how versatile are inositol phosphate kinases?

Biochem. J. (2004) 377, 265–280 (Printed in Great Britain) 265

REVIEW ARTICLEHow versatile are inositol phosphate kinases?Stephen B. SHEARS1

Inositol Signaling Section, Laboratory of Signal Transduction, NIEHS/NIH/DHSS Research Triangle Park, NC 27709, U.S.A.

This review assesses the extent and the significance of cata-lytic versatility shown by several inositol phosphate kinases:the inositol phosphate multikinase, the reversible Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase, and the kinases that synthesize diphospho-inositol polyphosphates. Particular emphasis is placed upon datathat are relevant to the situation in vivo. It will be shown thatcatalytic promiscuity towards different inositol phosphates is nottypically an evolutionary compromise, but instead is sometimesexploited to facilitate tight regulation of physiological processes.This multifunctionality can add to the complexity with which

inositol signalling pathways interact. This review also assessessome proposed additional functions for the catalytic domains,including transcriptional regulation, protein kinase activity andcontrol by molecular ‘switching’, all in the context of growinginterest in ‘moonlighting’ (gene-sharing) proteins [Jeffery (2003)Ann. Med. 35, 28–35].

Key words: diphosphoinositol polyphosphate, inositol phosphatekinase, InsP6 kinase, inositol phosphate multikinase (Ipmk), tris-phosphate, tetrakisphosphate.

INTRODUCTION

A large number of molecules result from the ability of cells toplace phosphate groups – and sometimes diphosphate groups –in combinatorial manner around the six-carbon myo-inositol ring.The multitudinous composition of this family challenges our ef-forts to understand the nature of the proteins that either metabolizethe inositol phosphates or mediate their intracellular functions.For example, consider situations in which a biologically activeinositol phosphate interacts with a target protein in a highlyspecific manner that excludes all other inositol phosphate isomers.What is the structural basis for this protein specificity? What arethe ligand recognition motifs on the inositol phosphate? The samequeries can also be posed for some of the enzymes that meta-bolize inositol phosphates, which, likewise, are highly specific.However, we should also ask about a contrasting situation: whatis the structural basis for promiscuity among some of the enzymesthat direct inositol phosphate turnover? We should also enquireinto its biological relevance; while the significance to the cell ofligand specificity is self-evident, the importance of promiscuouscatalytic activity towards several inositol phosphates is not alwaysobvious. This problem can be complicated by a sometimes mis-leading range of catalytic activities that can be achieved by a singleenzyme in vitro. This might come to light only when the experi-menter is endowed with extensive quantities of both recombinantenzyme and patience. Such enzyme multifunctionality does notnecessarily occur in intact cells. This is one of the issues that thisreview will address.

There are instances where diversity in enzyme function hasbeen proposed to go beyond an ability to convert one inositolphosphate into another. Recently, one of the inositol kinaseswas reported also to have protein kinase activity, and anotherhas been proposed to empower the activity of transcriptionalcontrol complexes. Are these ‘moonlighting’ activities [1] thatare independent of the protein’s ‘day jobs’? Or do they providea means to functionally couple inositol phosphate turnover to

Abbreviations used: CaMKII, calmodulin-dependent protein kinase II; Crac, cytosolic regulator of adenylate cyclase; Ipmk, inositol phosphate multi-kinase; PTEN, phosphatase and tensin homologue deleted on chromosome 10; TNF, tumour necrosis factor.

1 e-mail [email protected]

other cell-signalling events? General examples of both options –independent and interdependent multifunctionality – do occur [2].Interest in such versatility is growing as its structural, functional,evolutionary and therapeutic implications become more widelyappreciated [2–4]. As we shall see, when a protein is genuinelypleiotropic, it can be difficult to establish the relative importanceof each of its various activities to a physiological process.Nevertheless, it is the goal of this review to draw some conclusionsas to just how multifaceted these inositol kinases can be.

INOSITOL PHOSPHATE DIVERSITY AND NOMENCLATURE

In order to appreciate the similarities and differences in thestructures of the various inositol phosphates, it is necessaryto understand the nomenclature upon which their descriptiondepends. Agranoff’s ‘turtle’ [5] provides us with a useful, visualmnemonic. In this aide-memoire (Figure 1a), the hydrogens areconveniently ignored and the inositol ring is depicted in itsthermodynamically stable, so-called ‘chair’ conformation. Thestructure’s resemblance to a turtle is then apparent (Figure 1b).An International Nomenclature Committee [6] has stipulatedthat, when viewed from above, the turtle’s appendages can benumbered in an anticlockwise direction commencing with thefront right flipper (the D-stereochemical designation). The erectturtle’s head represents the axial (i.e. perpendicular) orientationof the 2-hydroxy group relative to the plane of the ring, whereasthe other five hydroxy groups (the four flippers and the tail) areequatorial (i.e. approximately in the same plane as the ring).This system permits a ready and unambiguous numbering ofthe various carbons around the inositol ring. Inositol 1,4,5-trisphosphate, for example, the well-known intracellular Ca2+-mobilizing signal [7], has three single phosphate groups, atpositions 1, 4 and 5. Ins(1,4,5)P3 and (1,4,5)IP3 are both widelyaccepted shorthand descriptions of this molecule (here, the ‘P’is italicized to designate phosphate rather than phosphorus, but

c© 2004 Biochemical Society

266 S. B. Shears

Figure 1 Agranoff’s turtle: a guide to inositol phosphate nomenclature

At physiological pH, inositol and its phosphorylated derivatives adopt a thermodynamicallystable, staggered chair conformation, with one axial and five equatorial groups [150], that hasbeen said to resemble a turtle [5]. The numbering of the carbons begins with the turtle’s frontright flipper and proceeds in an anticlockwise direction around the ring. This Figure is anexample of ‘turtle recall’ (R. Irvine, personal communication).

most journals no longer make this distinction). As for the goalof unambiguously describing the enzymes that direct inositolphosphate turnover, when the activity is specific, it is quite simple:the enzyme that adds a 3-phosphate to Ins(1,4,5)P3 is termedan ‘Ins(1,4,5)P3 3-kinase’. Unfortunately, but perhaps inevitably,enzymes that display a more complex catalytic repertoire areafflicted with a less systematic nomenclature (see below).

THE INOSITOL PHOSPHATE ‘MULTIKINASE’

Ins(1,4,5)P3 phosphorylation and the discovery of the ‘multikinase’

The first family of inositol phosphate kinases to be discoveredspecifically phosphorylate the 3-position of Ins(1,4,5)P3, yieldingIns(1,3,4,5)P4 [8,9]; no other naturally occurring inositolphosphates are substrates. Animal cells express three isoforms ofthis kinase family, which differ in their intracellular distributionand in the manner in which their catalytic activity is regulated[10]. Nevertheless, these particular enzymes are not themselvesmultifunctional. Their physiological role is only to phosphorylateIns(1,4,5)P3 to Ins(1,3,4,5)P4 [10]. This reaction serves to controlCa2+ mobilization [7,11], and provides precursor material formore highly phosphorylated inositol phosphates (Scheme 1). Thelatter anabolic pathway is usually near-saturated in non-stimulatedcells, fed by an ongoing basal rate of Ins(1,4,5)P3 synthesis andmetabolism. Thus overexpression of the 3-kinase does not altermass levels of Ins(1,3,4,5,6)P5 and InsP6 [12].

A different enzyme, which phosphorylates Ins(1,4,5)P3 at C-6,was discovered initially in peas [13] and detected subsequently inSaccharomyces cerevisiae [14] and Schizosaccharomyces pombe[15]. This enzyme received relatively little attention, however,until it was cloned from Sacch. cerevisiae by two groups workingindependently [16,17]. It was thereby revealed that this proteinalready had a 20-year history as a transcriptional regulator [18–22]; in that context it was known as Arg82 (and the nom deguerre ArgRIII). Odom et al. [17] re-christened this protein ‘Ipk2’,

Scheme 1 Metabolic pathways for phosphorylation of Ins(1,4,5)P3 inyeast, plants and animal cells in vivo

Shown are the quantitatively most important routes of Ins(1,4,5)P3 phosphorylation in vivo,based on data from several publications that are discussed in the text. Interestingly, even thoughthe yeast and mammalian forms of Ipmk (inositol phosphate multikinase) contribute differentcatalytic activities to dissimilar metabolic pathways, in each case this enzyme is essential tothe cells’ major pathway of Ins(1,3,4,5,6)P5 synthesis de novo. In contrast, plants and slimemoulds can synthesize Ins(1,3,4,5,6)P5 and InsP6 by alternative pathways that are independentof Ins(1,4,5)P3 [102,103]. However, it is not the role of this Scheme to show all inositol phosphateinterconversions (for a more complete picture, see [11] for example). It is assumed that thefailure to detect Ins(1,4,5)P3 3-kinases in plants [11] is because they do not exist. Plants areknown to contain PP-InsP5 and [PP]2-InsP4 [151,152], but PP-InsP4 has not been detected.Numbers refer to various enzymes, as follows: 1, Ipmk (also known as ipk2/Arg82/ArgRIII;EC 2.7.1.151); 2, Ins(1,4,5)P3 3-kinase (EC 2.7.1.127); 3, Ins(1,4,5)P3/Ins(1,3,4,5)P4 5-phosphatase (EC 3.1.3.56); 4, Ins(1,3,4)P3 6-kinase/Ins(3,4,5,6)P4 1-kinase (EC 2.7.1.134);5, PTEN (EC 3.1.3.67); 6, multiple inositol-polyphosphate phosphatase (EC 3.1.3.62); 7,Ins(1,3,4,5,6)P5 2-kinase or ‘ipk1’ (no EC number); 8, InsP6 kinase (EC 2.7.4.21), or Kcs1 inyeast; 9, diphosphoinositol-polyphosphate diphosphatase (EC 3.6.1.52); 10, PP-InsP5 kinase.This last enzyme has no EC number, and its molecular identity is unknown.

while Snyder’s group [16] preferred ‘Ipmk (inositol phosphatemultikinase)’. The use of Ipmk is frowned upon in the yeast field,where abbreviations with no more than three letters are de rigeur.However, in the opinion of the author, Ipmk is the most descriptiveof this protein’s aliases, and so this abbreviation is used here forclarity.

Homologues of Ipmk are present in animal and plant cells(Figure 2; [23–26]). Each of these isoforms has conservedcatalytic residues (Figure 2). A transcriptionally active poly(Asp)


How versatile are inositol phosphate kinases? 267

Figure 2 Domain organization of Ipmk in humans, yeasts and a plant

The Figure depicts a schematic alignment of Ipmk homologues from Arabidopsis thaliana (At, type α; GenBank AY147935), Schizosaccharomyces pombe (Sp; CAB63791), Saccharomycescerevisiae (Sc; CAA28945 as ArgRIII) and Homo sapiens (Hs; AAM97838) and a putative homologue from Aspergillus nidulans (An; accession number AN0156; http://www-genome.wi.mit.edu/annotation/fungi/aspergillus). Note that the predicted molecular mass of Schizosaccharomyces pombe Ipmk (30 kDa) is different from the estimated size of the purifiedprotein (41 kDa [15]). Numbers above arrows depict amino acid residues but, for clarity, scaling is approximate. The catalytic InsP-binding consensus sequence (PxxxDxKxG; in blue) was firstnoted in [17] and [16]; York’s group [17] were the first to confirm this with mutagenic studies. A putative ATP-binding site ([L/M][I/V]D[F/L]AH; in yellow) was first identified by Saiardi et al. [25] inmammalian Ipmk. The nuclear localization signal (NLS; in green) was characterized by Mayr and colleagues [23]. The alignment of Ipmk from A. thaliana was previously described by York’s group[26]. The SSLL-like motif is also conserved in InsP6 kinases, where it has been shown to be necessary for catalysis [25,61]. The poly(Asp) region ([D]17; in red) is required in Sacch. cerevisiae Ipmkfor transcriptional control over arginine-regulated genes [19]. The putative Aspergillus nidulans Ipmk has two acidic domains (poly[Asp/Glu]); perhaps not coincidentally, the mechanism by whicharginine induces the ornithine transaminase gene in Aspergillus nidulans shows similarities to the process in Sacch. cerevisiae [153]. A W65R mutant in Sacch. cerevisiae Ipmk compromises bothtranscriptional control [19] and inositol phosphate kinase activity in vivo [61]. Human Ipmk has 84 % sequence identity with rat Ipmk, 25 % identity with Arabidopsis thaliana Ipmk, 20 % identitywith Aspergillus nidulans Ipmk, and 16 % identity with Sacch. cerevisiae Ipmk ([24,26], and results not shown).

acidic domain [19] has a more restricted occurrence, and is so farknown to be present only in Sacch. cerevisiae, with possibly apotential homologue in Aspergillus nidulans (Figure 2). Ipmkis more widely distributed in Nature than is the Ins(1,4,5)P3

3-kinase family, which is apparently restricted to the animalkingdom [11]. In fact, a primordial Ipmk may have been theevolutionary precursor of the 3-kinases [11,16,17] and the InsP6

kinases [27], all of which contain the PxxxDxKxG catalyticdomain. The human IPMK gene contains six exons and is locatedon chromosome 10 [23]. The mRNA transcript is expressedubiquitously; in humans there are particularly high levels ofexpression in liver and skeletal muscle [24], whereas in therat the highest levels of expression were observed in kidney

and brain [25]. Northern analysis suggests there is alternativesplicing of mRNA transcripts for Ipmk [24,25]. The significanceof there being multiple mRNAs is typically interpreted in terms oftheir differing in either stability or translatability, in order tomodulate gene expression in a tissue- or developmental-stage-specific manner [28].

How many inositol phosphate kinase reactions are catalysedby Ipmk?

Recombinant yeast and plant Ipmks phosphorylate Ins(1,4,5)P3

mainly to Ins(1,4,5,6)P4 [17,26,29]. The Ins(1,4,5,6)P4 is phos-phorylated further by Ipmk at the 3-position, yielding


268 S. B. Shears

Figure 3 Diphosphoinositol polyphosphates

Ins(1,3,4,5,6)P5 and InsP6 can both be phosphorylated in an ATP-dependent manner by the‘InsP6 kinase’ (EC 2.7.4.21), which is known as Kcs1 in yeast. Hydrolysis of diphosphoinositolpolyphosphates is catalysed by specific phosphohydrolases [133,137]. The Figure shows thediphosphate group in PP-InsP5 attached to C-5, which is known to be the case in animal cells[154]; the same is assumed for PP-InsP4. Some 5-PP-InsP5 is also present in slime mouldssuch as Dictyostelium [154], but the 6-PP-InsP5 isomer predominates in these organisms[139,154]. See also Figure 6.

Ins(1,3,4,5,6)P5 [17,29]. The yeast enzyme will also formsome Ins(1,3,4,5)P4 from Ins(1,4,5)P3 [29], but studies with intactcells show that this is not a major pathway in vivo [17,29]. Thisassessment of the major route of Ins(1,4,5)P3 phosphorylationin yeasts and plants is summarized in Scheme 1. Recombinantmammalian Ipmk displays a different positional specificity to-wards Ins(1,4,5)P3, phosphorylating it primarily to Ins(1,3,4,5)P4

[23,25], although kinetic data indicate that this reaction has littlesignificance in vivo (see below). There are also some as yetunpublished results that indicate that the 3-kinase activity of Ipmkphosphorylates PtdIns(4,5)P2 to PtdIns(3,4,5)P3 (A. Resnick,A. Saiardi and S. H. Snyder, personal communication). Thepossibility that Ipmk might synthesize PtdIns(3,4,5)P3 representsa tantalizing potential addition to this enzyme’s versatility.

