Characterization of Metal Ion-induced [3H]Inositol Hexakisphosphate Binding to Rat Cerebellar Membranes*

The binding of [3H]inositol hexakisphosphate ([3H] InsP6) to rat cerebellar membranes has been charac-terized with the objective of establishing the role, if any, of a membrane protein receptor. In the presence of EDTA, we have previously identified an InsP6-bind-ing site with a capacity of -20 pmol/mg protein (Hawk-ins, Biochem. Biophys. Res. Com-nun. 167, 819-827). However, in the presence of 1 mM Mg2+, the capacity of [3H]InsP6 binding to membranes was increased -9-fold. This enhancing effect of Mg2+ was reversed by addition of 10 p~ of several cation chelators, suggesting that the increased binding required trace quantities of other metal cations. This is supported by experiments where it was possible to saturate binding by addition of excess membranes, despite not significantly depleting radioligand, pointing to removal of some other factor. Removal of endogenous cations from the binding assay by pretreatment with chelex resin also prevents the


The binding of [3H]inositol hexakisphosphate ([3H]
InsP6) to rat cerebellar membranes has been characterized with the objective of establishing the role, if any, of a membrane protein receptor. In the presence of EDTA, we have previously identified an InsP6-binding site with a capacity of -20 pmol/mg protein (Hawkins, P. T., Reynolds 167,[819][820][821][822][823][824][825][826][827]. However, in the presence of 1 mM Mg2+, the capacity of [3H]InsP6 binding to membranes was increased -9-fold. This enhancing effect of Mg2+ was reversed by addition of 10 p~ of several cation chelators, suggesting that the increased binding required trace quantities of other metal cations. This is supported by experiments where it was possible to saturate binding by addition of excess membranes, despite not significantly depleting radioligand, pointing to removal of some other factor. Removal of endogenous cations from the binding assay by pretreatment with chelex resin also prevents the Mg2+-induced potentiation. Consideration of the specificity of the chelators able to abolish this potentiation suggested involvement of Fe3+ or A13+. Both these ions (but not several others) were able to increase [3H]InsP6 binding to chelex-pretreated membranes at concentrations of 1 PM. It is possible to demonstrate synergy between Fe3+ and Mg2+ under these conditions. We propose that [3H] InsP6 may interact with membranes through non-protein recognition, possibly via phospholipids, in a manner dependent upon trace metals. The implications of this for InsP6 biology are considered.
Inositol hexakisphosphate is usually found a t concentrations of between 10 pM and 1 mM in most, if not all, plant and animal cells (cg. Refs. 1-7). However, its functions remain largely mysterious. Studies of its metabolism suggest * This work was supported in part by grants from Perstorp Pharma (to P. T that the synthesis of InsP6' is not directly linked to inositol phosphates or lipids involved in signal transduction (7). There are several reports which suggest that it has extracellular actions to excite nerve cells (8)(9)(10)23). Additionally, a number of intracellular roles have been proposed for it, such as acting as a phosphate store or antioxidant (11,12). In an effort to learn more about its biology, we and others (13,14) have carried out studies with [3H]In~P6 to see if it can bind to membranes. A membrane-binding site might be expected to mediate any physiological extracellular actions of InsP6 and might also be involved in a more general intracellular "housekeeping" role. In our previous study, working in a buffer containing 5 mM EDTA, we described a site of high capacity associated with most neuronal structures in the brain (13). Nicoletti et al. (14) also described a membrane-binding site for InsPs, in rat cerebral cortical membranes, which was similar to the site found in the cerebellum. In an effort to understand the biological importance of these InsP6-binding sites, we investigated the effects of various physiologically important cations (e.g. Mg2+ and Caz+) on [3H]In~Ps binding. This has led us to discover that the binding is extremely sensitive to trace quantities of certain metal ions and that this has important implications for assessing the significance of this binding.

MATERIALS AND METHODS
Membrane Preparation-Cerebella were removed from rats and homogenized in 10 volumes of 5 mM EDTA, 20 mM Tris, pH 7.7, and crude membrane fractions prepared by centrifugation (35,000 X g, 30 min). Membranes were resuspended, washed (1 volumes of 20 mM Tris, pH 7.7), and resuspended at -0.2 mg protein/ml (in 20 mM Tris, pH 7.7) together with other ions or chelators as required (see the figure legends). All operations were carried out a t 4 "C.
