Inositol pyrophosphates as multifaceted metabolites in the regulation of mammalian signaling networks

ABSTRACT Inositol pyrophosphates (PP-IPs) such as 5-diphosphoinositol pentakisphosphate (5-IP7) are inositol metabolites with high-energy phosphoanhydride bonds. The formation of PP-IPs is catalyzed by two groups of enzymes, the IP6 kinases and the PPIP5 kinases, which both phosphorylate inositol hexakisphosphate (IP6). In mammals, PP-IPs are implicated in diverse biological phenomena including cellular growth, vesicular trafficking, apoptosis, and metabolic homeostasis. Mechanistically, all the diverse actions of PP-IPs proceed in one of two ways: the PP-IPs modulate the activity of their target proteins either through allosteric binding or protein pyrophosphorylation. In this review, we highlight recent advances in our understanding of the pleiotropic functions of the mammalian PP-IPs and the metabolic enzymes that produce them. We also discuss some future challenges in the exploration of areas where PP-IPs play important but unknown roles in physiology and disease.


Introduction
Since the early 1990s, more than 30 different inositol pyrophosphates (PP-IPs) have been identified as phosphorylated derivatives of myo-inositol, which is itself an isomer of cyclohexanehexol. Many of these inositol pyrophosphates play regulatory roles in controlling a wide range of biological processes such as telomere length control, vesicular trafficking, chemotaxis, and insulin signaling (Tsui and York 2010;Chakraborty et al. 2011;Wilson et al. 2013;Shah et al. 2017). The most wellstudied PP-IPs in mammals are diphosphoinositol pentakisphosphate (PP-IP 5 or IP 7 ) and bis-diphosphoinositol tetrakisphosphate, ([PP] 2 -IP 4 or IP 8 ). These small, watersoluble molecules are produced by two classes of inositol kinases, the IP6 kinases (IP6Ks) and the PPIP5 kinases (PPIP5Ks), which add high-energy phosphate groups to inositol hexakisphosphate (IP 6 ). The steady-state cellular concentration of IP 7 in mammals varies from 0.5 to 5 μM (Stephens et al. 1993;Barker et al. 2004;Lin et al. 2009), and the level of IP 8 is typically less than half of the concentration of IP 7 (Glennon and Shears 1993;Albert et al. 1997). Due to their structural stability, high turnover rate, and high-energy phosphate bonds, PP-IPs have been referred to variously as 'metabolic messengers' (Shears 2009), 'regulators of cell homeostasis' (Wundenberg and Mayr 2012), and 'mammalian cell signals' (Chakraborty et al. 2011).

Dynamic turnover of PP-IPs
According to Shears and colleagues, 50% of the steadystate IP 6 pool and 20% of the steady-state IP 5 pool in mammalian cells are converted to PP-IPs every hour (Menniti et al. 1993). PP-IPs are also turned over rapidly. In primary hepatocytes, the IP 7 pool turns over every 4 min while the IP 6 turns over every 40 min (Glennon and Shears 1993). This presumably allows the levels of the PP-IPs to be tightly regulated in response to various cellular conditions. For example, ovarian cancer cells show a 4-fold increase in IP 7 upon treatment with staurosporine. In addition, several tumor cell lines show cell cycle-dependent modulation of IP 7 levels, and hepatocytes show a 2-fold elevation in IP 7 upon insulin stimulation. In mouse embryonic fibroblasts, serum starvation depletes IP 7 levels and hyperosmotic and thermal stress induce IP 8 levels (Pesesse et al. 2004). Thus, the regulation of the dynamic turnover of PP-IPs that is associated with various physiological stimuli suggests the PP-IPs play important roles in cellular and metabolic signaling.

