A. The Influence of Ins(1,4,5)P₃ Under Basal Conditions

In many cells, without any obvious receptor occupancy, basal levels of Ins(1,4,5)P₃ exist, and the same is true for pancreatic β-cells. Although degradation of higher inositol phosphates to Ins(1,4,5)P₃ can occur in special organisms (e.g., Dictyostelium) (Van et al., 1995), in the case of most mammalian cells, this Ins(1,4,5)P₃ pool is thought to reflect constitutive PLC activity, perhaps combined with a retarded breakdown resulting from the inhibition of the 5-phosphatase by glycolytic intermediates (Rana et al., 1986; Rana et al., 1987). What is the function of this basal Ins(1,4,5)P₃? During the investigation of a poorly glucose-responsive β-cell line, it was noted that an insulin secreting variant of HIT cells have significantly reduced basal [Ca²⁺]_i (Yu et al., 2003). Basal Ins(1,4,5)P₃ was artificially increased by expressing the normally endoplasmic reticulum (ER)–confined multiple inositol polyphosphate phosphatase (MIPP, Minpp1) (Craxton et al., 1997) in the cytosol (cyt-MIPP), where it could degrade higher inositol polyphosphates, mainly Ins(1,3,4,5,6)P₅ to Ins(1,4,5)P₃. Expression of cyt-MIPP restored [Ca²⁺]_i to normal basal levels (Yu et al., 2003), suggesting that basal Ins(1,4,5)P₃ could be an important element in setting basal [Ca²⁺]_i. However, more striking was the recovery of glucose-induced increases in [Ca²⁺]_i. The renewed functionality appeared to be dependent on the operation of both Ins(1,4,5)P₃- and ryanodine-sensitive stores in a process that resembled Ca²⁺-induced Ca²⁺ release (CICR) (Fig. 3A). In particular, by controlling basal [Ca²⁺]_I, Ins(1,4,5)P₃ was able to influence the threshold required by the ryanodine receptor to function in its capacity to drive Ca²⁺-induced Ca²⁺ release. An important message from these experiments is that the resting state of the intracellular Ca²⁺ handling machinery is an important platform for the development of complex signaling patterns in stimulated cells. However, the key point is that, even under basal conditions, Ins(1,4,5)P₃ has a profound influence on [Ca²⁺]_i and expands this role during β-cell stimulation, both via membrane-standing GPCRs and glucose.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

The role of Ins(1,4,5)P₃ in β-cell [Ca²⁺]_i regulation. How Ins(1,4,5)P₃ may be involved in the regulation of [Ca²⁺]_i under several different scenarios. (A) Basal. Under basal conditions, it is likely that the existing concentration of Ins(1,4,5)P₃, derived from the basal activity of PLC, is required to maintain the ER Ca²⁺ store in a condition of readiness for subsequent stimulatory events. For example, a minimum [Ca²⁺]_i is required to maintain the functionality of the ryanodine receptor (RyR). Under these conditions, the ER [Ca²⁺]_ER is high and there is a lack of coupling between the components of the capacitative Ca²⁺ entry pathway, namely Stim1 and Orai. (B) Stimulation of G-protein receptors by agonists (in this case acetylcholine) institutes activation of PLC and the formation of Ins(1,4,5)P₃. The receptor can also activate, in a G-protein–independent manner, an Na⁺ channel; however, the opening of this channel is insufficient to drive membrane depolarization to the extent that L-type voltage-dependent Ca²⁺ are opened. The inset depicts the complex interrelationship between G-protein coupled PLC activation (1) and Ca²⁺-driven PLC activation (2 and 3). Already within the first 30 seconds, there is sufficient Ca²⁺ liberated from intracellular stores to further drive PLC activity. (C) Glucose can also stimulate the formation of Ins(1,4,5)P₃, and this is thought to be secondary to Ca²⁺ influx through voltage-dependent Ca²⁺ channels (VDCCs). This means that glucose can stimulate Ins(1,4,5)P₃ generation in the absence of GPCR engagement. However, as with insulin secretion, there is a feedback loop, this time driven by the ATP released concomitantly with insulin exocytosis. This ATP feeds back on a different GPCR again driving Ins(1,4,5)P₃ signaling. In vivo, the signaling through M3 receptors would be either pre-empt or at least be concurrent with glucose action, and thus, the in vivo scenario regarding Ins(1,4,5)P₃ generation is very complex indeed. One important consequence of the simultaneous engagement of M3 receptors and glucose stimulation is that the activation of the Na⁺ current by the M3 receptor is now able to further enhance the depolarization driven by the closure of the K_ATP channels, leading to a stronger Ca²⁺ influx.

No regulatory event in the β-cell should be considered in isolation, and to aid our understanding of this complex system, we must first understand the role of the individual elements and, then, the role of the classic GPCRs in the regulation of intracellular [Ca²⁺]_i, particularly via Ins(1,4,5)P₃.

B. Ins(1,4,5)P₃ Generated from GPCRs and its Role in β-Cells

The pancreatic β cell, in common with all mammalian cells, has its intracellular Ca²⁺ stores regulated by the second messenger Ins(1,4,5)P₃. After the occupancy of a GPCR by its ligand, a sequence is initiated in which G-proteins are activated and subsequently drive the breakdown of PtdIns(4,5)P₂ by PLC, thus generating Ins(1,4,5)P₃ and diacylglycerol (DAG) (Berridge and Irvine, 1984). This canonical pathway, when activated in isolation, has, at best, a modest effect on the promotion of insulin secretion (Berggren and Barker, 2008; Tengholm and Gylfe, 2009; Yamazaki et al., 2010). Indeed, some have reported inhibitory effects on secretion when PLC isoforms are overexpressed (Ishihara et al., 1999), although such overexpression would concomitantly lead to a severe depletion of PtdIns(4,5)P₂, a necessary cofactor mediating exocytosis (Olsen et al., 2003). Regardless of these caveats, agonists operating through GPCRs have clear physiologic roles in the stimulus-secretion coupling of the β-cell. For example, they implement both neuronal stimulation, via acetylcholine, and autocrine feedback, via ATP (Gilon and Henquin, 2001; Tengholm and Gylfe, 2009; Jacques-Silva et al., 2010). However, under physiologic conditions, the GPCR pathways do not operate in isolation, but in combination with other insulin secretagogues, notably glucose. We examine this synergy in the next section of the review.However, to understand how Ins(1,4,5)P₃ generation by GPCR acts collaboratively with classic glucose stimulation, it is necessary to examine the role of GPCR-mediated Ins(1,4,5)P₃ production in isolation.

