [Ca2+]L can reflect a limitation of the method, suggesting that part of the signal comes from an intracellular compartment not connected with the caffeine- or TG-sensitive portion of the ER. Nonetheless, as we already discussed in our previous paper [43] it may also represent an intrinsic property of the ER store, when severe depletion of the later may inhibit further release through Ca2+release/k leakage channels.

Discussion

Here we report the first direct measurements of intraluminal Ca2+ dynamics in mammalian neurones during Ca2+ release produced by activation of InsP3Rs and RyRs. We found that these two types of Ca2+ release, the IICR and the CICR, not only coexist in sensory neurones but they share a common Ca2+ pool. Our suggestion contradicts a recent hypothesis postulating the existence of functionally separate ER Ca2+ pools in both excitable and non-excitable cells, including neurones [5]. The idea of separate Ca2+ pools was instigated by experiments employing direct imaging of intra-ER Ca2+ movements in astrocytes and atrial myocytes [15, 16, 17]. These studies concluded that excitable and non-excitable cells have at least two separate Ca2+ pools sensitive to InsP3 and caffeine, respectively. This was based on the finding that inhibition of SERCA pumps by TG and CPA depletes the InsP3-sensitive Ca2+ stores and abolishes responses to metabotropic agonists; however, this manipulation does not affect Ca2+ release triggered by caffeine. Although this finding explicitly implies the existence of a specific Ca2+-uptake pathway replenishing the caffeine-sensitive pool, the authors of the cited studies failed to hypothesize on it.

The experiments described in this paper do not support the hypothesis of separate Ca2+ pools. On the contrary, in a series of direct approaches, we have demonstrated that depletion of Ca2+ stores by either opening of RyRs or InsP3Rs or by SERCA inhibition is always complete, and it precludes any further Ca2+ release irrespective of its mechanism. The most direct evidence for a common Ca2+ pool was obtained from the protocols shown in Fig. 5. These experiments show that depletion of the Ca2+ store due to an activation of RyRs by caffeine or ryanodine completely abolishes InsP3-induced Ca2+ release, and vice versa, when the stores are depleted following InsP3Rs activation, the RyR-mediated Ca2+ release is fully blocked.

Our suggestion of the continuity of the neuronal ER Ca2+ store is in line with a multitude of morphological evidence which describes neuronal ER as a continuous interconnected network [44]. Furthermore, a direct approach aimed at investigating the continuity of the ER in Purkinje neurones with a lypophylic fluorescent dye travelling exclusively in ER membranes [48] has clearly demonstrated the continuity of the ER. Similarly, a wealth of experimental data obtained in non-neuronal cells favours, to a very large extent, the idea of ER continuity. For instance, relatively large molecules, such as ER-targeted GFP, were reported to rapidly diffuse within the ER luminal space [10, 45]. In addition, a series of refined experiments on pancreatic acinar cells [30] have convincingly demonstrated that (1) fluorescent Ca2+ probes can diffuse freely within the ER lumen, and (2) even more importantly that [Ca2+]L rapidly equilibrates within the ER lumen following local photorelease of caged calcium.

The existence of a continuous Ca2+ pool connected through the ER lumen could be very important for neuronal function. First, rapid Ca2+ diffusion through the ER "Ca2+ tunnels" [29] supports Ca2+ release in cell sub-compartments by preventing severe store depletion following intensive local stimulation. Second, the same intraluminal Ca2+ diffusion may be instrumental in conveying Ca2+ signals from distal neuronal processes toward the nucleus [34], as was suggested by recent findings showing the importance of ER Ca2+ uptake in nuclear Ca2+ signalling [18, 35]. Third, intra-ER Ca2+ equilibration could facilitate clearance of local excessive Ca2+ loads. Finally, the existence of a continuous ER Ca2+ store can be very important in guarding against profound store depletion (which may happen more easily in small, separated Ca2+ pools), thus protecting normal functioning of intraluminal chaperones and therefore supporting cell functioning.