Inositol trisphosphate and cyclic ADP-ribose release Ca2+ from the endoplasmic reticulum via inositol trisphosphate and ryanodine receptors, respectively. By contrast, nicotinic acid adenine dinucleotide phosphate may activate a novel Ca2+ channel in an acid compartment. We show, in two-photon permeabilized pancreatic acinar cells, that the three messengers tested could each release Ca2+ from the endoplasmic reticulum and also from an acid store in the granular region. The nicotinic acid adenine dinucleotide phosphate action on both types of store, like that of cyclic ADP-ribose but unlike inositol trisphosphate, depended on operational ryanodine receptors, since it was blocked by ryanodine or ruthenium red. The acid Ca2+ store in the granular region did not have Golgi or lysosomal characteristics and might therefore be associated with the secretory granules. The endoplasmic reticulum is predominantly basal, but thin extensions penetrate into the granular area and cytosolic Ca2+ signals probably initiate at sites where endoplasmic reticulum elements and granules come close together.
Neurotransmitter- or hormone-elicited Ca2+ release from intracellular stores into the cytosol is an important Ca2+ signalling mechanism, first described in salivary glands (Nielsen and Petersen, 1972). The subsequent discovery of inositol (1,4,5) trisphosphate [IP3 or Ins(1,4,5)P3] as the intracellular Ca2+-releasing messenger was based on experiments in permeabilized pancreatic acinar cells (Streb et al., 1983; Berridge, 1984). The crucial role of the IP3 receptor (IP3R) was established by the discovery of high-affinity IP3R-binding sites in the endoplasmic reticulum (ER) (Spät et al., 1986) and finally by the elucidation of the primary structure of this protein, which is both a Ca2+ channel and an IP3-binding protein (Furuichi et al., 1989). IP3-elicited Ca2+ release is now recognized as the principal mechanism for generation of Ca2+ signals in electrically non-excitable cells, but it also plays an important role in many different types of cells that possess voltage-gated Ca2+ channels (Berridge, 1993; Alvarez et al., 1999; Caroppo et al., 2004; Berridge et al., 2003). It is generally accepted that the ER is the most important organelle from which Ca2+ can be released (Meldolesi and Pozzan, 1998), through either IP3Rs or ryanodine receptors (RyR) (Berridge, 1993; Petersen et al., 1994; Pozzan et al., 1994; Ashby et al., 2003b; Bootman et al., 2002), but other organelles can also store and release Ca2+. The mitochondria do not primarily generate physiological Ca2+ signals, but the special and important role they play in compartmentalizing and shaping cytosolic Ca2+ signals, as well as in stimulus-metabolism coupling, has become increasingly clear (Pralong et al., 1992; Pralong et al., 1994; Rizzuto et al., 1993; Hajnoczky et al., 1995; Tinel et al., 1999; Pozzan et al., 2000; Gilabert et al., 2001; Collins et al., 2002; Villalobos et al., 2002; Johnson et al., 2003). The possible importance of Ca2+ release from other organelles such as the nuclear envelope (Nicotera et al., 1990; Malviya et al., 1990; Gerasimenko et al., 1995; Gerasimenko et al., 1996a; Gerasimenko et al., 2003; Gerasimenko and Gerasimenko, 2004), the Golgi complex (Pinton et al., 1998; Missiaen et al., 2004), the secretory granules (Fasolato et al., 1991; Gerasimenko et al., 1996c; Nguyen et al., 1998; Yoo et al., 2000; Quesada et al., 2001; Quesada et al., 2003) and the endosomes (Gerasimenko et al., 1998) is less clear.
