Introduction

Phosphatidylinositol (PI)-specific phospholipases C (PLCs) have a key role in calcium signalling in animal cells, where they catalyse the hydrolysis of PI-(4,5)-bisphosphate to diacylglycerol and d-myo-inositol-(1,4,5)-triphosphate (IP3). Diacylglycerol directly activates protein kinase C, while IP3 mediates the release of calcium from internal stores (Noh et al. 1995).

In budding yeast, a glucose-responding signal transduction pathway was reported, requiring the unique PI-specific PLC in Saccharomyces cerevisiae, Plc1 (Coccetti et al. 1998). In fact, inhibition of Plc1 activity or deletion of the PLC1 gene blocks the glucose-induced PI turnover and activation of the yeast plasma membrane H+-ATPase (Coccetti et al. 1998; Tisi et al. 2001). Phospholipase C is also involved in the generation of a transient elevation in cytosolic calcium concentration (Tisi et al. 2002), but very little is known so far about the mechanism that triggers this calcium signalling. Glucose phosphorylation and Gpr1/Gpa2 GPCR complex activation are involved in this pathway (Tisi et al. 2002); and the same has been observed for another well known glucose-responsive signal transduction pathway in yeast, the cAMP/protein kinase A pathway (Thevelein and de Winde 1999). Differing from mammalian cells, the calcium accumulation in yeast is most likely due to calcium influx from the external medium, mainly through the Cch1/Mid1 plasma membrane calcium channel (Tokes-Fuzesi et al. 2002).

Glucose-induced plasma membrane H+-ATPase activation also requires the yeast protein kinase C, Pkc1 (Brandão et al. 1994; Souza et al. 2001), and activation of the related MAP kinases cascade (de la Fuente and Portillo 2000). It is tempting to speculate that a pathway could exist in yeast also, connecting Plc1 activation to calcium and the Pkc1/Mpk1 signalling cascade. Nevertheless, it is not clear yet whether Pkc1 in yeast is actually responsive to a rise in calcium concentration. To date, the identified Pkc activators are cell wall stress sensors, such as Mid2 (Ketela et al. 1999), Wsc1/Hsc77/Slg1, Wsc2, Wsc3 (Delley and Hall 1999) and the Tor signalling proteins (Schmidt et al. 1997) via activation of the GTPase, Rho1 (Nonaka et al. 1995; Kamada et al. 1996), in addition to the yeast homologue of mammalian 3′-phosphoinositide-dependent kinase 1, Pkh2 (Inagaki et al. 1999). Consequently, it is possible that the phospholipase C-dependent calcium signalling pathway in yeast could lead to different effectors than in mammalian cells. At present, no physiological role has ever been proposed for glucose-induced calcium signalling; and the only indication about a possible effector of this signal is that both calcium signalling and H+-ATPase activation show a requirement for Plc1 activity.

No homologues of the mammalian IP3 receptors have been found in the yeast genome (Wera et al. 2001), but two IP3-responsive calcium channels, not homologous to the mammalian channels, are reported in Neurospora crassa (Silverman-Gavrila and Lew 2002). This suggests that novel types of IP3 receptor could exist in fungi. These observations raise the question of which mechanism could lead to calcium uptake after Plc1 activation and whether a second messenger (IP3 or diacylglycerol) is involved.

In yeast, a complete degradation pathway seems to exist that sequentially dephosphorylates IP3 to I(1,4)P2, IP and myo-inositol (York et al. 1999). Nevertheless, none of the phosphatases involved have been identified up to now, except for two inositol monophosphatases (Imp1, Imp2) highly homologous to animal and plant ones (Lopez et al. 1999). Surprisingly, inhibition or deletion of these inositol monophosphatases leads to a decrease in the IP3 level, in contrast to what is observed in mammalian cells (Navarro-Aviñó et al. 2003), indicating that IP3 metabolism in S. cerevisiae is quite unusual when compared with mammalian cells.