Recombinant mammalian Ipmk can phosphorylate Ins(4,5)P2

in vitro [25]. This is unlikely to be a significant reaction in vivo,where little Ins(4,5)P2 is generally seen, except in neuronal cellstreated with millimolar concentrations of lithium [30,31], whichis an experimental protocol designed to mimic the therapeuticapplication of this cation as an anti-manic agent. In any case,Ins(4,5)P2 does not compete well with Ins(1,4,5)P3 [25], whichitself is probably not the optimum substrate in vivo (see below).Another reaction catalysed by Ipmk in vivo is the phosphorylationof Ins(1,3,4,5,6)P5 to PP-InsP4, a diphosphorylated (Figure 3)inositol phosphate [25,27]. {The last two studies correctedan erroneous suggestion [16] that InsP6 was the product ofphosphorylation of Ins(1,3,4,5,6)P5 by Ipmk.} However, PP-InsP4 synthesis from Ins(1,3,4,5,6)P5 appears too slow to be ofmuch physiological significance [23,25,27].

The above discussion illustrates that Ipmk can perform re-actions in vitro that seem not to be relevant in vivo. This pointemphasizes the value of verifying Ipmk catalytic activities bymanipulating the levels of expression of the enzyme in intact cells.Since ipmk� strains of Sacch. cerevisiae are viable (in a nitrogen-rich medium), the impact of the gene deletion upon steady-statelevels of inositol phosphates can be determined [17,29,32]. Inthese experiments, intact cells were incubated with [3H]inositolso as to radiolabel the phosphate pools, the sizes of which were

then determined by separating the constituents of cell lysatesusing HPLC. The severity of the impediment to Ins(1,4,5)P3

phosphorylation in ipmk� cells was disclosed by the 170-foldelevation in the cellular levels of [3H]Ins(1,4,5)P3, compared withwild-type cells [17,29]. The steady-state levels of Ins(1,3,4,5,6)P5,the end-product of Ipmk activity (Scheme 1), were also reduced byapprox. 90% in ipmk� cells [29]. More strikingly, the normallyhigh level of InsP6 was reduced >100-fold in the ipmk� cells[17,29]. Thus Ipmk is virtually indispensable for the synthesis ofInsP6 in yeast (Scheme 1).

Studies with a human [24] and a plant [26] Ipmk recentlyindicated a very important addition to the enzyme’s versatility,with the demonstration that it is also an Ins(1,3,4,6)P4 5-kinase.Kinetic data obtained with recombinant human Ipmk in Majerus’laboratory argue strongly that the Ins(1,3,4,6)P4 5-kinase reactionis actually more physiologically active than its Ins(1,4,5)P3

3-kinase activity [24]. There is another reason for believingthat Ipmk does not normally provide a quantitatively importantpathway of Ins(1,4,5)P3 phosphorylation in vivo: the catalyticefficiency of this reaction is considerably lower than that achievedby the dedicated Ins(1,4,5)P3 3-kinases [24]. Of course, thesearguments are not irrefutable. There is always the possibility thatsome of the cellular Ins(1,4,5)P3 is in a metabolic compartment,possibly at a locally elevated concentration, with privileged accessto Ipmk. Actually, there are some older studies that concluded thatIns(1,4,5)P3 can be compartmentalized [33,34], but the inabilityto define the nature of the phenomenon (e.g. [35]) has contri-buted to a lack of current interest in this topic.

Mammalian Ipmk also has Ins(1,4,5,6)P4 3-kinase activity [24]that may be important. This seems likely to account for theIns(1,4,5,6)P4 3-kinase activity that was observed in cell lysates inearlier experiments performed in our laboratory [36,37]. This 3-kinase activity was competitively inhibited by Ins(1,3,4,6)P4 [37],which we now appreciate is consistent with what we should expectfrom Ipmk (it is unclear if this competition has any physiologicalsignificance). Other evidence that Ins(1,4,5,6)P4 3-kinase is anongoing reaction in intact cells comes from experiments in whichthe kinase was inhibited by reducing the levels of ATP, where-upon the concentration of Ins(1,4,5,6)P4 increased [36]. Thesource of that Ins(1,4,5,6)P4 is Ins(1,3,4,5,6)P5 [36]. That is, thereappears to be an ongoing 3-phosphatase/3-kinase substrate cyclebetween Ins(1,3,4,5,6)P5 and Ins(1,4,5,6)P4 (Scheme 1). Thusthe lack of general recognition given to the extent of cellularIns(1,4,5,6)P4 3-kinase activity reflects an under-appreciation ofthe catalytic versatility of Ipmk in vivo.

There are several reasons why it might prove productive tostudy this Ins(1,3,4,5,6)P5 3-phosphatase/Ins(1,4,5,6)P4 3-kinasecycle further. The cycle appears to be grossly perturbed by theSopE invasion factor from Salmonella typhimurium, leading to arapid and catastrophic net dephosphorylation of Ins(1,3,4,5,6)P5

to Ins(1,4,5,6)P4 [38]. Since SopE cannot itself dephosphoryl-ate Ins(1,3,4,5,6)P5, it must act by hijacking the host cell’sIns(1,3,4,5,6)P5 phosphatase and/or inhibiting the Ins(1,4,5,6)P4

3-kinase [38]. It will be interesting to determine whether this effectof SopE represents an example of structural mimicry, by whichpathogen virulence factors have evolved to imitate – and often toexaggerate – the activities of the host cell’s proteins [39]. As forthe nature of the Ins(1,3,4,5,6)P5 3-phosphatase, there is a verywell characterized and highly active enzyme with this activity(multiple inositol-polyphosphate phosphatase; Scheme 1), butthat is trapped inside the lumen of the endoplasmic reticulum, withapparently restricted access to inositol phosphates [40–43]. Morerecently, we demonstrated that PTEN (‘phosphatase and tensinhomologue deleted on chromosome 10’), classically recognizedto be an inositol lipid 3-phosphatase [44,45], is actually in



addition an efficient Ins(1,3,4,5,6)P5 3-phosphatase, both in vitroand in vivo [46]. This reactivity towards Ins(1,3,4,5,6)P5 islikely to be insignificant when PTEN is membrane-associated,whereupon the catalytic activity towards PtdIns(3,4,5)P3 (but notinositol phosphates) undergoes considerable interfacial activationby anionic inositol lipids [47]. However, a substantial proportionof endogenous PTEN is not actually bound to membranes [48]and is instead in the cytosol and the nucleus, and it is underthese conditions that it can have ready access to the relativelylarge (15–50 µM) cellular pool of Ins(1,3,4,5,6)P5. PTEN isunlikely to hydrolyse other inositol phosphates in vivo; InsP6,Ins(1,4,5,6)P4 and Ins(1,4,5)P3 are not substrates [45,46], andboth Ins(1,3,4,5)P4 and Ins(1,3,4)P3 are very poor substratesin vitro [45–47]. Perhaps Ins(1,3,4,5,6)P5 is a specific, regu-latable metabolic ‘clamp’ that impedes access of PTEN toPtdIns(3,4,5)P3 [46,49]. Moreover, Ins(1,3,4,5,6)P5 would appearto present a substantial impediment to PTEN’s already limitedcapacity to also use its catalytic site to dephosphorylate proteins[45]. In any case, PTEN and Ipmk represent the best candidatesfor the enzymes that participate in the ongoing metabolic cyclingbetween Ins(1,4,5,6)P4 and Ins(1,3,4,5,6)P5 [36]. It is possiblethat this ongoing cycle merely represents a metabolic costof using Ins(1,3,4,5,6)P5 to regulate PTEN activity towardsPtdIns(3,4,5)P3. Alternatively, the inositol phosphate cycle mightbe independently functional.

Rationalization of the different inositol phosphate kinase reactionscatalysed by Ipmk

The ability of Ipmk to phosphorylate both InsP3 and InsP4 is lostupon substitution of Asp-131 by Ala [17]. This result is con-sistent with there being a single active site on this protein. Onereaction mechanism that has been advocated [24] is that theIns(1,4,5)P3 substrate is held at the active site while the twophosphate groups are added, in either a stepwise or a concertedmanner. However, an observation that argues against this idea isthat the InsP4 intermediate is released into the bulk phase andaccumulates (see above).

Moreover, it is necessary to account for the ability of Ipmk toadd a phosphate group to various inositol phosphates at differingpositions around the inositol ring (see above). Well-evolvedenzymes do not typically indulge in haphazard regiochemistry [4],so instead one should consider whether the catalytic specificitycan be maintained by presenting key substrate recognition featuresto the active site in alternative binding orientations or ‘modes’[50]. This idea has previously helped to account for the catalyticversatility of native Ipmk that was partially purified from Schiz.pombe [15]. This model (summarized in Figure 4) proposes thatIpmk can accommodate its different substrates in three bindingmodes. In each of these modes, three recognition elements arepresented to Ipmk that are common both in nature and inorientation around the inositol ring: two phosphates and thehydroxy group that is phosphorylated (coloured red in Figure 4).These groups are likely to be essential for substrate recognition.Two binding modes for Ins(1,4,5)P3 are suggested, in order toaccount for its phosphorylation at both the 3- and 6-positions.Presumably, yeast Ipmk has a higher affinity for Ins(1,4,5)P3 inmode 2, as the enzyme mainly phosphorylates Ins(1,4,5)P3 toIns(1,4,5,6)P4 [17,29]. Conversely, mammalian Ipmk is suggestedto prefer to bind Ins(1,4,5)P3 in mode 1, since it phosphorylatesIns(1,4,5)P3 largely to Ins(1,3,4,5)P4 [23,24]. The ability of theenzyme to phosphorylate Ins(1,4,5,6)P4 and Ins(1,3,4,5)P4 isalso nicely accommodated by this model (Figure 4). The 1998study with Ipmk isolated from Schiz. pombe also predicted a

Figure 4 Proposed combinatorial recognition motifs for Ipmk

The diagram shows three proposed substrate-binding modes utilized by Ipmk, as originallyproposed in [15]. Note where the inositol ring has been rotated through vertical and/or horizontalplanes in order to align equivalent recognition motifs. The site of phosphorylation is indicatedby a yellow circle. Groups that are coloured red are common to all three binding modes andare proposed to be essential for high-affinity ligand recognition, but they are clearly insufficient(see the text). It is therefore proposed here that certain groups define substrate recognition ina mode-specific manner. Groups coloured purple, that are common to two binding modes, areconsidered most likely to play this role. Groups coloured green are deemed least importantto substrate recognition, as their position is common to only one binding mode. Note thatIns(1,4,5)P3 is proposed to bind to yeast Ipmk preferentially in mode 2, whereas mammalianIpmk prefers mode 1 (see the text for details).

third binding mode that could accommodate Ins(1,4,6)P3 andfacilitate its phosphorylation at the 5-position [15]. At that time,the physiological relevance of this observation was unknown,since Ins(1,4,6)P3 is not a significant constituent of the inositolphosphate metabolic pathway. Figure 4 now argues that theimportance of the third binding mode is for the Ins(1,3,4,6)P4

5-kinase activity of Ipmk, a key anabolic reaction in animal andplant cells (Scheme 1).

It was recognized previously that the three substituents on theinositol ring (coloured red in Figure 4) that are common to eachbinding mode are insufficient to define ligand specificity, since allthree groups are present on Ins(1,4)P2, which is not a substrate[15]. Attempts to rationalize the contributions made by othergroups around the inositol ring were previously confounded bythe belief that the enzyme also phosphorylated Ins(1,3,4,5,6)P5

to InsP6 [15]. However, we now know that the Ins(1,3,4,5,6)P5 2-kinase is a separate enzyme [51] that must have contaminated theearlier [15] preparations of native Ipmk. It is therefore proposedhere that substrate recognition occurs in a combinatorial manner.That is, mandatory substrate recognition features (coloured red in


270 S. B. Shears

Figure 4) are proposed to act in concert with certain other groupsthat define substrate recognition in a mode-specific manner.Note that there are groups on the inositol ring that are conserved interms of composition and orientation in two binding modes (col-oured purple in Figure 4). By virtue of this conservation, they aredeemed most likely to participate in mode-specific ligand recog-nition. Each of the remaining substituents, which are not con-served between binding modes (coloured green in Figure 4), areconsidered less likely to define substrate recognition, although apossible contribution cannot be excluded. This model for ligandbinding suggests that Ipmk might be one of those proteins [4]that utilizes rigid and flexible regions in its active site. Themore constrained amino acid residues could sustain specificinteractions with the mandatory substrate recognition features,whereas the more flexible regions of the protein would havethe adaptability to contribute to mode-specific interactions. Thishypothesis raises the possibility that a mutation in a flexible aminoacid residue that participates in mode-specific binding might onlycompromise ligand recognition in that particular mode. Perhapssuch a phenomenon underlies the nature of an Ipmk mutant,named 2–3 [17], in which Ins(1,4,5,6)P4 3-kinase activity (i.e.mode 1) is defective, even though Ins(1,4,5)P3 6-kinase activity(i.e. mode 2) is intact (see also [24]). All of these ideas can beexamined more closely when the structure of Ipmk is solved.

Ipmk and transcription: arginine-regulated genes in yeast

As mentioned above, before Ipmk was established to be an inositolphosphate kinase, the yeast protein had accumulated a 20-yearhistory as a transcriptional regulator, acting under its Arg82 alias.In this guise, Arg82 has become well known to participate in theprocess by which Sacch. cerevisiae prioritizes its use of alternativenitrogen sources. In the presence of arginine or ornithine, threeproteins (named Arg80, Arg81 and Mcm1) assemble into a trans-criptional complex (ArgR–Mcm1) which binds to the regulatoryelements of several co-regulated genes that encode enzymesinvolved in arginine metabolism [20]. As a consequence, twoenzymes which catabolize arginine and ornithine are induced(the CAR1 and CAR2 gene products respectively) and severalenzymes of arginine synthesis are repressed (the ARG1, ARG3,ARG5,6 and ARG8 gene products). These transcriptional controlprocesses also require Ipmk [52]. Until recently, it was generallythought that Ipmk is not a necessary constituent of the DNA-boundtranscriptional complex [19,21,53,54]. Instead, it was argued thatIpmk acts at an earlier stage, by stabilizing Mcm1 and Arg80,and also by facilitating their assembly into a multimeric complex[21,53,54]. Apparently critical to this process is a poly(Asp)-rich acidic domain between amino acid residues 282 and 303near the C-terminus of Ipmk (Figure 2). Evidence in supportof this idea includes the observation [54] that the stability ofMcm1 and Arg80 after cycloheximide treatment is impaired inthe ipmk�/ipmk�282−303 strain (i.e ipmk� cells transformed with aplasmid encoding ipmk�282−303). Two-hybrid assays indicated thatthe interaction between Arg80 and Mcm1 is also disrupted in theipmk�/ipmk�282−303 strain [54]. The absence of this acidic domainfrom the Ipmk from Schiz. pombe (Figure 2) appears to explainwhy it fails to rescue the defective regulation of arginine-sensitivegenes in the ipmk� strain of Sacch. cerevisiae [54].