In pancreatic acinar cells, the bulk of the ER Ca2+ store is located in the basal part of the cell surrounding the nucleus (Petersen et al., 1998; Petersen et al., 2001), but thin ER projections lumenally continuous with the bulk of the ER at the base (Park et al., 2000) are present in the secretory granule area (Gerasimenko et al., 2002). Although the lowest density of ER is in the secretory granule area near the apical membrane containing the Ca2+-acivated Cl– channels, this is precisely where agonist-elicited Ca2+ release is initiated (Osipchuk et al., 1990; Kasai and Augustine, 1990; Thorn et al., 1993; Park et al., 2001; Oshiro et al., 2005). Furthermore, intracellular application of IP3, cyclic ADP-ribose (cADPR) or nicotinic acid adenine dinucleotide phosphate (NAADP) produces repetitive cytosolic Ca2+ signals specifically localized in the apical granular pole (Thorn et al., 1993; Thorn et al., 1994; Cancela et al., 2002). Even selective activation of muscarinic receptors, specifically and exclusively located in a small region at the base, induces initiation of cytosolic Ca2+ signals in the apical part of the cell (Ashby et al., 2003a). Ca2+-induced Ca2+ release can only be initiated in the secretory granule area (Ashby et al., 2002). Whereas the IP3Rs are concentrated in the apical secretory pole (Nathanson et al., 1994; Lee et al., 1997), the RyRs are distributed throughout the cell (Fitzsimmons et al., 2000). The predominant localization of Ca2+ signals in the apical pole is most likely due to this being the only region with a high density of both IP3Rs and RyRs, since the apical local and repetitive Ca2+ signals depend on both functional IP3Rs and RyRs (Burdakov and Galione, 2000; Cancela et al., 2000). It would appear that the principal source of Ca2+ released into the cytosol in the apical pole is the basal ER, but Ca2+ is channelled through the functional ER tunnel to the apical ER terminals, where the Ca2+ release channels are concentrated (Mogami et al., 1997; Park et al., 2000; Petersen et al., 2001; Ashby and Tepikin, 2002).
There is controversy regarding the site of action of NAADP, a relatively novel Ca2+ liberating messenger, which was discovered in sea urchin eggs (Lee et al., 1989; Chini et al., 1995; Chini et al., 1996; Lee and Aarhus, 1995; Lee, 1997; Chini and De Toledo, 2002) and has been shown to release Ca2+ from internal stores in many different cell types, including pancreatic acinar cells (Genazzani and Galione, 1997; Johnson and Misler, 2002; Masgrau et al., 2003; Cancela et al., 1999; Cancela et al., 2000; Cancela et al., 2001; Cancela et al., 2002; Cancela et al., 2003; Petersen and Cancela, 1999; Brailoiu et al., 2003; Mitchell et al., 2003). In sea urchin eggs, NAADP mobilizes Ca2+ from a pool that does not have ER characteristics (Genazzani and Galione, 1996) and it has been suggested that the NAADP-sensitive store may be the so-called reserve granules, the functional equivalent of secretory lysosomes (Churchill et al., 2002). However, work in our laboratory shows that NAADP induces repetitive cytosolic Ca2+ spikes in the apical granular pole of pancreatic acinar cells (Cancela et al., 2002), exactly at the same site as IP3 or cADPR (Thorn et al., 1993; Thorn et al., 1994). Furthermore, recent work on isolated nuclei, which do not contain acidic Ca2+ stores and have ER Ca2+ transport characteristics, show that NAADP can release Ca2+ from the ER via RyRs, like cADPR, whereas IP3 releases Ca2+ via IP3Rs (Gerasimenko et al., 2003). The controversy about the precise mechanism of action of NAADP has been reviewed recently and several alternative hypotheses have been discussed (Galione and Petersen, 2005).
Yamasaki et al. (Yamasaki et al., 2004) have suggested that in pancreatic acinar cells NAADP specifically releases Ca2+ from lysosome-related acidic organelles located in the secretory granule area. Yamasaki et al. (Yamasaki et al., 2004) conclude that whereas NAADP exclusively elicits release of Ca2+ from these acid stores, IP3 and cADPR only liberate Ca2+ from the ER. In view of the apparent contrast between these results and those obtained from isolated pancreatic nuclei (Gerasimenko et al., 2003), we therefore decided to compare the Ca2+ releasing actions of NAADP to those of IP3 and cADPR, specifically focussing on the possible existence of separate functionally important intracellular Ca2+ stores that might be differentially controlled. We employed pancreatic acinar cells, the classical cell biological model for secretion and protein synthesis (Palade, 1975) and used permeabilized cells, a preparation that was crucial for the original identification of the Ca2+ releasing action of IP3 (Streb et al., 1983).
Two-photon microscopy is one of the most important recent technological advances in physiological imaging and continues to find an increasing number of applications in biology and medicine (Piston, 1999). We have now successfully developed a two-photon permeabilization technique to irreversibly permeabilize pancreatic acinar cells without damaging the intracellular Ca2+ stores.
