In sea urchin eggs, Ca2+ mobilization by nicotinic acid adenine dinucleotide phosphate (NAADP) potently self-inactivates but paradoxically induces long-term Ca2+ oscillations. We investigated whether NAADP-induced Ca2+ oscillations arise from the recruitment of other Ca2+ release pathways. NAADP, inositol trisphosphate (IP3) and cyclic ADP-ribose (cADPR) all mobilized Ca2+ from internal stores but only NAADP consistently induced Ca2+ oscillations. NAADP-induced Ca2+ oscillations were partially inhibited by heparin or 8-amino-cADPR alone, but eliminated by the presence of both, indicating a requirement for both IP3- and cADPR-dependent Ca2+ release. Thapsigargin completely blocked IP3 and cADPR responses as well as NAADP-induced Ca2+ oscillations, but only reduced the NAADP-mediated Ca2+ transient. Following NAADP-mediated release from this Ca2+ pool, the amount of Ca2+ in the Ca2+-induced Ca2+ release stores was increased. These results support a mechanism in which Ca2+ oscillations are initiated by Ca2+ release from NAADP-sensitive Ca2+ stores (pool 1) and perpetuated through cycles of Ca2+ uptake into and release from Ca2+-induced Ca2+ release stores (pool 2). These results provide the first direct evidence in support of a two-pool model for Ca2+ oscillations.
The sea urchin egg is highly responsive to IP3, cADPR and NAADP (Clapper and Lee, 1985; Clapper et al., 1987; Lee and Aarhus, 1995) and thus provides an excellent system in which to study the interaction between Ca2+ release pathways. Here we show that NAADP induces long-term Ca2+ oscillations by releasing Ca2+ from a separate store (pool 1) and is then taken up by CICR stores (pool 2). The shuttling of Ca2+ primes the CICR stores and results in cycles of Ca2+ overloading, release and re-uptake.
Results and discussion
NAADP but not IP3 or cADPR induces long-term Ca2+ oscillations
Lee and coworkers (Aarhus et al., 1996; Lee, 1997; Lee et al., 1997) reported that NAADP but not IP3 or cADPR resulted in Ca2+ oscillations in sea urchin eggs, and photorelease was much more effective than microinjection of NAADP (Aarhus et al., 1996). However, Swann and Whitaker (1986) reported that microinjection of IP3 also resulted in a second Ca2+ increase that occurred several minutes after the primary Ca2+ increase. Therefore, to investigate the mechanism by which NAADP gives rise to Ca2+ oscillations, we first compared the Ca2+ responses of sea urchin eggs to NAADP, IP3 and cADPR.
The photorelease of supramaximal amounts of messenger revealed that NAADP released more Ca2+ than cADPR, which in turn released more than IP3 (Figures 1 and 6A). That NAADP evokes a larger increase than either IP3 or cADPR supports the notion that NAADP is releasing Ca2+ from both a common store and a store responsive to only NAADP, as suggested by the data from sea urchin homogenates (Lee and Aarhus, 1995; Genazzani and Galione, 1996). NAADP induced long-lasting (at least 80 min) Ca2+ oscillations in 26 of 29 experiments (Figure 1), as demonstrated previously (Aarhus et al., 1996; Genazzani et al., 1996; Lee, 1997). NAADP-induced Ca2+ oscillations persisted in the absence of extracellular Ca2+ (data not shown), indicating that the oscillations involved mobilization from intracellular Ca2+ stores, as reported previously (Aarhus et al., 1996; Lee et al., 1997). In contrast to the high proportion of eggs exhibiting Ca2+ oscillations after photorelease of NAADP, Ca2+ oscillations were induced less frequently after photorelease of cADPR (eight of 17 experiments, p = 0.004) and IP3 (one of nine experiments, p <0.0001; Figure 1). Additionally, the magnitude of the Ca2+ oscillations was much larger after NAADP-mediated Ca2+ release than after either cADPR- or IP3-mediated Ca2+ release (Figure 1). Taken together, it is clear that Ca2+ oscillations are more robust and frequent after a Ca2+ increase mediated by NAADP than after either cADPR or IP3.
The Ca2+ increase that occurs 1–3 min after the NAADP-mediated Ca2+ increase occurs before Ca2+ recovers to baseline and appears as either a spike clearly separated from the primary increase or as only a shoulder during recovery from the primary increase (Figure 1) (Aarhus et al., 1996; Genazzani et al., 1996; Lee, 1997). To enable quantitative comparisons of data with and without a discernible Ca2+ spike, we determined the duration that Ca2+ was elevated above its half-maximal concentration (Figure 6B). This measure validly described all treatments except for the effect of removing extracellular Ca2+ (Figure 6B), in which case it is an underestimate because the first oscillation is separated from the NAADP-induced Ca2+ increase to the extent that Ca2+ decreases below its half-maximal concentration before the first oscillation occurs (Aarhus et al., 1996; Lee et al., 1997). Duration data show that the initial Ca2+ increase was more prolonged when initiated by NAADP than when initiated by either IP3 or cADPR (Figures 1 and 6B).
