Eukaryotic cells have ubiquitously utilized the myo-inositol backbone to generate a diverse array of signalling molecules. This is achieved by arranging phosphate groups around the six-carbon inositol ring. There is virtually no biological process that does not take advantage of the uniquely variable architecture of phosphorylated inositol. In inositol biology, phosphates are able to form three distinct covalent bonds: phosphoester, phosphodiester and phosphoanhydride bonds, with each providing different properties. The phosphoester bond links phosphate groups to the inositol ring, the variable arrangement of which forms the basis of the signalling capacity of the inositol phosphates. Phosphate groups can also form the structural bridge between myo-inositol and diacylglycerol through the phosphodiester bond. The resulting lipid-bound inositol phosphates, or phosphoinositides, further expand the signalling potential of this family of molecules. Finally, inositol is also notable for its ability to host more phosphates than it has carbons. These unusual organic molecules are commonly referred to as the inositol pyrophosphates (PP-IPs), due to the presence of high-energy phosphoanhydride bonds (pyro- or diphospho-). PP-IPs themselves constitute a varied family of molecules with one or more pyrophosphate moiety/ies located around the inositol. Considering the relationship between phosphate and inositol, it is no surprise that members of the inositol phosphate family also regulate cellular phosphate homoeostasis. Notably, the PP-IPs play a fundamental role in controlling the metabolism of the ancient polymeric form of phosphate, inorganic polyphosphate (polyP). Here we explore the intimate links between phosphate, inositol phosphates and polyP, speculating on the evolution of these relationships.
The fine organization of a biological system is constantly challenged by the always-changing environment. In order to monitor, manage and react to these changes, cells have developed an intricate network of signalling pathways. The covalent attachment of phosphate groups (PO3−) to the inositol ring provides a vast artillery of molecules to be employed by these signalling processes.
Whereas Ins(1,4,5)P3 induced Ca++ release is by far the best recognized paradigm of inositol phosphate signalling, in fact it represents just one of the many signalling pathways regulated by inositol phosphates. The surprise discovery of inositol derivatives possessing more phosphates than the six carbons of the inositol ring, the so-called inositol pyrophosphates (PP-IPs), gives further complexity to an already intricate network. Undeniably, the association between inositol and phosphate group has been heavily exploited by evolution. However, inositol is not unique in its exploitation of phosphate; this group plays a crucial role in the structure of macromolecules, information transfer and the storage of energy. Since the phosphate moiety plays such a fundamental role in cellular biology, its bioavailability must be strictly controlled. Intriguingly, a number of inositol phosphates have been implicated in this process. PP-IPs may be the main player of this regulation, since in yeast they regulate the metabolism of the linear polymer of phosphate groups, inorganic polyphosphate (polyP). The objective of the present review is not to extensively evaluate the literature on inositol phosphates, nor polyP; for this, we suggest the recent excellent reviews [1–4]. Instead, we aim to highlight the complex interactions between phosphate and inositol, speculating on the ability of PP-IPs to regulate, phosphate, polyP and signalling.
In life, the essential element phosphorus exists in its ionic phosphate form, either free, or as Pi (or PO3−) or, more commonly, associated with other compounds. Two main characteristics have allowed phosphate to play a primary role in biology. First, phosphate at physiological pH carries two negative charges. Second, the 3D geometry of phosphate allows it to co-ordinate hydrogen bonds through its oxygen groups. The importance of phosphate in biology is demonstrated by the variety of functions it plays from structural to signalling. The phosphodiester bond plays a crucial structural role in the charged backbone of the nucleic acids RNA and DNA. Attachment of a phosphate group to serine, threonine or tyrosine represents one of the principal regulatory mechanisms in biology . However, it is the high energy stored in the phosphoanhydride bond that gives phosphate its most vital role. These so-called ‘high-energy’ phosphates release their stored energy through the hydrolysis of the phosphoanhydride bond. In modern cells, the hydrolysis of the phosphoanhydride bonds, usually in the form of ATP, drives virtually all biochemical reactions in living organisms. Life as we know it depends on phosphate .