In this preparation, we have compared the characteristics of Ca2+ release from intracellular stores elicited by the three Ca2+ releasing messengers NAADP, cADPR and IP3 and also studied their interactions. All three messengers release Ca2+ from two separate intracellular stores (one of these is thapsigargin sensitive, whereas the other is acidic and thapsigargin insensitive). Our data show that NAADP, like cADPR, acts on both stores by activation of RyRs, since the NAADP effect is abolished by the RyR blockers ryanodine and ruthenium red, but not by inhibition of the IP3Rs. The thapsigargin-sensitive store clearly corresponds to the ER and is mainly localized in the basolateral part of the cells, whereas the acidic thapsigargin-insensitive store is located in the apical secretory pole. The apical store could in principle be composed of the secretory (zymogen) granules and/or other acidic organelles such as the lysosomes and the Golgi complex. Because of the insensitivity of messenger-elicited release from this store to Gly-Phe-β-naphthylamide (GPN) and brefeldin, we conclude that the acid store is most likely located in the secretory (zymogen) granules. Our data show that all three Ca2+ releasing messengers investigated act on both stores. NAADP and cADPR activate RyRs in both stores, whereas IP3 activates IP3Rs, also in both stores.
The action of Ca2+ releasing messengers
For measurements of [Ca2+] changes in intracellular stores, cells were loaded with the low affinity Ca2+ sensitive dyes Mag Fura-2 or Fluo-5N in their membrane permeable AM forms (Hofer et al., 1996). Following two-photon laser permeabilization of the cell membrane (Fig. 1A-D), the intracellular Ca2+ releasing messengers IP3 (10 μM, n=25), cADPR (10 μM, n=16), or NAADP (100 nM, n=30) were added to the intracellular solution and each produced a substantial decrease in the intra-organellar [Ca2+] (Fig. 1E-G). Addition of IP3, following NAADP-elicited Ca2+ release, produced a second phase of Ca2+ liberation from the stores (Fig. 1H, n=7). Similarly, application of NAADP, following IP3-induced Ca2+ release, also produced a second Ca2+ release from the stores (Fig. 1I, n=5). These data could be interpreted as indirect evidence for two separate Ca2+ pools. In what follows, this proposition will be tested.
Two separate intracellular stores from which messengers can release Ca2+
The experiments shown in Fig. 2 were designed to compare the effects of the Ca2+ releasing messengers, applied in the continued presence of the sarco/endoplasmic reticulum Ca2+-ATPase (SERCA) pump inhibitor thapsigargin, in the granular and basal areas of the same cells. The normal polarity is well preserved in a single isolated permeabilized cell and it is easy to identify regions of interest specifically in the granular and basal areas (Fig. 1). Fig. 2A shows the result of an experiment in which IP3 (10 μM) and then NAADP (100 nM) evoked normal Ca2+ release responses. Thereafter, thapsigargin (10 μM) induced a large Ca2+ release. In the continued presence of thapsigargin, [Ca2+] in the intracellular stores remained low, but when NAADP (100 nM) was added on top of thapsigargin a further Ca2+ release was observed (Fig. 2Aa). The regional analysis (Fig. 2Ab,c,d) showed that the NAADP-elicited Ca2+ release in the presence of thapsigargin could only be observed in the granular area (n=20, P<0.001), indicating that the thapsigargin-insensitive Ca2+ store is localized exclusively in this part of the cell. Very similar results were obtained with IP3 (Fig. 2Ba-c; n=10, P<0.001) and cADPR (Fig. 2Ca-c; n=8, P<0.002). In the continued presence of thapsigargin, the Ca2+ releasing messengers can only elicit further Ca2+ liberation in the granular, but not the basal part of the cells. By contrast, the Ca2+ release elicited by thapsigargin, as well as by the messengers applied in the absence of thapsigargin, can be observed in both the basal and granular areas. These experiments suggest that, in addition to the well known thapsigargin-sensitive ER Ca2+ store – which is present in both the basal and granular parts of the cells (Gerasimenko et al., 2002) and is functionally fully connected (Mogami et al., 1997; Park et al., 2000) – there is a non-ER type Ca2+ store, found exclusively in the granular (apical) pole, from which NAADP, IP3 and cADPR can also release Ca2+. The granular area, as the name implies, is dominated by zymogen granules (Palade, 1975; Gerasimenko et al., 2002), which contain large amounts of Ca2+ (Gerasimenko et al., 1996c).