The inositol phosphates phosphomonoesterase activity is outweighed by the 100-fold higher IP3 kinase activity (Robinson et al. 1996). IP3 kinase activity in yeast is represented by the ARG82 (also called ARGRIII or IPK2) gene product, Arg82, a dual kinase which is active on both IP3 and IP4, leading to IP5 production. Arg82 was reported to be involved in arginine-responsive transcription regulation (Saiardi et al. 2000). In fact, the regulation of arginine metabolism requires the integrity of four regulatory proteins, ArgRI, ArgRII, Arg82/ArgRIII and Mcm1. Arg82 seems to be involved in maintaining the stability of ArgRI and Mcm1, but this protein surely has some additive functions that are not Mcm1-dependent (El Bakkoury et al. 2000).

The I(1,3,4,5,6)P5 is then phosphorylated to IP6 by the kinase encoded by IPK1 (also called GSL1), which is required together with Arg82 for efficient mRNA export from the nucleus (York et al. 1999).

In this paper, we analyse the IP3 level upon glucose stimulation and the effect of the deletion of IPK1 and ARG82 genes on Plc1-dependent calcium signalling. Our findings suggest that IP3 production could be sufficient as a signal for calcium increase and no further phosphorylation is required. This is the first report suggesting a role for IP3 in S. cerevisiae, and further work is required to clarify many aspects about the signal transduction mechanism and the physiological effectors of this signalling pathway.

Materials and methods

Yeast strains and plasmids

The wild-type strain W303-1a (MATa ura3-1 his3-11,15 leu2-3,112 trp1-1 ade2-1 can1-100 GAL suc2) and the isogenic mutant plc1Δ (plc1::URA3) were kindly provided by Laura Popolo (Università di Milano, Milan, Italy). The W303-1a strain carrying the multicopy plasmid pWP101, containing the PLC1 coding sequence, was described by Coccetti et al. (1998). The SWY1659 strain (Matα ipk1::KanMX2) and the isogenic wild type, W303-1b, were a gift from Susan Wente and James York (Washington University School of Medicine, St. Louis, Mo., USA; York et al. 1999). Strain PJ69-4A (Clontech) and its derivative arg82::KanMX2 (Saiardi et al. 2000) were kindly given by James Caffrey and Stephen Shears (National Institute of Environmental Health Sciences, Research Triangle Park, N.C., USA).

Plasmid YEplac195-PDE1 was described by Ma et al. (1999); and the wild-type strain FDL33.6B (leu2 ura3 trp1 arg4 ade) and its derivatives, cdc25::HIS3 sdc25::HIS3 YEpRAS2 Ile152 and cdc25::HIS3 sdc25::HIS3 YEpTPK1, were obtained from M. Jacquet, Orsay (Colombo et al. 1998).

In vivo monitoring of cytosolic free calcium concentration after nutrient starvation

The cytosolic free calcium concentration was measured as described by Tisi et al. (2002). Briefly, exponentially growing cells in liquid rich medium (YPD; Tisi et al. 2002) at 24 °C were harvested by filtration, washed three times with water and resuspended in 0.1 M Mes/Tris buffer, pH 6.5. Cells were incubated for 1.5 h at 24 °C, collected by centrifugation and resuspended in the same nutrient-free buffer (2.5×109 cells ml−1). To reconstitute functional aequorin, the cell suspension was incubated in the dark for 20 min at 24 °C, with 50 μM coelenterazine (stock solution at 1 mg ml−1, dissolved in methanol; Molecular Probes). After washing the cells three times with Mes/Tris buffer, a suspension containing 2.5×107 cells in the same buffer was transferred to a luminometer tube and 10 mM CaCl2 (final concentration) was added. Light emission was monitored with a Berthold Lumat LB 9501/16 luminometer at intervals of 10 s for 1 min before and for at least 6 min after the addition of glucose to a final concentration of 100 mM (or diacylglycerol, 2 mM final concentration; Sigma-Aldrich); and it was reported in relative luminescence units (RLU) s−1 . The figures represent typical RLU lines that were reproduced in at least three different experiments. At the end of each experiment, the aequorin expression and activity were tested by lysing cells with 0.5% Triton X-100. Typically, less than 10% of the total light emission (TLE) was expended during the calcium response assay.