A clear indication of the physiologic relevance of PLC-mediated Ins(1,4,5)P₃ generation in β-cells comes from two studies. In the first, acetylcholine-activated, PLC-linked M3 receptors were selectively knocked out in mouse pancreatic β-cells (Gautam et al., 2006) and, in the second, depletion of the Ins(1,4,5)P₃ receptor after the knockout of ankyrin B (Healy et al., 2010). The phenotype of both these interventions was a marked curtailment of glucose-induced insulin secretion. Although it is important that, downstream of M3 receptor occupancy, lies a myriad of signal transduction pathways (Kong and Tobin, 2011), and not just the production of Ins(1,4,5)P₃, the second study emphasizes that Ins(1,4,5)P₃ plays an important role. These studies are an important reminder of the significance of classic-receptor occupancy in a cell in which glucose action has been considered to be pre-eminent.

Recent studies on the generation of Ins(1,4,5)P₃ in pancreatic β-cells have relied on the indirect measurement of Ins(1,4,5)P₃ by green fluorescent protein (GFP)/enhanced GFP–tagged reporter constructs based on the PtdIns(4,5)P₂-binding PH-domain of PLCδ (Tengholm and Gylfe, 2009; Wuttke et al., 2010). Under basal conditions, the construct binds to plasma membrane (PM) PtdIns(4,5)P₂. When endogenous PLC is activated and PtdIns(4,5)P₂ breaks down to Ins(1,4,5)P₃, fluorescence is lost from the PM. This loss of fluorescence is measured using either confocal- or total internal reflection fluorescence microscopy. A concern about this approach is its indirectness. Although it is a good assay of PLC activity, it does not directly report on the subsequent changes in Ins(1,4,5)P₃ concentration that are regulated by both PLC generation and subsequent Ins(1,4,5)P₃ metabolism. Unfortunately, direct reliable assays using fluorescence resonance energy transfer generation after Ins(1,4,5)P₃ binding have had poor sensitivity, although a more recent candidate looks to be more promising (Tanimura et al., 2009). Despite the aforementioned caveat about the assay’s indirect measurement of Ins(1,4,5)P₃, the data on Ins(1,4,5)P₃ changes obtained by this technique are quite similar to historical data gained by direct Ins(1,4,5)P₃ measurement (Tengholm and Gylfe, 2009). The clear advantage of this fluorescent reporter approach is to the ability to monitor Ins(1,4,5)P₃-related events in single cells simultaneously with other important parameters, especially the monitoring of [Ca²⁺]_i. Studies using this approach have revealed new levels of complexity, even at the level of the simple and classic GPCR-mediated liberation of Ca²⁺ from Ins(1,4,5)P₃-sensitive stores. Even greater complexity arises when we consider glucose stimulation, which we discuss in the next section.

The original studies on the engagement of β-cell GPCRs with maximal agonist concentrations, in line with other cell systems, demonstrate parallel, biphasic Ins(1,4,5)P₃ and [Ca²⁺]_i increases. The initial phase is marked by both a rapid increase in Ins(1,4,5)P₃ and [Ca²⁺]_i, and a lower sustained phase of both molecules (Fig. 3B)

Some caution needs to be applied when considering data based on maximal agonist concentration, because Cobold and Cuthbertson only discovered agonist-dependent changes in the frequency of [Ca²⁺]_i oscillations after GPCR activation of PLC, when using physiologic agonist concentrations (Cobbold and Cuthbertson, 1990). Physiologic concentrations are often dramatically lower than the maximum effective concentrations extrapolated from the dose-response curve. Unfortunately, there is less information in β-cells about the oscillatory behavior of [Ca²⁺]_i when stimulated with lower, physiologic concentrations of agonists. However, because the majority of published studies use high agonist concentrations, most information on the PLC system in β-cells comes from them.

Studies using the aforementioned GFP-PLCδ-PH domain reporter technique, in conjunction with classic Ca²⁺-sensitive fluorescent dyes, also indicate two phases of PLC activity, consistent with the older published data using direct Ins(1,4,5)P₃ measurements (Tengholm and Gylfe, 2009). There is an initial rapid increase in Ins(1,4,5)P₃ peaking at ∼30 seconds, followed by a second lower sustained phase Fig. 3B. In both phases, there is some dependency on the Ca²⁺ released from intracellular stores to drive PLC. However, with use of either 2-APB or La³⁺, blockers of store-mediated Ca²⁺ entry or nonspecific Ca²⁺ entry, the sustained but not the initial phase of PLC activity is blocked (Thore et al., 2005). There is therefore a feed-forward activation of PLC, through both release of Ca²⁺ from intracellular stores and capacitative Ca²⁺ entry. In the β-cell, entry through store-operated channels is modest (Tengholm and Gylfe, 2009), and it has been proposed that the efficacy of this Ca²⁺ is enhanced by a close proximity between Ca²⁺ entry channels and PLC (Tengholm and Gylfe, 2009). The modest increase in [Ca²⁺]_i from store entry may reflect a reduced level of STIM1 in β-cells. STIM1, together with Orai channels, mediate the store-operated Ca²⁺ entry in cells (Srikanth and Gwack, 2012), including β-cells (Tamarina et al., 2008; Tengholm and Gylfe, 2009). There is no direct proof of reduced STIM1 expression in β-cells, but recent data from other cell types suggest a reciprocal relationship between store-mediated Ca²⁺ entry and entry through Ca_v1.2 voltage-dependent Ca²⁺ channels. STIM1 blocks the voltage-dependent Ca²⁺ channels, making them silent (Park et al., 2010; Wang et al., 2010). The fact that the β-cell has active and functionally critical voltage-dependent channels suggests that STIM1 concentration is lower in these cells. In other cell systems, physical coupling between TRP channels and PLC has been shown, but how this is arranged in the β-cell is currently poorly characterized.

In terms of the PLC isoform(s) that may be activated, classically, PLCβ is coupled to GPCRs, whereas PLCδ, because of its higher Ca²⁺ sensitivity, is proposed to drive Ca²⁺-stimulated PLC activation. However, at least in some insulin-secreting cell lines, there is also evidence for Ca²⁺-driven activation of the PLCβ isoform (Kelley et al., 2001). These data should be considered carefully, however, because the conclusions about isoform involvement are based on the transfection of exogenous PLCs. Nonetheless, PLC β, δ, and γ have been detected in β-cells (Tengholm and Gylfe, 2009; Yamazaki et al., 2010). Although PLCγ is not activated directly by Ca²⁺, it seems to play a role in [Ca²⁺]_i regulation (Patterson et al., 2005).

