Signaling by Higher Inositol Polyphosphates

Evidence has accumulated that inositol pyrophosphates (diphosphoinositol pentakisphosphate (PP-InsP5) and bisdiphosphoinositol tetrakisphosphate ([PP]2-InsP4)) are intracellular signals that regulate many cellular processes including endocytosis, vesicle trafficking, apoptosis, and DNA repair. Yet, in contrast to the situation with all other second messengers, no one studying multicellular organisms has previously described a stimulus that acutely and specifically elevates cellular levels of PP-InsP5 or [PP]2-InsP4. We now show up to 25-fold elevations in [PP]2-InsP4 levels in animal cells. Importantly, this does not involve classical agonists. Instead, we show that this [PP]2-InsP4 response is a novel consequence of the activation of ERK1/2 and p38MAPα/β kinases by hyperosmotic stress. JNK did not participate in regulating [PP]2-InsP4 levels. Identification of [PP]2-InsP4 as a sensor of hyperosmotic stress opens up a new area of research for studies into the cellular activities of higher inositol phosphates.


From the Inositide Signaling Group, NIEHS, National Institutes of Health, Department of Health and Social Services, Research Triangle Park, North Carolina 27709
Evidence has accumulated that inositol pyrophosphates (diphosphoinositol pentakisphosphate (PP-InsP 5 ) and bisdiphosphoinositol tetrakisphosphate ([PP] 2 -InsP 4 )) are intracellular signals that regulate many cellular processes including endocytosis, vesicle trafficking, apoptosis, and DNA repair. Yet, in contrast to the situation with all other second messengers, no one studying multicellular organisms has previously described a stimulus that acutely and specifically elevates cellular levels of PP-InsP 5

or [PP] 2 -InsP 4 . We now show up to 25-fold elevations in [PP] 2 -InsP 4 levels in animal cells. Importantly, this does not involve classical agonists. Instead, we show that this [PP] 2 -InsP 4 response is a novel consequence of the activation of ERK1/2 and p38MAP␣/␤ kinases by hyperosmotic stress. JNK did not participate in regulating [PP] 2 -InsP 4 levels. Identification of [PP] 2 -InsP 4 as a sensor of hyperosmotic stress opens up a new area of research for studies into the cellular activities of higher inositol phosphates.
Pivotal to the signaling functions of PP-InsP 5 1 and [PP] 2 -InsP 4 are their highly negative electrostatic potential and the considerable free energy of their hydrolysis (for review, see Ref. 1). The former attribute facilitates their functionally significant binding to PH (2) and other protein domains (3,4). Addi-tionally, the high energy phosphate groups may phosphorylate proteins (5). These effects are believed to underlie the roles of inositol pyrophosphates in regulating clathrin assembly by adaptor proteins (3), apoptosis (6), vesicle trafficking (7), chemotaxis (2), and DNA repair (5). It has also been pointed out (e.g. Ref. 2) that many of the important cellular functions currently attributed to InsP 6 may in vivo be more effectively performed by PP-InsP 5 and [PP] 2 -InsP 4 . The Dictyostelids have uniquely exploited the physico-chemical properties of the inositol diphosphates by synthesizing near-millimolar levels of these molecules (8). However, animal cells have 300-fold lower levels of PP-InsP 5 and [PP] 2 -InsP 4 (1). It has therefore generally been anticipated that, in animal cells, there should be substantial, stimulus-dependent increases in cellular levels of PP-InsP 5 and [PP] 2 -InsP 4 to elevate levels of these polyphosphates to a functionally significant threshold. However, no such phenomenon has been described previously. Instead, only quite small stimulus-dependent changes in PP-InsP 5 and [PP] 2 -InsP 4 turnover have been reported, and in each case the result was a decrease rather than an increase in cellular levels of the pyrophosphates. For example, depletion of endoplasmic reticulum Ca 2ϩ stores by thapsigargin transiently decreases PP-InsP 5 levels by 50% (9). Activation of ␤-adrenergic receptors decrease levels of [PP] 2 -InsP 4 by up to 40% (10).
This absence of a substantial, regulatory context for PP-InsP 5 and [PP] 2 -InsP 4 has been addressed in the current study, in which we selected a biological paradigm of fundamental and widespread significance, namely, defense against osmotic stress. The maintenance of osmotic balance across plasma membranes is an ongoing cellular burden in the face of controlled changes in both cell size and intracellular hydration as well as exchange of metabolites and ions across plasma membranes (11). There are also clinically relevant osmotic challenges that can impose strain upon the cytoskeleton, perturb chromatin structure, damage DNA, and inhibit DNA repair (12,13). One way to examine how cells sense and adapt to hyperosmotic stress is to add sorbitol to the medium (11). In this report, we show that the addition of 0.2 M sorbitol can elevate [PP] 2 -InsP 4 levels up to 25-fold, and we also study the molecular mechanisms that are involved.