The nature of the granular thapsigargin-insensitive Ca2+ pool
The potential stores of Ca2+ in the granular area could consist of the secretory (zymogen) granules themselves, or of lysosomes or endosomes. Since these organelles are acidic, we tested to what extent Ca2+ release responses from the thapsigargin-insensitive stores could be inhibited by breaking down the transmembrane H+ gradient (Camello et al., 2000), either by using a protonophore or by reducing the activity of the vacuolar H+ ATPase with bafilomycin A1 (Fig. 3A-H). Nigericin (7 μM), added after emptying the ER with thapsigargin (10 μM) induces slow Ca2+ release from the acidic store (data not shown, n=4). When the permeabilized cells were pre-treated with 10 μM thapsigargin and 7 μM nigericin (Fig. 3A,C) neither NAADP nor IP3 were able to evoke any Ca2+ release, suggesting involvement of acidic Ca2+ stores (Fig. 3A,C IP3, n=7; NAADP, n=8). However, nigericin alone, in the absence of thapsigargin, (Fig. 3B,D) was unable to block NAADP–induced or IP3-induced Ca2+ release, confirming previous findings concerning NAADP-induced as well as IP3-induced Ca2+ release from the thapsigargin-sensitive ER Ca2+ store in the nuclear envelope (Gerasimenko et al., 2003). Bafilomycin A1 is known to disrupt the acid pH in organelles by specifically blocking the vacuolar H+ ATPase at a concentration of 100 nM (Bowman et al., 1988). We have confirmed this with ratiometric measurements of endosomal pH (not shown). We tested the effects of bafilomycin A1 at both 100 nM and 1 μM, which gave similar results. After bafilomycin A1 pre-treatment for 30 minutes in the presence of thapsigargin, NAADP (n=7) and IP3 (n=7) were unable to evoke any Ca2+ release (Fig. 3E,G). In the absence of thapsigargin, bafilomycin A1 did not prevent responses to NAADP (Fig. 3F, n=10), or IP3 (Fig. 3H), consistent with Ca2+ release from the ER. All these data indicate that NAADP can release Ca2+ from both the ER and from an acidic Ca2+ store.
We also made an attempt to differentiate further between different acidic organelles as candidates for the acidic Ca2+ store. We estimate that the averaged response to Ca2+ releasing messengers from the acidic Ca2+ store is much smaller than the response from the ER in the whole cell [IP3: 6.5±2.8% mean ± s.e.m., n=10; NAADP: 10.5±4.3%, n=20; cADPR: 8.3±3.2%, n=8; different from control (P<0.005); Fig. 4A]. Calibrations with 10 μM ionomycin in the presence of 7 μM nigericin, after emptying the ER with thapsigargin, show that in the whole cell the acidic store is approximately half the size of the ER (47±5.3%, n=7, Fig. 4A). However, in the secretory granule area alone, there is less difference in the sizes of the stores and the responses, and the acidic store is 30±5% (mean ± s.e.m.) larger than the ER Ca2+ store (n=8). We estimate that the free [Ca2+] in the acidic store is 300±70 μM, whereas the free [Ca2+] in the ER is 120±50 μM (n=3).
We tested the effect of GPN, a specific substrate for cathepsin C, which accumulates inside lysosomes and leads to their collapse (Jadot et al., 1984). GPN (50 μM) induced Ca2+ release on its own (n=27) and both the cholecystokinin (CCK; 2 pM) and acetylcholine (ACh; 10 nM) responses were inhibited in the presence of GPN (Fig. 4B, n=10). The GPN-induced Ca2+ release could be blocked or substantially reduced by pre-treatment with thapsigargin (not shown, n=11). In contrast to the long-standing belief that lysosomal enzymes are harmless at neutral pH, it has recently been shown that even partial permeabilization of lysosomes induces apoptosis or necrosis, depending on the extent of the permeabilization (Li et al., 2000; Bursch et al., 2001; Turk et al., 2002; Guicciardi et al., 2004). We therefore used cathepsin inhibitor 1 (CI-1) to reduce the damaging effect of GPN. CI-1 is known to inhibit cathepsin B, L and S as well as papain, but not cathepsin C. Therefore, permeabilization of lysosomes by GPN should remain unaffected by CI-1. In our experiments, CI-1 substantially reduced or abolished the GPN-induced Ca2+ release and preserved the responses to CCK (2 pM) and ACh (10 nM) (Fig. 4C, n=10). To test whether CI-1 reduced the effect of GPN on lysosomes, we used LysoTracker Red staining and observed that the apparent destruction of lysosomes by GPN remained at the same level as without CI-1 (Fig. 4D, n=5, P>0.6). We have also labelled lysosomes with fluorescent pepstatin A (Molecular Probes), a fluorescent probe for cathepsin D based on a selective cathepsin D inhibitor, which is a major lysosomal aspartic endopeptidase (Chen et al., 2000). We have shown that GPN, both alone or in the presence of CI-1, destroyed lysosomal integrity (Fig. 4D, n=6, P>0.4). Similar results were obtained with the fluorescent general protease/caspase substrate (Molecular Probes), which shows the same level of protease activation (P>0.4) in the cytoplasm after destruction of lysosomes by GPN alone (17%±1.5, n=5) or GPN with CI-1 (17%±1.4, n=5). Whereas GPN alone inhibited NAADP-induced Ca2+ release (Fig. 4E, n=7), as reported previously by Yamasaki et al. (Yamasaki et al., 2004), destruction of lysosomes in the presence of CI-1 did not inhibit the NAADP-induced Ca2+ release from the acidic store in the granular area, in the majority (67%) of cases (Fig. 4Fa-c, n=15, P<0.002). These results indicate that lysosomes are unlikely candidates for the NAADP-sensitive acidic Ca2+ store in the secretory granule area.
