Throughout the biological kingdom, the cyclitol myo-inositol and its metabolites are involved in diverse physiological processes such as growth regulation, membrane biogenesis and osmotolerance, apart from their significance in eukaryotic signal transduction. The biosynthesis of myo-inositol follows an evolutionarily conserved pathway [1]. All myo-inositol-producing organisms studied to date generate the cyclitol from Glc6P (glucose 6-phosphate) via an internal oxidoreduction and aldol cyclization reaction catalysed by MIPS (inositol-3-phosphate synthase; formerly known as L-myo-inositol-1-phosphate synthase; EC 5.5.1.4) in an NAD⁺-dependent manner [2]. Subsequent dephosphorylation of the product leads to unsubstituted inositol. The MIPS enzyme has been reported from a host of diverse sources (e.g. higher plants, animals, green algae, fungi, bacteria and archaea) and has been considered to be an ancient protein/gene [1]. The structural gene for MIPS, termed ino1, was first identified and cloned in Saccharomyces cerevisiae and has meanwhile been sequenced from evolutionarily diverse organisms. Deduced amino acid sequences reveal that eukaryotic MIPS are remarkably conserved throughout their length (≥45% identity) and that the following stretches of amino acid residues are highly conserved: GWGGNNG, LWTANTERY, NGSPQNTFVPGL and SYNHLGNNDG. The GXGGXXG motif (Rossman fold) is involved in NAD⁺ binding and is typical for an oxidoreductase.

The reaction catalysed by MIPS is the committed and rate-limiting step in the synthesis of all inositol-containing compounds, making it an ideal target for genetic manipulation in our ongoing studies of the metabolism and functions of inositol phosphates/phosphoinositides in Dictyostelium discoideum. The typal characteristics of this cellular slime mould prompted the NIH to select it as a model organism for functional analysis of sequenced genes (http://www.nih.gov/science/models/d_discoideum/). D. discoideum has a complex life cycle with independent vegetative growth and multicellular developmental phases. Indeed this organism has unique advantages for the study of fundamental cellular processes resembling those in higher eukaryotes, which are absent or less accessible from other protists, such as yeast [3]. Examples in the unicellular organism are cell motility coupled with alterations in the actin/myosin-cytoskeleton, membrane trafficking especially associated with endocytosis, and signal transduction, as well as developmental aspects after initiation of differentiation, processes in which inositol derivatives such as inositol phosphates and phosphoinositides are most likely to be involved [4,5].

myo-Inositol is not an essential ingredient of well-defined culture medium [6], so cells are able to synthesize the compound de novo. Evidence given below supports the involvement of MIPS. Tritiated myo-inositol transferred into the cytoplasm by electroporation is incorporated into PtdIns within minutes [7]. Phosphorylation of PtdIns is accomplished by a family of five specific kinases [PIK1 (phosphoinositide kinase 1)–PIK5], whose corresponding genes have been cloned [8]. Site-directed mutagenesis provokes defects in the organization of the cytoskeleton and perturbations in differentiation. Certain double knockouts appear to be lethal. The phosphoinositide signalling cascade is involved in the processing of chemotactic stimuli (e.g. cAMP and folate). Stimulation of phospholipase C induces the formation of the second messengers DAG (diacylglycerol) and Ins(1,4,5)P₃ within seconds [7]. An inositol-polyphosphate 5-phosphatase participates in signal termination. Recently, a group of four inositol-polyphosphate 5-phosphatases from D. discoideum was cloned and characterized [9]. Hints about their substrate specificity were derived from overexpressed catalytic domains in Escherichia coli, and their possible physiological functions were implicated in the regulation of the signalling molecules PtdIns(4,5)P₂, PtdIns(3,4,5)P₃, Ins(1,4,5)P₃ and Ins(1,3,4,5)P₄. Further inositol phosphate metabolism seems to be very complex, with approx. 25 identified compounds in this organism and very little information about the enzymes involved [7]. Remarkable are the relatively large concentrations of InsP₆ (~0.3–0.6 mM) in the exponential growth phase while being cultured either on bacteria or in axenic medium. InsP₆ and the compounds containing energy-rich pyrophosphate groups, 6-PP-InsP₅ (InsP₇) and 5,6-bis-PP-InsP₄ (InsP₈), are accumulated in stationary cells and in spores, where each of these reaches nearly millimolar concentrations [10]. In line with such observations, it may be speculated that these compounds may fulfil storage functions, serving as a unimolecular source of inositol, phosphate and also metal ions, because of their complexation properties, so that they correspond to the role of phytic acid (mixed calcium/magnesium salt of InsP₆) in plant seeds. This should certainly be important for ancient organisms like amoebae, with a resistant stage in their life cycle.