A paper by Odom et al. [17] offers a dramatic expansion inthe versatility of Ipmk (which in their paper was named Ipk2; seeabove). It is important to note that this new model does not refuteany of the earlier work showing that Ipmk is required to assemblethe ArgR–Mcm1 complex (see above). This new hypothesisargues that, in addition to the kinase-independent mechanismsdescribed above, the kinase catalytic domain of Ipmk is proposed

to be “. . . required to properly execute transcriptional control asmeasured by growth on ornithine as the sole nitrogen source”. Forcells to survive and grow on such medium, ornithine has to inducethe expression of the CAR2 gene product, ornithine transaminase;this process requires a functional ArgR–Mcm1 transcriptionalcomplex [55,56]. When cells are grown on nitrogen-rich media,the requirement for CAR2 induction is bypassed. Therefore adefect in Ipmk that impairs ArgR–Mcm1 function might revealitself as a growth-impaired phenotype that is specific to ornithinebeing offered as the sole nitrogen source. In their study, Odomet al. [17] transformed ipmk� cells with an ipmkD131A plasmid(which encodes a ‘kinase-dead’ mutant). When plated on an orni-thine-based medium, the rate of growth of the ipmk�/ipmkD131A

cells was reduced relative to growth of the wild-type strain [17].This result led to the proposal [17] that the kinase activity ofIpmk is necessary for the ArgR–Mcm1 transcriptional complexto induce CAR2 expression. However, this interpretation hasbeen disputed by others [32], who have reported that the growthdefect exhibited by ipmk�/ipmkD131A cells was not nutritionallydependent, and instead was general in nature, since, in theirhands, it was observed irrespective of whether ornithine, ammoniaor glutamate was offered as the sole nitrogen source ([32];E. Dubois and F. Messenguy, personal communication). More-over, measurements of the cellular levels of arginase (the CAR1gene product) did not indicate that expression of that particu-lar gene was controlled by the kinase activity of Ipmk [32]. Oneway to move forward on this difference of opinion concerning theextent to which Ipmk kinase drives CAR2 expression might be forits proponents to assay the induction of the CAR2 promoter–lacZfusion reporter construct described previously [57]. Incidentally,it should be noted that there is agreement that the repression of theARG5,6 gene by ArgR–Mcm1 does not require the kinase activityof Ipmk ([32]; J. D. York, personal communication).

So as not to base their hypothesis entirely upon cell growthdata, Odom et al. [17] also studied the role of Ipmk by monitoringplating efficiency, which describes the proportion of dispersedsingle cells that divide and form colonies; this essentiallymeasures cell viability [58]. In an ornithine-based medium, theipmk� and ipmk�/ipmkD131A strains showed <2% plating ef-ficiency [17]. This almost complete lack of cell viability wasblamed on their inability to obtain nitrogen from the ornithine; thatis, the CAR2 gene product which converts ornithine into glutamatewas proposed not to be induced. However, when glutamate isoffered as an alternative nitrogen source, the need for CAR2induction is bypassed. Both ipmk� and ipmk�/ipmkD131A strainswere reported to show near-normal cell viability when glutamatereplaced ornithine as the sole nitrogen source ([17]; J. D. York,personal communication). Thus it was proposed that the inositolkinase activity of Ipmk was necessary for CAR2 gene induction.

Unfortunately, the quantification of plating efficiency is notalways an exact science. There are a number of factors that can in-troduce variability [58]. For example, the results that are obtainedcan be influenced by the methods used to disperse the originalcells [58]. In addition, individual viable cells may vary greatly inthe lag time before discernable growth and division can be ob-served. Thus it may be difficult to know how long one shouldobserve a cell before pronouncing it non-viable [58]. Even plateage and moisture content have been suggested to affect the results(J. D. York, personal communication). Some variability betweenexperiments is therefore to be expected. Indeed, the plating ef-ficiency of the ipmk�/ipmkD131A strain grown on ornithine (cal-culated relative to that of cells grown on nitrogen-rich media),originally reported to be <2% [17], has been recentlyre-evaluated, and revised upward to 20–40% (J. D. York, personalcommunication). Another group (E. Dubois and F. Messenguy,



personal communication) obtained a plating efficiency of 50%.This worrisome variability advises us to assess the data cautiously.A nutritionally dependent contribution to the reduced cell viabilityof the ipmk�/ipmkD131A strain was observed by E. Dubois andF. Messenguy (personal communication). That is, the numberof surviving ipmk�/ipmkD131A cells increased when they wereoffered a nitrogen source (e.g. glutamate) that was more ‘user-friendly’ than ornithine, although this was a much smaller effectthan that reported by Odom et al. [17]. These data should also beinterpreted with care, since even wild-type cells exhibit improvedcell viability when grown on glutamate or ammonia, comparedwith poorer-quality nitrogen sources such as ornithine or proline(E. Dubois and F. Messenguy, personal communication).

Odom et al. [17] also transformed ipmk� cells with theIpmk mutant named 2–3 (see above). These cells phosphoryl-ate Ins(1,4,5)P3 to Ins(1,4,5,6)P4, but cannot phosphorylateIns(1,4,5,6)P4 further (see also [24]). The plating efficiency ofthe ipmk�/ipmk2−3 strain was indistinguishable from that of wild-type cells, irrespective of the nitrogen source [17]. In other words,both ArgR–Mcm1 function and cell viability were reported to benormal in the absence of any inositol phosphates more highlyphosphorylated than Ins(1,4,5,6)P4. This observation indicatesthat Ins(1,3,4,5,6)P5 and InsP6 are not responsible for any de-fects that arise when the kinase activity of Ipmk is mutated,but one aspect of this finding is arguably a little puzzling. Theabsence of InsP6 would, from other studies, be expected toimpair mRNA export from the nucleus [29,59], cause vacuolarfragmentation [60,61], compromise endocytosis [62] and impaircell wall integrity [61]; it would not have been surprising if any oneof these defects had impaired the viability of the ipmk�/ipmk2−3

strain to a certain extent.Despite all of these difficulties, this idea that the kinase acti-

vity of Ipmk drives transcription [17] has been a very influentialhypothesis (see below). It has been theorized that transcriptionallydormant ArgR–Mcm1 complexes might be silently poised onDNA regulatory elements, awaiting activation by the kinase acti-vity of Ipmk [17,63]. This idea dovetails nicely with an emerginghypothesis that certain inducible or repressible promoters mayrecruit regulatory enzymes that are maintained in an inactive state,until they are awakened by an appropriate stimulus [64]. Abruptchanges in the levels of signalling molecules such as inositolphosphates, with a concomitant activation of inositol phosphatekinase activity, provides an attractive model for rapid stimulationof the transcriptional process [64]. Messenguy et al. [20] haveidentified the ArgR–McM1 binding sites in the inducer elementsof the CAR2 gene. As yet, there no evidence that a transcriptionallysilent version of the ArgR–Mcm1complex can actually bind to thispromoter.

Ipmk and transcription: phosphate-responsive genes in yeast

Transcription of PHO5, the gene encoding a secreted phosphatase,is regulated in response to phosphate availability by the reversiblephosphorylation of Pho4 [65]. During conditions of phosphatestarvation, Pho4 is activated by dephosphorylation, and thenbinds to the PHO5 promoter, and the phosphatase is expressed[66]. There was a dramatic reduction in the starvation-dependentinduction of PHO5 mRNA levels in cells lacking the inositolkinase activity of Ipmk, most notably in ipmk�/ipmkD131A cells[67]. One of the key new developments in this particular study isthat, by recording PHO5 mRNA levels, the authors performed amore direct analysis of the effect of Ipmk upon transcription thancan be obtained by measuring cell growth and viability.

Research into PHO5 is particularly significant because it has be-come a paradigm for understanding the relationship between tran-

scription and the structure of DNA–protein complexes (i.e. chro-matin) [68]. Chromatin undergoes transitions between closed, re-pressive states and remodelled, transcriptionally permissive states.Molecular mechanisms that switch between these twostates lie at the very heart of eukaryotic gene regulation. O’Sheaand colleagues [67] measured the accessibility of a ClaI restrictionsite in the PHO5 promoter in isolated nuclei when wild-type cellswere shifted to PHO5-inducing conditions. This increased ClaIaccessibility was impaired in ipmk�/ipmkD131A cells [67]. In thisway, the kinase activity of Ipmk was shown to be important forchromatin remodelling. Another key experiment was the demon-stration, using chromatin immunoprecipitation, that the amountof Pho4 bound to the PHO5 promoter was considerably reducedin the ipmk� strain [67]. This appeared to reflect defectiveoperation of the SWI/SNF and INO80 chromatin remodellingcomplexes. These data have reinforced the concept [17] that thekinase activity of Ipmk has the functional versatility to regulatetranscription. Moreover, mammalian cells contain homologues ofIpmk (see above) and SWI/SNF transcriptional complexes [69],so chromatin remodelling is now receiving attention as possiblybeing a widespread action of Ipmk.

O’Shea and colleagues [67] recognized the possibility thatloss of Ipmk catalytic activity could have affected cell functionsindirectly, by depleting higher inositol phosphates (Scheme 1).They therefore studied control of PHO5 expression in two mutantyeast strains. In one of these, the Ins(1,3,4,5,6)P5 kinase (Ipk1)was deleted so that cells could not synthesize InsP6. In theother strain, the InsP6 kinase (Kcs1) was deleted and cells couldnot make diphosphoinositol polyphosphates. There was a slightdecrease in PHO5 mRNA levels in both the ipk1� and kcs1�strains, suggesting an element of functional redundancy, but thiseffect was much less pronounced than in the ipmk� cells [67].Thus the major effect of Ipmk upon chromatin remodelling doesappear to represent a direct effect of its own kinase activity.

Is the effect of the kinase activity of Ipmk on PHO5 trans-cription completely independent of the separate transcriptionallyactive acidic domain in the Ipmk from Sacch. cerevisiae [54],and possibly also Aspergillus nidulans (Scheme 1)? There arecertainly other examples of functionally separate domains withina single protein acting in an independent manner [2,4]. However,deletion of the poly(Asp) domain in Ipmk from Sacch. cerevisiaereduced InsP6 levels in vivo by approx. 25% [61]. This sug-gests some, albeit limited, link between the function of this acidicdomain and the control of inositol phosphate turnover. Alter-natively, the ipmk�282−303 protein might have a modified tertiarystructure that has non-specific consequences for the activity of thekinase domain.

Some cautionary notes may still be warranted. There are far-reaching effects upon many mRNA species in yeast strains inwhich the catalytic activity of Ipmk is compromised [54], andit will be important to separate primary regulatory events fromsecondary consequences. Secondly, metabolic homoeostasisutilizes regulatory processes that control the expression of genesencoding metabolic enzymes. Thus, as McKnight [70] recentlyreminded us, there are many links between gene regulation andmetabolic status. It might be considered inevitable that controlover phosphate supply to yeast, and the regulation of PHO5 ex-pression, must be intertwined with regulation of inositol phos-phate synthesis, which of course is a phosphate-consumingprocess. The big question, therefore, is whether this apparenteffect of Ipmk upon chromatin remodelling reflects a rathermundane metabolic control process, or instead is this really partof a wider conspiracy that utilizes inositol phosphate turnover tocontrol gene expression? Answers to these questions may comefrom efforts to determine whether or not the catalytic domain of


272 S. B. Shears

Impk regulates transcription in other organisms. Given that ad-vances in organism complexity walk hand-in-hand with increasedsophistication of transcriptional control processes [71], there areno guarantees that Ipmk has been retained throughout evolutionas an important regulator of gene expression. Hopefully, futurework on the role of Ipmk in higher eukaryotes will address theseissues.

Towards understanding the mechanism by which the inositolkinase activity of Ipmk might regulate transcription: directroles for inositol phosphates?

To address directly the extent of versatility of the kinase activityof Ipmk, it must be asked whether its postulated transcriptionaleffects are intrinsic to the catalytic domain of the protein, orare instead promoted in a secondary manner by the inositolphosphates themselves. A recent study is relevant to this question;that report indicated that certain inositol phosphates can remodelchromatin in vitro [72]. InsP6 provided the main focus for thiswork, but it is a different and more minor part of the study –the effects of Ins(1,4,5,6)P4 and Ins(1,3,4,5,6)P5 – that hasgarnered the most attention [64,67]. It was suggested thatthe effects of Ipmk upon PHO5 expression are due to directstimulation of ‘nucleosome sliding’ by Ins(1,4,5,6)P4 andIns(1,3,4,5,6)P5 [67,72]. Nucleosomes are the basic repetitive unitof chromatin: histone octomers around which is wrapped approx.150 bp of DNA [73]. Regulatory elements can be exposed whennucleosomes are nudged along the DNA helix by ATP-consum-ing nucleosome remodelling factors. Wu and colleagues [72]studied nucleosome movement along a Drosophila hsp70 DNAfragment driven by the yeast SWI/SNF chromatin remodellingcomplex. It was reported that 500 µM Ins(1,4,5,6)P4 orIns(1,3,4,5,6)P5 stimulated this nucleosome sliding [72]. Lowerconcentrations have also been studied, but the effects werequite small. For example, 100 µM Ins(1,4,5,6)P4 elicited onlya 12% effect, and 50 µM was ineffective (C. Wu, personalcommunication). How do these concentrations compare withthe levels of Ins(1,4,5,6)P4 and Ins(1,3,4,5,6)P5 in intact yeastcells? These have not been assessed directly, but we can obtainreasonable estimates from cells that have been labelled with[3H]inositol to isotopic equilibrium. In such experiments, thelevels of [3H]Ins(1,4,5,6)P4 and [3H]Ins(1,3,4,5,6)P5 in Sacch.cerevisiae were, respectively, 0.6% and 2% of [3H]InsP6 levels[29]. Cellular levels of InsP6 in yeasts are approx. 40 µM [74].This leads us to conclude that total cellular Ins(1,4,5,6)P4 levelsin yeast will not exceed 0.3 µM; the corresponding value forIns(1,3,4,5,6)P5 is 1 µM. Thus Wu and colleagues [72] studied theeffects of inositol phosphates at concentrations that are 500–1700times higher than estimated cellular levels. In such circumstances,it is difficult to accept that these effects are physiologicallyrelevant.

Wu and colleagues [72] argued that compartmentalization ofIns(1,4,5,6)P4 in the nucleus might elevate its levels locally. Towhat extent might this be possible? The nucleus is where most(although not all) of the cells’ Ipmk is located [23,75]. How-ever, if we assume Ins(1,4,5,6)P4 cannot escape the nucleus [eventhough Ins(1,4,5)P3 must get in], and further assuming the yeastnucleus comprises 10% of the total cell volume, then the nu-cleus could not concentrate Ins(1,4,5,6)P4 more than 10-foldabove its average cellular level of <0.3 µM. This theoreticalmaximum concentration of 3 µM is still 30-fold less than aconcentration of Ins(1,4,5,6)P4 that only affected nucleosomesliding to a very marginal extent (12%; see above). In any case,there is also a fundamental kinetic constraint that will prevent theaccumulation of locally elevated levels of Ins(1,4,5,6)P4; even low

micromolar Ins(1,4,5,6)P4 levels will discourage its own syn-thesis, by virtue of it being an alternative substrate of Ipmk (seeabove) that will competitively inhibit Ins(1,4,5)P3 phosphoryl-ation. As for Ins(1,3,4,5,6)P5, the end product of Ipmk activity, itis freely dispersed around the cytosol, at least in human cells[35]. Thus the study by Wu and colleagues [72] does notprovide sufficient evidence that Ins(1,4,5,6)P4 or Ins(1,3,4,5,6)P5

can modulate chromatin remodelling in vivo. Finally, if it turnsout that Ipmk is also an active PtdIns(4,5)P2 3-kinase, as unpub-lished work suggests (A. Resnick, A. Saiardi and S. H. Snyder,personal communication), PtdIns(3,4,5)P3 might then become analternative candidate for regulating chromatin remodelling.