The later Ca2+ increases that occur 5–15 min after the NAADP-induced Ca2+ increase exhibit irregular patterns (Figure 1) (Aarhus et al., 1996; Genazzani et al., 1996; Lee, 1997). The time of onset, duration, amplitude and frequency of the Ca2+ increases is decidedly variable (Figure 1, NAADP). Often there are small Ca2+ transients of short duration (∼1 min) superimposed on larger prolonged Ca2+ transients (∼5–15 min). This suggests that there are at least two independent oscillatory mechanisms operating simultaneously. Ca2+ oscillations with a similar ‘signature’ occur after fertilization of sea urchin eggs with increases occurring before nuclear envelope breakdown, metaphase–anaphase transition and cytokinesis (Poenie et al., 1985; Whitaker and Larman, 2001).
In view of our recent data showing that NAADP desensitization is reversible in intact eggs (Churchill and Galione, 2001), we determined whether re-sensitization could explain NAADP-induced Ca2+ oscillations. Under the conditions used in the current study the eggs remained insensitive to NAADP for at least 2 h (data not shown), probably due to the use of supramaximal concentrations of NAADP. Thus, Lee and colleagues' original argument that the long-term Ca2+ oscillations cannot be due to NAADP because of desensitization (Aarhus et al., 1996; Lee, 1997; Lee et al., 1997) remains valid.
NAADP-induced Ca2+ oscillations require both IP3- and cADPR-dependent Ca2+ release pathways
To determine the potential roles of IP3- and cADPR-dependent Ca2+ release pathways in the later oscillations (≥5 min after NAADP photorelease), the effects of IP3 and cADPR antagonists on NAADP-induced Ca2+ oscillations were investigated. The involvement of IP3 was probed with heparin, which inhibits IP3 binding to its receptor (Taylor and Broad, 1998). NAADP induced oscillations in four of six experiments (p = 0.20) in the presence of 2.5 mg/ml heparin and in three of seven experiments (p = 0.07) in the presence of 5 mg/ml heparin (Figure 2). Heparin, however, inhibited the first Ca2+ oscillation after the NAADP-induced Ca2+ increase and shortened the duration of this Ca2+ transient (Figures 2 and 6B). The involvement of cADPR was probed with 8-amino-cADPR (8NH2, Figure 6), which inhibits cADPR binding to its receptor (Walseth and Lee, 1993). The presence of 8-amino-cADPR (5 μM) did not block Ca2+ oscillations in six of six experiments (Figure 2, p >0.99). Nevertheless, 8-amino-cADPR, as did heparin, decreased the amplitude and frequency of the Ca2+ oscillations (Figure 2), indicating a partially inhibitory effect. In contrast to the variable and partially inhibitory effects of either heparin or 8-amino-cADPR alone, the combination of both compounds eliminated NAADP-induced later Ca2+ oscillations in seven of seven experiments (Figure 2, p <0.0001). In the presence of both inhibitors, NAADP released more Ca2+ than it did in the presence of either inhibitor alone (Figures 2 and 6A), possibly due to prevention of a basal leak of Ca2+ from a common store, which would lead to greater loading of this common store. These data demonstrate that both IP3- and cADPR-mediated pathways for Ca2+ release are required for NAADP-induced Ca2+ oscillations.
That NAADP-induced Ca2+ oscillations could only be blocked completely by inhibition of both CICR pathways is similar to the findings reported for inhibition of fertilization-induced Ca2+ waves in sea urchin eggs (Galione et al., 1993; Lee et al., 1993), NAADP-induced Ca2+ release in starfish eggs (Santella et al., 2000) and for amplification of NAADP-mediated Ca2+ waves in sea urchin eggs (Churchill and Galione, 2000). The situation is different from that in pancreatic acinar cells, where NAADP-induced Ca2+ oscillations can be blocked by inhibiting either the IP3 or cADPR pathway (Cancela et al., 1999). In the case of pancreatic acinar cells, the interaction was suggested to occur at the level of CICR, with Ca2+ released via the NAADP-mediated pathway serving as ‘trigger Ca2+’, which feeds forward on to either IP3 or ryanodine receptors thereby amplifying the response (Cancela et al., 1999). Such a mechanism cannot explain the Ca2+ oscillations that occur after NAADP-induced Ca2+ mobilization because the NAADP receptors are completely desensitized to NAADP for at least 2 h after the initial release using this particular protocol (data not shown). Indeed, the first Ca2+ oscillation, which is sometimes distinct and at other times appears as a shoulder after the initial NAADP-mediated Ca2+ increase (Figure 1) (Aarhus et al., 1996; Genazzani et al., 1996; Lee, 1997), occurs during recovery from the NAADP-induced Ca2+ transient when Ca2+ is decreasing. This is at odds with a cytosol-based CICR mechanism, which would provide amplification during the increase in Ca2+.