In certain experiments the membranes and radioligand were treated with chelex resin to remove endogenous cations. Membranes (40 ml, see above) were incubated with chelex resin slurry (10-ml packed volume) for 15 min at 4 "C, followed by removal of resin by a brief centrifugation (2000 X g, 5', 4 " c ) . [3H]InsP6 (1 ml of 50 nM [3H]InsPG) was incubated with chelex resin (0.2-ml packed volume) for 45 min a t room temperature and the supernatant recovered by brief centrifugation through a small plastic column (Kontes, 10-ml plastic column with 0.45-PM filter attached).
Binding Assays-Binding was carried out a t 4 "C for 90 min in 20 mM Tris, pH 7.7, with additions as described under "Results" section and terminated by centrifugation, as described earlier (13). Routine assays were performed in a final volume of 1.0 ml with 0.5 nM [3H] InsPG (-90,000 dpm/assay) and -0.2 mg/ml membrane protein. All metal ion solutions were initially dissolved to a final concentration The abbreviations used are: InsPB, InsP1, InsP6, InsPs, inositol tris-, tetrakis-, pentakis-and hexakisphosphate (isomers are numbered according to IUPAC recommendations (38)); HPLC, high pressure liquid chromotography.  (13,15). The resulting radioligands had specific activities of approximately 80 Ci/mmol. The unlabeled InsPS isomers were prepared as described previously (15).
HPLC Analysis of pH]InSP6 Metabolism-This was done as described previously (13).
Analysis of Binding Data-Best values f standard errors of the parameters were obtained from non-linear regression analysis using the Hanvell Library routine VBQllA (27) and the following equation.

RESULTS
In our initial study (13), the binding of [ Fig. 1. This rate plot resembles that seen in 5 mM EDTA, 100 mM KC1 (13) in that both association and dissociation rates appear t o be biphasic, with the more rapid components occurring too quickly to be measured accurately by a microfuge binding assay. The specificity of binding in the presence of 1 mM Mg2+ is shown in  (15,24); the potency order for inhibition of InsPs binding is DL-Ins(1,2,3,5,6)P5   (1) taken at the end of the binding experiments indicated that [3H]InSP~ was not significantly metabolized under these conditions (data not shown). It should be noted that the binding curves for the InsP5 isomers all have Hill slopes significantly less than unity which suggests the presence of multiple or interacting binding sites.
As can be seen from Table I, in the presence of 1 mM M e there is a 9.4-fold increase in the amount of InsPs which is membrane-associated. Since [3H]In~Ps is present in these experiments at concentrations well below its apparent Kd (estimated at 60 n M in 5 mM EDTA, Ref. 13), the increased binding could, in principle, be due to either an increase in the affinity or the capacity of the sites, or some combination of these effects. However, although a contribution from a modest shift in affinity cannot be ruled out, the apparent IC50 of this site(s) for InsPG in the presence of 1 mM Mg2+ was found to be -100 nM (Table 11), and thus it is likely that M$+ increases the total capacity of the InsPs-binding site(s).

Metal Ion-potentiated fH]InsP6 Binding
A variety of pharmacologically active substances were screened to see if any of them could alter [3H]In~Ps binding in the presence of 1 mM Mg2+. Only isoprenaline inhibited binding (legend, Table 111). This was not acting via 6-adrenoceptors as deduced from three pieces of evidence: (i) the effect was not stereospecific (difference between inhibition produced by (+)isoprenaline and (+)isoprenaline = 9.3 + 5.4%); (ii) the effect was not blocked by propanolol; and (iii) the structure activity relationship for the effect was not that predicted for activation of 6-receptors (Table 111). Indeed, the biologically active part of the molecule was the catechol moiety. The dose-response curves for isoprenaline and catechol are almost superimposable (Fig. 2 Compounds such as catechol, with two vicinal hydroxyl groups are good chelators of divalent and trivalent metal ions  (17). Clearly chelation of magnesium is unlikely to explain the effectiveness of these compounds, since they are active at concentrations which remove only 1% of this metal. However, in the presence of 1 mM Mg2+, a second ion which is present in trace quantities could be essential for the observed increase in binding. Catechol would therefore produce its effect by removing this hypothetical second cation. To test this possibility, a variety of structurally distinct metal ion chelators were added to incubations at concentrations of 1-100 p~, to see if they could inhibit the Mg2+ potentiated binding. As can be seen in Table IV, all of the compounds tested were active at 10 p~ to inhibit this binding. So far as we are aware, the only property they have in common is their ability to chelate metal ions. While EGTA, tetrakis-2-pyridylmethylethylenediamine and maltol are relatively non-selective, desferrioxamine shows a marked preference for trivalent cations such as Fe3+ and A13+ (25).