INS(1,3,4)P3/INS(3,4,5,6)P4 KINASE

Discovery

The enzyme activity that phosphorylates Ins(1,3,4)P3 toIns(1,3,4,6)P4 was discovered in 1987 by two groups workingindependently [76,77]. At about the same time, a third group[78] discovered another enzyme activity that phosphoryl-ates Ins(3,4,5,6)P4 to Ins(1,3,4,5,6)P5. [In the latter paper, D-Ins(3,4,5,6)P4 is named L-Ins(1,4,5,6)P4. The use of the ‘L’-stereochemical designation was perfectly logical and appropriate,but these days it is considered more convenient to use the ‘D’designation exclusively – usually implicitly – when describingany myo-inositol phosphate.] The separate purification andcharacterization of both the Ins(1,3,4)P3 and Ins(3,4,5,6)P4 kinaseactivities led to the discovery that each enzyme was stronglyinhibited by the other’s substrates [79,80]. These observationsraised the possibility that both activities might be performedby enzymes that are, at the very least, highly similar. Majerus’laboratory cloned the human Ins(1,3,4)P3 6-kinase in 1996 [81].Subsequently, we [82] independently obtained a human brainkinase cDNA as an EST (expressed sequence tag) clone, ex-pressed recombinant protein in Escherichia coli, and we showedthat this single 46 kDa enzyme catalysed both Ins(1,3,4)P3 5/6-kinase and Ins(3,4,5,6)P4 1-kinase activities. This catalytic ver-satility is physiologically relevant in vivo (see below).

Kinase mRNA transcripts are ubiquitous in animal cells, withespecially high levels of expression in the brain and the heart[81]. It is possible that different mRNA transcripts might beformed by alternative splicing of the 5′ non-translated region [82].There appears to be only one isoform of this kinase expressedin mammalian cells (Genbank NP 055031). Alignments withgenomic sequence (Genbank NT 026437) indicate that the mRNAtranscript arises from 12 exons on chromosome 14. [Unfortu-nately, the HUGO-approved name for this gene is ITPK1 (inositoltrisphosphate kinase), which predates the discovery of its multiplecatalytic activities and misleadingly implies similarities withITPKa, ITPKb and ITPKc, the very different genes for theIns(1,4,5)P3 3-kinase family.]

This Ins(1,3,4)P3 kinase/Ins(3,4,5,6)P4 kinase is widely dis-tributed across the phylogenetic spectrum. For example, thereare apparent homologues in the mouse (GenBank NM 172584;94% identical with human) and Xenopus (BC054977; 78%identity). Surprisingly, in view of its relatively small genome, themustard weed Arabidopsis appears to possess three homologues(AY050408, AY096412 and AY096566; 16–29% identity).Entamoeba histolytica, a tissue-lysing amoeboid pathogen, alsoreportedly contains a homologue, but this enzyme is atypical inthat it also phosphorylates Ins(1,4,5)P3 [83]. Actually, a broaderaspect of the inositol phosphate enzymology of this organism isatypical, in that it has apparently adapted to the synthesis of largequantities of neo-inositol phosphates instead of the more familiar



myo-inositol-based family [84]. Yeasts apparently do not have theIns(1,3,4)P3 kinase/Ins(3,4,5,6)P4 kinase.

Biological significance of Ins(3,4,5,6)P4

The kinase that phosphorylates Ins(3,4,5,6)P4 is of particular bio-logical interest, because Ins(3,4,5,6)P4 is an intracellular signal-ling molecule. For example, Ins(3,4,5,6)P4 inhibits conductanceof those Cl− channels in the plasma membrane that are activated byCa2+ and CaMKII (calmodulin-dependent protein kinase II) [85–87]. This forms part of the process by which salt and fluid secretionis regulated in the lungs, exocrine glands and gastrointestinaltract [88]. The effects of Ins(3,4,5,6)P4 are likely to be more far-reaching. For example, Ca2+-activated Cl− channels also regulateneurotransmission [89] and smooth muscle contraction [90], bothof which should be considered as potential Ins(3,4,5,6)P4 targets.Ins(3,4,5,6)P4 is probably also an important signal in plants; thepolyphosphate modulates pollen tube growth through an effectupon Cl− channels [91].

Ins(3,4,5,6)P4 regulates the conductance of at least one othertype of Cl− channel – a ClC channel in insulin-secretinggranules inside pancreatic β-cells [92]. ClC provides charge-shuntneutralization to support a functionally important acidification ofthe granules by H+-ATPase [92,93]. This is a general process thatcontrols acidification of various types of intracellular vesicles[94]. The ATP-driven delivery of H+ into the vesicle interior iselectrogenic, and if this were not charge-compensated, a largetransmembrane electrical gradient would develop. This wouldprovide an electrochemical impediment to further H+ pumping,even before a substantial acidification of the vesicular interiorcould take place [94–96]. Cl− flux into the vesicles, throughClC, provides the electrical neutralization that is necessary toprevent an inhibitory electrochemical gradient from forming [94–96]. It has been proposed that intraluminal acidification drives thepriming of insulin granules, so that they become competent tofuse with the plasma membrane and release their cargo [93]. Insupport of this idea, we showed that Ins(3,4,5,6)P4 reduces insulinsecretion by decreasing the degree of ClC-dependent acidificationof insulin granules [92].

There are many ClC family members [97]. Antibodies againstthe type 3 ClC protein have disclosed its association with insulingranules [93], although other ClC channels could also be present.So it is not clear which particular ClC proteins are regulated byIns(3,4,5,6)P4. Further work will also be needed in order to deter-mine whether Ins(3,4,5,6)P4 regulates ClC-dependent acidifica-tion of additional intracellular vesicles, the roles of which in-clude endocytosis, enzyme targeting, H+-coupled uptake of smallmolecules (such as neurotransmitters), and optimization of theproteolytic activities of, for example, prohormone processingenzymes [94].

The action of Ins(3,4,5,6)P4 upon Cl− channels can be viewedas being dynamic in nature. The open probability for individualchannels will be determined from a dynamic integration of sep-arate inputs that are either inhibitory [e.g. Ins(3,4,5,6)P4] or stimu-latory (e.g. CaMKII). The strength of input from Ins(3,4,5,6)P4

can be varied in both a temporal and a spatial fashion by virtue oflocalized changes in its concentration. In addition, the efficacyof Ins(3,4,5,6)P4 decreases as the intracellular Ca2+ concentrationrises [98] or when protein phosphatase activities are reduced[98,99]. By regulating the number of Cl− channels in a singlecell that are open at any instant, the whole-cell Cl− current canbe manipulated in a quantitative rheostatic manner. Localizedchanges in the strength of competing inhibitory and stimulatoryinputs may also be co-ordinated reciprocally. That is, an increasein CaMKII activity may accompany a decrease in Ins(3,4,5,6)P4

concentration; the overall effect of such integrative signallingwould be to amplify the signal output [100]. One of the keyplayers in this protein-based computational circuitry is the reversi-ble kinase/phosphatase that establishes the prevailing cellularconcentration of Ins(3,4,5,6)P4 (see below).

There is one very important aspect of Ins(3,4,5,6)P4 that wedo not yet understand, namely its mechanism of action. It hasbeen argued that Ins(3,4,5,6)P4 acts as a channel blocker, sinceits effects upon a recombinant Cl− channel could be observed“. . . in an [experimental] system devoid of additional intracellularsignalling components” [101]. It is argued here that the latterinterpretation is not correct. The cloned Cl− channel in questionwas expressed in Xenopus oocytes, but only at low levels, andit was membrane fragments (not purified recombinant protein)that were extracted from the oocytes and incorporated into lipidbilayers in order to study the action of Ins(3,4,5,6)P4 [101]; manyancillary membrane-associated proteins can safely be assumed tohave contaminated the experiments. Therefore those experimentsdo not exclude Ins(3,4,5,6)P4 acting indirectly upon Cl− channelsthrough intermediary proteins. Indeed, we have shown that theaction of Ins(3,4,5,6)P4 upon epithelial Cl− channels dependsupon a very important class of signalling proteins, namely proteinphosphatases [98,99]. Single-channel recordings also directlydemonstrated that Ins(3,4,5,6)P4 is not a channel blocker [98].

Significance of the catalytic versatility of theIns(1,3,4)P3/Ins(3,4,5,6)P4 kinase

One important function of the Ins(1,3,4)P3 6-kinase activity isto supply precursor material for the synthesis of Ins(1,3,4,5,6)P5,InsP6 and the diphosphorylated inositol phosphates (Scheme 1).In mammals this kinase is absolutely essential to this anabolicprocess; there are no other known pathways that synthesizethese highly phosphorylated metabolites. However, plants [102]and slime moulds [103] have active alternative synthetic routes.Indeed, the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase seems to make onlya minor contribution to InsP6 synthesis in maize, since when thekinase gene is compromised, InsP6 levels fall by only 30% [104].Yeasts use these alternative synthetic routes exclusively; they donot contain any Ins(1,3,4)P3 6-kinase activity.

Another function of the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase is tocontrol the signalling strength of the Ins(3,4,5,6)P4 messenger. Akey aspect of this control process is the reversibility of the kinase.That is, the enzyme also acts as an Ins(1,3,4,5,6)P5 1-phosphatase[105]. The phosphate group that is removed from Ins(1,3,4,5,6)P5

need not be released into the bulk phase, and instead it seemsthat the phosphate can be retained by the enzyme and offered toeither Ins(3,4,5,6)P4 or Ins(1,3,4)P3 [105]. Elevations in the levelsof Ins(1,3,4)P3 will therefore competitively inhibit Ins(3,4,5,6)P4

phosphorylation and, by promoting the dephophorylation of theputative phosphoryl-enzyme intermediate, Ins(1,3,4)P3 augmentsIns(1,3,4,5,6)P5 dephosphorylation [105]. Thus the addition tocells of a cell-permeant analogue of Ins(1,3,4)P3 led to a specificincrease in the cellular levels of Ins(3,4,5,6)P4 [106]. This is animportant point because, in vivo, the cellular levels of Ins(1,3,4)P3

increase by mass-action effects whenever Ins(1,4,5)P3 synthe-sis is activated (see Scheme 1) following receptor-dependentphospholipase-C-mediated hydrolysis of PtdIns(4,5)P2 [107].Thus receptor-activated increases in cellular Ins(1,3,4)P3 levelsare inevitably coupled to increases in Ins(3,4,5,6)P4 levels.

We demonstrated that the enzyme operates as anIns(1,3,4,5,6)P5 1-phosphatase in vivo by stably transfecting T84

cells with the gene for the human enzyme. This did notaffect ‘resting’ levels of Ins(3,4,5,6)P4, presumably because bothIns(3,4,5,6)P4 1-kinase and Ins(1,3,4,5,6)P5 1-phosphatase


274 S. B. Shears

activities were elevated to the same extent [105]. However, duringreceptor activation, the accumulation of Ins(1,3,4)P3 inhibitsthe Ins(3,4,5,6)P4 1-kinase and stimulates the Ins(1,3,4,5,6)P5

1-phosphatase (see above). This shift in the dynamic balancebetween phosphatase and kinase activities leads to a propor-tionately greater overexpression of the phosphatase activity inreceptor-activated cells. Thus the enzyme-transfected T84 cellsshow amplified receptor-dependent increases in Ins(3,4,5,6)P4

levels, relative to vector-transfected cells [105]. These experi-ments confirm that control over the expression of the reversiblekinase/phosphatase can regulate Ins(3,4,5,6)P4 signalling. Thesestudies also show that the strength and duration of theIns(3,4,5,6)P4 signal is inevitably tied to the degree and durationof receptor-dependent phospholipase C activation. This is clearlya situation where promiscuous catalytic activity of an inositolphosphate kinase is biologically important in regulating signaltransduction.

Rationalization of the different inositol phosphate kinase reactionscatalysed by the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase

The structure of this enzyme is not yet known. Indeed, theamino acid residues that contribute to the catalytic site of thisenzyme have not even been identified, in no small part becausethe enzyme has no major sequence similarities with any otherinositol phosphate kinases. However, the Km values for eitherIns(3,4,5,6)P4 and Ins(1,3,4)P3 as substrate (0.1–0.5 µM) areidentical with the K i values determined when each substrateinhibits phosphorylation of the other [79,82,106,108]. It is thus areasonable assumption that the enzyme uses a single catalytic siteto perform all of its inositol phosphate kinase activities. We cantherefore rationalize the enzyme’s promiscuous catalytic activity(Figure 5) using some of the same principles that we apply to theinositol phosphate kinase activities of Ipmk (Figure 4). Thus we[98] recently proposed a model in which substrates can bind tothe active site of the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase in threedifferent orientations (Figure 5). The model offers two modesof Ins(1,3,4)P3 and Ins(1,2,4)P3 binding, to explain why bothsubstrates are phosphorylated at the 5-hydroxy and 6-hydroxygroups [80–82,109]. Finally, Figure 5 accounts for the ability ofthe kinase to phosphorylate the 1-hydroxy group of Ins(3,4,6)P3

[105], a non-physiological substrate (in mammals at least).In Figure 5, groups attached to the inositol ring that are

coloured red have a common nature and orientation in all threebinding modes; these are likely to be the most important of thegroups designating substrate recognition. Yet, by themselves,these groups provide insufficient ligand recognition elements,because they are all present in Ins(1,4)P2, which is not a substrate[105]. Thus, as with Ipmk (Figure 4), it is proposed that eachbinding mode utilizes different sets of substrate recognitionmotifs in a combinatorial manner. Further support for thiscombinatorial recognition hypothesis comes from the followingobservation. The enzyme apparently has a preference for bindingIns(1,3,4)P3 in mode 2 (since it is phosphorylated predominantlyat the 6-hydroxy group [81,82]). In contrast, mode 3 bindingis preferred for Ins(1,2,4)P3 (it is phosphorylated predominantlyat the 5-hydroxy group [109]). These different substrate-bindingpreferences must be dictated by the phosphate and hydroxy groupsat the 2- and 3-positions of Ins(1,3,4)P3 and Ins(1,2,4)P3, eventhough these groups are only conserved in one (coloured green inFigure 5) or two (coloured purple) of the three modes. As withIpmk (see above), an active site that utilizes both flexible andrigid regions [4] would seem to be particularly well suited to thiscombinatorial model for ligand recognition. Further structural andmutagenic work will hopefully clarify this hypothesis.