A co-ordinated inositol phosphate/phosphoinositide metabolism depends on the supply of inositol. Our efforts to manipulate the central biosynthetic pathway by insertional mutagenesis combined with an extensive analysis of phenotypic and metabolic changes in the mutants generated should yield new information about the biological importance of some inositol derivatives and their metabolic interrelationship. For instance practically nothing is known about possible interconnections between the metabolism of the signalling molecules at low concentrations, and the highly concentrated InsP₆, InsP₇ and InsP₈.

D. discoideum were also cultured in suspensions of E. coli B/r [10¹⁰ cells/ml in 40 mM phosphate buffer (containing potassium phosphate), pH 6.8; 120 rev./min, 22 °C] or in association with bacteria on SM agar plates [12]. Cells were harvested by centrifugation (500 g, 5 min and 4 °C). The cell density and the cell size of D. discoideum were determined with a Coulter® counter (model Z2). Chemotaxis was analysed by the agar-well assay [13]. Differentiation was initiated by plating washed, exponentially growing cells on non-nutrient agar plates (15 g/l agar in 40 mM phosphate buffer, pH 6.8).

Aliquots of the cell suspension (equal to 1–5×10⁶ cells) were added to 10 ml of HL5 containing 500 μM myo-inositol and transferred to 80 mm Petri dishes. After 24 h, the medium was changed and the cells were kept in HL5 with 500 μM myo-inositol and 10 μg/ml blasticidin S (pUCBsr-INO1) or G418 (pDEX-INO1) for 8 days with a change of medium every 2–3 days. Potential mutants were screened by Northern blots and by enzymatic assays. Additionally, their growth in the presence or absence of myo-inositol was examined.

The metabolite 2,3-BPG was purified by HPLC on Mono Q (details described in the Analysis of inositol phosphates and ATP subsection) from an HClO₄ extract of 10⁹ Ddino1Δ1 cells (where Dd is Dictyostelium discoideum) pre-incubated for 24 h on a medium without inositol. After freeze-drying, this method yielded 50–75 nmol of 2,3-BPG. The homogeneity of the compound was confirmed as described below. 2,3-BPG (30 nmol) was subjected to a 4 h enzymatic degradation with 250 m-units of phosphoglycerate mutase (EC 5.4.2.1; Sigma). The reaction was carried out at room temperature (20 °C) in 1.5 ml of 75 mM Bis-Tris (pH 7.4), and in the presence of 1.7 mM 2-phosphoglycolate (Sigma). The enzyme works in the presence of phosphoglycolate as a 2-phosphatase. The reaction was stopped by boiling, the suspension was cleared by centrifugation, and aliquots were analysed by HPIC-CD (0 min, 0.2 mM KOH; 5 min, 0.2 mM KOH; 15 min, 15 mM KOH; 25 min, 40 mM KOH; 45 min, 55 mM KOH; 55 min, 70 mM KOH; 70 min, 90 mM KOH; flow rate 1 ml/min). This gradient was also applied to separate and quantify other glycolysis metabolites from cellular extracts (3-phosphoglycerate, 25.2 min; phosphoglycolate, 25.7 min; Fru(1,6)P₂ (fructose 1,6-bisphosphate), 31.7 min; 2,3-BPG, 37 min).

The lipid samples were deacylated by the method described previously [18]. Prior to HPIC-CD, hydrophobic contaminations were removed by an RP/H Maxi-Clean cartridge (Alltech). Aliquots of 2×10⁷ cells were used for analysis. To elute glycerophosphoinositol and the glycerophosphoinositol phosphates, we used a water/methanol mixture (75:25, v/v) with a KOH gradient (0 min, 5 mM KOH; 5 min, 5 mM KOH; 15 min, 30 mM KOH; 25 min, 30 mM KOH; 35 min, 60 mM KOH; 45 min, 60 mM KOH; 50 min, 90 mM KOH; 60 min, 90 mM KOH; flow rate 0.85 ml/min). Compounds and their corresponding retention times: glycerophosphoinositol, 4.54 min; glycerophosphoinositol 4-phosphate, 23.54 min; glycerophosphoinositol 3-phosphate, 24.83 min; glycerophosphoinositol 4,5-bisphosphate, 42.95 min.