Calibration of aequorin-dependent light signal

The raw data obtained from the luminometer were recorded and transferred to a personal computer for analysis. In order to make different experiments comparable, variations in the cytosolic free calcium concentration [Ca2+]i were evaluated from the ratio of the aequorin light intensity of the yeast cells vs TLE, calculated from the total light yield obtained by lysing an aliquot of the same cells with 0.5% Triton X-100, referring to a standard curve reported in the literature (Matsumoto et al. 2002). The Ca2+-independent luminescence of cells carrying the corresponding empty vector was subtracted as a blank value from all other data.

Determination of IP3

Nutrient-starved cells were resuspended at room temperature at a density of 93.75 mg (wet weight) ml−1 in Mes/Tris buffer. At time-point zero, glucose was added to a final concentration of 100 mM. Treatment with 3-nitrocoumarin (synthesised as described by Tisi et al. 2001) was performed in Mes/Tris buffer at a concentration of 50 mg l−1 for 15 min before the stimulus. Then, 1-ml samples, containing 75 mg of cells, were quenched in 4 ml of 60% methanol pre-cooled in a dry-ice/acetone bath and subsequently harvested by centrifugation. Ice-cold perchloric acid (1 ml) and 1.5 g of glass beads (0.5 mm diameter) were added in order to break the cells by vigorous shaking at 4 °C. The cell homogenates were centrifuged, after which the soluble fraction (extract) was used to determine the IP3 content. IP3 was measured in neutralised cell extracts, as described by Bergsma et al. (2001), using an IP3 bovine adrenal-binding protein-based radioligand assay (Amersham Pharmacia Biotech). As demonstrated by Bergsma et al. (2001), this test was highly specific for IP3; and the glycerophosphoinositol (phosphate) produced upon glucose stimulation (Hawkins et al. 1993) did not interact with this radioligand assay. Briefly, this assay was based on competitive binding to an insoluble protein between a radioactive IP3 and the cold IP3 contained in the samples.

Calculation of IP3 content

The raw data obtained from the radioligand assay were the levels of radioactivity bound to the IP3 bovine adrenal-binding protein when it was incubated with nearly 100 Bq of [3H]IP3 together with the appropriate volume of cell extract. These raw data were treated as follows.

The IP3 content (in relative units) was calculated with the formula: IP3=[(B 0−X)/(X−NSB)], where B 0 was the raw value obtained when water substituted the cell extract, X was the raw value obtained with the sample and NSB (for not specifically bound) was the raw value obtained when 120 pmol of cold IP3 substituted the cell extract. The IP3 contents (in relative units) of at least four different standards were calculated, using known amounts of cold IP3. These standards were utilised to establish a linear correlation between the IP3 (in relative units) and the real IP3 content in the assays. The error in the IP3 content calculation was estimated from the standard deviations in X, NSB, B 0 and standard raw data, all of which were always performed in duplicate. Each curve here reported is typical of at least three different experiments.

Results

Mutants impaired in IP3 metabolism show an increase in calcium response

In order to clarify whether IP3 production could be sufficient as a signal for calcium peak stimulation or whether further phosphorylation is required, we utilised yeast strains deleted either in IPK1 or ARG82. In both strains, the calcium response was not only detectable but was higher than in the corresponding wild-type strain (Figs. 1, 2). In the ipk1Δ strain, the calcium accumulation was slightly increased, as was observed previously in the strain carrying PLC1 on a multicopy plasmid (Tisi et al. 2002). In the arg82Δ strain, the calcium concentration increase was much higher, even at saturating concentration of glucose (100 mM), and the calcium peak appeared broader than in the wild-type strain, suggesting that the lack of IP3/IP4 kinase could enhance the accumulation of calcium in the cytosol.