Figure 5 Proposed combinatorial recognition motifs for the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase

The diagram shows three proposed substrate-binding modes utilized by the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase, as originally outlined in [105]. Note where the inositol ring has beenrotated through vertical and/or horizontal planes in order to align equivalent recognition motifs.Groups that are coloured red are common to all three binding modes and are proposed tobe essential for high-affinity ligand recognition. The site of phosphorylation is indicated by ayellow circle. Groups coloured purple are common to two binding modes and are consideredmost likely to play a mode-specific role in substrate recognition. Groups coloured green aredeemed least important to substrate recognition, as their position is common to only one bindingmode.

Protein kinase activity: additional catalytic versatility?

Predictive modelling studies have led to the suggestion that theIns(1,3,4)P3/Ins(3,4,5,6)P4 kinase may have a structure similar tothat exhibited by the ‘ATP-grasp’ family of protein kinases, inwhich ATP is held between two antiparallel β-sheets [110]. Thiscomparison suggests that there is a relatively close evolutionarylink between protein kinases and the Ins(1,3,4)P3/Ins(3,4,5,6)P4

kinase. One group [111,112] has suggested that there is also afunctional similarity. Both the native kinase purified from bovinebrain and the baculovirus-expressed recombinant human kinasehave been reported to phosphorylate Ser and Thr residueson several transcription factors [ATF-2 (activating transcriptionfactor-2), c-jun, IκBα and p53], at least in vitro [111,112]. So far,there is no evidence that, in vivo, the Ins(3,4,5,6)P4/Ins(1,3,4)P3

kinase can either phosphorylate any of these proteins or affecttheir transcriptional functions. In the specific case of IκBα,both its nuclear translocation and its transcriptional activity re-mained unmodified following overexpression of the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase in HEK cells [113]. The recombi-nant Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase can phosphorylate itselfin vitro [112], but this observation in itself does not shed light on



whether or not the enzyme has protein kinase activity in vivo;many ATP-binding proteins that are not protein kinases stillundergo autophosphorylation [114]. For a kinase of this size, andconsidering the modelling prediction of a single ATP-bindingsite [110], it could be anticipated that the inositol phosphatekinase and protein kinase catalytic sites would overlap. However,it was reported that protein kinase activity was unaffected byIns(3,4,5,6)P4 [112]. Thus there is as yet no suggestion thatinositol phosphates might modulate this protein kinase activity,and the significance of this proposed multifunctionality remainsunclear at present.

Majerus and colleagues [111] have presented evidence that theIns(1,3,4)P3/Ins(3,4,5,6)P4 kinase associates with COP9 duringits purification from bovine brain. COP9 is a key player in theacronym-heavy ubiquitination field. The COP9 multimeric sig-nalosome (SCN) regulates the activity of SCF-type E3 ubiquitinligases [115,116]; SCF itself comprises four subunits: an SKP1(S-phase kinase-associated protein 1)-like protein, a ‘cullin’ pro-tein, a so-called ‘F-box protein’ (FBP) and an Rbx1/Hrt1/Roc1subunit [116]. ‘Neddylation’ (i.e. attachment of the Nedd8 pro-tein) to cullin stimulates the ability of SCF to ubiquitinateproteins, thereby targeting them for proteolytic destruction. SinceCOP9 deneddylates the cullins, the entire complex can down-regulate the activities of cullin-based ligases [115]. While bio-chemical evidence supports this notion, genetic data indicate thatCOP9 is also a positive regulator of cullin function [115]. Thereis still further evidence that a protein kinase activity co-purifieswith native COP9; this kinase has been reported to phosphorylatecertain transcription factors – including p53 and IκBα – therebytargeting them for ubiquitination-initiated degradation [117,118].Majerus and colleagues [111] proposed that this protein kinaseactivity in COP9 was catalysed by the Ins(1,3,4)P3/Ins(3,4,5,6)P4

kinase. However, Dubiel’s group [117] have purified the COP9complex from human erythrocytes in such a way that all of theIns(1,3,4)P3/Ins(3,4,5,6)P4 kinase activity was removed, yet pro-tein kinase activity was retained, and attributed to protein kinase Dand casein kinase [117]. The absence of inositol phosphate kinaseactivity was determined by Western blotting [117] and by enzymeassays (S. B. Shears and W. Dubiel, unpublished work). The latterobservations dispute the idea that the Ins(1,3,4)P3/Ins(3,4,5,6)P4

kinase is “the COP9-associated kinase”, as suggested by Majerus’laboratory [111]. On the other hand, an association of the kinasewith COP9 in vitro may yet prove to be more than coincidental.Many proteins do bind to COP9, so it has been suggested that thesignalosome acts as a scaffold, not just for the constituents ofthe proteolytic process, but also for some of the doomed substrates[116]. Perhaps the cellular levels of Ins(1,3,4)P3/Ins(3,4,5,6)P4

kinase are regulated by COP9. In fact, experiments with adrenalglomerulosa cells performed some years ago [119] suggestthat the levels of cellular Ins(1,3,4)P3 6-kinase activity mightbe controlled by a cAMP-mediated, phosphorylation-dependentprocess. Other more direct protein–protein interactions of COP9with the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase may also influence itsactivity [111]. We do not know of any other regulatory processesthat might modulate Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase activity.For example, it is not affected by either Ca2+ (plus or minuscalmodulin or CaMKII), protein kinase A or protein kinase C([79]; X. Yang and S. B. Shears, unpublished work).

Overexpression of the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase inHEK cells has recently been reported to protect against TNFα(tumour necrosis factor α)-induced apoptosis [113]. Conversely,reduced expression of the kinase by RNAi (RNA interference)increased the susceptibility of HeLa cells to TNFα-inducedapoptosis [113]. However, changes in the levels of expressionof this kinase can have a multitude of effects. The Ins(1,3,4)P3/

Ins(3,4,5,6)P4 kinase is a central player in the pathway of synthesisof many highly phosphorylated inositol phosphates (Scheme 1),any of which might be responsible for the modified sensitivity toapoptosis. Thus the study of Sun et al. [113] does not yet constituteevidence for multifunctionality of the Ins(1,3,4)P3/Ins(3,4,5,6)P4

kinase itself.

DIPHOSPHOINOSITOL POLYPHOSPHATE KINASES

Discovery and characteristics

Two independent reports in 1993 described enzyme activitiesin cell lysates that converted Ins(1,3,4,5,6)P5 and InsP6 intodiphosphorylated inositol phosphates [120,121]. Since then, muchof what we have since learnt about the proteins that synthesizethese diphosphates has come from the work of Snyder, Saiardi andtheir colleagues. Three isoforms of these kinases, ranging in sizefrom 46 to 49 kDa, have been cloned from animals [16,122,123].Homologues in yeast (120 kDa; [16]) and Dictyostelium (72 kDa;[124]) have also been identified. All of these kinases share twoimportant features with Ipmk (Figure 2), namely the PxxxDxKxGcatalytic domain, and a remote, catalytically essential Ser-Ser-Leu-Leu (or similar; see Figure 2) tetrapeptide. Yeast andDictyostelium each possess only one isoform of InsP6 kinase,so disruption of the corresponding gene reduces the syn-thesis of diphosphoinositol polyphosphates nearly completely[60,124,125]. There is an under-appreciated point that emergesfrom these experiments. In spite of being InsP6 kinase null, andeven though these cells contain active phosphohydrolase acti-vities towards the diphosphoinositol polyphosphates [126], theynevertheless retain small steady-state amounts of these diphos-phates [60,61,124]. What might be the source of the remainingsynthetic activity? Are such data a warning that it mightbe premature to dismiss as non-physiological the diphospho-inositol polyphosphate synthase activity of Ipmk (see above)?Alternatively, it has been reported that the ‘PP-InsP5 kinase’in Dictyostelium can phosphorylate InsP6 [84]. Perhaps the stilluncloned PP-InsP5 kinase from yeast and other organisms (seeScheme 1 and [127,128]) might also utilize InsP6 as a substrate.

The yeast InsP6 kinase (Kcs1) is endowed with twoleucine zipper motifs. These are amphipathic heptad repeats[(LxxxxxxL)n, where the value of n is typically 4 or 5] which formcoiled-coil structures that dimerize; leucine zippers are found inproteins that regulate transcription [129]. Indeed, there is evidencethat the yeast InsP6 kinase activity does influence gene expression(see above and [54]). These leucine-based heptad repeats arenot conserved in the mammalian InsP6 kinases, but there isindirect evidence that these enzymes may still have a nuclearfunction. When the type 2 InsP6 kinase was overexpressed in HEKcells as a fusion construct with green fluorescent protein, it wasconcentrated predominantly in the nucleus, although the proteindoes not yet have a recognized nuclear localization signal [122];in similar experiments, the type 1 and 3 kinases were expressedin both the nucleus and the cytosol [122].

The InsP6 kinase in Dictyostelium is activated when thefree-living amoeboid form of this organism exhausts its foodsupply and aggregates into a multicellular organism; in thehours immediately following starvation, the cellular levels ofPP-InsP5 increase 10-fold [130] through the actions of the G-protein-coupled cAMP receptor [124]. This apparently regulateschemotaxis (see below). Release of Ca2+ from the endoplasmicreticulum inhibits cellular InsP6 kinase activity [131], but theunderlying mechanisms have not been elucidated. There is littleother information concerning how the activities of the InsP6

kinases are regulated by the cell.


276 S. B. Shears

Catalytic versatility

The mammalian type 1 and 3 kinases phosphorylateIns(1,3,4,5,6)P5 and InsP6 with nearly equal affinities and similarVmax values [60,122]. Thus the type 1 and 3 kinases appear tobe indifferent to the presence or absence of a 2-phosphate onthe inositol ring. Despite this substrate versatility, the enzymesare almost universally known as ‘InsP6 kinases’. (It has beensuggested that these enzymes might be more appropriately named‘diphosphoinositol polyphosphate synthases’ [132]; this idea hasbeen met with a degree of enthusiasm that might benevolently bedescribed as underwhelming.) The fact that both Ins(1,3,4,5,6)P5

and InsP6 are physiologically significant substrates in vivo canbe shown by perturbing the dynamic equilibrium of the kinase/phosphatase substrate cycles in which they participate (seeScheme 1). This can be achieved simply by adding fluoride tointact cells to inhibit the diphosphoinositol polyphosphate phos-phohydrolases [133], exposing the underlying activities of thekinases. Cellular levels of both PP-InsP4 and PP-InsP5 increasesubstantially, and this is accompanied by decreases in the levels ofthe Ins(1,3,4,5,6)P5 and InsP6 precursors [120,134]. On the otherhand, the type 2 kinase is less versatile; it is kinetically morespecialized to metabolize InsP6 {the affinity for which is 20-foldgreater than that for Ins(1,3,4,5,6)P5; [60]}.

The physiological significance of the kinases that fabricatediphosphate derivatives from both Ins(1,3,4,5,6)P5 and InsP6

is unclear. At least in Sacch. cerevisiae, the two diphosphatesin question, PP-InsP4 and PP-InsP5 respectively (Figure 3),have some functional redundancy. This conclusion arises froman analysis of vacuolar biogenesis in cells that lack either theInsP6 kinase (kcs1�) or the Ins(1,3,4,5,6)P5 kinase (ipk1�).There is fragmented vacuolar morphology in kcs1� cells, inwhich synthesis of diphosphorylated inositol phosphates iscompromised [60–62]. The ipk1� cells cannot synthesize InsP6,and thus they also do not contain any PP-InsP5 or [PP]2-InsP4

(see Scheme 1). Yet the ipk1� cells possess normal vacuolarmorphology [62]. The explanation for these observations isthat the ipk1� cells do still contain active Kcs1, which nowcompensates for the absence of InsP6 by instead phosphorylatingIns(1,3,4,5,6)P5 to PP-InsP4 (see Scheme 1), and to materialtentatively determined (see below) to be [PP]2-InsP3 [62]. ThusIns(1,3,4,5,6)P5 di-phosphorylation functionally compensates forloss of InsP6 di-phosphorylation, as least as far as vacuolarbiogenesis is concerned.

The InsP6 kinases also have the capacity to be reversibleunder in vitro assay conditions that are physiologically relevant[135]. That is, when purified enzyme is incubated with PP-InsP5

and ADP, then InsP6 and ATP are formed [135]. The substrateaffinities and Vmax values for the ‘forward’ (PP-InsP5 synthase)and ‘reverse’ (PP-InsP5 phosphatase) reactions were determinedto be very similar [135]. These data led to the proposal that PP-InsP5 might be a ‘high-energy’ phosphate donor, but any suchrole must be very limited, since there is only 1–5 µM PP-InsP5

in cells [136]. It has also not been possible to demonstrate ifkinase reversibility may be significant in vivo. Independentlyof such a process, cells already have considerable capacity todephosphorylate PP-InsP5 through the actions of a family ofdedicated phosphatases [133,137].

A demonstration that purified recombinant InsP6 kinase canphosphorylate PP-InsP5 further, in vitro at least [60], has inspiredspeculation that the elusive ‘PP-InsP5 kinase’ activity (seeScheme 1) might in fact be catalysed by the InsP6 kinase in vivo[124]. This idea has to be reconciled with the demonstration thatcellular PP-InsP5 kinase activity can be purified away from all cel-lular InsP6 kinase activity ([128]; T. Zhang and S. B. Shears,

Figure 6 Versatility of the ‘InsP6 kinases’: bisdiphosphoinositol poly-phosphate or triphosphate?

Recombinant ‘InsP6 kinase’ can further phosphorylate PP-InsP5; the product has tentativelybeen proposed to be [PP]2-InsP4 [60], but it has not been excluded that the product mightinstead be a triphosphate, i.e. PPP-InsP3. The Figure shows these two possibilities, with theposition of the disputed phosphate group coloured red for clarity. An analogous problem appliesto the nature of the product of PP-InsP4 phosphorylation by the InsP6 kinase (see the text).

unpublished work). There is also a problem that a maximum ofonly approx. 5–10% of added PP-InsP5 can be phosphorylatedby the recombinant InsP6 kinase, irrespective of the quantityof enzyme in the incubations, or the degree of patience of theexperimenter [60]. One possibility is that the InsP6 kinase mightassociate with a regulatory protein that switches its reactivity, butif this were the case, both substrate specificity and the positionalpreference of the kinase would need to be modified.

Another potential complication is that the phosphorylationof PP-InsP5 by recombinant InsP6 kinase in vitro mightconceivably yield a triphosphate, PPP-InsP5, rather than the bis-diphosphoinositol, [PP]2-InsP4 (Figure 6). The latter is certainlya natural product, according to NMR analysis of slime moulds[138–140] and enzymic assays in mammalian cell lysates [127].In contrast, it is not known if PPP-InsP5 is normally present incells. Equally, the InsP6 kinases can also add an additionalphosphate to PP-InsP4 [60]. The product was designated [PP]2-InsP3 [60]. However, the possibility has not been excluded that theproduct might instead be PPP-InsP4. Overall, there remain someimportant unanswered questions concerning the true catalyticversatility of the family of InsP6 kinases in vivo.

Other roles for the InsP6 kinases?

Diphosphoinositol polyphosphates represent the most extremeexample of the degree to which phosphate groups can beclustered in high density around the inositol ring. Indeed, it isbelieved that a considerable free-energy change accompaniestheir dephosphorylation, due to the relief of severe electrostaticand steric constraints within these molecules [121]. This‘high-energy’ concept has prompted Snyder’s group to studywhether diphosphoinositol polyphosphates might phosphorylateproteins directly. However, the results obtained to date are onlypreliminary and have not yet been published (see [125]).