In another set of experiments, designed to identify the acidic Ca2+ store, we used brefeldin A, which is known to disrupt the Golgi (Lippincott-Schwartz et al., 1989). The NAADP-elicited Ca2+ release was similar to control in brefeldin A-treated cells, in which it is known that the Golgi has been disassembled (Dolman et al., 2005) (Fig. 4Ga-c, n=7, P<0.0007), which would seem to exclude this organelle from the list of candidates for the acidic Ca2+ store. The only remaining organelles with a substantial Ca2+ content in this (granular) area of the cell are the secretory granules, which have previously been shown to release Ca2+ in response to stimulation with IP3 or cADPR in pancreatic acinar cells (Gerasimenko et al., 1996c), tracheal goblet cells (Nguyen et al., 1998), mast cells (Quesada et al., 2001; Quesada et al., 2003) and neuroendocrine cells (Yoo et al., 2000; Mitchell et al., 2001).
The nature of the Ca2+ release channels
Fig. 5A shows that a second addition of NAADP, following the first NAADP-induced Ca2+ release, resulted in Ca2+ liberation from the stores with undiminished amplitude (n=8). Preincubation with ruthenium red (RR; 10 μM) abolished the responses to NAADP (Fig. 5B, n=9) and cADPR (Fig. 5C, n=4), whereas the responses to IP3 were preserved (Fig. 5D, n=5). A high concentration of ryanodine (100 μM) also abolished the NAADP- and cADPR-induced responses (Fig. 5E, n=4; Fig. 5F, n=5, respectively), leaving the IP3-induced responses unchanged (Fig. 5H, n=3). Fig. 5G shows that normally a second addition of cADPR, following the first cADPR-elicited Ca2+ liberation, reproduced the first Ca2+ release response (n=5). These results indicate that NAADP induces Ca2+ release in both internal stores by activating RyRs.
The effect of clamping [Ca2+] at 100 nM on NAADP-elicited Ca2+ release
We used a Ca2+/BAPTA mixture with a high concentration of BAPTA (10 mM) to clamp [Ca2+] in the bath solution, and therefore in the cytosol, at ∼100 nM. This would substantially reduce any fast changes of the cytosolic [Ca2+] close to the Ca2+ release channels and effectively inhibit Ca2+-induced Ca2+ release (CICR) (Mogami et al., 1998). In this situation, the NAADP- and IP3-induced responses (Fig. 6A) were similar to those obtained under the low buffering condition (100 μM EGTA, 50 μM CaCl2, free [Ca2+] ∼100 nM) shown in Fig. 1E,F,H.
When we emptied the ER, by inhibiting the SERCA pumps with thapsigargin, in a highly buffered condition ([Ca2+] clamp, with [Ca2+] ∼100 nM), the subsequent responses induced by IP3 (Fig. 6C, n=6, P<0.001) and cADPR (Fig. 6D, n=5, P<0.005) were similar to those obtained in the standard low buffer condition (see Fig. 2F,H). However, the NAADP-induced Ca2+ release was blocked by the [Ca2+] clamp in the presence of thapsigargin (Fig. 6B, n=17, P>0.5). These results suggest that the mechanism of NAADP-elicited Ca2+ release in the secretory granule area is different from that in the ER and indicates that the NAADP-induced Ca2+ liberation in the secretory granule area depends on CICR. Interestingly, the NAADP-induced Ca2+ release in the secretory granule area depends on CICR whereas the cADPR response does not, indicating that the mechanisms of activation of RyRs by these two messengers are different in the secretory granule area.