Deacylated PtdIns could not be quantified by HPIC-CD, because further substances co-eluted with it (one of these is probably deacylated phosphatidylglycerol [19]). To determine cellular PtdIns concentrations, the deacylation product glycerophosphoinositol was recovered after HPIC-CD and converted in two steps into myo-inositol.

Freeze-dried samples derived from 5×10⁷ cells were dissolved in 400 μl of water, and 50 μl of 90 mM NaIO₄ was added to oxidize the glycerol backbone (20 min, room temperature, darkness). Excess NaIO₄ was destroyed by addition of 20 μl of 700 mM Na₂SO₃ (20 min, room temperature, darkness). Dimethylhydrazine [40 μl; 1% (w/v); reagent was adjusted to pH 4.5 with formic acid prior to use] was added, and the reaction mixture was incubated for 4 h to convert the resulting glycolaldehyde into Ins1P. Ins1P was dephosphorylated by treatment with 1 ml of 50 mM Mops (pH 9.0) and 10 units of alkaline phosphatase for 4 h at 37 °C. The reaction was stopped by heating at 95 °C for 2 min. The samples were deionized on a mixed bed ion exchanger (Amberlite® MB-3; Merck), and inositol was quantified by HPIC-IPAD.

ATP was determined after application of the samples to a Source 15Q HR 5/5 column, which was eluted by a linear gradient of HCl (0 min, 0.2 mM HCl; 15 min, 0.5 M HCl; 20 min, 0.5 M HCl; flow rate 1.5 ml/min). UV detection was performed at 254 nm (ATP, 13.5 min).

Activity was determined by a modification of the method of Van Lookeren Campagne et al. [23]. The assay was carried out on a microplate in a solution composed of the following components (per cavity): 10 μl of buffer (160 mM Bis-Tris, pH 7.0, 800 mM sucrose, 1 mM EDTA and 20 mM MgCl₂), 10 μl of water or a solution of the inhibitor 2,3-BPG (0.03–2 mM), 10 μl of the corresponding protein fraction and 10 μl of 0.64 mM Ins(1,4,5)P₃ solution. For further details, please consult one of our earlier publications [21].

A stringent requirement for generating vital inositol-auxotrophic mutants is the uptake of external myo-inositol that will bypass the deficiency in inositol biosynthesis. Our experiments and previous studies [24] monitoring the incorporation of myo-[³H]inositol showed that this is possible for D. discoideum by pinocytosis. Gene disruption by homologous recombination requires additionally that only a single copy of the gene exists. Southern-blot analysis of genomic DNA digested with restriction enzymes not present in the cDNA sequence of FC-AA11 detected only one hybridizing fragment with a ³²P-labelled cDNA probe. It is not possible to get mutants by a gene knockout if functional ino1 is essential for cell viability. In this case, it is preferable to choose the antisense strategy for gene silencing. For both strategies, vectors were constructed and expressed in E. coli. Transformation of AX2 cells results in three independent transformants by gene disruption (Ddino1Δ1–Ddino1Δ3) and 18 independent transformants by antisense mutagenesis (Ddino1as1–Ddino1as18). Two strains (Ddino1Δ1 and Ddino1as1) showing the most pronounced effects on MIPS expression were selected for further studies. Their molecular defects were confirmed by enzyme activity assays and Northern blots (Figure 1).

Figure 1 Molecular defects of the mutants Ddino1Δ1 and Ddino1as1 as demonstrated by MIPS activity assays and Northern-blot analysis

The method applied to determine MIPS activity is based on the complete dephosphorylation of the reaction products and quantification of myo-inositol after HPIC-IPAD. As a further benefit, it allowed the partial purification and the preliminary characterization of MIPS from the parent strain AX2 (for details on HPIC-IPAD and the enzyme characteristics, see Supplementary data at http://www.BiochemJ.org/bj/397/bj3970509add.htm).