EXPERIMENTAL PROCEDURES
Cellular levels of inositol phosphates were determined as previously described (10). InsP 6 and PP-InsP 5 kinase activities were assayed as described (15). Several colleagues kindly supplied the following cDNA

RESULTS AND DISCUSSION
Cells are constantly adapting to the osmotic challenges brought on by normal cellular activities, such as the breakdown of macromolecules, controlled changes in both cell size and intracellular hydration, as well as exchange of metabolites and ions across plasma membranes (11). We investigated whether inositol diphosphates might be regulated by hyperosmotic stress, which we introduced by the addition of 0.2 M sorbitol. It is notable that many earlier studies into hyperosomotic stress have frequently employed much higher doses of sorbitol (0.5-0.7 M). Cells adapt successfully to 0.2 M sorbitol, since it failed * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. even after 24 h to affect viability of either HEK cells (101 Ϯ 2%) or the DDT 1 MF 2 vas deferens smooth muscle line (96 Ϯ 1%) compared with non-sorbitol controls. A novel, immediate and selective response of DDT 1 MF 2 cells and HEK cells to 0.2 M sorbitol is a dramatic elevation of [PP] 2 -InsP 4 levels, in the range of 10 -25-fold ( Fig. 1). It is worth emphasizing the novelty of this observation; there is no previous report of any extracellular stimulus that can specifically and acutely increase levels of an inositol diphosphate.
To assess the mechanisms underlying the sorbitol-dependent increase in [PP] 2 -InsP 4 levels, we considered how this enzyme might interface with other signaling entities. Hyperosomotic stress is known to activate Ser/Thr phosphoprotein phosphatases (17) and Tyr phosphoprotein phosphatases (18), but inhibitors of either pathway (20 nM okadaic acid or 10 M phenylarsine oxide, respectively) did not modify the [PP] 2 -InsP 4 response to sorbitol (data not shown). The JNK MAP kinase pathway can be activated by 0.5 M sorbitol (19). In agreement with this earlier study, we also found that 0.5 M sorbitol activated JNK ( Fig. 2A). However, JNK was not activated by 0.2 M sorbitol ( Fig. 2A), yet this dose of sorbitol was sufficient to maximally elevate [PP] 2 -InsP 4 levels (Fig. 1C). Thus, we conclude that JNK activation is not necessary for [PP] 2 -InsP 4 levels to be increased by hyperosmotic stress.
The ERK and p38 MAP kinase pathways are activated by osmotic stress (Fig. 2, B and C, and Refs. 20 and 21). Pretreatment of cells with the MEK inhibitors U0126 (10 M) or PD098059 (50 M) inhibited ERK1/2 phosphorylation (Fig. 2, B and C, and Ref. 22). This was accompanied by a 45-50% reduction of the sorbitol-dependent increase in [PP] 2 -InsP 4 levels (Fig. 3A). The similarity of the results obtained with two structurally unrelated MEK inhibitors makes it unlikely that reversal of the sorbitol effect results from unsuspected nonspecific actions of U0126 or PD098059. Pretreatment of cells with SB203580 inhibited phosphorylation of p38␣/␤ (Fig. 2C and Ref. 23) and also elicited a 23% attenuation of the sorbitol-dependent increase in [PP] 2 -InsP 4 levels (Fig. 3A). The relatively mild effect of SB203580 when it was added alone may be in part explained by its ability to non-specifically activate ERK1/2 We eliminated the possibility that the MAP kinase inhibitors might non-specifically block synthesis of the diphosphoinositol polyphosphates: native preparations of PP-InsP 5 kinase showed 103 Ϯ 2% activity when incubated with 10 M SB203580, 105 Ϯ 3% with 10 M U0126, and 103 Ϯ 2% with 50 M PD098059 (n ϭ 3), all relative to vehicle controls. The recombinant InsP 6 kinase showed 97 Ϯ 6% activity when incubated with 10 M SB203580, 105 Ϯ 2% with 10 M U0126 and 96 Ϯ 5% with 50 M PD098059 (n ϭ 3-4), all relative to vehicle controls. Although MAP kinase inhibitors did not affect PP-InsP 5 synthesis in vitro, the level of PP-InsP 5 in cells treated with either PD098059 or UO1276 was 25% lower than control cells (Fig. 3B). These data suggest that the ERK activity that is ongoing even in resting cells (Fig. 2B) may weakly enhance InsP 6 kinase activity. Consistent with this explanation, the sorbitol-dependent decrease in PP-InsP 5 levels was not reversed by the MEK inhibitors (Fig. 3B). The absence of any significant effect of the MAP kinase inhibitors upon PP-InsP 5 levels in sorbitol-treated cells (Fig. 3B) is of additional interest; this observation indicates that the attenuation of sorbitol-mediated increases in [PP] 2 -InsP 4 levels by the MAP kinase inhibitors is not a secondary consequence of a reduction in substrate supply for the PP-InsP 5 kinase.