Other evidence supports the notion that the key factor in the Mg2+ potentiated binding is a second metal cation found in the assay buffer and ligand preparation. In Fig. 3a it can be seen that the [3H]InsP6 binding saturated at concentrations of membranes greater than 1 mg/ml, despite the fact that the bulk (>80%) of the radioligand remained unhydrolyzed and available for binding (assessed by HPLC chromatography of the supernatant at the end of the binding assay, Fig. 3b). This is consistent with the second ion being limiting, such that it was depleted by adding more tissue before the free concentration of [ 3 H ] h~P s was itself significantly reduced. If this "depleted supernatant was removed from the membranes after the binding assay and added to fresh membranes (pretreated with 10 PM isoprenaline, 1 mM MgC12, to remove any endogenous source of the second ion), binding was greatly reduced compared to that seen when fresh supernatant and [3H]InSP6 were added, (Table VA). In contrast, if fresh [3H]In~Ps and assay buffer were added to membranes apparently saturated with bound radioligand (see Fig. 3a), then it was possible to get a further increase in binding, consistent with the addition of more of the limiting second ion (Table VA). If the membranes and the solution containing the radioligand were treated with chelex resin (to remove endogenous cations) prior to use in a binding study, then the observed M$+ enhancement was only 16 k 2% of that seen normally (Table VB). It is interesting to note that if only the radioligand solution was chelex treated prior to addition to the binding assay, then the Mg2+ potentiation was still reduced by 50% (Table VB). This suggests that the radioligand solution itself may contribute the second ion in these experiments. Because [3H]InsP6 is prepared by lyophilization from 2 M ammonium formate (15), it is possible that the hypothetical ion becomes concentrated at this stage. (For example, using data provided for BDH Analar Grade ammonium for-  To investigate the nature of the second ion involved in M$+-potentiated [3H]In~Ps binding, various ions were added t o membranes at concentrations between 10 l M and 1 mM, in the absence of Mg2+, to see if they could promote binding. Zn2+, AI3+, Pb", and Fe3+ were particularly effective in promoting up to 8-fold increases in binding a t concentrations 5 1 0 PM (Table VI and data not shown), whereas Cu2+, Ni", Co2+, Ba2+, and Ca2+ were effective only a t much higher concentrations (data not shown). The mechanism of action of the cations under these conditions is difficult to interpret; they could be mimicking the effect of the unknown cation, or synergizing with it in a more complicated interaction. In an attempt to provide a better reconstituted system, the membranes and radioligand were each pretreated with chelex resin t o remove endogenous cations, and the effects of readdition of various cations was examined. When added back at 1 PM, only A13+ and Fe3+ were able to cause a substantial increase in [3H]In~Ps binding (Table VI). Furthermore, it was possible t o demonstrate a synergy between Mg2+ and Fe3+. As can be seen in Fig. 4, the interactions between the two cations are complex. At high concentrations of Fe3+ (10 PM), addition of Mg2+ makes little difference to the increase in binding. However, at lower concentrations of Fe3+ (<1 l~) , Mg2+ increases the binding markedly. In the absence of added Fe3+, simply increasing the Mg2+ concentration will increase binding by -2-fold. However, it is unclear whether this is a direct effect InsP~ in the assay of 0.5 nM) was added to some of the samples at this stage. All samples were incubated for 90 min a t 4 "C, after which time they were pelleted in a microfuge and the supernatants separated from the membrane pellets (see "Experimental Procedures"). For the assays which contained [3H]InSP~, the membrane pellets were counted to determine the amounts of [3H]InSP6 bound and the supernatants were used as "depleted supernatant in the next stage of the experiment (the "depleted" supernatants contained 44,900 f 100 cpm/ml of [3H]InsP6 at this stage). The membrane pellets from samples which did not contain [3H]InsP6 and which were originally resuspended in buffer containing 1 mM EGTA were used as the source of "resuspended membranes" in the next stage of the experiment. The resuspended membranes were then incubated in a final volume of -1 ml with either "depleted supernatants" or "fresh supernatants" (fresh supernatants were prepared using incubation buffer containing 0.    responsible for Mg2+-potentiated [3H]In~P6 binding in the non-chelex-pretreated binding assay. However, this explanation does not explain all the observations. Remarkably, addition of the iron chelator desferrioxamine (10 p~) does not inhibit the Fe3+-or A13+-induced increase in binding in a chelex-pretreated binding assay but actually increases the binding (data not shown). This phenomenon implies a complex interaction between different metal ions, InsP6, and the membrane-binding site.