Figure 2 Growth of Ddino1Δ1 and the parent strain AX2 at different myo-inositol concentrations

Figure 3 Morphological differences between Ddino1Δ1 and the parent strain AX2

In the first 5–6 h of starvation (half the doubling time), the cells increase in number in parallel with the inositol-supplemented control, suggesting that their internalized reserves are sufficient. Then the cell number enters a plateau, but the cells remain viable for the next 18–20 h, as demonstrated by their ability to resume growth after myo-inositol addition. Afterwards the cells lose their viability (Figure 4). Therefore the time period of 24 h seems to be optimal for observing pronounced effects of inositol starvation on the phenotype.

Figure 4 Survival of Ddino1Δ1 after sustained inositol deficiency

Inositol-auxotrophic cell lines from diverse organisms, including S. cerevisiae [25], Neurospora crassa [26] and mammalian cells [27], share a phenomenon called ‘inositolless death’. Under conditions that otherwise support growth, the cells die when deprived of myo-inositol. Shatkin and Tatum [28] argued that ‘inositolless death’ is caused by an imbalance between the rate of membrane growth and the synthesis of cytoplasmic constituents. In agreement with their hypothesis, life span of Ddino1Δ1 is prolonged when protein biosynthesis is blocked by cycloheximide or when they are additionally deprived of glucose, their main carbon source (results not shown).

In their natural environment, D. discoideum live on phagocytosis and digestion of bacteria. The parent strain can also be grown on agar plates in association with bacteria or in bacterial suspension. The mutants were unable to do so. This could be a consequence of the bacteria E. coli B/r and Klebsiella aerogenes we used, because these prokaryotes contain no myo-inositol. It seems puzzling that, in contrast with the growth defect observed on axenic medium, this was independent of myo-inositol supplementation, but it is known that pinocytosis rates of cells cultured on bacteria are almost negligible [29]. A deficient myo-inositol supply may cause a disturbed inositol phosphate/phosphoinositide turnover with potential impacts on vital processes, such as utilization of nutrients [30]. This aspect was studied further (see below).

Ddino1Δ1 showed only minor defects in chemotaxis and differentiation. In the small droplet chemotaxis assay, the migration of the cells towards cAMP on agar plates with or without myo-inositol was examined. Independently of the substrate chosen, gene inactivation does not affect chemotaxis (results not shown). In comparison with the parent strain plated on non-nutrient phosphate-buffered agar plates to initiate multicellular development, Ddino1Δ1 forms greater multiple-tipped mounds (Figure 3B), developing into fruiting bodies with the typical morphology (Figure 3C). On an average, the mutants need 20% more time to differentiate. These defects cannot be suppressed by inositol supplementation. Mounds with multiple tips were also observed for a double knockout of two PI3Ks (phosphoinositide 3-kinases) (DdPIK1 and DdPIK2) closely related to mammalian p110 PI3K [8]. Another report describes the identification of a developmentally regulated PI4K that is highly active in post-aggregate stages [31]. Altogether, these results give hints of a relationship between phosphoinositide metabolism and multicellular development.

In the course of development, D. discoideum cells do not ingest nutrients from their environment; cells of Ddino1Δ1 therefore have to access their internal reserves, especially for inositol. These resources seem to be sufficient to allow differentiation, even when the cells are inositol-starved for 24 h (middle trace of Figure 3).

Inoculated on medium without myo-inositol, spores are also capable of germinating, but the resulting cells die within the next few hours. In the presence of myo-inositol, however, the cells resume growth normally, reflecting once more their need of inositol for vegetative growth. It can be concluded that the continuous generation of at least one inositol-containing compound is essential for vegetative growth.

Figure 5 Fluid-phase uptake and phagocytosis of bacteria

The inositol-auxotrophic mutants are unable to grow on bacterial lawns or in suspensions of bacteria. Is this a consequence of a disturbed phagocytosis? In the presence of myo-inositol, phagocytosis of Ddino1Δ1 cells is comparable with that of the parent strain. No significant difference was observed for the internalization of E. coli B/r, K. aerogenes or latex beads (Figure 5B, shown only for FITC-labelled E. coli B/r). At 16 h of myo-inositol starvation, the uptake rate of bacteria and the phagocytosis capacity of the cells are reduced. After 24 h myo-inositol starvation, the cells completely lose their ability to phagocytose (Figure 5B).