An interesting perspective on acetylcholine/carbachol regulated [Ca²⁺]_i and Ins(1,4,5)P₃ has come from the characterization of the islets from mice in which the β₃ subunit of the L-type voltage-dependent Ca²⁺ channel has been globally knocked out (Berggren et al., 2004). There are many interesting facets to this phenotype, because the overall result is apparently an enhanced sensitivity to glucose stimulation and subsequent release of insulin. Other aspects are discussed later, but one key observation is relevant to our current discussion. In comparing the islets from these knockout animals with those of controls, carbachol-stimulation led to both an enhanced initial increase in [Ca²⁺]_i and an enhanced increase in Ins(1,4,5)P₃ within the same period (30 seconds stimulation). Removal of extracellular Ca²⁺ normalized the response to that of the islets from the control animals; thus, some form of Ca²⁺ entry is presumed to be necessary (unpublished data). Furthermore, the Ca²⁺ entry could be blocked with a specific L-type Ca²⁺ channel inhibitor, nimodipine (unpublished data). At first sight, the participation of the L-type Ca²⁺ channel seems to be important, but under standard conditions, no changes in Ca²⁺ channel parameters were observed in the knockout animal. Historically, engagement of muscarinic receptors in β-cells has been shown to slightly depolarize cells after activation of an Na⁺ current (Henquin et al., 1988). However, under the basal glucose concentration that was used in the β₃ knockout study, the depolarization would not be sufficient to open voltage-gated channels (Henquin et al., 1988). What other possibilities exist to explain these data? One suggestion is that nimodipine is blocking another Ca²⁺ entry pathway. Indeed, dihydropyridines, such as nimodipine, can block store- or receptor-operated Ca²⁺ channels (Harper et al., 2003), but this usually occurs at higher doses of the inhibitor than was used in the knockout study. Furthermore, as we have stated earlier, store-operated Ca²⁺ entry occurs after the initial increase in Ca²⁺ and Ins(1,4,5)P₃ and, thus, is unlikely to contribute to the increase in [Ca²⁺]_i occurring within the first 30 seconds. Several other possible explanations can be advanced to explain how β₃ knockout can be enhancing the carbachol response. For example, the lifetime of the G_qα subunit, which drives PLC activity, could be sustained by the increased binding of the inactivating Gβγ to either the alpha subunit of the L-type channel (Yang and Berggren, 2005) or some other intracellular target (Berggren et al., 2004). The increased binding to the channel or another unidentified protein is made possible because of the vacation of the binding site by β₃. In any event, the β₃ knockout model is intriguing, because it starts to expose little understood aspects of classic GPCR activation of PLC and the mobilization of intracellular Ca²⁺, indicating that there is still much to learn about even this basic system.

C. Changes in Ins(1,4,5)P₃ in Response to Glucose Stimulation

When we consider Ins(1,4,5)P₃ concentration increases driven by glucose stimulation, a new level of complexity emerges as both GPCR- and Ca²⁺-driven PLC activation occur, without addition of exogenous GPCR ligands. We describe how this occurs below.

From the earliest work in this field, it was clear that glucose alone could drive the increase in Ins(1,4,5)P₃ formation (Best and Malaisse, 1983; Biden et al., 1987), although this was often a modest response, compared with, for example, carbachol-driven changes. Although there were a number of suggestions of how glucose might increase intracellular Ins(1,4,5)P₃, the major explanation to emerge was the idea that the Ins(1,4,5)P₃ was derived from Ca²⁺ activation of PLC, secondary to Ca²⁺ influx through voltage-gated Ca²⁺ channels (Barker et al., 2002). However, the original experiments suggested that this modest increase in Ins(1,4,5)P₃ was not wholly dependent on the presence of extracellular Ca²⁺ (Biden et al., 1987). This glucose-dependent, Ca²⁺-independent generation of Ins(1,4,5)P₃ may be derived from either the arrested breakdown of Ins(1,4,5)P₃ by glycolytic metabolites (Rana et al., 1986; Rana et al., 1987) or depolarization-induced activation of PLC (Henquin et al., 1998; Bergsten, 2002; Thore et al., 2004; Tamarina et al., 2005; Tengholm and Gylfe, 2009). Because glucose stimulation alone can drive PLC activity, the same series of events shown Fig. 3B with the Ins(1,4,5)P₃ and the Ca²⁺ stores can occur with glucose stimulation alone (Fig. 3C).

Glucose stimulation is more complex than the engagement of the M3 muscarinic receptor (see section I.B; Fig. 3B), because the influx of Ca²⁺ through voltage-dependent Ca²⁺ channels also leads to the release of both insulin and ATP from insulin-containing granules. This ATP, in turn, can act through purinergic GPCRs to also stimulate PLC and generate Ins(1,4,5)P₃, again leading to the complex interplay of stores and capacitive Ca²⁺ entry (Tengholm and Gylfe, 2009). These events induced by ATP are temporally separated from the events stimulated by glucose. Therefore, the Ins(1,4,5)P₃ generated and the Ca²⁺ subsequently mobilized may influence signaling events distinct in time from those generated by activation of M3 GPCRs and/or glucose-induced Ca²⁺ influx. The link between released ATP and PLC activation is mediated through P2Y receptors in rodents (Petit et al., 2009). An added complication is that, in contrast to rodents, human β cells have ionotropic P2×3 receptors (Jacques-Silva et al., 2010) and not P2y receptors. P2×3 receptors are not linked to PLC activation but are membrane ion channels that, on engagement, mediate the influx of extracellular Ca²⁺ (Jacques-Silva et al., 2010). Ca²⁺ entry through plasma membrane standing-channels could influence a local environment that is different from the one that Ca²⁺ mobilized from intracellular Ca²⁺ stores reaches. Therefore, it is possible that there is some diversity in downstream signaling between rodent and human islets. Indeed purinergic stimulation in rodents has been described to both stimulate and inhibit insulin secretion (Petit et al., 2009), whereas in humans, it seems to solely enhance insulin secretion (Jacques-Silva et al., 2010).

The final issue to consider in relation to glucose-stimulated increases in Ins(1,4,5)P₃ is how these glucose-induced changes act in concert with the Ins(1,4,5)P₃ produced by the acetylcholine-mediated activation of M3 GPCRs (Gilon and Henquin, 2001). This scenario is more representative of the physiologic situation when the β-cell responds to multiple inputs after the consumption of food. When acetylcholine or its analog, carbachol, are applied simultaneously with glucose stimulation, there is a synergistic effect on insulin secretion (Biden et al., 1987). One reason for this is the muscarinic-driven opening of the Na⁺ channel, which partially depolarizes the membrane (Henquin et al., 1988). As stated in section I.B, under basal glucose conditions, this depolarization is insufficient to drive Ca²⁺ entry through voltage-dependent Ca²⁺ channels. However, this modest depolarization becomes significant in combination with the glucose-induced depolarization, resulting in the closure of K_ATP channels. The combined depolarization is greater than that elicited by blockade of the K_ATP channels alone, resulting in an enhanced Ca²⁺ entry through voltage-dependent Ca²⁺ channels. Thus, a synergistic effect is created between GPCR activation and glucose metabolism. It has been suggested that part of the Na⁺ current may be attribtuable to Na⁺ entry through store-operated Ca²⁺ channels, which are not fully selective for Ca²⁺ (Tengholm and Gylfe, 2009). However, an independent Na⁺ tetrodotoxin-insensitive channel has now been identified, sodium leak channel, nonselective (NALCN), which at least in part, is likely to be responsible for this phenomenon (Rolland et al., 2002b; Swayne et al., 2010) and is directly coupled to M3 receptors (Swayne et al., 2010).