Ionomycin releases Ca2+ from the ER but not from acidic stores
It has been shown previously, that ionomycin alone does not release Ca2+ from acidic stores without collapse of the pH gradient (Fasolato et al., 1991). We have also used ionomycin to empty the ER Ca2+ store and study the responses from acidic stores. Application of ionomycin (10 μM) induced Ca2+ release from the internal stores in both regions of interest: granular (Fig. 7Aa) and basal (Fig. 7Ac). Following addition of NAADP (100 nM) we observed additional Ca2+ release, but only in the granular area (Fig. 7Aa,b) and without any change in the basal region (Fig. 7Ac). The response in the granular area was significantly different from that in the basal region (Fig 7Ab, n=6, P<0.005). Similar responses were obtained after application of 10 μM IP3 (Fig. 7B, n=5, P<0.002). Nigericin (14 μM) also induced a Ca2+ response in the presence of 10 μM ionomycin (Fig. 7C, n=4, P<0.005).
Thapsigargin does not discharge acidic Ca2+ stores in intact cells
We also performed experiments in which we slowly emptied the ER in cells loaded with Fluo-4 in AM form in the absence of external Ca2+. We used a protocol where we avoided CICR (which could potentially induce Ca2+ release from acidic stores; see Ca2+ clamp experiments in Fig. 6), by inducing a very slow Ca2+ loss from the thapsigargin-sensitive stores. Subsequent additions of low (10 nM, for 10 minutes) and high (10 μM, for another 10 minutes) doses of thapsigargin in the absence of external Ca2+ induced a large, but slow rise of the cytosolic [Ca2+] (Fig. 7D), confirming Ca2+ release from the ER as a result of inhibition of thapsigargin-sensitive Ca2+-ATPases. After the cytosolic [Ca2+] had returned to the resting level (>20 minutes in the presence of thapsigargin), application of 1 nM CCK induced a small rise in the cytosolic [Ca2+], which was highly significant in comparison to the pre-stimulation level (Fig. 7E), according to t-test analysis (P<0.0005, n=8; Fig. 7D,E).
In a classical cell biological model, the pancreatic acinar cell, we have demonstrated two separate functional Ca2+ stores: the already known thapsigargin-sensitive ER store, present in both the basal and granular areas, and a thapsigargin-insensitive acidic store exclusively located in the apical secretory granule area. Both stores can be completely and independently emptied: the ER by thapsigargin and the acidic store by nigericin or bafilomycin. When the stores are intact, they can both respond to any of the three Ca2+ releasing messengers IP3, cADPR or NAADP. The ER is the quantitatively dominant store. The acidic Ca2+ store accounts for only approximately 15% (amplitude comparison) of the total ER Ca2+ release. However, in the secretory granule area, the acidic Ca2+ store is ∼30% larger than the ER. Whereas lysosomes have not generally been considered as organelles with a high Ca2+ content, the secretory granules are known for their high Ca2+ concentration (Clemente and Meldolesi, 1975; Nakagaki et al., 1984; Nicaise et al., 1992; Thirion et al., 1995; Gerasimenko et al., 1996c). Our experiments with brefeldin A would appear to exclude the Golgi, and the results with GPN to exclude the lysosomes as possible candidate organelles for the thapsigargin-insensitive Ca2+ store. This leaves only one plausible candidate, namely the secretory (zymogen) granules. Such granules have been shown previously to liberate Ca2+ in response to Ca2+ releasing messengers (Gerasimenko et al., 1996c; Nguyen et al., 1998; Yoo et al., 2000; Quesada et al., 2001; Mitchell et al., 2001; Quesada et al., 2003). However, we cannot exclude completely the possibility that other acidic organelles such as the Golgi, the endosomes or the lysosomes, including the secretory lysosomes (Hirano, 1991; Grondin, 1996; Cerny et al., 2004; Yamasaki et al., 2004; Malosio et al., 2004) make some contribution. In any case, the acidic store is highly sensitive to all the three Ca2+ releasing messengers tested, as is the ER in the basal area (Fig. 8).