In contrast with pinocytosis under advanced inositol deficiency, the mutants are defective in the uptake of particles. The results suggest that the processes have different dependencies on inositol-containing compounds. Phagocytosis and subsequent digestion consist of a number of stages, including the binding of a particle to the cell surface, engulfment of the particle by pseudopod extension, and fission and fusion reactions to form phago-lysosomes. It is well known that phosphoinositides are required for the internalization process in a number of cell types [32,33] and that they play a role in phagosomal maturation in D. discoideum [30]. This aspect will be taken up again when considering the results of the metabolite analyses.

Table 1 Cellular concentrations of several inositol metabolites and 2,3-BPG in the strains AX2, Ddino1Δ1 cultured on medium with 500 μM myo-inositol (+) and Ddino1Δ1 after 24 h of myo-inositol starvation (−)

The level of Ins3P, the actual product of the conversion of Glc6P by MIPS, was determined by HPIC-CD. It is maintained between 0.6 and 1.0 μM. It should be noted that an achiral method was used, and the enantiomer Ins1P could be present as well. Furthermore, other metabolic pathways might lead to the formation of Ins1P and Ins3P.

An isotope dilution assay was used to quantify Ins(1,4,5)P₃. No significant alteration of the Ins(1,4,5)P₃ concentration (0.6–0.8 μM) occurs after inositol starvation.

As found for plant seeds, some highly phosphorylated inositol phosphates, particularly inositol hexakisphosphate, may fulfil storage functions and serve as a unimolecular source of myo-inositol, phosphate and also metal ions on account of their complexation properties [34]. Typical intracellular concentrations of InsP₆ (up to 600 μM), InsP₇ (up to 100 μM) and InsP₈ (up to 300 μM) in D. discoideum [10] are high enough, so that complete dephosphorylation could compensate for a deficiency of intracellular myo-inositol over a longer period. HPLC–MDD analysis shows only minor fluctuations in the levels of inositol hexakisphosphate and the diphosphoinositol phosphates after inositol starvation (Table 1 and Figure 6), but InsP₆ concentration tends to decrease, whereas InsP₇ and InsP₈ concentrations increase. Taking into account that the mean cellular volume of the mutants is smaller, especially after myo-inositol starvation (see legend of Table 1), the apparent concentration of a metabolite that is not subject to variations will increase. A small proportion of the InsP₆ pool is actually decomposed, but the phenotypic changes and the inositol analyses indicate that this is not enough to compensate for the inositol demand.

Figure 6 HPLC–MDD analysis of inositol phosphates from Ddino1Δ1 extracts

Various explanations are conceivable for why such high concentrations of the highly phosphorylated inositol phosphates are conserved. A simple explanation may be that their metabolism is too slow to meet the needs for inositol in vegetative growth. From metabolic labelling studies of amoebae with [³H]myo-inositol, it is known that the marker accumulates relatively slowly with the compounds InsP₆, InsP₇ and InsP₈ [24]. However, some essential function could be fulfilled by those compounds later in development. As we have shown, differentiation is not fundamentally impaired even if Ddino1Δ1 cells are inositol-starved for 24 h (Figure 3). In contrast, the cells die afterwards, when development is not initiated. Two recent publications describing the gene inactivation of the inositol phosphate kinases involved in the biosynthesis of highly phosphorylated inositol phosphates in mammals confirm the physiological relevance of all or some of these metabolites in development [35,36], and this may be also the case for D. discoideum [10]. However, deletion of the InsP₆-kinase abolished the synthesis of InsP₇ and InsP₈ and resulted in a strain with accelerated aggregation but unimpaired differentiation [37], so it is unlikely that the pyrophosphorylated compounds are significantly involved. InsP₆ or the pentakisphosphates remain to be considered. We are currently studying an inositol pentakisphosphate 3/5-kinase [38] that participates in InsP₆ homoeostasis, and hope to obtain further information about a possible function of such a kind soon by inactivating the corresponding gene.