The modes of action of the inositol pyrophosphates
Two molecular mechanismsallosteric non-covalent binding and covalent protein pyrophosphorylationare thought to account for all of the diverse roles the inositol pyrophosphates play in modulating protein function and directing sequential biological events ( Figure 2). The direct binding of an inositol phosphate to a specific protein effector was originally demonstrated in the discovery that the IP3-gated Ca2+ release channel acts as an IP3 receptor (Mikoshiba et al. 1993). Both 1-IP7 and 5-IP7 were found to modulate their targets through allosteric interactions (Wilson et al. 2013). As the inositol pyrophosphates cycle between binding and hydrolysis, their charges may also interfere with other protein-metabolite interactions (Shears 2015). For example, 5-IP7 interacts with proteins containing pleckstrin homology (PH) domains, and PH domains are known to interact primarily with PI(3,4,5)P3. Akt kinase is the most well-studied 5-IP7binding protein. Akt kinase is an important regulator of cellular growth and metabolism that is acutely activated by insulin and growth factors (Manning and Cantley 2007). To fully activate Akt signaling cascade, PI(3,4,5)P3 binds and recruits Akt to the plasma membrane. 5-IP7 inhibits Akt by competing with PI(3,4,5)P3 for binding to Akt's PH domain at an IC50 of 1 μM (Bhandari et al. 2008;Chakraborty et al. 2010). Using a photocage-based method, 5-IP7 release was recently found to suppress Figure 1. Inositol pyrophosphate biosynthetic pathway. This diagram shows the metabolic pathways of inositol pyrophosphate synthesis in mammalian cells. IP3 3-kinases or IPMK phosphorylates IP 3 to generate IP 4 [Ins(1,3,4,5)P 4 or Ins(1,4,5,6)P 4 ]. Subsequent phosphorylation of IP 4 and IP 5 ultimately yields IP 6 . IP6Ks pyrophosphorylate the 5 position, generating 5-PP-IP 4 from IP 5 and 5-PP-IP 5 (5-IP 7 ) from IP 6 . PPIP5Ks act upon the 1 or 3 position to produce 1-/3-PP-IP 5 (shown as 1-IP 7 ) from IP 6 . the translocation of Akt from the plasma membrane to the cytoplasm . The PH domains of stressactivated map kinase-interacting protein 1 (SIN1) and general receptor for phosphoinositides 1 (GRP1) also bind 5-IP7 with higher affinity than 1-IP7 and IP8 (Gokhale et al. 2013). In addition, inositol pyrophosphates have been found to interact with protein partners that lack PH domains. In response to DNA damage, 5-IP7 produced by IP6K2 binds to the TTT complex and modulates its binding to casein kinase 2 (CK2) (Rao et al. 2014). This stabilizes both the DNA-dependent protein kinase catalytic subunit and ATM kinase. This, in turn, increases p53 phosphorylation and activates apoptosis (Rao et al. 2014). 5-IP7 also affects synaptic vesicle fusion during exocytosis through its direct interaction with synaptotagmin1 (Syt1). Syt1 is an essential calcium sensor in presynaptic termini that triggers the membrane fusion events required for neurotransmitter release (Rizo and Rosenmund 2008). Single-molecule imaging studies using reconstituted synaptic vesicles revealed that 5-IP7 is 10fold more efficient in suppressing vesicle fusion than 1-IP7 or other inositol phosphates. Mechanistically, the dynamic, high-affinity binding of 5-IP7 to the polybasic region of Syt1 appears to hold Syt1 in a nonfunctional configuration ). In budding yeast, 1-IP7 but not 5-IP7 interacts selectively with Pho80-Pho85-Pho81 in the regulation of phosphate homeostasis (Lee et al. 2007). No 1-IP7-specific protein targets in mammals have been discovered yet.
Inositol pyrophosphates also modulate protein function via the pyrophosphorylation of their target protein (Saiardi et al. 2004;Bhandari et al. 2007). Initially, the inositol pyrophosphates were thought to act as high-energy phosphate donors like ATP because of the energy in their phosphoanhydride bonds that is released along with their β-phosphates (Wilson et al. 2013). Indeed, the inositol pyrophosphates can transfer their β-phosphates nonenzymatically to pre-phosphorylated serine residues on nearby acidic amino acids of target proteins in the presence of magnesium. When incubated with cellular extracts prepared from yeast, mouse, and flies, 32 Plabelled IP 7 is capable of transferring 32 P to target proteins. Follow-up mass spectrometry studies revealed a role for CK2 in target protein pyrophosphorylation. CK2 provides phosphoprotein substrates for subsequent pyrophosphorylation because acidic amino acids fit nicely into the consensus motif for CK2. The pyrophosphate bonds formed on serine residues in target proteins appear acid labile, but they are resistant to most protein phosphatases. Further studies will be required to measure the reversibility of protein pyrophosphorylation and isolate any relevant bona fide phosphatases. Remarkably, protein pyrophosphorylation has already been linked to the regulation of diverse biological phenomena ranging from glycolysis , rRNA synthesis (Thota et al. 2015), viral particle release (Azevedo et al. 2009), and dynein-driven trafficking (Chanduri et al. 2016). Although 5-IP 7 , 1-IP 7 , and even IP 8 show similar capacities for effecting protein pyrophosphorylation in vitro, the fact that 1-IP 7 , but not 5-IP 7 pyrophosphorylates and activates interferon regulatory factor 3 (IRF3) suggests some degree of stereo-selectivity between the various PP-IPs (Pulloor et al. 2014).