In fulfilling these varied roles, phosphate moieties circulate between organic molecules, generally being passed between their organic partners via phosphotransfer reactions involving a nucleotide. Ultimately, this cascade of events generates free phosphate, which mitochondrial ATPases subsequently utilize to regenerate ATP, thus completing the cycle. Given the fundamental roles played by phosphate, the cellular abundance of Pi must be tightly regulated. Whereas a specific transporter regulates the cellular entry of phosphate , its concentration in the cell also depends upon the metabolic turnover of the nucleotides pool. However, the importance of fine-tuning this balance necessitates a further buffering system. It is the polymeric form of phosphate, polyP (Figure 1), that plays this fundamental role in controlling cellular phosphate homoeostasis. Thus, the balance between polyP polymerization and depolymerization is policed by the cellular need for free phosphate. The polymer polyP, in complex with counterions, neutralizes the negative osmotic effect that high levels of free Pi would cause to the cell. In addition, polyP, a chelator of metal ions, also functions as a buffer for cations, regulating their cellular availability.
polyP structure and endo- and exo-polyphosphatase enzymatic site of action
PolyP is the simplest biological polymer and is ubiquitously distributed across all kingdoms . Its synthesis from ATP has been elucidated in bacteria and some lower eukaryotes [9,10]; however, polyP synthesis remains uncharacterized in both archaea and higher eukaryotes. Two classes of enzymes control the catabolism of polyP: the endo-polyphosphatases that cleave internal phosphoanhydride bonds, producing shorter polyP chains; and the exo-polyphosphatases, which remove the terminal phosphate releasing Pi (Figure 1). The best-characterized exo-polyphosphatases are the yeast protein Ppx1  and the human homologue h-Prune . Two polyP endo-polyphosphatases have been characterized, Ppn1 and Ddp1, plus its mammalian homologues, the diphosphoinositol polyphosphate phosphohydrolases (DIPPs) . In fact, Ddp1 and DIPPs are also able to metabolize several diverse polyphosphate substrates , including the PP-IPs, of which we will hear more later. Aside from simply buffering the level of free phosphate, a number of specific signalling properties have been attributed to polyP. It regulates bacterial stress response and survival , controls pathogen virulence , certain cancers  and even blood coagulation . Despite this array of signalling properties, little is known about how the polyP is regulated, particular in mammalian cells; thus, the recent discovery that PP-IPs might regulate polyP metabolism is of particular interest .
The PP-IPs are the most-phosphate-rich members of a large family of signalling molecules derived from the sequential phosphorylation of the six-carbon sugar, inositol. Not only are the PP-IPs and polyP catabolized by the same enzymes, but PP-IPs are also able to regulate the levels of polyP in yeast , as well as in human platelets . These observations indicate a metabolic coupling between PP-IPs and polyP.
Inositol, the building block of the PP-IPs has an intimate relationship with phosphate, demonstrated by the wide-ranging biological functions of inositol phosphates. The direct biosynthetic route of inositol from glucose is conserved between archaea and humans . A single, irreversible isomerization reaction converts glucose into the much more stable form of inositol (Figure 2). This provides a metabolically inert and versatile sugar, the ideal canvas to be decorated with phosphates. The stability of inositol also makes this sugar an ideal osmolyte. In fact, in archaea, this is the primary function of inositol, present in its di-inositol phosphate form, allowing archaea to adapt to extreme environments .