Surprisingly, in the metabolism of inositol-starved mutants, there accumulates to a considerable extent a substance that is not a primary inositol metabolite (Figure 6A). On the basis of co-elution of the purified compound with authentic standards under several sets of HPLC conditions and a specific enzymatic degradation in vitro with concomitant identification of the products, we were able to identify it unequivocally as 2,3-BPG. Both Ddino1Δ1 cells and those of the antisense mutant Ddino1as1, which were generated by a completely different method that targets ino1 expression, accumulate 2,3-BPG under inositol starvation (the effect on cellular 2,3-BPG for Ddino1as1 is approx. 1/3 of that described in Table 1). The potential physiological relevance of this observation is discussed later.

PtdIns is the exclusive precursor for all PtdInsPs that play a decisive role in cellular signalling. The concentration of the downstream metabolite PtdIns4P is nearly unaffected after inositol starvation (Table 1). This also holds for the PtdIns3P concentration. The level of the Ins(1,4,5)P₃ precursor PtdIns(4,5)P₂ decreases, but not so drastically as PtdIns. Unfortunately, the sensitivity of the HPLC system does not permit the direct determination of other biologically relevant PtdInsPs, especially of PtdIns(3,4,5)P₃. It can be concluded that their concentrations are clearly below 0.2 μM. In comparison with PtdIns, the intracellular levels of the PtdInsPs investigated are not as strongly affected. It seems that a mechanism exists that stabilizes the levels in response to internal disturbances of inositol homoeostasis. In a genetically engineered strain with inositol biosynthesis destroyed, a compensation of this kind can be only temporary. Therefore, after 24 h, re-synthesis of PtdInsPs should only be possible to a very limited degree. Experiments with macrophages indicate that phagocytosis is accompanied by localized biphasic changes in PtdIns(4,5)P₂ concentration, which contribute to cytoskeletal remodelling [32]. The observed failure in phagocytosis after 24 h of inositol starvation (see Figure 5B) could be a consequence of an impaired PtdIns(4,5)P₂ metabolism. Thereafter the cells also lose their vitality (see Figure 4). Of course this may happen for several reasons, but one possible explanation could again be PtdIns(4,5)P₂. Recent publications on the physiological functions of PtdIns(4,5)P₂ indicate that continuous synthesis of PtdIns(4,5)P₂ is essential to maintain the integrity of the Golgi apparatus [41]. Disintegration of this organelle will inevitably lead to cell exitus.

There is at least one ancient function of 2,3-BPG in glycolysis that makes it indispensable for all eukaryotic cells. To maintain the active form of phosphoglycerate mutase, catalytic amounts of 2,3-BPG are needed. In vitro, 2,3-BPG has often been used as a competitive inhibitor of inositol-polyphosphate 5-phosphatases in enzyme assays, and in the case of insulin-secreting pancreatic islets the physiological relevance of this inhibition was postulated [43]. There are indications of a reciprocal action: that inositol phosphates, especially Ins(1,4,5)P₃, interact with aldolase [44], and that the Ins(1,4,5)P₃ receptor associates with glyceraldehyde-3-phosphate dehydrogenase [45]. These literature results encouraged us to analyse the time course of the changes in cellular inositol, Ins(1,4,5)P₃ and phosphoinositide concentrations in relation to the 2,3-BPG concentration, as provoked by myo-inositol starvation in the first 24 h (Figures 7A and 7B). This might be an indicator of a causal relationship.

Figure 7 Kinetics of the concentrations of selected inositol metabolites and 2,3-BPG in the strain Ddino1Δ1 subjected to inositol starvation