We have already described above the data from the β-cell–specific M3 receptor knockout mouse, which underscores the importance of the synergy between parasympathetic input and glucose increases in driving insulin secretion (Gautam et al., 2006). A more recent report reinforces this idea from a different perspective. In both cases, an impairment of glucose-stimulated insulin secretion results in a type 2 diabetic (T2D) phenotype. The latest report comes from work on the ankyrin B knockout mouse (Healy et al., 2010). Ankyrin B is only expressed in the pancreatic β-cells and not in other islet cells, which aids interpretation of the results. This knockout mouse has a similar phenotype to the β-cell–specific M3 receptor knockout mouse with regard to its dampened cholinergic augmentation of glucose-stimulated insulin release. In this case, the IP3R specifically is involved. Because homozygous ankyrin B knockout mice do not survive to adulthood, experiments were performed using heterozygous animals with half the copy number of ankyrin B. In these animals, there was a severely blunted and delayed increase in [Ca²⁺]_i in response to carbachol, both in the presence and absence of extracellular Ca²⁺. There was a similarly blunted response to the carbachol-enhanced, glucose-stimulated secretion of insulin in dynamic insulin release assays. Strikingly, even glucose-stimulated insulin secretion alone was compromised, highlighting the importance in glucose-induced Ins(1,4,5)P₃ signaling in exocytosis. A key defect in these cells was a halved amount of the IP3R. This suggests that, when ankyrin-B is absent, it is no longer able to target the IP3R correctly, leading to its destruction. Furthermore, a loss of function mutation of ankyrin B (R1788W) was found in a significant number of patients with T2D. This study is helpful in our thinking about the role of Ins(1,4,5)P₃, because the knockout of the M3 receptor affects several different signal transduction pathways besides the generation of Ins(1,4,5)P₃. However, the ankyrin B study suggests that the downstream production of Ins(1,4,5)P₃ from the M3 receptor and its modulation of intracellular Ca²⁺ stores is likely to be a key element of the M3 knockout phenotype.

Another aspect of Ins(1,4,5)P₃ involvement in the regulation of Ca²⁺ comes with the observation that Ins(1,4,5)P₃ can indirectly activate a Ca²⁺-activated, inwardly rectifying K⁺ channels (Ammala et al., 1991). This becomes more significant when we discuss the influence of Ins(1,4,5)P₃ on Ca²⁺ oscillations in the next section. This topic of oscillatory phenomena in the regulation of β-cells represents the final level of complexity involving Ins(1,4,5)P₃.

D. A Role for Ins(1,4,5)P₃ in [Ca²⁺]_i Oscillations

Oscillations in cellular messengers, including Ca²⁺, are an established phenomenon in cell biology. It has particular importance in the β-cell, because besides [Ca²⁺]_i, electrical activity, metabolism, cAMP, and insulin, are subject to dynamic change in a synchronized manner (Tengholm and Gylfe, 2009). We consider specifically the relationship of Ins(1,4,5)P₃ to [Ca²⁺]_i oscillations in this section.

A notable breakthrough in mammalian Ca²⁺ signaling was the observation made in hepatocytes that, at lower physiologic concentrations of agonists, the oscillation frequency (but not the amplitude) of Ca²⁺ spikes was directly proportional to agonist concentration (Cobbold and Cuthbertson, 1990). An important question, of course, is whether oscillations in [Ca²⁺]_i are driven by parallel oscillations in Ins(1,4,5)P_3. If this is the case, it would imply an oscillatory behavior of PLC or an oscillatory behavior of Ins(1,4,5)P₃ metabolism. In the literature, there are examples of [Ca²⁺]_i oscillations with or without parallel oscillations in Ins(1,4,5)P₃ (Berridge, 2007). This seems to be dependent on the system examined (Berridge, 2007), although, as we have noted above, some of the evidence has come from the indirect measurement of PLC activity with use of the GFP-PLCdelta PH domain (Tengholm and Gylfe, 2009). In summary, it is now generally understood that Ins(1,4,5)P₃ does not need to oscillate to generate oscillations in [Ca²⁺]_i, even when we consider the simplest system without the involvement of voltage-dependent Ca²⁺ entry. The oscillatory behavior can arise from the complex interplay between store-filling, the IP3R sensitivity to luminal Ca²⁺, and the process of Ca²⁺-induced Ca²⁺ release (CICR). Berridge has called this the “store- or luminal-loading model of [Ca²⁺]_i oscillations” (Berridge, 2007, 2009). We consider this model later, because we believe that it has some bearing in trying to understand what is happening in pancreatic β-cells.

Narrowing our focus back to pancreatic β-cells, an early observation was that oscillations in Ins(1,4,5)P₃ could accompany oscillations in [Ca²⁺]_i (Barker et al., 1994). The β-cell is considered to have two types of oscillation in [Ca²⁺]_i : fast oscillations, which are strongly linked to plasma membrane electrical activity, and slow oscillations, which correlate better with oscillations in metabolism and, more importantly, oscillations in insulin release (Hellman et al., 1992; Henquin et al., 1998; Bergsten, 2002; Rolland et al., 2002a; Thore et al., 2004; Tamarina et al., 2005). In these early experiments, a population of isolated pancreatic β-cells was artificially synchronized, after which several parallel slow oscillations in [Ca²⁺]_i and Ins(1,4,5)P₃ were observed (Barker et al., 1994). Unfortunately, the system did not have enough temporal resolution to clearly delineate whether Ins(1,4,5)P₃ oscillations drove [Ca²⁺]_i oscillations or vice versa. More recently, the use of GFP-PLCdelta PH domain reporters has led to the observation of similar parallel oscillations in Ins(1,4,5)P₃ and [Ca²⁺]_i under both slow and faster oscillatory patterns (Henquin et al., 1998; Bergsten, 2002; Thore et al., 2004; Tamarina et al., 2005; Beauvois et al., 2006; Tengholm and Gylfe, 2009). In some of these studies, the temporal resolution has been sufficiently good to demonstrate that [Ca²⁺]_i oscillations drive Ins(1,4,5)P₃ oscillations and not the other way round. Does this mean that in the β-cell Ins(1,4,5)P₃ and, thus, that the mobilization of Ca²⁺ from intracellular stores is completely subservient to Ca²⁺ entry? A consideration of the two previous sections of this article would indicate that the situation is unlikely to be so clear-cut because of the intricate feedback loops existing in even the simplest scenarios involving Ins(1,4,5)P₃, Ca²⁺ stores and Ca²⁺ entry.