Genetic disruption of the turnover of diphosphoinositol poly-phosphates has implicated these polyphosphates as being activein several biological arenas. These include effects upon yeastvacuolar biogenesis [60,62], endocytosis [62], stress responses[61], DNA hyper-recombination [125] and apoptosis [141]. In



order to rationalize the diverse nature of these responses, it maybe proposed that diphosphoinositol polyphosphates participatein some as yet uncharacterized fundamental regulatory processthat is applicable to many biological situations. For example, thepolyphosphates might be ligands that modify the functions ofcertain proteins [16,135,142]. In such an event, synthesis and/ormetabolism of the polyphosphates can be envisaged to comprise amolecular switch [16,135,143], akin to G-proteins that function asa binary switch between two interconvertible (GTP-bound activeand GDP-bound inactive) states. Such a process would providedirect roles for the InsP6 kinases (and the diphosphoinositolpolyphosphate phosphohydrolases) in many biological processes.However, there is only circumstantial evidence in support ofthis idea. For example, PP-InsP5 has been found to bind withhigh affinity to a number of cellular proteins in vitro, includingmyelin proteolipid protein [144], coatomer [145,146], as wellas the clathrin-assembly adaptors AP2 [127] and AP3/AP180[62,147]. Unfortunately, as pointed out previously [143], thephysiological relevance of these in vitro binding assays remainsunclear. Typically, such experiments have been conducted in theabsence of bivalent cations, which normally neutralize some ofthe polyphosphate’s high negative charge density. Non-specificunphysiological interactions between PP-InsP5 and proteins aremore likely when bivalent cations are absent.

A recent report [124] does provide a more convincing demon-stration of PP-InsP5 being a regulatory ligand, in this caseby binding to Crac (cytosolic regulator of adenylate cyclase)in the slime mould, D. discoideum. PP-InsP5 competed withPtdIns(3,4,5)P3 for binding to the pleckstrin homology domainof Crac, and this phenomenon appeared to influence chemotaxis[124]. Whether PP-InsP5 is a physiologically relevant competitorfor PtdIns(3,4,5)P3 binding proteins in other organisms remainsto be seen. The slime moulds are unusually suited to this ligandcompetition, as they contain concentrations of diphosphoinositolpolyphosphates in excess of 100 µM [130,138]. The only otherorganisms known to have such high levels of these compoundsare the pathogen Entamoeba histolytica and the free-livingPhreatamoeba balamuthi [84]. In contrast, yeasts and mammaliancells contain only 1–5 µM PP-InsP5 [74,136], so competition withPtdIns(3,4,5)P3 will be less likely in those organisms.

A specific role for type 1 mammalian InsP6 kinase emergedfrom a study that identified a new guanine nucleotide exchangefactor for Rab3a, named GRAB. The neuron-specific Rab3a is oneof over 40 mammalian Rabs, which regulate vesicle traffickingprocesses and are members of the Ras superfamily of smallGTPases [148]. It was found that the ability of GRAB to promoteGDP release from Rab3A was reduced in cells in which thetype 1 mammalian InsP6 kinase was overexpressed [149]. Thisseemed to result from competition between the kinase and Rab3afor binding to GRAB [149]. This effect was independent of thekinase catalytic domain, since a catalytically inactive mutantof the type 1 kinase was equally effective [149]. Thus there isno evidence that this effect is tied to the turnover of inositolphosphates. Finally, a recent study, using DNA microarrays andNorthern analysis, identified 30 genes in Sacch. cerevisiae whoseexpression was influenced by the InsP6 kinase [54]. Of course,it will be useful to follow up these microarray experiments withstudies that separate direct primary effects of the kinase fromsecondary indirect events.

CONCLUSIONS

In this review, the case has been made that certain inositol phos-phate kinases do indeed have versatile catalytic domains that can

metabolize a range of inositol phosphate substrates in vivo. Anew combinatorial ligand recognition hypothesis is put forwardin order to explain this versatility. Sometimes it is clear thatthis metabolic promiscuity has been exploited to facilitate tightregulation of physiological processes. Such multifunctionalitycan add to the complexity with which signal transductionpathways interact. The latter point is best illustrated by theIns(1,3,4)P3/Ins(3,4,5,6)P4 kinase. However, there are other caseswhere it is still unclear what the significance might be of akinase (e.g. the ‘InsP6 kinase’) phosphorylating two or moreinositol phosphates. The proposed significance of the proteinkinase activity of the Ins(1,3,4)P3/Ins(3,4,5,6)P4 kinase also needsfurther exploration. As for the postulated ‘moonlighting’ [1] rolesof transcriptional regulation by Ipmk, the function of the acidictranscriptionally active domains in Ipmk from certain yeasts has asolid basis in 20 years of research. More recent work that focuseson transcriptional functions for the kinase activity of Ipmk isless well developed. It seems fair to conclude that the proposedrole of the kinase catalytic domain in the regulation of arginine-metabolizing genes in yeast is still controversial. Nevertheless, thehypothesis still survives, in the more recent guise of the regulationof phosphate-responsive genes. In any case, it seems that ourappreciation of the versatility of inositol phosphate kinases islikely to increase, as further work pursues some of the ideas andquestions outlined in this review.

I am indebted to Dr Evelyne Dubois, Dr Terry Cooper, Dr Susan Wente, Dr Carl Wu, Dr JohnYork, Dr Adolfo Saiardi and Dr Sol Snyder for taking the time to answer patiently my manydetailed questions; it was especially helpful that some of these colleagues were happy tolet me cite some of their unpublished results. I am also grateful to Dr Robin Irvine for hisuseful comments, and for convincing me to submit this review to the Biochemical Journal.I also express my appreciation to the past and present members of the Inositol PhosphateSignaling Group, and to colleagues at NIEHS and other institutions, who have contributedto the data and the concepts discussed in this review. Naturally, despite the assistance Ihave been given, any remaining mistakes and omissions are entirely my responsibility.

REFERENCES

1 Jeffery, C. J. (1999) Moonlighting proteins. Trends Biochem. Sci. 24, 8–112 Jeffery, C. J. (2003) Multifunctional proteins: examples of gene sharing. Ann. Med. 35,

28–353 Jeffery, C. J. (2003) Moonlighting proteins: old proteins learning new tricks.

Trends Genet. 19, 415–4174 Copley, S. D. (2003) Enzymes with extra talents: moonlighting functions and catalytic

promiscuity. Curr. Opin. Chem. Biol. 7, 265–2725 Agranoff, B. W. (1978) Textbook errors: cyclitol confusion. Trends Biochem. Sci. 3,

N283–N2856 Nomenclature Committee of the International Union of Biochemistry (1989) Numbering

of atoms in myo-inositol. Recommendations 1988. Biochem. J. 258, 1–27 Streb, H., Irvine, R. F., Berridge, M. J. and Schulz, I. (1983) Release of Ca2+ from a

nonmitochondrial store in pancreatic cells by inositol-1,4,5-trisphosphate.Nature (London) 306, 67–68

8 Irvine, R. F., Letcher, A. J., Heslop, J. P. and Berridge, M. J. (1986) The inositoltris/tetrakisphosphate pathway – demonstration of Ins(1,4,5)P3 3-kinase activity inanimal tissues. Nature (London) 320, 631–634

9 Takazawa, K., Perret, J., Dumont, J. E. and Erneux, C. (1991) Molecular cloning of ahuman brain inositol 1,4,5-trisphosphate 3-kinase. Biochem. Biophys. Res. Commun.174, 529–535

10 Dewaste, V., Moreau, C., De Smedt, F., Bex, F., De Smedt, H., Wuytack, F., Missiaen, L.and Erneux, C. (2003) The three isoenzymes of human inositol-1,4,5-trisphosphate3-kinase show specific intracellular localization but comparable Ca2+ responses ontransfection in COS-7 cells. Biochem. J. 374, 41–49

11 Irvine, R. F. and Schell, M. (2001) Back in the water: the return of the inositolphosphates. Nat. Rev. Mol. Cell Biol. 2, 327–338

12 Balla, T., Sim, S. S., Baukal, A. J., Rhee, S. G. and Catt, K. J. (1994) Inositolpolyphosphates are not increased by overexpression of ins(1,4,5)P3 3-kinase but showcell-cycle dependent changes in growth factor-stimulated fibroblasts. Mol. Biol. Cell 5,17–28


278 S. B. Shears

13 Chattaway, J. A., Drobak, B., Watkins, P. A. C., Dawson, A. P., Letcher, A. J., Stephens,L. R. and Irvine, R. F. (1992) An inositol 1,4,5-trisphosphate 6-kinase activity in pearoots. Planta 187, 542–545

14 Estevez, F., Pulford, D., Stark, M. J. R., Carter, A. N. and Downes, C. P. (1994) Inositoltrisphosphate metabolism in Saccharomyces cerevisiae: identification, purification andproperties of inositol 1,4,5-trisphosphate 6-kinase. Biochem. J. 302, 709–716

15 Ongusaha, P. P., Hughes, P. J., Davey, J. and Michell, R. H. (1998) Inositolhexakisphosphate in Schizosaccharomyces pombe: synthesis from Ins(1,4,5)P3 andosmotic regulation. Biochem. J. 335, 671–679

16 Saiardi, A., Erdjument-Bromage, H., Snowman, A., Tempst, P. and Snyder, S. H. (1999)Synthesis of diphosphoinositol pentakisphosphate by a newly identified family of higherinositol polyphosphate kinases. Curr. Biol. 9, 1323–1326

17 Odom, A. R., Stahlberg, A., Wente, S. R. and York, J. D. (2000) A role for nuclear inositol1,4,5-trisphosphate kinase in transcriptional control. Science 287, 2026–2029

18 Bercy, J., Dubois, E. and Messenguy, F. (1987) Regulation of arginine metabolism inS. cerevisiae: expression of the three ARGR regulatory genes and cellular localizationof their products. Gene 55, 277–285

19 Qiu, H. F., Dubois, E., Broen, P. and Messenguy, F. (1990) Functional analysis of ARGRIand ARGRIII regulatory proteins involved in the regulation of arginine metabolism inS. cerevisiae. Mol. Gen. Genet. 222, 192–200

20 Messenguy, F., Dubois, E. and Boonchird, C. (1991) Determination of the DNA-bindingsequences of the ARGR proteins to arginine anabolic and catabolic promoters.Mol. Cell. Biol. 11, 2852–2863

21 Dubois, E. and Messenguy, F. (1994) Pleiotropic function of ArgIIIp (Arg82p), one of theregulators of arginine expression in Saccharomyces cerevisiae. Role in expression ofcell-type-specific genes. Mol. Gen. Genet. 243, 315–324

22 Amar, N., Messenguy, F., El Bakkoury, M. and Dubois, E. (2000) ArgRII, a component ofthe ArgR-Mcm1 complex involved in the control of arginine metabolism inSaccharomyces cerevisiae, is the sensor of arginine. Mol. Cell. Biol. 20, 2087–2097

23 Nalaskowski, M. M., Deschermeier, C., Fanick, W. and Mayr, G. W. (2002) The humanhomologue of yeast ArgRIII protein is an inositol phosphate multikinase withpredominantly nuclear localization. Biochem. J. 366, 549–556

24 Chang, S.-C., Miller, A. L., Feng, Y., Wente, S. R. and Majerus, P. W. (2002) The humanhomologue of the rat inositol phosphate multikinase is an inositol 1,3,4,6-tetrakisphosphate 5-kinase. J. Biol. Chem. 277, 43836–43843

25 Saiardi, A., Nagata, E., Luo, H. R., Sawa, A., Luo, X., Snowman, A. M. and Snyder, S. H.(2001) Mammalian inositol polyphosphate multikinase synthesizes inositol 1,4,5-trisphosphate and an inositol pyrophosphate. Proc. Natl. Acad. Sci. U.S.A. 98,2306–2311

26 Stevenson-Paulik, J., Odom, A. R. and York, J. D. (2002) Molecular and biochemicalcharacterization of two plant inositol polyphosphate 6-/3-/5-kinases. J. Biol. Chem.277, 42711–42718

27 Zhang, T., Caffrey, J. J. and Shears, S. B. (2001) The transcriptional regulator, Arg82, isa hybrid kinase with both monophosphoinositol- and diphosphoinositol-polyphosphatesynthase activity. FEBS Lett. 494, 208–212

28 Edwalds-Gilbert, G., Veraldi, K. L. and Milcarek, C. (1997) Alternative poly(A) siteselection in complex transcription units: means to an end? Nucleic Acids Res. 25,2547–2561

29 Saiardi, A., Caffrey, J. J., Snyder, S. H. and Shears, S. B. (2000) Inositol polyphosphatemultikinase (ArgRIII) determines nuclear mRNA export in Saccharomyces cerevisiae.FEBS Lett. 468, 28–32

30 Hughes, P. J. and Drummond, A. H. (1987) Formation of inositol phosphate isomers inGH3 pituitary tumour cells stimulated with thyrotropin releasing hormone. Acute effectsof lithium ions. Biochem. J. 248, 463–470

31 Jenkinson, S., Challiss, R. A. J. and Nahorski, S. R. (1992) Evidence for lithium-sensitive inositol 4,5-bisphosphate accumulation in muscarinic cholinoceptor-stimulated cerebral cortex slices. Biochem. J. 287, 437–442

32 Dubois, E., Dewaste, V., Erneux, C. and Messenguy, F. (2000) Inositol polyphosphatekinase activity of Arg82/ArgRIII is not required for the regulation of the argininemetabolism in yeast. FEBS Lett. 486, 300–304

33 Baron, C. B., Pompeo, J. N. and Azim, S. (1992) Inositol 1,4,5-trisphosphatecompartmentalization in tracheal smooth muscle. Arch. Biochem. Biophys. 292,382–387

34 Zhao, H., Khademazad, M. and Muallem, S. (1990) Agonist-mediated Ca2+ release inpermeabilized UMR-106–01 cells. Transport properties and generation of inositol1,4,5-trisphosphate. J. Biol. Chem. 265, 14822–14827

35 Stuart, J. A., Anderson, K. L., French, P. J., Kirk, C. J. and Michell, R. H. (1994) Theintracellular distribution of inositol polyphosphates in HL60 promyeloid cells.Biochem. J. 303, 517–525

36 Oliver, K. G., Putney, Jr, J. W., Obie, J. F. and Shears, S. B. (1992) The interconversion ofinositol 1,3,4,5,6-pentakisphosphate and inositol tetrakisphosphates in AR4-2J cells.J. Biol. Chem. 267, 21528–21534

37 Craxton, A., Erneux, C. and Shears, S. B. (1994) Inositol 1,4,5,6-tetrakisphosphate isphosphorylated in rat liver by a 3-kinase that is distinct from inositol 1,4,5-trisphosphate 3-kinase. J. Biol. Chem. 269, 4337–4342

38 Zhou, D., Chen, L.-M., Hernandez, L., Shears, S. B. and Galan, J. E. (2001) ASalmonella inositol polyphosphatase acts in conjunction with other bacterial effectors topromote host-cell actin cytoskeleton rearrangements and bacterial internalization.Mol. Microbiol. 39, 248–259