We have investigated ['HH]InsPs binding to membranes prepared from a number of different rat tissues. In every tissue examined (heart, liver, kidney, spleen, lung, and brain (cerebellum, forebrain, hindbrain, and cortex)), we found significant InsP6-displaceable [%]InsPs binding (data not shown). Evidence that similar binding sites were being measured in these tissues is provided by the similar amounts of radioligand displaced by 100 nM InsPs (between 32 and 48%). The degree of metal ion-potentiated binding appeared to vary quite widely between different tissues (e.g. isoprenaline inhibited binding by 88% in cerebellum but by only 44% in kidney). Autoradiographic examination of [3H]InsP6-binding sites in rat brain, in the presence of 1 mM Mg", showed a distribution similar to that described previously in 5 mM EDTA (13, data not shown), suggesting that Mg2+ did not create new sites in regions not previously expressing [%]InsP6 binding. However, without knowing the endogenous concentrations of the relevant metal ions or their buffering capacity in the various tissues it is obviously impossible to arrive at a more quantitative estimate of the degree of metal ion-potentiated binding in each tissue.

Specific [%H]
InsPs binding to cerebellar membranes can be dramatically potentiated by Mg2+ (1 mM), and this effect is dependent on limiting quantities of at least one further trace metal ion. The "metal ion potentiated" binding resembles the binding previously described in the presence of 5 mM EDTA (13), in that it has a similar affinity for InsPs and has similar kinetics. The mechanism of this potentiation is uncertain, but it is likely to result from an increase in the capacity of the [3H]InsP6-binding sites. Furthermore, with the addition of increasing amounts of cations it cannot be shown to be saturable. These results are in broad agreement with an observation made by Nicoletti et al. (14), who reported that various divalent cations (25 p M ) potentiated [3H]In~Ps binding to membranes prepared from primary cultures of rat cerebellar granule cells and may explain why they were unable to observe saturable binding at high membrane protein concentrations (their membranes were not prepared in the presence of cation chelators).
The relationship between specific [3H]InsPs binding in the presence of 5 mM EDTA and the metal ion potentiated binding is unclear. Given that InsPs is an excellent chelator of cations (e.g. 1,18,19), InsP6 may be complexed to endogenous metal ions even in the presence of excess EDTA, and accordingly it might be that all InsPs-membrane interactions require some form of metal ion participation (this would be analogous to kinase recognition of Mg ATP). However, chelators such as desferrioxamine, although they can reduce InsPs binding, do not abolish it. Therefore, the small contribution of a metal ion-independent binding site will be masked by the much greater metal ion-potentiated binding. A strong argument in favor of a separate membrane protein being responsible for [3H]In~P6 binding in the presence of EDTA is the recent purification of an inositol polyphosphate-binding protein from solubilized rat cerebellar membranes (26). This purified protein exhibits similar recognition characteristics in the presence of EDTA to [3H]In~Ps binding to intact membranes. We do not know whether InsPs binding to this purified protein is potentiated by transition metal ions, although we think it is unlikely that metal ion-potentiated InsPs binding is mediated by a specific protein (see below).
The data suggest multivalent cations influence the interactions of InsP6 with biological membranes. The mechanism of the metal ion InsPs membrane association is unexpectedly complex. When chelex resin is used to remove endogenous cations from binding solutions and membrane preparations, then relatively high concentrations of Fe3+ and A13+ (1 p~) can enhance InsP6 binding. It may be that some divalent cations on their own are also able to promote binding when present at concentrations in excess of 1 mM. However, we have obtained clear evidence that in the case of 1 mM Mg2+, the enhanced binding in non-chelex-treated binding assays is possibly due to small quantities of a second ion, perhaps Fe3+ or Al". Although readdition of Fe3+ to chelex-treated membranes can mimic some aspects of the situation, we find in natural membranes that there are paradoxical potentiation effects of the iron chelator desferrioxamine, which indicate that the reconstituted ionic conditions may not match physiological conditions.