Figure 4 shows the Berridge model of store- or luminal-loading model of calcium oscillations (Berridge, 2007, 2009), which we will now try and integrate into our understanding of β-cell [Ca²⁺]_i oscillations. This model is dependent on the following experimentally verified elements: the IP3R luminal Ca²⁺ sensor, store-operated Ca²⁺ entry, the mobilization of Ca²⁺ from intracellular stores and the process of CICR. The way that Berridge has envisioned this is as follows. Under basal conditions and at low agonist doses, there is not sufficient Ins(1,4,5)P₃ to mobilize Ca²⁺, but there is enough agonist to promote Ca²⁺ entry and a filling of the stores. Increasing Ca²⁺ in the stores leads to spontaneous Ca²⁺ release (spikes), because the increased Ca²⁺ changes the conformation of the IP3R through its luminal Ca²⁺ sensor. This structural change makes the IP3R more sensitive to Ins(1,4,5)P₃, leading to a release of Ca²⁺. There is then an amplification of the increasing phase of the Ca²⁺ spike driven by CICR. What then is the explanation that Berridge offers for the change in frequency of the oscillations seen after higher agonist concentrations? He suggests that the increased agonist concentrations lead to a higher rate of store filling, because only at a high [Ca²⁺]_ER will Ca²⁺ spiking occur. There is a significant amount of literature that supports the idea that the intracellular stores are important for at least sustaining Ca²⁺ oscillations, but this proposal takes this idea one step further by attempting to also explain changes in oscillatory frequency. The key question for our current review is whether this has a role to play in understanding of the regulation of [Ca²⁺]_i oscillations in pancreatic β-cells and islets.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

The luminal Ca²⁺ model of Ins(1,4,5)P₃-induced oscillations in [Ca²⁺]_i. The figure shows the luminal Ca²⁺ model of generating oscillations in [Ca²⁺]_i proposed by Berridge (2007). For a detailed discussion about how this may impact on β-cell Ca²⁺ oscillations, see Section III.D. In brief, the model proposes that, after GPCR engagement with low, but physiologic concentrations of agonist, Ins(1,4,5)P₃ is released, but does not achieve sufficient concentrations to activate the IP3R. Therefore, there is lag phase (L). During this time, there is a gradual build up in luminal [Ca²⁺]_i through increased uptake into the stores, from extracellular Ca²⁺ entry. When a threshold [Ca²⁺]_ER is reached, the sensitivity of the IP3R to Ins(1,4,5)P₃ is enhanced, leading to activation of the receptor and the release of Ca²⁺ from the stores in the form of a spike. The luminal concentration consequently decreases, and the cycle is repeated as the stores refill. The change in frequency from slow to fast is then dependent on the time that it takes to refill the stores. More efficient store filling will lead to faster oscillations.

A good place to start, perhaps, is an examination of those islets that oscillate, compared with those that do not, in an otherwise identical preparation. Such a scenario was investigated relatively recently (Jahanshahi et al., 2009). First, islets that did not oscillate had high basal [Ca²⁺]_i and reduced Ca²⁺ responses to glucose stimulation. However, the crux of the difference appeared to be a decreased content of ER [Ca²⁺]_ER, rather than in any defects in mitochondrial membrane potential, plasma membrane channels (including L-type Ca²⁺ channels and K_ATP channels), or cell death rates. Of interest, in other systems, it has been reported that Ins(1,4,5)P₃-regulated ER stores are important for the sustaining of oscillations (Yano et al., 2004). These data seem to fit well with the aforementioned Berridge hypothesis.

Another scenario that we have already introduced is the β₃ knockout mouse with its enhanced frequency of [Ca²⁺]_i oscillations. In this model, an increased Ca²⁺ mobilization is observed after stimulation with the GPCR agonist carbachol. Although this enhanced mobilization is dependent on extracellular Ca²⁺, suggesting that Ca²⁺ entry plays a role, whatever the source, the greater increase in [Ca²⁺]_i could easily lead to a more rapid filling of stores and, thus, a faster onset of the next oscillation. Examination of the Ca²⁺ stores of these mice revealed that the rate of loss of Ca²⁺ from the stores was significantly slower than that of islets from control animals (unpublished data). This could reflect less efficient mobilization (which is unlikely, given the general phenotype) or a more efficient refilling of the pools, shortening the time to the next oscillation. Of interest, when inhibitors of the IP3R are applied to the β₃ knockout islets, the [Ca²⁺]_i oscillation frequency of the β₃ knockout islets decreases (Berggren et al., 2004).

In summary, Ins(1,4,5)P₃ is likely to be important for the sustaining of [Ca²⁺]_i oscillations in pancreatic β-cells, and although it is most likely that Ins(1,4,5)P₃ oscillations follow [Ca²⁺]_i oscillations rather than vice versa, the Berridge model suggests that Ins(1,4,5)P₃ could play a role. The major caveat with this theory is that there is much evidence to suggest that it is the electrical excitability of the plasma membrane and voltage-driven events that underscore β-cell [Ca²⁺]_i oscillations (Tengholm and Gylfe, 2009). The current common consensus is that the time-keeping and initiation of oscillations is driven by these plasma membrane events and that Ins(1,4,5)P₃-dependent Ca²⁺ events enhance the fundamental oscillations already established by the classic oscillations in electrical activity. Therefore, any information that could link Ins(1,4,5)P₃ to membrane electrical activity would save this inositol phosphate from just being a follower of oscillatory events in excitable cells. Of interest, a study from more than 20 years ago addresses this issue specifically.

The end of the [Ca²⁺]_i oscillation is marked by a repolarization of the plasma membrane and an effective silencing of voltage-dependent Ca²⁺ entry. This repolarization is believed to be performed by inwardly rectifying Ca²⁺-activated K⁺ channels. A family of such channels exists in the PM of pancreatic β-cells. Twenty years ago, one channel that was insensitive to the triethylamine inhibition, typical of these channels, was found to be specifically sensitive to the Ins(1,4,5)P₃ -promoted release of Ca²⁺ from intracellular stores (Ammala et al., 1991). Thus, Ins(1,4,5)P₃ could indirectly influence the electrically driven oscillations in [Ca²⁺]_i. Certainly, evidence from the study of other inwardly rectifying K⁺ channels suggests that their inhibition can change the frequency of [Ca²⁺]_i oscillations in pancreatic β-cells. These studies leave open the debate about the centrality of Ins(1,4,5)P₃ in the regulation of β-cell [Ca²⁺]_i oscillations and reveal an important area for future investigations.