To substantiate the inhibitor experiments, we next transiently transfected cells with dominant negative (DN) MAP kinase cDNA constructs. It is first important to note that environmental stresses provide "lateral" as well as "top-down" input into MAP kinase responses (25). The DN-ERK constructs block the MAP kinase pathway at a point that is downstream of the site of action of the MEK inhibitors, so these two empirical approaches could yield quantitatively different degrees of attenuation of the sorbitol response if there were any lateral input between MEK and ERK. When co-expressed, DN-ERK1 (K71R) and DN-ERK2 (K52R) (26) reduced the effect of sorbitol upon [PP] 2 -InsP 4 levels by 30% (Fig. 4). Neither DN construct had any significant effect when added alone (data not shown). Also when co-expressed, the TGY-to-AGF DN-p38␣ and DN-p38␤ proteins (27,28) reduced the effect of sorbitol upon [PP] 2 -InsP 4 levels by 30% (Fig. 4). Again, neither DN construct had any significant effect when added alone (data not shown). Transient transfection of DDT 1 MF 2 cells with all four DN constructs reduced by 50% the effect of hyperosmotic stress upon [PP] 2 -InsP 4 levels (Fig. 4). This is approximately the maximum effect that could be anticipated, given the limited transfection efficiency (50%; determined by GFP co-transfection). We found that the activity of the native PP-InsP 5 kinase was unaffected by recombinant, phosphorylated ERK1/2 (data not shown). It therefore seems likely that another signaling protein that lies downstream of ERK mediates its activation of cellular [PP] 2 -InsP 4 accumulation. Nevertheless, [PP] 2 -InsP 4 signaling offers a new repertoire for MAP kinase actions upon cellular physiology. Its unusual nature is also of interest; MAP kinases do not typically recruit diffusible second messengers.
There is a growing appreciation that the inositol diphos- phates can bind in a functionally significant manner to certain proteins through largely delocalized (i.e. nonspecific) electrostatic interactions, most notably to the majority of PH domains (2,29), but other protein domains are certainly involved (3,4). In these circumstances, the addition of even just one pyrophosphate group to InsP 6 can increase ligand affinity by at least 30 -100-fold (2). Elevations in [PP] 2 -InsP 4 levels following hyperosmotic stress would therefore be expected to increasingly compete with PtdIns(4,5)P 2 , another important ligand for PH domains that helps recruit proteins into multimeric signaling complexes (signalosomes) (29). Indeed, dynamic competition between soluble inositol pyrophosphates and membrane-bound inositol lipids has been shown by others to act as a logic gate to determine whether an inositide-binding protein is recruited to signalosomes (2,4). If there were autonomous spatio-temporal regulation of the levels of the competing lipid and pyrophosphate ligands, this would facilitate stimulus-specific control over the membership of these multimeric complexes. We have evidence of this metabolic autonomy; [PP] 2 -InsP 4 levels in either control or sorbitol-treated cells were unaffected by inhibition of PtdIns(4,5)P 2 hydrolysis by 10 M U73122, by inhibition of PI 3-kinases with100 nM wortmannin, or by specific activation of PtdIns(3,4,5)P 3 synthesis through transient overexpression of PIPKH (data not shown). Thus, [PP] 2 -InsP 4 turnover is completely independent of inositol lipid turnover.
Of importance to the subject of stimulus specificity, MAP kinase activation by the inflammatory cytokine tumor necrosis factor ␣ (10 ng/ml for 30 min) had no effect upon [PP] 2 -InsP 4 (data not shown). Furthermore, even though EGF is well known to activate ERK and p38 ( Fig. 5 and Ref. 20), it barely elevated [PP] 2 -InsP 4 levels (0.7-fold; Fig. 5) and had no effect upon PP-InsP 5 levels (Fig. 5). These data indicate that activation of both ERK and p38 are not, by themselves, sufficient to elevate [PP] 2 -InsP 4 levels. Perhaps [PP] 2 -InsP 4 turnover is regulated downstream of yet another signaling pathway that is selectively activated by hyperosmotic stress and not by EGF. Alternately, hyperosmotic stress (but not EGF) may attenuate an inhibitory process that normally constrains [PP] 2 -InsP 4 accumulation in cells. These questions touch on an enigma that pervades MAP kinase research: how does a cell elicit a specific output from the MAP kinase pathway when it is utilized by so many extracellular and intracellular stimulii? Much additional work may be required to solve this problem, but in any case, our data are still significant in their demonstration that it is stress-dependent and not agonist-initiated increases in [PP] 2 -InsP 4 levels that provides a regulatory context for the actions of inositol pyrophosphates. It is intriguing that hyperosmotic shock can damage DNA and affect vesicle trafficking processes (13), since these are both processes that can be regulated by inositol pyrophosphates (see Introduction). In any case, our demonstration of a stimulus that can specifically elevate one of the higher inositol polyphosphates, in a specific and acute manner, is an unprecedented observation that opens up a new area of research for inositide signaling.