Although a number of models may explain how the different metal ions can act together to promote InsP6 binding, a significant constraint on these models is our ignorance of the nature of the [3H]InsP6-binding site. The metal ion-potentiated site is of very high capacity; we have been unable to obtain convincing data that it can be saturated, and it seems t o be ubiquitous in membranes from mammalian tissues. Taken together, these data argue against a specific membrane protein being the site of metal ion-potentiated InsP, binding, and argues in favor of a more abundant membrane component, possibly negatively charged lipids or derivatives of them. This suggests that one explanation for the role of metal ions in InsP, binding is that they act as "bridges" between the InsP, and the negatively charged phospholipid phosphate groups of membranes. The more heavily charged trivalent metal ions might be expected to be particularly effective in taking part in the phosphate-metal ion phosphate complex required by this scheme (16)(17)(18). M$+ could interact directly with the InsPs, increasing its affinity for the second ion, or perhaps allowing it to take part more readily in phosphate bridge complexes with the membranes. InsP, is an excellent chelator of metal ions, and in some cases the binding of one ion may increase its affinity for a second (20). However, an alternative mechanism is for large quantities of M$+ to saturate metal ion-binding sites of the membranes, in the process increasing the effective concentration of the second ion in solution. In this scenario, in the absence of Mg2+, the second ion is all bound to the membrane at sites which are unable to allow the formation of the InsPs-ion sandwich. InsPs by itself is unable to strip the second ion from these non-productive binding sites because it does not have a sufficiently high affinity for this metal ion. The "second ion" promoted potentiation of InsP6 binding is consistent with most of our data. However, we cannot yet provide an explanation of why chelators can promote InsP, binding under conditions when certain ions are added back. A full explanation of the effects of metal ions on InsPs binding will require a much more extensive study of both the occurrence and concentrations of various metal ions in the binding assays and the chemistry of metal ion-InsPs interactions.
The biological significance of the metal ion-potentiated binding of [3H]InsP6 cannot be addressed without further study. It is not clear whether, under physiological conditions, traces of appropriate metal cations would allow significant amounts of InsPs to become membrane-bound. It seems likely, however, that this metal ion-potentiated binding may affect certain in vitro experiments. It may, for example, confound attempts to assess the true intracellular distribution of InsPs or it may lead to the unwitting introduction of InsP, (and/or its associated cations) into assays with membranes (e.g. as an inhibitor of inositol phosphate phosphatases, Ref. 21).
This work emphasizes the ability of InsPs to act as a quite remarkable ion chelator. It seems that certain metal ions can significantly modify the physiochemical properties of InsPs and the role of InsP, as a putative physiological chelating agent should be borne in mind when considering the biology of metals such as A13+ or Fe3+. In this regard, there are gaps in our knowledge of how cells handle iron which require a low molecular weight iron-binding molecule to shuttle iron between transport and storage proteins (e.g. transferrin and ferritin) and their ultimate destinations in the cell, the proteins which require iron to function (see for example, Refs. 28-30). A number of molecules have been postulated to exert such a role, e.g. nucleotides, citrate, glycine, and glucose, but the evidence seems somewhat unconvincing in view of their relatively low affinity and specificity for iron and their proven roles in other major areas of metabolism. InsPs would seem t o be an attractive candidate for such a role since it both possesses a high affinity for iron (our preliminary experiments based on the solubilization of Fe(OH)3 precipitates and the decolorization of various Fe3+ ligand complexes suggest that the affinity constant of InsP, for Fe3+ is in the range 1025-1030 and the stoichiometry of binding is 4-5 Fe3+/InsP6, data not shown) and prevents the bound iron from participating in potentially damaging free radical reactions (12,31). A further attractive feature of such speculation is that the rapid, "futile cycling" of specific phosphate groups on InsP, which is seen in cells (7) would open the door to rapid, directional transport of bound metal ions mediated by controlled and localized phosphorylation and dephosphorylation reactions.
InsPs has been reported to have a number of extracellular actions (8-10, 23, 32-34). Furthermore, two possible sites of intracellular action have recently been identified 1) a highly selective interaction with the G-protein receptor regulatory protein, arrestin (35,36) and 2) a novel inositol polyphosphate receptor which appears to be a gated potassium channel (37). The ubiquitous metal ion-dependent binding site described here is unlikely to mediate any physiological response to this compound, but will certainly mask any lower capacity binding site which might be involved in producing these responses. Consequently, radioligand binding to crude membranes may be limited in its applicability to the analysis of the membrane actions of InsPs. A more productive approach may be the purification of inositol polyphosphate-binding proteins from detergent-solubilized membranes (26). However, the interaction of InsPs with cations does raise questions about studies on InsP6-induced cellular 45Ca accumulation (10,34). Recently it has been claimed that metabolically dead cells accumulate 45Ca when treated with InsPs (22). This may be another instance of the formation of a metal ion-InsPs-membrane complex involving Ca2+. In this regard, Nicoletti et al. (14) have noted a good correlation between the ability of divalent cations to both potentiate InsP6-stimulated 45Ca accumulation and InsPs binding. More generally, it is now clear that experiments designed to investigate membrane actions of InsPs should carefully control for the possibility that InsPs can bind to cell membranes via metal ions and thus alter their biological properties, in an apparently very specific, yet probably unphysiological manner.