A Ca2+ store that is thapsigargin insensitive and NAADP sensitive is present in intact eggs
If NAADP-induced Ca2+ oscillations are due to overloading and spontaneous Ca2+ release from CICR stores, then Ca2+ would have to be moved into the CICR stores from an NAADP-sensitive Ca2+ store that is functionally and physically separate. The NAADP-sensitive Ca2+ store is pharmacologically and physically distinct in sea urchin egg homogenates (Lee and Aarhus, 1995; Genazzani and Galione, 1996). The presence of such a store in intact sea urchin eggs, however, was unknown until a recent report that demonstrated spatial separation of the NAADP-sensitive Ca2+ stores from the IP3- and cADPR-sensitive Ca2+ stores in intact sea urchin eggs whose organelles were stratified by centrifugation (Lee and Aarhus, 2000). We used a pharmacological approach to determine whether NAADP-sensitive Ca2+ stores in intact sea urchin eggs were functionally distinct from IP3- and cADPR-sensitive Ca2+ stores. Treatment of intact eggs with the Ca2+ pump inhibitor thapsigargin (2 μM) for 30 min or more completely eliminated the response to both IP3 and cADPR (Figures 3A and 6A), as reported previously for homogenates (Genazzani and Galione, 1996). In contrast, thapsigargin reduced but did not eliminate the initial response to NAADP (Figures 3A and 6A). This result demonstrates that in the intact egg NAADP targets Ca2+ stores that are functionally separate from the CICR stores. This conclusion is consistent with that of Lee and Aarhus (2000), who demonstrated recently that the NAADP-sensitive store was separable from the endoplasmic reticulum and segregated with mitochondria. However, it is unlikely that mitochondria are the target for NAADP because NAADP-sensitive stores are distinct from cytochrome C oxidase, a mitochondrial marker, in fractionated sea urchin egg homogenates (Lee and Aarhus, 1995), and mitochondria are sinks not sources for Ca2+ in non-fertilized eggs (Eisen and Reynolds, 1985). Therefore, NAADP probably targets an organelle that remains to be characterized.
Pretreatment with thapsigargin eliminated NAADP- induced Ca2+ oscillations in eight of eight experiments (Figure 3B; p <0.001 compared with the NAADP control), demonstrating a requirement for IP3- and cADPR-sensitive Ca2+ stores. These data are consistent with the elimination of NAADP-induced Ca2+ oscillations obtained with heparin and 8-amino-cADPR (Figure 2), and consistent with a mechanism in which the CICR stores (IP3- and cADPR-sensitive) play a role in the NAADP-induced Ca2+ oscillations.
CICR stores contain more Ca2+ after an NAADP-mediated Ca2+ transient
Next we directly assessed the effect of an NAADP-mediated Ca2+ release on the amount of Ca2+ in the CICR stores. As only the CICR stores are sensitive to thapsigargin (Genazzani and Galione, 1996 and Figure 3), the size of the Ca2+ transient induced by thapsigargin was used as a measure of CICR store loading. Figure 4 shows that thapsigargin (5 μM) induced a significantly (p = 0.01) larger Ca2+ transient after an NAADP-induced Ca2+ transient (maximum amplitude of 2.0 ± 0.27 ratio units, n = 10) than in control eggs (maximum amplitude of 1.3 ± 0.07 ratio units, n = 6). These data indicate that a substantial amount of the Ca2+ released by NAADP is taken up by the CICR stores. In contrast, the amplitude of Ca2+ release upon photorelease of NAADP was not increased after Ca2+ mobilization by either IP3 (maximum amplitude of 2.2 ± 0.14 ratio units, t-test, p = 0.53) or both IP3 and cADPR (maximum amplitude of 2.2 ± 0.13 ratio units, t-test, p = 0.62). The mechanism for the preferential accumulation of NAADP-released Ca2+ into the CICR stores is unknown. A possible mechanism could be faster removal of Ca2+ by thapsigargin-sensitive Ca2+ pumps than by thapsigargin-insensitive Ca2+ pumps.