Fig. 1
figure 1

Glucose-induced calcium signalling in an ipk1-deleted strain. Apoaequorin-expressing cells were starved for all nutrients for 2 h and then treated with coelenterazine as described in the Materials and methods. At time zero, 100 mM glucose was added. White squares Wild-type strain (W303-1b), black squares ipk1Δ strain

Full size image
Fig. 2
figure 2

Glucose-induced calcium signalling in an arg82-deleted strain. Apoaequorin-expressing cells were starved for all nutrients for 2 h and then treated with coelenterazine as described in the Materials and methods. At time zero, 100 mM glucose was added. White squares Wild-type strain (PJ69-4a), black squares arg82Δ strain

Full size image

Since the addition of diacylglycerol, the other product of Plc1 activation, was reported to activate the plasma membrane H+-ATPase (Brandão et al. 1994), we also tested whether this compound had any effect on calcium levels, but the exposition of cells to 2 mM diacylglycerol did not cause any change in the cytosolic calcium concentration (data not shown).

IP3 levels increase after glucose stimulation of nutrient-starved cells

In the wild-type strain, a small but reproducible accumulation of IP3 was observed after glucose stimulation of starved cells (Fig. 3). The low IP3 content of yeast cells (about 1 nmol g−1 wet weight) and the small amount of IP3 accumulated upon glucose stimulation could explain why it was reported in the literature that only nitrogen, but not glucose, stimulation induces IP3 production (Schomerus and Kuntzel 1992). The IP3 increase was small but significant and showed a characteristic oscillatory behaviour, which we observed in all experiments done and also with a different wild-type strain (see Fig. 5).

Fig. 3
figure 3

Inositol (1,4,5) triphosphate (IP 3 ) content after glucose addition to wild-type, PLC1-overexpressing and plc1-deleted strains. Cells were starved for all nutrients for 2 h. At least three samples were taken for determining the basal level of IP3 content, then 100 mM glucose was added and samples were taken at the indicated times. The IP3 relative increase was calculated by taking the basal level as 1. White squares Wild-type strain (W303-1a), black squares W303-1a strain carrying pWP101 (multicopy PLC1), black triangles plc1Δ strain

Full size image

Overexpression of PLC1 caused a small increase in the calcium response (Tisi et al. 2002). The accumulation of IP3 observed in this strain was also slightly higher than in the wild-type strain (Fig. 3) with an increase in the amplitude of oscillations, suggesting that the two events could be correlated. In contrast, no IP3 production was observed in the plc1Δ strain, which is consistent with the complete inhibition of calcium signalling. Moreover, the IP3 level decreased immediately after glucose addition (Fig. 3); and a similar lack of response was observed in a wild-type strain after the addition of 3-nitrocoumarin (data not shown).

The amplitude of IP3 oscillations increased also in the ipk1Δ strain (Fig. 4), but the greatest accumulation was observed in the arg82Δ strain (Fig. 5). Even if the relative increase in IP3 after glucose addition was not very high, this corresponded to a really high amount of IP3 accumulation, which correlated with the higher accumulation of calcium in the cytosol (Fig. 2). In this mutant strain, the oscillatory behaviour was less evident and a more persistent increase of IP3 was observed.

Fig. 4
figure 4

IP3 content after glucose addition in an ipk1-deleted strain. Cells were starved for all nutrients for 2 h. At least three samples were taken for determining the basal level of IP3 content, then 100 mM glucose was added and samples were taken at the indicated times. The IP3 relative increase was calculated by taking the basal level as 1. White squares Wild-type strain (W303-1a), black squares ipk1Δ strain

Full size image
Fig. 5
figure 5

Inositol (1,4,5) triphosphate content after glucose addition in an arg82-deleted strain. Cells were starved for all nutrients for 2 h. At least three samples were taken for determining the basal level of IP3 content, then 100 mM glucose was added and samples were taken at the indicated times. White squares Wild-type strain (PJ69-4a) black squares arg82Δ strain

Full size image

The arg82Δ strain showed a nearly 5-fold increase in the basal level of IP3, likely depending on its slow conversion in the absence of the Arg82 kinase. Saiardi et al. (2000) indicated a much higher IP3 basal level in this strain. However, those authors tested the IP3 content in cells grown in YPD after a 2-h shift to a non-permissive temperature (37 °C) whereas, in our experimental conditions, cells were starved of nutrients for 2 h. These completely different conditions could explain the contrast in our results.