Ddino1Δ1 cells cultured on synthetic medium supplemented with 500 μM myo-inositol show artificially high cellular myo-inositol levels due to filled endocytic compartments. Thus, to get a realistic picture before the experiment was started, the cells were pre-incubated for 2 h, just the time needed to reach a new equilibrium between pinocytosis and exocytosis and when the myo-inositol levels were still slightly higher than the basal level of the parent strain (40–60 μM). The intracellular myo-inositol concentration of the mutant falls very far below this value in the first 2 h (Figure 7A). Afterwards it decreases only slightly, reaching a plateau of approx. 3–4 μM at 14 h of starvation. Directly opposed are the alterations in the glycolysis metabolite 2,3-BPG. Its concentration in the parent strain and Ddino1Δ1 (+) varies between 2 and 3 μM. Coinciding with a deficit in cellular myo-inositol, after 2 h the level rises 4-fold and then continuously to a maximum at 14 h, when it remains at a level up to 50 times that of basal. In contrast, the cellular Ins(1,4,5)P₃ concentration fluctuates only within the error margins over the whole 22 h (0.6–0.8 μM nearly equal to [Ins(1,4,5)P₃]_i in AX2, see Table 1). On starvation, the PtdIns level decreased below the basal value in the first 2 h, and the decline came to rest after approx. 14 h at a low level of approx. 6 μM (Figure 7B). Thus the time dependencies of the myo-inositol and the PtdIns concentrations show striking similarities in the progression and the minimal value finally reached, which is in both cases approx. 1/10 of the value found in the parent strain AX2. Despite the considerable loss of PtdIns, the concentration of the downstream metabolite PtdIns4P is subject to only minor variations over the 22 h examined. This also applies for the PtdIns3P concentration (results not shown). The level of the Ins(1,4,5)P₃ precursor PtdIns(4,5)P₂ decreases for 6 h to approx. 1/3 of the initial value with a slight tendency to recover afterwards. Of the metabolites that vary, the PtdIns(4,5)P₂ level normalizes fastest after a renewed myo-inositol supplementation. It needs only 4 h, followed by PtdIns and 2,3-BPG (~8 h; results not shown).

We hypothesized that the homoeostasis of the Ins(1,4,5)P₃ level and the limited breakdown of PtdIns(4,5)P₂, only lasting approx. 6 h, is associated with the increase in 2,3-BPG acting in vivo as an inhibitor of inositol-polyphosphate 5-phosphatases. Vice versa, the observed delayed 2,3-BPG degradation in response to myo-inositol addition to starved cells may facilitate the rapid replenishment of the PtdIns(4,5)P₂ pool. A purified inositol-polyphosphate 5-phosphatase [22], highly expressed in the growth phase, with the characteristics of the recently cloned Dd5P4 [9], was subjected to inhibition studies. It is a type II 5-phosphatase catalysing the dephosphorylation of soluble Ins(1,4,5)P₃ [less effective with Ins(1,3,4,5)P₄] and the membrane-bound PtdIns(4,5)P₂ [less effective with PtdIns(3,4,5)P₃]. The enzyme possess a K_M value of 40.2±3.8 μM for Ins(1,4,5)P₃ and is inhibited by 2,3-BPG in a competitive manner [22]. After additional inhibition studies (Figure 8), the K_i was determined according to [46]. A K_i value of 40.6±2.3 μM indicates that the substrate and the competitive inhibitor are bound with nearly equal affinities to this enzyme. The cellular concentrations of substrate and inhibitor as mentioned above suggest that this inhibition might be relevant in vivo. Further experiments are needed to confirm our hypothesis that 2,3-BPG variations influence cellular levels of signalling molecules by acting on the activity of 5-phosphatases. First of all, however, there are serious questions of how these metabolic pathways are interconnected and what triggers the drastic 2,3-BPG increase as a consequence of the inositol deficiency.

Figure 8 Study on the inhibition of an inositol-polyphosphate 5-phosphatase (Dd5P4) by 2,3-BPG

Recently, a further unexpected link between glycolysis and myo-inositol metabolism was found in S. cerevisiae [47]. Mutants with a loss-of-function allele for triose phosphate isomerase exhibited inositol-auxotrophy and showed an ‘inositolless death’ phenotype. A proposed molecular explanation was presented; the mutants accumulate cellular dihydroxyacetone phosphate 30-fold over basal, which itself inhibits MIPS in vitro. Unfortunately, data for the cellular myo-inositol levels are lacking, and myo-inositol depletion was merely deduced from the observed inositol-auxotrophic phenotype. Nevertheless, considering these results and those presented here, there do exist unexplored interdependencies between glycolysis and myo-inositol metabolism. Further research on this topic may for instance explain why cellular alterations in inositol levels have been implicated in the aetiology of diabetic complications [48].

We are grateful to Dr W. V. Turner (University of Wuppertal) for a critical reading of this paper. We are indebted to all of the teams involved in the D. discoideum sequencing projects, particularly ‘The Dictyostelium cDNA project in Japan’. We thank Dr H. Lemoine (University of Duesseldorf, Duesseldorf, Germany) for carrying out one of the isotope dilution assays and advising us on this area. This work was supported by the Deutsche Forschungsgemeinschaft (grant VO 348/3-1).