E. The β-Cell Ins(1,4,5)P₃ Receptor

The IP3R was already featured several times in our discussions. In fact, comparatively few studies have been performed on the β-cell IP3Rs; therefore, most of our information is from extrapolation from work done on IP3Rs in other systems. We do not discuss these here but highlight what is known specifically about the β-cell IP3R receptors and suggest that the reader refers to several excellent reviews covering this subject (Patterson et al., 2004; Mikoshiba, 2007; Taylor and Prole, 2012).

Evidence for type 1–3 IP3Rs exists in pancreatic β cells. Originally, there was a disproportionate focus on the type 3 receptor, because it was believed to be specifically enriched in β-cells (Hagar and Ehrlich, 2000). However, this perception arose from the use of the rat RIN m5F β-cell line, which is enriched in this receptor and was reinforced by subsequent studies on primary rat β-cells; these had the same dominance as the type 3 receptor (Lee et al., 1999). However, in mouse pancreatic β-cells, the type 1 is as abundant as type 3 (Lee and Laychock, 2001). The actual proportions of these different isoforms may be important in trying to understand β-cell [Ca²⁺]_i oscillations, because in other cells, deletion of one or the other IP3R subtype has resulted in different effects on [Ca²⁺]_i oscillations (Zhang et al., 2011).

Another controversial aspect of β-cell IP3Rs is the notion that IP3Rs are in insulin-containing vesicles and regulate vesicle Ca²⁺. This area is contentious, because the original study showing this (Blondel et al., 1994) was subsequently questioned. The debate centered around the reliability of the antibody used to localize the receptor (Ravazzola et al., 1996). In other cell systems, there now seems to be a clear indication of IP3R in secretory granules (Yoo and Hur, 2012), which at least leaves this idea open.

Most of the focus related to the IP3R in the β-cell, in common with many other cell types, has been directed at the regulation of [Ca²⁺]_i. However, it is sobering to realize that the Ca²⁺ pore-forming domain represents only ∼5% of the total sequence of the protein (Patterson et al., 2004; Mikoshiba, 2007). Although there is a portion of the molecule that is exposed to the ER lumen, the vast majority of the protein forms a massive porous superstructure, which in addition to possessing many sites for the functional regulation of the channel, has a large number of other proteins associated with it. One yeast 2 hybrid screen trapped more than 300 proteins (Patterson et al., 2004), and although such approaches can give false positives, to date, more than 25 proteins have been rigorously demonstrated to be associated with the channel. This includes structural proteins, regulatory proteins, and proteins that seem to use the IP3R as a docking station from which they are liberated to implement diverse cellular functions. When considering this aspect of the IP3R, it is clear that it is a unique master conductor of cellular regulation and may exert many thus far uncharacterized events around cellular [Ca²⁺]_i regulation. With few exceptions, there has been little characterization of partner proteins in β-cells. However, the data from both the ankyrin B (Healy et al., 2010) and β₃-knockout mice (Berggren et al., 2004), suggest that there is much to explore in this subject area.

An important consideration when thinking about the pancreatic β-cell is the quest to identify the components in the stimulus-secretion coupling that may contribute to or be affected by the T2D phenotype. We have already described the role that the IP3R may have in human diabetes, which was revealed by studies using the ankyrin B–deficient mouse. Are there other examples of links between IP3Rs and diabetes? There are two examples to which we can refer, both relatively recent. The first links the gene for the type 3 IP3R (ITPR3) to type 1 diabetes (T1D) in the Swedish population (Swedes have one of the highest incidence of T1D worldwide). The original publication suggested an independence of the association from any MHC class II gene (Roach et al., 2006) and, thus, a clear likelihood of the involvement of the IP3R gene. Unfortunately, another group in a subsequent study (Qu et al., 2008) questioned these findings and, in their study, found that the ITPR3 gene polymorphism was attributable to a linkage disequilibrium with HLA class II. A second recent example relates to T2D. After investigation of a transgenic mouse line that exhibited a progressive glucose intolerance, it was shown that the transgenic insertion disrupted 10 loci in chromosome 6, one of which was IP3R1 (Ye et al., 2011). Under a regular diet, the heterozygous transgenic mice developed glucose intolerance, but not insulin resistance, and a decrease in second-phase, glucose-stimulated blood insulin levels, consistent with a reduced insulin release in islets. In this context, of importance, Ca²⁺ mobilization from Ins(1,4,5)P₃-sensitive stores has been reported to help drive the replacement of the readily releasable pool (RRP) of insulin vesicles, and thus second-phase secretion (Gromada et al., 1999). Subjecting these same animals to a high fat diet (Ye et al., 2011) resulted in the additional development of hyperglycemia and insulin resistance. Therefore, the current data suggest at least a close association of the IP3R with T2D. Further work is, however, required to decide whether IP3R3 is associated with T1D.

These above examples suggest that Ins(1,4,5)P₃, along with its receptor, represents an important dimension of both the physiology and the pathophysiology of the β-cell and is an obvious candidate in the search for new pharmacological approaches to the treatment of diabetes.

A. Inositol 3,4,5,6-Tetrakisphosphate as a Novel Second Messenger

When we look beyond Ins(1,4,5)P₃ for a second messenger in the realm of the higher inositol polyphosphates, there is perhaps only one serious candidate: inositol 3,4,5,6-tetrakisphosphate [Ins(3,4,5,6)P₄] (Shears et al., 2012). Other higher inositol polyphosphates have been claimed in this role (Shears et al., 2012); however, because of their existing high concentrations in cells, it is more likely that they function as cellular regulators rather than messengers. Of interest, although Ins(3,4,5,6)P₄ is metabolically distinct from Ins(1,4,5)P₃, it is nonetheless regulated in mammalian cells by inositol 1,3,4-trisphosphate [Ins(1,3,4)P₃], a downstream product of Ins(1,4,5)P₃ metabolism. It does this in a unique and elegant fashion (Shears et al., 2012). We first focus on the unique mechanism of regulation of Ins(3,4,5,6)P₄ and, then, return to the discussion of its second messenger activity.