39 Stebbins, C. E. and Galan, J. E. (2001) Structural mimicry in bacterial virulence.Nature (London) 412, 701–705

40 Craxton, A., Caffrey, J. J., Burkhart, W., Safrany, S. T. and Shears, S. B. (1997)Cloning and expression of rat hepatic multiple inositol polyphosphate phosphatase.Biochem. J. 328, 75–81

41 Chi, H., Yang, X., Kingsley, P. D., O’Keefe, R. J., Puzas, J. E., Rosier, R. N., Shears, S. B.and Reynolds, P. R. (2000) Targeted deletion of Minpp1 provides new insight into theactivity of multiple inositol polyphosphate phosphatase in vivo. Mol. Cell. Biol. 20,6496–6507

42 Ali, N., Craxton, A. and Shears, S. B. (1993) Hepatic Ins(1,3,4,5)P4 3-phosphatase iscompartmentalized inside endoplasmic reticulum. J. Biol. Chem. 268, 6161–6167

43 Nogimori, K., Hughes, P. J., Glennon, M. C., Hodgson, M. E., Putney, Jr, J. W. andShears, S. B. (1991) Purification of an inositol (1,3,4,5)-tetrakisphosphate3-phosphatase activity from rat liver and its substrate specificity. J. Biol. Chem. 266,16499–16506

44 Leslie, N. R. and Downes, C. P. (2002) PTEN: The down side of PI 3-kinase signalling.Cell. Signalling 14, 285–295

45 Maehama, T. and Dixon, J. E. (1999) The tumor suppressor, PTEN/MMAC1,dephosphorylates the lipid second messenger, phosphatidylinositol3,4,5-trisphosphate. J. Biol. Chem. 273, 13375–13378

46 Caffrey, J. J., Darden, T., Wenk, M. R. and Shears, S. B. (2001) Expanding coincidentsignaling by PTEN through its inositol 1,3,4,5,6-pentakisphosphate 3-phosphataseactivity. FEBS Lett. 499, 6–10

47 McConnachie, G., Pass, I., Walker, S. M. and Downes, C. P. (2003) Interfacial kineticanalysis of the tumour suppressor phosphatase, PTEN. Evidence for activation byanionic phospholipids. Biochem. J. 374, 947–955

48 Sulis, M. L. and Parsons, R. (2003) PTEN: from pathology to biology. Trends Cell Biol.13, 478–483

49 Shears, S. B. (2003) Ins(1,3,4,5,6)P5, a signal transduction hub. In The Handbook ofCell Signaling (Bradshaw, R. and Dennis, E., eds.), vol. 2, pp. 233–235, Academic Press,San Diego

50 Wilcox, R. A., Safrany, S. T., Lampe, D., Mills, S. J., Nahorski, S. R. and Potter, B. V. L.(1994) Modification at C2 of myo-inositol 1,4,5-trisphosphate produces inositoltrisphosphates and tetrakisphosphates with potent biological activities. Eur. J. Biochem.223, 115–124

51 Ives, E. B., Nichols, J., Wente, S. R. and York, J. D. (2000) Biochemical and functionalcharacterization of inositol 1,3,4,5,6-pentakisphosphate 2-kinases. J. Biol. Chem. 275,36575–36583

52 Bechet, J., Grenson, M. and Wiame, J. M. (1970) Mutations affecting the repressibility ofarginine biosynthetic enzymes in Saccharomyces cerevisiae. Eur. J. Biochem. 12, 31–39

53 El Bakkoury, M., Dubois, E. and Messenguy, F. (2000) Recruitment of the yeast MADS-box proteins, ArgRI and Mcm1 by the pleitropic factor ArgRIII is required for theirstability. Mol. Microbiol. 35, 15–31

54 El Alami, M., Messenguy, F., Scherens, B. and Dubois, E. (2003) Arg82p is a bifunctionalprotein whose inositol polyphosphate kinase activity is essential for nitrogen and PHOgene expression but not forMcm1p chaperoning in yeast. Mol. Microbiol. 49, 457–468

55 Magasanik, B. (1992) Regulation of nitrogen utilization. In The Molecular and CellularBiology of the Yeast Saccharomyces: Gene Expression (Jones, E. W., Pringle, J. R. andBroach, J. R., eds.), pp. 283–317, Cold Spring Harbor University Press, Cold SpringHarbor

56 Wiame, J. M. (1971) The regulation of arginine metabolism in Saccharomycescerevisiae: exclusion mechanisms. Curr. Top. Cell. Regul. 4, 1–38

57 Messenguy, F., Vierendeels, F., Scherens, B. and Dubois, E. (2000) In Saccharomycescerevisiae, expression of arginine catabolic genes CAR1 and CAR2 in response toexogenous nitrogen availability is mediated by the Ume6 (CargRI)-Sin3 (CargRII)-Rpd3(CargRIII) complex. J. Bacteriol. 182, 3158–3164

58 Pringle, J. R. and Mor, J. R. (1975) Methods for monitoring the growth of yeast culturesand for dealing with the clumping problem. Methods Cell Biol. 11, 131–168

59 York, J. D., Odom, A. R., Murphy, R., Ives, E. B. and Wente, S. R. (1999) Aphospholipase C-dependent inositol polyphosphate kinase pathway required forefficient messenger RNA export. Science 285, 96–100

60 Saiardi, A., Caffrey, J. J., Snyder, S. H. and Shears, S. B. (2000) The inositolhexakisphosphate kinase family: catalytic flexibility, and function in yeast vacuolebiogenesis. J. Biol. Chem. 275, 24686–24692



61 Dubois, E., Scherens, B., Vierendeels, F., Ho, M. W. Y., Messenguy, F. and Shears, S. B.(2002) In Saccharomyces cerevisiae, the inositol polyphosphate kinase activity of Kcs1pis required for resistance to salt stress, cell wall integrity and vacuolar morphogenesis.J. Biol. Chem. 277, 23755–23763

62 Saiardi, A., Sciambi, C., McCaffery, J. M., Wendland, B. and Snyder, S. H. (2002)Inositol pyrophosphates regulate endocytic trafficking. Proc. Natl. Acad. Sci. U.S.A. 99,14206–14211

63 York, J. D., Guo, S., Odom, A. R., Spiegelberg, B. D. and Stolz, L. E. (2001) An expandedview of inositol signaling. Adv. Enzyme Regul. 41, 57–71

64 Rando, O. J., Chi, T. H. and Crabtree, G. R. (2003) Second messenger control ofchromatin remodeling. Nat. Struct. Biol. 10, 81–83

65 Kaffman, A., Herskowitz, I., Tjian, R. and O’Shea, E. K. (1994) Phosphorylation of thetranscription factor PHO4 by a cyclin-CDK complex, PHO80-PHO85. Science 263,1153–1156

66 Carroll, A. S. and O’Shea, E. K. (2002) Pho85 and signaling environmental conditions.Trends Biochem. Sci. 27, 87–92

67 Steger, D. J., Haswell, E. S., Miller, A. L., Wente, S. R. and O’Shea, E. K. (2003)Regulation of chromatin remodeling by inositol polyphosphates. Science 299, 114–116

68 Svaren, J. and Horz, W. (1997) Transcription factors vs nucleosomes: regulation of thePH05 promoter in yeast. Trends Biochem. Sci. 22, 93–97

69 Martens, J. A. and Winston, F. (2003) Recent advances in understanding chromatinremodeling by Swi/Snf complexes. Curr. Opin. Genet. Dev. 13, 136–142

70 McKnight, S. (2003) Gene switching by metabolic enzymes – how did you get on theinvitation list? Cell 174, 150–152

71 Levine, M. and Tjian, R. (2003) Transcription regulation and animal diversity. Nature(London) 424, 147–151

72 Shen, X., Xiao, H., Ranallo, R., Wu, W.-H. and Wu, C. (2003) Modulation of ATP-dependent chromatin-remodeling complexes by inositol polyphosphates. Science 299,112–114

73 Becker, P. B. and Horz, W. (2002) ATP-dependent nucleosome remodeling.Annu. Rev. Biochem. 71, 247–273

74 Ingram, S. W., Safrany, S. T. and Barnes, L. D. (2003) Disruption and overexpression ofthe Schizosaccharomyces pombe aps1 gene and the effects on growth rate, morphology,and intracellular diadenosine 5′,5′ ′ ′-P1, P5-pentaphosphate and diphosphoinositolpolyphosphate concentrations. Biochem. J. 369, 519–528

75 Xia, H.-J., Brearley, C., Elge, S., Kaplan, B., Fromm, H. and Mueller-Roeber, B. (2003)Arabidopsis inositol polyphosphate 6-/3-kinase is a nuclear protein that complements ayeast mutant lacking a functional ArgR-Mcm1 transcription complex. Plant Cell 15,449–463

76 Shears, S. B., Parry, J. B., Tang, E. K. Y., Irvine, R. F., Michell, R. H. and Kirk, C. J.(1987) Metabolism of D-myo-inositol 1,3,4,5-tetrakisphosphate by rat liver, includingthe synthesis of a novel isomer of myo-inositol tetrakisphosphate. Biochem. J. 246,139–147

77 Balla, T., Guillemette, G., Baukal, A. J. and Catt, K. J. (1987) Formation of inositol1,3,4,6-tetrakisphosphate during angiotensin II action in bovine adrenal glomerulosacells. Biochem. Biophys. Res. Commun 146, 199–205

78 Stephens, L. R., Hawkins, P. T., Morris, A. J. and Downes, C. P. (1988) L-myo -Inositol1,4,5,6-tetrakisphosphate (3-hydroxy)kinase. Biochem. J. 249, 283–292

79 Tan, Z., Bruzik, K. S. and Shears, S. B. (1997) Properties of the inositol 3,4,5,6-tetrakisphosphate 1-kinase purified from rat liver. Regulation of enzyme activity byinositol 1,3,4-trisphosphate. J. Biol. Chem. 272, 2285–2290

80 Abdullah, M., Hughes, P. J., Craxton, A., Gigg, R., Desai, T., Marecek, J. F., Prestwich,G. D. and Shears, S. B. (1992) Purification and characterization of inositol 1,3,4-trisphosphate 5/6-kinase from rat liver using an inositol hexakisphosphate affinitycolumn. J. Biol. Chem. 267, 22340–22345

81 Wilson, M. P. and Majerus, P. W. (1996) Isolation of inositol 1,3,4-trisphosphate5/6-kinase, cDNA cloning, and expression of recombinant enzyme. J. Biol. Chem. 271,11904–11910

82 Yang, X. and Shears, S. B. (2000) Multitasking in signal transduction by a promiscuoushuman Ins(3,4,5,6)P4 1-kinase/Ins(1,3,4)P3 5/6-kinase. Biochem. J. 351, 551–555

83 Field, J., Wilson, M. P., Mai, Z., Majerus, P. W. and Samuelson, J. (2000) An Entamoebahistolytica inositol 1,3,4-trisphosphate 5/6-kinase has a novel 3-kinase activity.Mol. Biochem. Parasitol. 108, 119–123

84 Martin, J.-B., Laussmann, T., Bakker-Grunwald, T. and Vogel, G. (2000) neo-Inositolpolyphosphates in the amoeba Entamoeba histolytica. J. Biol. Chem. 275, 10134–10140

85 Ho, M. W. Y. and Shears, S. B. (2002) Regulation of calcium-activated chloride channelsby inositol 3,4,5,6-tetrakisphosphate. Curr. Top. Membr. 53, 345–363

86 Carew, M. A., Yang, X., Schultz, C. and Shears, S. B. (2000) Ins(3,4,5,6)P4 inhibits anapical calcium-activated chloride conductance in polarized monolayers of a cysticfibrosis cell-line. J. Biol. Chem. 275, 26906–26913

87 Xie, W., Kaetzel, M. A., Bruzik, K. S., Dedman, J. R., Shears, S. B. and Nelson, D. J.(1996) Inositol 3,4,5,6-tetrakisphosphate inhibits the calmodulin-dependent proteinkinase II-activated chloride conductance inT84 colonic epithelial cells. J. Biol. Chem.271, 14092–14097

88 Petersen, O. H. (1992) Stimulus-secretion coupling: cytoplasmic calcium signals andthe control of ion channels in exocrine acinar cells. J. Physiol. (Cambridge, U.K.) 448,1–51

89 Kenyon, J. L. and Scott, R. H. (2002) Ca2+-activated Cl− channels as Ca2+ sensors withparticular reference to the modulation of neuronal excitability. In Calcium-ActivatedChloride Channels (Fuller, C. M., ed.), pp. 135–166, Academic Press, San Diego

90 Large, W. A., Greenwood, I. A. and Piper, A. S. (2002) Recent advances on the propertiesand role of Ca2+-activated chloride currents in smooth muscle. In Calcium-ActivatedChloride Channels (Fuller, C. M., ed.), pp. 99–118, Academic Press, San Diego

91 Zonia, L., Cordeiro, S., Tupy, J. and Feijo, J. A. (2002) Oscillatory chloride efflux at thepollen tube apex has a role in growth and cell volume regulation and is targeted byinositol 3,4,5,6-tetrakisphosphate. Plant Cell 14, 2233–2249

92 Renstrom, E., Ivarsson, R. and Shears, S. B. (2002) Ins(3,4,5,6)P4 inhibits insulingranule acidification and fusogenic potential. J. Biol. Chem. 277, 26717–26720

93 Barg, S., Huang, P., Eliasson, L., Nelson, D. J., Obermuller, S., Rorsman, P., Thevenod, F.and Renstrom, E. (2001) Priming of insulin granules for exocytosis by granular chlorideuptake and acidification. J. Cell Sci. 114, 2145–2154

94 Nishi, T. and Forgac, M. (2002) The vacuolar (H+)-ATPases – nature’s most versatileproton pumps. Nat. Rev. Mol. Cell. Biol. 3, 94–103

95 Jentsch, T. J., Stein, V., Weinreich, F. and Zdebik, A. A. (2002) Molecular structure andphysiological function of chloride channels. Physiol. Rev. 82, 503–568

96 Al-Awqati, Q., Barasch, J. and Landry, D. (1992) Chloride channels of intracellularorganelles and their potential role in cystic fibrosis. J. Exp. Biol. 172, 245–266

97 Estevez, R. and Jentsch, T. J. (2002) CLC chloride channels: correlating structure withfunction. Curr. Opin. Struct. Biol. 12, 531–539

98 Ho, M. W. Y., Kaetzel, M. A., Armstrong, D. L. and Shears, S. B. (2001) Regulation of ahuman chloride channel: a paradigm for integrating input from calcium, CaMKII andIns(3,4,5,6)P4. J. Biol. Chem. 276, 18673–18680

99 Xie, W., Solomons, K. R. H., Freeman, S., Kaetzel, M. A., Bruzik, K. S., Nelson, D. J. andShears, S. B. (1998) Regulation of Ca2+-dependent Cl− conductance in T84 cells,cross-talk between Ins(3,4,5,6)P4 and protein phosphatases. J. Physiol. (Cambridge,U.K.) 510, 661–673

100 Ferrell, Jr, J. E. (1996) Tripping the switch fantastic: how a protein kinase cascadecan convert graded inputs into switch-like outputs. Trends Biochem. Sci. 21,460–465