Enzymatic conversion of glucose-6P to inositol
Although eukaryotic cells also use inositol in this way , they have exploited the stability of inositol to generate a multifaceted array of phosphorylated signalling molecules. The six-carbon ring of inositol can be considered as a 6-bit code with the potential to ‘encode’ 64 unique phosphorylation molecules . The majority of these phosphorylated molecules, commonly referred to as inositol phosphates, have been identified in the eukaryotic cell. The assortment of phosphorylated inositol molecules is further enhanced by the presence of phosphorylated inositol lipids commonly referred to as phosphoinositides . Higher complexity still is derived from the ability of carbon atoms in the inositol ring to host more than one phosphate, generating the so-called PP-IPs [4,23]. Thus, the combinatorial arrangements of up to nine (or possibly even more) [24,25] phosphate groups around the inositol ring generate a very large collection of molecules. Representative members of each subfamily of these molecules are depicted in Figure 3.
Illustrating representative members of three subfamilies of inositol phosphate-derived molecules
Although molecules in this family share some common properties, each member possesses its own unique hallmarks. The negative charge of the phosphate group at physiological pH allows several protein domains to specifically recognize the molecular patterns of the individual inositol phosphates. Protein recognition of the unique stereochemical arrangement of phosphate groups on the inositol ring by PH, PX, FYVE and PHD domains represents the accepted mechanism of action of phosphoinositides [26,27] and also of some inositol phosphates. In fact, the concept of protein recognition of phosphorylated inositol species has influenced much of the research on inositol signalling.
To put this into context, we can consider that myo-inositol was initially purified from muscle, hence the name, in the middle of the nineteenth century. Its fully phosphorylated form, phytic acid (IP6 or inositol hexakisphosphate), was then purified from plant seeds in 1919 . However, these molecules did not garner significant attention until the demonstration that the receptor-dependent release of Ins(1,4,5)P3 mobilizes calcium (Ca++) from intracellular stores . The intense research that followed has revealed the fundamental importance of phosphoinositides in cell biology, elucidating their role in defining membrane identity  and the importance of PtdIns(3,4,5)P3 in controlling signalling pathways regulating growth and differentiation. Unfortunately, the progress in understanding the role of phosphoinositides has not been matched for their molecular cousins, the inositol phosphates. In part, this is due to the higher complexity of possible phosphorylation patterns and the lack of experimental tools to easily study them. However, this gap in our knowledge can perhaps also be attributed to the historical emphasis on action through receptors.
Phosphate and inositol relationship
Whereas the discovery of the IP3-mediated calcium release exponentially stimulated research in this field, the evolutionary conservation of inositol phosphate signalling suggests that an alternative, more ancient, mechanism of action may be at play. Clues about this more ancient mechanism can be derived from molecular evolution studies. A proven, genuine inositol trisphosphate receptor (IP3R) is not present in the majority of protozoa genomes and is entirely absent from other kingdoms of life such as Plantae . Furthermore, the occurrence in the genome of metazoan of IP3-3K enzymes, which remove the Ins(1,4,5)P3 second messenger, thereby ‘switching off’ Ca++ signalling, is coincident with the presence of a Ins(1,4,5)P3 regulated calcium signalling through the IP3R . These observations indicate that Ins(1,4,5)P3-mediated Ca++ signalling is a relatively recent addition to the repertoire of inositol phosphate functions in metazoans. Therefore, analysing the evolutionary history of inositol phosphate kinases might give further insight into how the large family of inositol phosphates evolved and what are the most basic aspects of their function.
Surprisingly, until now, four unrelated protein-folding domains have been shown to possess inositol phosphate kinase activity [32–35]. Three of those, ITPK1, IP5-2K (IP5-2Kinase; also known as IPPK) and PP-IP5K (Vip1-like kinase), possess a specific enzymatic activity limited to one position on the inositol ring. However, the fourth inositol kinase fold, characterized by the PxxxDxKxG motif (Pfam PF03770), makes up a gene family comprising three diverse enzymes. This gene family includes the inositol hexakisphosphate kinases (IP6Ks), the inositol polyphosphate multikinase (IPMK) and the IP3-3K mentioned above.