Ca2+ oscillations can be induced by photorelease of Ca2+
With a Ca2+-overloading mechanism, any means of overloading Ca2+ stores should induce long-term Ca2+ oscillations (Berridge and Galione, 1988; Berridge, 1991). Photoreleasing Ca2+ (NP-EGTA) (Ellis-Davies and Kaplan, 1994) resulted in an increase equivalent to that induced by NAADP (Figures 5 and 6A), and induced long-term Ca2+ oscillations in four of 10 experiments (Figure 5). The pattern of the oscillations generated by photoreleasing Ca2+ (Figure 5) was similar but not identical to that generated by NAADP (Figure 1). The amount of caged Ca2+ remaining after the first period of photorelease was determined by two additional periods of photorelease, one immediately after recovery to basal levels and one 70 min after the first photorelease (Figure 5). These additional UV illuminations failed to increase Ca2+, demonstrating that the first period of photorelease destroyed most of the NP-EGTA. The amount of Ca2+ can be approximated from this information. Given that NP-EGTA was at a concentration of 500 μM and that the resting Ca2+ (∼100 nM) is approximately at the Kd (80 nM) for NP-EGTA (Ellis-Davies and Kaplan, 1994), there would be ∼250 μM caged Ca2+ (and 250 μM free NP-EGTA). Complete photorelease of the caged Ca2+ would liberate ∼250 μM Ca2+. Although an increase in Ca2+ is sufficient for inducing Ca2+ oscillations in sea urchin eggs about half the time, it does not induce Ca2+ oscillations as often as NAADP, and fails to induce the higher frequency component. Therefore, Ca2+ alone can partially, but not completely, mimic NAADP-induced Ca2+ oscillations.
Mechanism for NAADP-induced Ca2+ oscillations
The mechanism by which NAADP induces Ca2+ oscillations and a summary of the evidence for this mechanism are presented in Figure 7. The initial Ca2+ release mediated by NAADP is largely independent of any additional processes (pathway 1, Figure 7), except for an amplification by CICR pathways, as described previously (Churchill and Galione, 2000). A portion (approximately one-third) of the Ca2+ mobilized by NAADP comes from a thapsigargin-resistant store that is distinct from the CICR stores. This released Ca2+ is then taken up into CICR stores. Ca2+ uptake continues until the stores overload and spontaneously release Ca2+ as demonstrated previously (Galione et al., 1991; Missiaen et al., 1991). A necessary feature of this mechanism is that the released Ca2+ is taken up into thapsigargin-sensitive stores more quickly than into thapsigargin-insensitive (NAADP-sensitive) stores, but how this is accomplished is unknown. The first Ca2+ oscillation after the NAADP-induced transient requires IP3-dependent Ca2+ release (pathway 2, Figure 7). Both IP3- and cADPR-sensitive Ca2+ stores participate in the later cycles of Ca2+ uptake, overload and release (pathway 3, Figure 7). This Ca2+-overloading mechanism can also explain the occasional induction of long-term Ca2+ oscillations by IP3 (Figure 1 and Swann and Whitaker, 1986) and cADPR (Figure 1), in which overloading would occur via Ca2+ influx. This mechanism meets the criteria of a two-pool model for Ca2+ oscillations (Berridge, 1988; Berridge and Galione, 1988; Tsien and Tsien, 1990) because one Ca2+ store is the source of priming Ca2+ (NAADP sensitive) and a separate Ca2+ store mediates the oscillations (CICR), and oscillations can be triggered by any increase in Ca2+ that primes the CICR stores. Therefore, this mechanism represents the first direct demonstration of the two-pool model for Ca2+ oscillations.
Materials and methods
Microinjection and Ca2+ imaging were performed as described previously (Churchill and Galione, 2000). Briefly, sea urchin (Lytechinus pictus) eggs (Marinus, Long Beach, CA) were obtained by intracoelomic injection of 0.5 M KCl, shed into artificial sea water (in mM, NaCl 435, MgCl2 40, MgSO4 15, CaCl2 11, KCl 10, NaHCO3 2.5, EDTA 1), dejellied by passing through 90 μm nylon mesh, and then washed twice by centrifugation. Eggs were transferred to polylysine-coated glass coverslips, pressure microinjected (Picospritzer, World Precision Instruments) with Oregon Green 488 BAPTA Dextran (Molecular Probes) and caged compounds and/or inhibitors. The Ca2+-sensitive dye was imaged by laser-scanning confocal microscopy (TCS NT, Leica) and caged compounds were photolysed with an ultraviolet laser. Images were processed with the software NIH Image to create a self-ratio by dividing each image by an image acquired before stimulation. Where appropriate, parametric data were subjected to analysis of variance with means separated by Fisher's least significant difference test with significance taken as p ≤0.05. Proportions (e.g. number of eggs exhibiting oscillations versus number not) were analysed by creating a 2 × 2 contingency table, with NAADP compared with another treatment and then subjected to Fisher's exact probability test.
We thank F.Boittin, R.Masgrau, Y.Okada, S.Patel and J.Thomas for helpful discussion. This work was funded by the Wellcome Trust.