101 Ismailov, I. I., Fuller, C. M., Berdiev, B. K., Shlyonsky, V. G., Benos, D. J. and Barrett,K. E. (1996) A biologic function for an “orphan” messenger: D-myo-inositol3,4,5,6-tetrakisphosphate selectively blocks epithelial calcium-activated chloridecurrent. Proc. Natl. Acad. Sci. U.S.A. 93, 10505–10509

102 Brearley, C. A. and Hanke, D. E. (1996) Metabolic evidence for the order of addition ofindividual phosphate esters to the myo-inositol moiety of inositol hexakisphosphate inthe duckweed Spirodela polyrhiza L. Biochem. J. 314, 227–233

103 Stephens, L. R. and Irvine, R. F. (1990) Stepwise phosphorylation of myo-inositolleading to myo-inositol hexakisphosphate in Dictyostelium. Nature (London) 346,580–583

104 Shi, J., Wang, H., Wu, Y., Hazebroek, J., Meeley, R. B. and Ertl, D. S. (2003) The maizelow-phytic acid mutant Ipa2 is caused by mutation in an inositol phosphate kinase gene.Plant Physiol. 131, 507–515

105 Ho, M. Y. H., Yang, X., Carew, M. A., Zhang, T., Hua, L., Kwon, Y. U., Chung, S. K., Adelt,S., Vogel, G., Riley, A. M. et al. (2002) Regulation of Ins(3,4,5,6)P4 signaling by areversible kinase/phosphatase. Curr. Biol. 12, 477–482

106 Yang, X., Rudolf, M., Carew, M. A., Yoshida, M., Nerreter, V., Riley, A. M., Chung, S. K.,Bruzik, K. S., Potter, B. V., Schultz, C. and Shears, S. B. (1999) Inositol1,3,4-trisphosphate acts in vivo as a specific regulator of cellular signaling by inositol3,4,5,6-tetrakisphosphate. J. Biol. Chem. 274, 18973–18980

107 Berridge, M. J. (1983) Rapid accumulation of inositol trisphosphate reveals thatagonists hydrolyse polyphosphoinositides instead of phosphatidylinositol. Biochem. J.212, 849–858

108 Hughes, P. J., Kirk, C. J. and Michell, R. H. (1994) Inhibition of porcine brain inositol1,3,4-trisphosphate kinase by inositol polyphosphates, other polyol phosphates,polyanions and polycations. Biochim. Biophys. Acta 1223, 57–70

109 Adelt, S., Plettenburg, O., Dallmann, G., Ritter, F. P., Shears, S. B., Altenbach, H. J. andVogel, G. (2001) Regiospecific phosphohydrolases from Dictyostelium as tools for thechemoenzymatic synthesis of the enantiomers D-myo-inositol 1,2,4-trisphosphate andD-myo-inositol 2,3,6-trisphosphate: non-physiological, potential analogues ofbiologically active D-myo-inositol 1,3,4-trisphosphate. Bioorg. Med. Chem. Lett. 11,2705–2708


280 S. B. Shears

110 Cheek, S., Zhang, H. and Grishin, N. V. (2002) Sequence and structure classification ofkinases. J. Mol. Biol. 320, 855–881

111 Sun, Y., Wilson, M. P. and Majerus, P. W. (2002) Inositol 1,3,4-trisphosphate 5/6-kinaseassociates with the COP9 signalosome by binding to CSN1. J. Biol. Chem. 277,45759–45764

112 Wilson, M. P., Sun, Y., Cao, L. and Majerus, P. W. (2001) Inositol 1,3,4-trisphosphate5/6-kinase is a protein kinase that phosphorylates the transcription factors c-jun andATF-2. J. Biol. Chem. 276, 40998–41004

113 Sun, Y., Mochizuki, Y. and Majerus, P. W. (2003) Inositol 1,3,4-trisphosphate 5/6-kinaseinhibits tumor necrosis factor-induced apoptosis. J. Biol. Chem. 278, 43645–43653

114 Hunter, T. (1995) When is a lipid kinase not a lipid kinase? When it is a protein kinase.Cell 83, 1–4

115 Cope, G. A. and Deshaies, R. J. (2003) COP9 signalosome: a multifunctional regulatorof SCF and other cullin-based ubiquitin ligases. Cell 114, 663–671

116 Schwechheimer, C. and Deng, X.-W. (2001) COP9 signalosome revisited: a novelmediator of protein degradation. Trends Cell Biol. 11, 420–426

117 Uhle, S., Medalia, O., Waldron, R., Dumdey, R., Henklein, P., Bech-Otschir, D., Huang,X., Berse, M., Sperling, J., Schade, R. and Dubiel, W. (2003) Protein kinase CK2 andprotein kinase D are associated with the COP9 signalosome. EMBO J. 22, 1302–1312

118 Bech-Otschir, D., Kraft, R., Huang, X., Henklein, P., Kapelari, B., Pollmann, C. andDubiel, W. (2001) COP9 signalosome-specific phosphorylation targets p53 todegradation by the ubiquitin system. EMBO J. 20, 1630–1639

119 Balla, T., Baukal, A. J., Hunyady, L. and Catt, K. J. (1989) Agonist-induced regulation ofinositol tetrakisphosphate isomers and inositol pentakisphosphate in adrenalglomerulosa cells. J. Biol. Chem. 264, 13605–13611

120 Menniti, F. S., Miller, R. N., Putney, Jr, J. W. and Shears, S. B. (1993) Turnover of inositolpolyphosphate pyrophosphates in pancreatoma cells. J. Biol. Chem. 268, 3850–3856

121 Stephens, L. R., Radenberg, T., Thiel, U., Vogel, G., Khoo, K. H., Dell, A., Jackson, T. R.,Hawkins, P. T. and Mayr, G. W. (1993) The detection, purification, structuralcharacterization and metabolism of diphosphoinositol pentakisphosphate(s) andbisdiphosphoinositol tetrakisphosphate(s). J. Biol. Chem. 268, 4009–4015

122 Saiardi, A., Nagata, E., Luo, H. R., Snowman, A. M. and Snyder, S. H. (2001)Identification and characterization of a novel inositol hexakisphosphate kinase.J. Biol. Chem. 276, 39179–39185

123 Schell, M. J., Letcher, A. J., Brearley, C. A., Biber, J., Murer, H. and Irvine, R. F. (1999)PiUS (Pi uptake stimulator) is an inositol hexakisphosphate kinase. FEBS Lett. 461,169–172

124 Luo, H. R., Huang, Y. E., Chen, J. C., Saiardi, A., Iijima, M., Ye, K., Huang, Y., Nagata, E.,Devreotes, P. and Snyder, S. H. (2003) Inositol pyrophosphates mediate chemotaxis inDictyostelium via pleckstrin homology domain-PtdIns(3,4,5)P3 interactions. Cell 114,559–572

125 Luo, H. R., Saiardi, A., Yu, H., Nagata, E., Ye, K. and Snyder, S. H. (2002) Inositolpyrophosphates are required for DNA hyperrecombination in protein kinase C1 mutantyeast. Biochemistry 41, 2509–2515

126 Safrany, S. T., Ingram, S. W., Cartwright, J. L., Falck, J. R., McLennan, A. G., Barnes,L. D. and Shears, S. B. (1999) The diadenosine hexaphosphate hydrolases fromSchizosaccharomyces pombe and Saccharomyces cerevisiae are homologues of thehuman diphosphoinositol polyphosphate phosphohydrolase, Overlapping substratespecificities in a MutT motif. J. Biol. Chem. 274, 21735–21740

127 Shears, S. B., Ali, N., Craxton, A. and Bembenek, M. E. (1995) Synthesis andmetabolism of bis-diphosphoinositol tetrakisphosphate in vitro and in vivo.J. Biol. Chem. 270, 10489–10497

128 Huang, C.-F., Voglmaier, S. M., Bembenek, M. E., Saiardi, A. and Snyder, S. H. (1998)Identification and purification of diphosphoinositol pentakisphosphate kinase, whichsynthesizes the inositol pyrophosphate bis(diphospho)inositol tetrakisphosphate.Biochemistry 37, 14998–15004

129 Vinson, C., Myakishev, M., Acharya, A., Mir, A. A., Moll, J. R. and Bonovich, M. (2002)Classification of human B-ZIP proteins based on dimerization proerties. Mol. Cell. Biol.22, 6321–6335

130 Laussmann, T., Pikzack, C., Thiel, U., Mayr, G. W. and Vogel, G. (2000) Diphospho-myo-inositol phosphates during the life cycle of Dictyostelium and Polysphondylium.Eur. J. Biochem. 267, 2447–2451

131 Glennon, M. C. and Shears, S. B. (1993) Turnover of inositol pentakisphosphates,inositol hexakisphosphate and diphosphoinositol polyphosphates in primary culturedhepatocytes. Biochem. J. 293, 583–590

132 Abel, K., Anderson, R. A. and Shears, S. B. (2002) Phosphatidylinositol and inositolphosphate metabolism. J. Cell Sci. 114, 2207–2208

133 Safrany, S. T., Caffrey, J. J., Yang, X., Bembenek, M. E., Moyer, M. B., Burkhart, W. A. andShears, S. B. (1998) A novel context for the ‘MutT’ module, a guardian of cell integrity, ina diphosphoinositol polyphosphate phosphohydrolase. EMBO J. 17, 6599–6607

134 Safrany, S. T. and Shears, S. B. (1998) Turnover of bis-diphosphoinositoltetrakisphosphate in a smooth muscle cell line is regulated by β2-adrenergic receptorsthrough a cAMP-mediated, A-kinase-independent mechanism. EMBO J. 17, 1710–1716

135 Voglmaier, S. M., Bembenek, M. E., Kaplin, A. I., Dorman, G., Olszewski, J. D.,Prestwich, G. D. and Snyder, S. H. (1996) Purified inositol hexakisphosphate kinase isan ATP synthase: diphosphoinositol pentakisphosphate as a high-energy phosphatedonor. Proc. Natl. Acad. Sci. U.S.A. 93, 4305–4310

136 Shears, S. B. (1996) Inositol pentakis- and hexakisphosphate metabolism addsversatility to the actions of inositol polyphosphates. Novel efects on ion channels andprotein traffic. In myo-Inositol Phosphates, Phosphoinositides and Signal Transduction(Biswas, B. B. and Biswas, S., eds.), pp. 187–226, Plenum Press, New York andLondon

137 Caffrey, J. J., Safrany, S. T., Yang, X. and Shears, S. B. (2000) Discovery of molecularand catalytic diversity among human diphosphoinositol polyphosphatephosphohydrolases: an expanding NUDT family. J. Biol. Chem. 275, 12730–12736

138 Laussmann, T., Hansen, A., Reddy, K. M., Reddy, K. K., Falck, J. R. and Vogel, G. (1998)Diphospho-myo-inositol phosphates in Dictyostelium and Polysphondylium:identification of a new bisdiphospho-myo-inositol tetrakisphosphate. FEBS Lett. 426,145–150

139 Laussmann, T., Reddy, K. M., Reddy, K. K., Falck, J. R. and Vogel, G. (1997)Diphopho-myo-inositol phosphates from Dictyostelium identified as D-6-diphospho-myo-inositol pentakisphosphate and D-5,6-bisdiphospho-myo-inositoltetrakisphosphate. Biochem. J. 322, 31–33

140 Laussmann, T., Eujen, R., Weisshuhn, C. M., Thiel, U., Falck, J. R. and Vogel, G. (1996)Structures of diphospho-myo-inositol pentakisphosphate and bisdiphospho-myo-inositol tetrakisphosphate from Dictyostelium by NMR analysis. Biochem. J. 315,715–725

141 Morrison, B. H., Bauer, J. A., Kalvakolanu, D. V. and Lindner, D. J. (2001) Inositolhexakisphosphate kinase 2 mediates growth suppressive and apoptotic effects ofinterferon-β in ovarian carcinoma cells. J. Biol. Chem. 276, 24965–24970

142 Safrany, S. T., Caffrey, J. J., Yang, X. and Shears, S. B. (1999) Diphosphoinositolpolyphosphates: the final frontier for inositide research? Biol. Chem. 380, 945–951

143 Shears, S. B. (2001) Assessing the omnipotence of inositol hexakisphosphate.Cell. Signalling 13, 151–158

144 Yamaguchi, Y., Ikenaka, K., Niinobe, M., Yamada, H. and Mikoshiba, K. (1996) Myelinproteolipid protein (PLP), but not DM-20, is an inositol hexakisphosphate-bindingprotein. J. Biol. Chem. 271, 27838–27846

145 Ali, N., Duden, R., Bembenek, M. E. and Shears, S. B. (1995) The interaction ofcoatomer with inositol polyphosphates is conserved in Saccharomyces cerevisiae.Biochem. J. 310, 279–284

146 Fleischer, B., Xie, J., Mayrleitner, M., Shears, S. B., Palmer, D. J. and Fleischer S. (1994)Golgi coatomer binds, and forms K+-selective channels gated by inositolpolyphosphates. J. Biol. Chem. 269, 17826–17832

147 Ye, W., Ali, N., Bembenek, M. E., Shears, S. B. and Lafer, E. M. (1995) Inhibition ofclathrin assembly by high-affinity binding of specific inositol polyphosphates to thesynapse-specific clathrin assembly protein AP-3. J. Biol. Chem. 270, 1564–1568

148 Lin, R. C. and Scheller, R. H. (2000) Mechanisms of synaptic vesicle exocytosis.Annu. Rev. Cell Dev. Biol. 16, 19–49

149 Luo, H. R., Saiardi, A., Nagata, E., Ye, K., Yu, H., Jung, T. S., Luo, X., Jain, S., Sawa, A.and Snyder, S. H. (2001) GRAB: A physiologic guanine nucleotide exchange factor forRab3A, which interacts with inositol hexakisphosphate kinase. Neuron 31, 439–451

150 Barrientos, L. G. and Murthy, P. P. N. (1996) Conformational studies of myo-inositolphosphates. Carbohydr. Res. 296, 39–54

151 Flores, S. and Smart, C. C. (2001) Abscisic acid-induced changes in inositolmetabolism in Spirodela polyrrhiza. Planta 211, 823–832

152 Lemtiri-Chlieh, F., MacRobbie, E. A. C. and Brearley, C. A. (2000) Inositolhexakisphosphate is a physiological signal regulating the K+ inward rectifyingconductance in guard cells. Proc. Natl. Acad. Sci. U.S.A. 97, 8687–8692

153 Dzikowska, A., Kacprzak, M., Tomecki, R., Koper, M., Scazzocchio, C. and Weglenski, P.(2003) Specific induction and carbon/nitrogen repression of arginine catabolism geneof Aspergillus nidulans – functional in vivo analysis of the otaA promoter.Fungal Genet. Biol. 38, 175–186

154 Albert, C., Safrany, S. T., Bembenek, M. E., Reddy, K. M., Reddy, K., Falck, J., Brocker,M., Shears, S. B. and Mayr, G. W. (1997) Biological variability in the structures ofdiphosphoinositol polyphosphates in Dictyostelium discoideum and mammalian cells.Biochem. J. 327, 553–560

Received 18 September 2003/14 October 2003; accepted 20 October 2003Published as BJ Immediate Publication 20 October 2003, DOI 10.1042/BJ20031428


how versatile are inositol phosphate kinases?

Documents