cAMP signalling is not required for calcium signalling

Since we previously showed that both the Gpr1/Gpa2 system (Tisi et al. 2002) and hexokinase/glucokinase activity (Tisi et al. 2002; Tokes-Fuzesi et al. 2002) were required for glucose-induced calcium signalling, the calcium response appears to have the same activation requirements as the cAMP signalling pathway (Rolland et al. 2000). Therefore, we tested whether cAMP signalling could be a prerequisite for calcium signalling. A strain was used carrying the phosphodiesterase-encoding gene PDE1 on a multicopy plasmid, which was reported to completely abolish cAMP accumulation upon glucose stimulation (Ma et al. 1999). This strain was not defective in the glucose-induced calcium peak (Fig. 6). Moreover, we also tested a strain lacking both Ras-guanine nucleotide exchange factors (GEFs), Cdc25 and Sdc25, which was therefore completely impaired in glucose-induced cAMP signalling (Van Aelst et al. 1990). The CDC25 deletion is lethal in a wild-type background, so this strain needs a suppressor of lethality. We used two different suppressors: overexpression of the catalytic subunit of protein kinase A, Tpk1, or expression of a mutated version of Ras, Ras2Ile152, which was reported not to require a GEF (Camonis and Jacquet 1988). Both strains were able to generate a normal glucose-induced calcium peak (data not shown).

Fig. 6
figure 6

Glucose-induced calcium signalling in a cAMP signalling-impaired strain. Apoaequorin-expressing cells were starved for all nutrients for 2 h, then treated with coelenterazine as described in the Materials and methods. At time zero, 100 mM glucose was added. White squares Wild-type strain (W303-1a), black squares wild-type strain carrying YEplac195-PDE1

Full size image

Discussion

Since Plc1 is required for both glucose-induced calcium signalling (Tisi et al. 2002), glucose-induced activation of PI turnover and the stimulation of H+-ATPase (Coccetti et al. 1998), it is reasonable to suppose that a phospholipase C-dependent signalling pathway could also exist in yeast, involving a classic second messenger for PLC signalling, which is IP3. Unfortunately, the accumulation of IP3 is quite difficult to detect in yeast, because IP3 is quickly converted to inositol polyphosphates (York et al. 1999; Saiardi et al. 2000), rather then dephosphorylated to inositol monophosphate and inositol, following a peculiar scheme of inositol phosphate metabolism (Robinson et al. 1996). Moreover, IP3 has a very low basal level ( about 1 nmol g−1 wet weight).

It was reported that the addition of a nitrogen source to nitrogen-starved cells induces the generation of a peak of IP3 (Schomerus and Kuntzel 1992; Bergsma et al. 2001), whereas glucose addition to nutrient-starved cells was reported either to induce (Kaibuchi et al. 1986) or not to induce (Schomerus and Kuntzel 1992) IP3 accumulation. Surprisingly, nitrogen-induced IP3 accumulation is independent of Plc1 activity (Bergsma et al. 2001), indicating that IP3 can be produced through at least two different biosynthetic pathways. This could account for the basal level of IP3 in the plc1Δ strain being not dramatically lower than in the wild type, in our experimental conditions. The same effect on IP3 basal level in a Plc1-deficient background was observed in another organism, Dictyostelium, where IP3 can be produced in a Plc1-independent pathway (Van Dijken et al. 1995).

In our experimental system, it was possible to reveal a glucose-induced accumulation of IP3 (Fig. 3). The increase in IP3 was small and showed a characteristic oscillatory behaviour that was observed in all experiments and in two different wild-type yeast strains (W303-1a, PJ69-4a). Oscillations could be related to either a positive or a negative feedback mechanism, coupled to a non-linear behaviour caused either by a threshold or by a delay in response to a stimulus (Goldbeter 2002). At present, the basis for this oscillatory behaviour is not clear, but the indication that the oscillations are much less evident in the arg82 deletion mutant would suggest that IP3 phosphorylation could be relevant for this phenomenon.