Our new data are important also in relation to the on-going debate about the mechanism of action of the relatively novel Ca2+ releasing messenger NAADP. Arising from experimental work on sea-urchin eggs, it has been suggested that NAADP acts on a separate, novel and hitherto unidentified Ca2+ release channel, which should be located exclusively in reserve granules or lysosome-related organelles, but not in the ER (Chini et al., 1995; Genazzani and Galione, 1996; Churchill et al., 2002; Churchill et al., 2003; Yamasaki et al., 2004). By contrast, we – and others – have proposed that NAADP, like cADPR, acts on RyRs, but via a separate binding site (protein) (Mojzisova et al., 2001; Hohenegger et al., 2002; Gerasimenko et al., 2003; Langhorst et al., 2004; Guse, 2004; Dammermann and Guse, 2005). With regard to pancreatic acinar cells, we have previously shown that NAADP can release Ca2+ into the nucleoplasm in isolated nuclei that contain only the ER-type Ca2+ store and do not contain lysosomes or other acidic organelles (Gerasimenko et al., 2003). The NAADP-induced Ca2+ release from the ER in these isolated nuclei (Gerasimenko et al., 2003) is abolished by RyR inhibitors. Our new data, using two-photon permeabilized pancreatic acinar cells, show that the messenger action on the acidic store in the secretory granule area can also be explained by the presence of just two types of Ca2+ release channels, namely IP3Rs and RyRs. Indeed, a high concentration of ryanodine (100 μM), as well as ruthenium red, specifically blocks cADPR- and NAADP-induced Ca2+ release from both stores without affecting the responses to IP3.
Data obtained using isolated pancreatic nuclei (Gerasimenko et al., 2003), when compared with studies of local Ca2+ spikes in intact pancreatic acinar cells (Cancela et al., 2000; Cancela et al., 2002) and our new data on permeabilized acinar cells, show that local control of Ca2+ release operates differently in the basal and secretory granule areas. Local Ca2+ spiking in the secretory region (as well as the global Ca2+-induced Ca2+ waves that are initiated in the apical granular pole) are highly dependent on co-operative interactions between IP3Rs and RyRs, mediated by Ca2+ (Cancela et al., 2000; Cancela et al., 2002; Ashby et al., 2002). However, we have found no evidence for interaction between IP3Rs and RyRs in the nucleus. In this part of the cell, the receptors appear to function independently, each allowing similar Ca2+ release responses (Gerasimenko et al., 2003).
As demonstrated here, the secretory granule area has two separate Ca2+ stores, the ER extensions (connected to the basal ER) and the acidic store (Fig. 8). These organelles are extremely close in the secretory granule area (separation of less than a few hundred nanometers, i.e. at the level of confocal resolution), which should allow functional interactions under appropriate conditions. With regard to receptor interactions, we know that the concentration of IP3Rs is much the highest in the secretory granule area (Nathanson et al., 1994; Lee et al., 1997), whereas RyRs are distributed more or less homogeneously throughout the cell (Leite et al., 1999; Fitzsimmons et al., 2000). The probability of interaction between IP3Rs and RyRs, and therefore the probability of CICR is therefore highest in the secretory granule area, in complete agreement with studies of the effects of local Ca2+ uncaging (Ashby et al., 2002).
Materials and Methods
Mag-Fura Red AM, Mag-Fura-2 AM, Fluo-5N AM, Fluo-4 AM, Fura Red AM, Texas Red dextran (Mr 3×103), BODIPY FL-Pepstatin A and general protease/caspase substrate were obtained from Molecular Probes (Invitrogen, UK). FFP-18 (K+) salt was obtained from TEFLabs (USA). Ryanodine was purchased from MP (UK). The rest of the chemicals were purchased from Sigma (UK) or Calbiochem (Merck, UK).
Mouse pancreatic acinar cells were isolated by collagenase digestion as described previously (Gerasimenko et al., 1996b). The solution for cell isolation contained (in mM): NaCl, 140; KCl, 4.7; Hepes-KOH, 10; MgCl2, 1; CaCl2, 1; (pH 7.2). After isolation, cells were loaded with low affinity Ca2+-sensitive dyes: Mag Fura-2 AM (5 μM in a solution containing 0.01% Pluronic F-127) or Fluo-5N AM (2.5 μM) by incubation for 30-45 minutes at 36.5°C. Cells were attached to poly-L-lysine-coated coverslips in a flow chamber. All experiments were performed at room temperature. Before and during two-photon permeabilization, cells were bathed in intracellular solution, which contained (in mM): KCl, 128; NaCl, 20; Hepes-KOH, 10; ATP, 2; MgCl2, 1; EGTA, 0.1; CaCl2, 0.075; (pH 7.2). After permeabilization, cells were perfused with intracellular solution for 10 minutes to wash out the cytosolic component of the fluorescent dye. Experiments shown in Fig. 1 were conducted in the same solution as above except that 0.05 mM CaCl2 was used. In experiments with calcium clamp (Fig. 6), 10 mM BAPTA was substituted for EGTA and 3.3 mM of CaCl2 was used.