Although systematic phylogenetic distribution studies of all four types of kinases have not been performed, it is recognized that the PP-IPs synthesizing enzymes IP6K and PP-IP5K are evolutionary conserved [36,37]. Phylogenetic analysis of the enzymes possessing the PxxxDxKxG motif suggests that this gene family arose from a primitive IP6K ancestor . This unconventional view is supported by the recently resolved structure of Entamoeba histolytica IP6K . This hybrid kinase, besides synthesizing PP-IPs (InsP7 or PP-IP5), can also metabolize Ins(1,4,5)P3 to two diverse InsP4s . The substrate promiscuity of this ancient IP6K can be explained by a partial separation of the substrate-binding sites for Ins(1,4,5)P3 and InsP6. Loss of the InsP6-binding determinants would generate an IPMK enzyme. It seems likely that the evolution of InsP3 as a key Ca++ second messenger then drove the transformation of this IPMK into a highly specialized IP3-3K [35,38].
These considerations may suggest that PP-IPs represent the most ancient members of the inositol phosphate family of molecules. This view seems at odds with the fact that synthesis of InsP7 (diphosphoinositol pentakisphosphate, PP-IP5) (Figure 3), the most abundant and easily detectable PP-IPs, first requires the synthesis of precursor of inositol phosphates up to InsP6. However, the elegant demonstration that pyrophosphate derivatives of InsP3 and InsP4 can exist in today's cells  is supportive of the possibility that a primitive inositol pyrophosphate could be derived from less phosphorylated inositol phosphates. If the hydrolysis of the high-energy phosphoanhydride bond gives PP-IPs their unique mode of action, then these ancient forms could be more than adequate evolutionary precursors to today's InsP6-derived PP-IPs. Nevertheless, the steric hindrance of a fully phosphorylated inositol ring does increase the energy stored in the phosphoanhydride, which may explain why evolution has favoured the InsP6-derived PP-IPs that are most commonly detected in modern cells.
Many roles have been assigned to PP-IPs and several recent reviews have been written, describing their functions [4,23,40]. What has emerged from the last 20 years or so of research is that the PP-IPs regulate a highly disparate array of cellular functions, from vesicular trafficking  to epigenetic mechanisms [42,43]. Meanwhile, phenotypic analysis of the three IP6K mice knockouts revealed the importance of PP-IPs in controlling obesity , cancer  and motor co-ordination . This variety of roles indicates that PP-IPs regulate a very basic function, a finding that is consistent with the hypothesis that these molecules are part of an evolutionarily ancient signalling mechanism. Thus, it has been proposed that PP-IPs are ‘metabolic messengers’ , able to regulate cellular homoeostasis. The ability of PP-IPs to control the cellular level of ATP  strengthens this hypothesis. Since ATP is the central molecule of intermediary metabolism, it is easy to understand how the ability of PP-IPs to alter cellular levels of ATP would be manifested in such diverse processes.
Furthermore, the previously noted ability of PP-IPs to regulate polyP levels offers another intriguing link to energetic metabolism and phosphate homoeostasis. As stated above, cellular Pi homoeostasis, buffered by polyP, is key to cellular wellbeing, since phosphate is essential to ATP synthesis. In fact, evidence linking PP-IPs to cellular phosphate homoeostasis predates the cloning of the IP6Ks. During a search for a mammalian intestinal phosphate transporter, the gene PiUS (Pi Uptake Stimulator) was identified . PiUS does not encode a putative transmembrane domain and thus is not a canonical phosphate transporter. Nevertheless, injection of PiUS cDNA into the Xenopus oocyte stimulates the uptake of radiolabelled phosphate from the medium. Hence, it was called Pi Uptake Stimulator. We now know that this gene corresponds to the mammalian IP6K2 gene [37,49]. Surprisingly, there has been no further attempt to examine how PiUS/IP6K2 stimulates the uptake of phosphate. However, if we exclude non-enzymatic roles for IP6K2 (which, given the original observations, were made by injecting mammalian cDNA into Xenopus oocytes seems a reasonable assumption), then the increase of phosphate entry must depend upon the synthesis of InsP7 as synthesized by the exogenous IP6K2.