Ins(3,4,5,6)P₄ is generated by a idiosyncratic enzyme, inositol tetraphosphate kinase, which was originally identified as an Ins(1,3,4)P₃ 5/6 kinase (Abdullah et al., 1992) and, then, an Ins(3,4,5,6)P₄ 1-kinase (Tan et al., 1997). In fact, it performs a multitude of functions, including the formation of Ins(1,3,4,5,6)P₅ from Ins(1,3,4)P₃ through either Ins(1,3,4,6)P₄ or Ins(1,3,4,5)P₄ or the formation pf Ins(1,3,4,5,6)P₅ from Ins(3,4,5,6)P₄ and vice versa, thus acting as both a kinase and a phosphatase (Fig. 1)(Ho et al., 2002). However, it performs these functions in an atypical manner; thus, for example, the phosphatase activity is not inhibited by ATP as a normal reverse of a kinase reaction would be. The complete mechanism was only recently elucidated after detailed structural information from a crystallographic study of the human enzyme, which has important differences from the structures of the equivalent plant or amoeba enzymes that have been determined previously (Chamberlain et al., 2007). Thus, only the human and, presumably, other mammalian enzymes can use Ins(1,3,4)P₃ to regulate the concentration of Ins(3,4,5,6)P₄ (Chamberlain et al., 2007). The Ins(1,3,4)P₃ is derived sequentially from the metabolism of the second messenger Ins(1,4,5)P₃ through Ins(1,3,4,5)P₄, and thus, its concentration is directly coupled to receptor occupancy. The elevation in Ins(1,3,4)P₃ then drives increases in the metabolically unrelated Ins(3,4,5,6)P₄ (Yang and Shears, 2000). The mechanism of this regulation is believed to be a specialized intersubstrate phosphate transfer, mediated by a nucleotide deeply imbedded in the active site of the enzyme. In short, the 1-phosphate of Ins(1,3,4,5,6)P₅ is donated to the ADP tightly bound to the enzyme, forming Ins(3,4,5,6)P₄ and ATP, the former being released. In a linked subsequent step, this bound ATP donates a phosphate to Ins(1,3,4)P₃, and the Ins(1,3,4,6)P₄ or Ins(1,3,4,5)P₄ that is formed is released. Thus, Ins(1,3,4)P₃ acts as an acceptor that enables the cycling of the enzyme back into a form that can again degrade Ins(1,3,4,5,6)P₅ to Ins(3,4,5,6)P₄. The increase in Ins(3,4,5,6)P₄ after receptor occupancy is seen in all mammalian cells studied, including insulin-secreting cells (Shears et al., 2004). In β-cells, stimulation with glucose without receptor occupancy can generate Ins(1,4,5)P₃ (see section II.C and Fig. 3C). Therefore, it is possible that, in β-cells, Ins(3,4,5,6)P₄ can be generated without the engagement of GPCRs. This brings us to the second messenger function of Ins(3,4,5,6)P₄.

Shears et al. (2004, 2010) established that Ins(3,4,5,6)P₄, within its defined physiologic range and with great specificity, inhibited the CaMKII-regulated chloride channels found largely in the PM of cells. In mammalian cells, this Cl- channel has been identified as the ClC-3 subtype. These channels regulate salt and fluid secretion in epithelial cells and cell volume or electrical excitability in muscle and neuronal cells (Shears et al., 2004, 2012). Unfortunately, the precise mechanism by which Ins(3,4,5,6)P₄ inhibits these channels is still unknown, but it is unlikely to be a direct interaction with the channel. In β-cells and, perhaps, in other cell systems, Ins(3,4,5,6)P₄ can regulate the homologous CaMKII-activated chloride channels found in the membranes of secretory vesicles (Renstrom et al., 2002). The role of these channels is to provide a counter ion and, thus, charge neutralization during the acidification of the vesicle by the electrogenic H⁺-ATPase. This acidification is considered to be an important step to produce fusion-competent vesicles in the final stages of exocytosis (Hou et al., 2009). This means that Ins(3,4,5,6)P₄ acts to inhibit acidification and, thus, exocytosis (Renström et al., 2002). Two studies showing the importance of the ClC-3 channel for insulin exocytosis were published in Cell Metabolism (Deriy et al., 2009; Li et al., 2009). These two studies reported that the complete knockout of the ClC-3 gene led to impaired glucose and K⁺-stimulated insulin secretion. However, these studies were questioned on several grounds, including conflicting information between them and a dispute about the presence of CIC-3 in insulin-secretory vesicles, rather than other vesicle types (Jentsch et al., 2010). Both the original groups were, however, able to provide a robust defense, including additional data supporting their positions in the same issue of the journal. These studies highlight that the CIC-3 is involved in several distinct processes leading to secretion, including insulin processing, granular acidification, and ultimately exocytosis. This means that increases in Ins(3,4,5,6)P₄ could regulate any or all of these events. Because the increase in Ins(3,4,5,6)P₄ is not rapid, its action is likely to influence second rather than first phase insulin release. It may also play a pathologic role during periods of hyperglycemia. This is because Ins(3,4,5,6)P₄ production is enhanced when energy metabolism is compromised. In addition, chronic stimulation of PLC that can occur under high glucose would lead to prolonged increases in Ins(3,4,5,6)P₄ and, thus, a longer term impairment of insulin exocytosis.

B. Inositol Hexakisphosphate Regulation of Voltage-Gated Ca²⁺ Channels and Membrane Trafficking

Inositol hexakisphophophate (InsP₆) is the most abundant inositol polyphosphate in a number of cellular systems, ranging from plants to mammalian cells, and it is the major natural organic phosphate in the biosphere (Turner et al., 2002). In this review, we focus particularly on mammalian cells and primarily on pancreatic β-cells, but it is important to recognize the overwhelming presence of InsP₆ in biology. We return to this in the next section.

Studies in the pancreatic β cell exposed one of the first clear physiologic roles for InsP₆ in mammalian cells (Fig. 5). A more detailed description of the role of InsP₆ in β-cells can be found elsewhere (Berggren and Barker, 2008). Here, we describe more briefly the processes in which InsP₆ is involved.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Role of higher inositol phosphates in β-cell function. There are many different arenas in which higher inositol polyphosphates operate, particularly InsP₆ (inositol hexakisphosphate) and InsP₇ (5-diphosphoinositol pentakisphosphate, PP-InsP₅). InsP₆ stimulates L-type VDCC activity, increasing phosphorylation of the channel by both inactivating serine threonine protein phosphatases and stimulating serine threonine kinases (PKC and indirectly PKA through activation of adenylyl cyclase) (Yang et al., 2001). The enhanced Ca²⁺ influx driven by InsP₆ will also increase exocytosis, but InsP₆ can also enhance secretion more directly through the activation of PKCε. InsP₆ also targets dynamin-mediated endocytosis through a complex process that involves the dephosphorylation of dynamin through calcineurin and InsP₆’s ability to both stimulate PKCε and inhibit synaptojanin (SJ), a PtdIns(4,5)P₂ 5-phosphatase. The inhibition of the lipid phosphatase leads to enhanced levels of PtdIns(4,5)P₂, an important element in the endocytotic process. Finally, through IP6K1, InsP₆ is converted to PP-InsP₅ (InsP₇/IP₇), which regulates the cell's exocytotic capacity at the level of the readily releasable pool of vesicles (RRP).