Conclusions
Over the past 20 years, substantial evidence has accumulated implicating PP-IPs and the enzymes that produce them in coordinating a diverse group of signaling and metabolic events. Despite significant advances in our understanding of the physiological functions of inositol pyrophosphates, many questions remain unanswered. For example, can we generalize the mode of interaction between the various PP-IPs and their protein targets? Wu et al. recently took a novel approach with chemically modified, non-hydrolysable IP affinity reagents to carry out a large-scale search for inositol polyphosphatebinding proteins. Using this approach, they identified vacuolar membrane polyphosphate polymerase (Vtc4) as a potential 5-IP 7 binding partner . Wild et al. further elucidated that Vtc4's positively charged SPX domain binds 5-IP 7 , leading to polyphosphate synthesis in yeast and plant (Wild et al. 2016). Developing innovative technology will be a great challenge for us to identify the endogenous PP-IP interactome in mammals.
The generation and analysis of mouse models harboring tissue-specific deletions of the IP6Ks or PPIP5Ks will surely improve our understanding of the roles inositol pyrophosphates play in the control of mammalian physiology. These genetic approaches, however, should be accompanied with more mechanistic studies of the known functions of the PP-IPs, IP6K1/2/3, and PPIP5Ks using cultured cells and other in vitro models. In addition to the more typical methods for analyzing the impact of elevated PP-IPs by overexpressing IP6Ks or PPIP5Ks in mammalian cells over several days, it has recently become possible to synthesize and deliver photocaged PP-IPs into live cells. This and other similar technological advances will allow a more comprehensive dissection of the cellular responses that depend on fine-tuned production and localization of the various PP-IPs.
Recently, SNP variations have been discovered in PP-IP metabolic pathways in disease patients (Crocco et al. 2016). Defining the functional contribution of such genetic variations in PP-IP metabolism will be critical to translate our knowledge into clinical applications. The quantification of inositol poly-and pyrophosphates as potential metabolic biomarkers has so far been possible only by monitoring radiolabeled [ 3 H]-inositol or [ 32 P]phosphate via anion exchange HPLC (Azevedo and Saiardi 2006). In the future, we hope to improve the detection of PP-IPs directly from patient cells and tissue samples. It will be possible for us to apply and further develop the techniques of Saiardi et al. who showed that PP-IPs are efficiently enriched on titanium oxide beads and can be visualized using the fluorescent dye 4 ′ ,6-diamidino-2-phenylindole (DAPI) (Wilson et al. 2015).
The multifunctionality of PP-IP metabolic pathway form a complex regulatory web when we include signaling actions of IPMK, another promiscuous inositol kinase involved in the production of IP 4 , IP 5 , and PIP 3 (Chakraborty et al. 2011;Kim et al. 2016). For example, IP6Ks and their product, 5-IP 7 , participate in the negative feedback regulation of Akt, whereas IPMK's lipid kinase contributes to the full activation of Akt via PIP 3 . The extensive interface between IP6K2 and IPMK can be also found from the regulation of tumor suppressor p53. In contrast to IPMK that directly binds to p53 and activates its transactivation functions in both apoptosis and cell cycle arrest, IP6K2-mediated 5-IP 7 favors p53-dependent transcription toward apoptosis than cell cycle arrest. It will be thus important to clarify the extent to which signaling and metabolic networks related to the IP6Ks, PPIP5Ks, and IPMK interact or overlap.
As further efforts are made to study PP-IPs in various tissuesboth normal and diseasedand as more of their mechanisms of signaling action are revealed, we will eventually discover the full range of the physiological functions of the PP-IPs and the metabolic enzymes that produce them. We strongly believe that this improved understanding of the roles for PP-IPs in mammalian cellular signaling networks will direct the development of novel therapeutic strategies and diagnostic tools for major human diseases like diabetes and cancer.