Pancreatic acinar cells can be permeabilized by localized perforation of the membrane using two-photon (tuned to 740-750 nm) high intensity laser pulses. We found that this permeabilization technique is superior to the chemical permeabilization with saponin or digitonin (see below). The cell membrane was stained with the `near membrane' Ca2+ indicator FFP-18 (K+ salt, 10 μM) to help the formation of a single, site-specific perforation in the cell membrane using the two-photon laser beam. Two-photon light, when directed at high intensity to a small area of a cell membrane can temporarily perforate the membrane and allow successful intracellular delivery of foreign DNA (Tirlapur and Konig, 2002). We have modified this technique to achieve permanent permeabilization of pancreatic acinar cells (Fig. 1A-D). A high intensity two-photon laser beam in pulse mode at 740-750 nm from Spectraphysics (8W Millenia femto-second laser) was applied to a small area of the cell membrane (Fig. 1A). This resulted in heating of this small membrane area and subsequent pore formation. Permeabilization was confirmed by monitoring the fluorescence of Texas Red dextran added to the extracellular medium. Upon permeabilization Texas Red dextran penetrated into the cytoplasm of the targeted cell (Fig. 1B). Before permeabilization, the cells were perfused by a K+ rich, low Ca2+ (EGTA-containing), intracellular type solution. Subsequent washing of Texas Red dextran from the extracellular solution resulted in the wash out of Texas Red dextran from the cytoplasm, confirming stable cell permeabilization (Fig. 1C). After perforation, the cells were able to respond to intracellular Ca2+ releasing messengers (IP3, NAADP or cADPR, Fig. 1E-G). We compared responses to the messengers in cells permeabilized by two-photon light with responses obtained from saponin-permeabilized acinar cells. We found that two-photon permeabilization has two principal advantages, namely better preserved morphology (including polarity) and responsiveness. The amplitudes of the responses to IP3 were approximately 1.5 times higher in the two-photon permeabilized cells than in cells permeabilized by saponin. Two-photon permeabilization resulted in a hole with a diameter of ∼2 μm at any site selected on the surface of the cell. This hole did not close after permeabilization, as in previously published work (Tirlapur and Konig, 2002), perhaps because of the larger size and/or the exposure of the cell to an intracellular solution before the two-photon pulse. The success rate with two-photon permeabilization was high (>50%) and we propose this technique as a reliable and convenient method to permeabilize cells.
Fluorescent [Ca2+] measurements
Fluorescent images were obtained using a Leica SP2 MP dual two-photon confocal microscope with a ×63 1.2 NA objective. For Mag Fura-2, excitation and emission wavelengths were 430 nm (2-5% power) and 460-590 nm, respectively. Alternatively, for excitation of Mag Fura-2, the two-photon wavelength 745 nm was used. For Fluo-5N and Fluo-4, excitation and emission wavelengths were 488 nm (argon ion laser, 1-2% power) and 510-590 nm, respectively. Fluorescent images were collected with a frequency of 0.6-1.0 frame/second. Texas Red dextran was excited at 543 nm and emission collected at 580-650 nm. For the purpose of calculating free Ca2+ concentrations the Kd of Fluo-5N for Ca2+ was assumed to be 90 μM; the Kd of Fluo-4: 350 nM. The pH dependence of the Kd of Fluo-5N for Ca2+ was tested. There was no significant difference between the results at pH 7.2 and at pH 6 (P>0.9; n=6). The calibration procedure was performed by applying ionomycin (10 μM) and nigericin (7 μM) with 2 mM EGTA or 10 mM CaCl2. Statistical analysis was performed using Microsoft Excel software. P values were determined for statistical significance between sets of data using Student's t-test.
This work was supported by an MRC Programme Grant (G8801575). O.H.P. is an MRC Research Professor.