Due to its rapid conversion to inositol polyphosphate in a wild-type strain, this accumulation was more clearly detected either in mutants impaired in IP3 phosphorylation or (with an increase in the amplitude and/or duration of the oscillations) in the Plc1-overexpressing strain (Figs. 4, 5). All these strains also showed a higher accumulation of calcium in the cytosol (Figs. 1, 2; Tisi et al. 2002), suggesting a possible involvement of this second messenger in the generation of a calcium pulse. In addition, it can be inferred that further phosphorylation of IP3 is evidently not required to elicit the calcium signal.

Interestingly, the higher basal level of IP3 in the arg82Δ strain does not involve a serious defect in the basal level of calcium. This could suggest that either IP3 is not the signalling molecule that triggers calcium entry from the external medium or the IP3 accumulated in the arg82Δ strain due to its genetic defect has no effect on calcium levels because of a different compartmentalisation. The former hypothesis leaves an open question: since diacylglycerol addition does not evoke any effect on the calcium level (data not shown) and IP3 is not responsible for stimulating calcium uptake, should a different mechanism be considered which does not involve any second messenger, e.g. a direct interaction between Plc1 and the calcium channel. However, the hypothesis of a proper IP3 localisation requirement for calcium signalling could also justify the different cellular behaviour following nitrogen stimulation. In fact, the IP3 generated independently of Plc1 does not trigger any calcium mobilisation (Bergsma et al. 2001), which could indicate that its accumulation resides in a different location from the IP3 generated via Plc1 activity after glucose stimulation.

PLC1 deletion causes the complete inhibition of not only the glucose-induced calcium entry (Tisi et al. 2002) but also the IP3 accumulation (Fig. 3), in agreement with the report that PI turnover was also completely inhibited (Coccetti et al. 1998). Further, the level of IP3 following glucose addition falls to undetectable levels, suggesting that not only could Plc1-mediated PIP2 cleavage be induced, but also inositol phosphate kinase activity could be stimulated. That could explain why the IP3 accumulation in a wild-type strain is so faint, due to the immediate transformation of IP3 into inositol polyphosphates.

The importance of inositol polyphosphates was recently stressed as regulators of the activity of the Swi/Snf ATP-dependent chromatin remodelling complex (Rando et al. 2003; Shen et al. 2003; Steger et al. 2003). This could suggest a possible role for the Plc1 pathway in transcriptional regulation induced by glucose. It is interesting to note that the Swi/Snf complex is involved in the induction of phosphate-responsive genes such as PHO5 and PHO84. Moreover, the plc1 null mutant can be partially rescued by activation of the PHO pathway either by genetic modification or by growth on a low-phosphate medium (Flick and Thorner 1998). This is consistent with a putative requirement of Plc1 to generate the inositol phosphates implicated in transcriptional activation of PHO pathway-controlled genes.

We previously found that glucose-induced calcium signalling requires the activation of the Gpr1/Gpa2 GPCR system (Tisi et al. 2002). As demonstrated for cAMP signalling (Rolland et al. 2000), glucose transport and phosphorylation are also required for the glucose induction of calcium uptake (Tisi et al. 2002; Tokes-Fuzesi et al. 2002). These observations prompted us to investigate whether cAMP signalling could be involved in the generation of calcium signalling. However, neither Ras activation by its exchange factors, Cdc25 or Sdc25, nor cAMP accumulation were required for cytosolic calcium elevation. This suggests that the two pathways immediately diverge after initial activation by the same glucose-sensing system. This finding is also in agreement with the previous observation that glucose-induced PI turnover (requiring Plc1 activity) was not dependent upon the Cdc25/Ras pathway in yeast (Frascotti et al. 1990).

The physiological significance of glucose-induced IP3/calcium signalling is still unknown. Since Pkc1p is required for glucose-induced H+-ATPase activation (Brandão et al. 1994; Souza et al. 2001), a connection could exist between Plc1 and Pkc1 activation upon glucose stimulation. These could provide different contributions to the general rearrangement occurring in yeast metabolism and physiology during the shift from a condition of nutrient starvation or slow growth in respiratory conditions to a condition of nutrient availability or rapid growth in fermentable conditions.