It is not only at the cellular level that IP6Ks seem to play a key role in controlling phosphate homoeostasis, but also at the level of the whole organism. In fact, IP6K3 is one of only seven genes identified by a genome-wide association study to be important to regulate the phosphate level in human blood serum .
The exact molecular mechanism/s behind the ability of PP-IPs to regulate phosphate homoeostasis remains unknown. It is possible that it may be through either direct or indirect control of polyP metabolism [2,51]. It is intriguing, however, to speculate on the possibility that one of the initial functions of inositol phosphate signalling, perhaps carried out by primitive PP-IPs species, was to regulate phosphate homoeostasis. The evolution of the link between phosphate and inositol appears to coincide closely with the origin of eukaryotic life forms. In a further evolutionary coincidence, the formation of calcium-rich organelles, and ultimately the utilization of Ca++ as a signalling molecule, was concurrent with the emergence of eukaryotic cells. Prior to this development this ion was primarily excluded from the cell . The regulated extrusion of Ca++ from prokaryote cells, as well as the eukaryotic cytosol, is a fundamental property of life. Calcium, in the presence of both organic and inorganic phosphates, at neutral or basic pH typical for the eukaryote cytosol, has a propensity to form calcium-phosphate precipitates, something not compatible with biochemical processes [53,54]. Thus, it is captivating to consider that although the Ins(1,4,5)P3/Ca++ signalling paradigm is a recent evolutionary development, the most ancient calcium signalling may stem from the ability of a primitive PP-IPs to modulate phosphate homoeostasis in an early eukaryotic cell. The evolution of primitive PP-IPs able to regulate phosphate concentration may have permitted the transient presence of Ca++ in the cytosol as it is less likely to result in precipitation with free phosphate. This relaxation of Ca++ exclusion could have opened the door for the appropriation of Ca++ as a signalling molecule. This hypothesis, supported by the coincident evolution of these cellular properties, provides the tantalizing possibility that inositol phosphate biology and calcium signalling are in fact uniquely coupled. Evolving in parallel, this coupling has provided us with the famed Ins(1,4,5)P3/Ca++ we see in today's metazoan cells.
It took 40 years of painstaking research to outline the now classical Ins(1,4,5)P3/Ca++ signalling paradigm . From there the number of known inositol phosphates has grown along with the appreciation that they are regulating virtually every aspect of cellular physiology. The PP-IPs, with their peculiar structure, add further complexity and have driven further attention to this signalling network. Many aspects of inositol pyrophosphate modus operandi remain to be fully understood. In the present review, we have explored their links to ‘phosphate’.
Our proposed origin of inositol phosphate signalling may appear speculative, but we believe that to properly appreciate and understand the elaborate signalling network generated by inositol and phosphate, we should not only look from our own human cell's perspective. Rather, we should remember that much can be learnt by looking from below. Understanding the evolutionary origin of the combinatorial attachment of phosphate to the inositol ring will surely help shed light on the selective pressures that have produced such an intricate metabolic network of signalling molecules. After all, we must not forget the maxim that ‘nothing in biology makes sense except in the light of evolution’ .
This work was supported by the Medical Research Council (MRC) core support to the MRC/UCL Laboratory for Molecular Cell Biology University Unit [grant number MC_UU_1201814].
diphosphoinositol polyphosphate phosphohydrolases
Fab1, YOTB, Vac 1 and EEA1 homology
inositol trisphosphate receptor
inositol hexakisphosphate kinase
inositol polyphosphate multikinase
Pi Uptake Stimulator
Signalling 2015: Cellular Functions of Phosphoinositides and Inositol Phosphates: Held at Robinson College, University of Cambridge, Cambridge, U.K., 1–4 September 2015.