When considering functions for InsP₆ in the wake of the Ca²⁺-mobilizing second messenger, Ins(1,4,5)P₃, it was natural to consider Ca²⁺ regulation as a putative target area, especially because it was a known Ca²⁺ chelator (Gersonde and Weiner, 1981). In the β-cell, in keeping with many other neuroendocrine or neuronal cells, mobilization of Ca²⁺ from intracellular stores plays a more secondary role to influx of Ca²⁺ through PM-standing, voltage-dependent Ca²⁺ channels (Yang and Berggren, 2005). In β-cells, these channels are physically coupled to the exocytotic machinery (Yang et al., 1999) and are responsible for providing the Ca²⁺ that drives the final steps of exocytosis and, thus, insulin secretion (Yang and Berggren, 2005). In an article in Science (Larsson et al., 1997), it was established that InsP₆ enhanced L-type Ca²⁺ channel activity and that, although other pentakisphosphates could also mediate this effect, they were less potent than was InsP₆. Two other studies, one on hippocampal neurons (Yang et al., 2001) and one on vascular smooth muscle cells (Quignard et al., 2003), also showed the enhancing effect of InsP₆ on L-type voltage-dependent Ca²⁺ channels, although in the smooth muscle cells, there was not a marked difference between the potency of InsP₆ and that of Ins(1,3,4,5,6)P₅. The L-type voltage-dependent Ca²⁺ channel is controlled through protein phosphorylation; in β-cells, a mechanism of action of InsP₆ was to inhibit serine threonine phosphatase activity, and thus, a likely mechanism of action would be to promote channel activity through increased phosphorylation (Larsson et al., 1997). The studies on hippocampal neurons showed that InsP₆ could regulate channel phosphorylation, not only by inhibition of serine threonine phosphatases, but also by the activation of PKA. This, in turn, was an indirect effect of increasing cAMP by the stimulation of adenylate cyclase activity (Yang et al., 2001). In studies in the β-cell, it was shown that PKC isoforms can also be activated by InsP₆ (Efanov et al., 1997; Hoy et al., 2003). Thus, InsP₆ regulates L-type voltage-dependent Ca²⁺ channel phosphorylation both by the inhibition of serine threonine phosphatases and by the activation of serine threonine kinases, namely PKA and PKC.

Because of the high basal InsP₆ concentration in β-cells (50–60 µM) (Larsson et al., 1997), this inositol phosphate could serve to establish the set point of channel phosphorylation and, thus, its activity. Glucose stimulation further increased InsP₆ global concentrations; this increase was modest (10–15%), but it occurred within the period in which a large increase in [Ca²⁺]_i can be observed. To have a functional significance, this small increase in InsP₆ concentration needs to occur in a microdomain close to the channel. Although there are precedents for other β-cell channels, such as the K_ATP channel, existing in microdomains, there is no direct evidence that InsP₆ is associated with such a domain. In other cell systems, however, there is evidence for a pool of InsP₆ associating with the plasma membrane (Poyner et al., 1993), which would represent a separate compartment. In such a plasma membrane-associated pool, it is possible that the small changes in InsP₆ could have physiologic significance to adjacent channels. Compartmentalization of InsP₆ is also indicated by recent work showing that the Ins(1,3,4,5,6)P₅-2-kinase, responsible for its synthesis, can have distinct cellular localizations. These are in the nucleus (Brehm et al., 2007) and in cilia structures (Sarmah et al., 2007).

As we have stated above, the Ca²⁺ channel is tightly coupled to the exocytotic machinery, which suggests that InsP₆, at least indirectly, could be coupled to exocytosis. However, work on neuronal cells had indicated a negative effect of InsP₆ on neurotransmission by binding to the key exocytotic protein synaptotagmin (Llinas et al., 1994), with the caveat that this effect required a rather high InsP₆ concentration. Nonetheless, there are significant differences between synaptic vesicle fusion events and the large dense core vesicle fusions that underscore insulin secretion. Therefore, in contrast to the neuronal systems, investigations in permeablized insulin-secreting cells (Efanov et al., 1997) revealed that InsP₆ could stimulate Ca²⁺-independent exocytosis and prime Ca²⁺-dependent exocytosis, possibly through a PKC- or a PKC-related kinase mechanism. Additional studies using electrophysiological techniques, which exploit an increase in cell capacitance as an index of secretion, showed that InsP₆ enhanced glucose-stimulated insulin secretion through activation of PKC epsilon (Hoy et al., 2003), an isoform believed to be associated with insulin secretory vesicles (Mendez et al., 2003). These electrophysiological studies also revealed another facet of the role of InsP₆ in vesicle trafficking. Initial experiments using the capacitance technique showed that InsP₆ decreased rather than increased exocytosis. However, with use of a technique that simply reports changes in cell surface area, the final measurement was found to be a composite of exo- and endocytosis. In fact, it was revealed that InsP₆ could drive endocytosis more efficiently than could exocytosis (Hoy et al., 2002); thus, when endocytosis was inhibited, InsP₆ exocytotic function was exposed. This led to a more detailed study of the role of InsP₆ in β-cell endocytosis. InsP₆ targets dynamin-mediated endocytosis through a complex process that involves the dephosphorylation of dynamin through calcineurin and the ability of InsP₆ to both stimulate PKC and inhibit synaptojanin, a PtdIns(4,5)P₂ 5-phosphatase. The inhibition of the lipid phosphatase led to enhanced levels of PtdIns(4,5)P₂, an important element in the endocytotic process (Hoy et al., 2002). These studies expose a novel cross-talk between inositol polyphosphate and lipid metabolism.

C. Novel Roles of InsP₆ Outside the β-Cell

When considering the role of InsP₆ outside the β-cell, a large number of functional roles have been suggested, especially in mammalian cells (Irvine and Schell, 2001; Irvine, 2005). This vast number of roles provoked Shears to entitle a review on the subject “Assessing the Omnipotence of Inositol Hexakisphosphate” (Shears, 2001). In this review, Shears eliminated many of the putative roles on the basis of a rigorous application of logic and a careful examination of the physiologic relevance of the conditions in which InsP₆ was tested. We mainly focus on those roles that were not eliminated by this review. We should clarify that the importance of InsP₆ stems both from its own role and as a platform from which other regulatory molecules are derived. Examples of the latter include the inositol pyrophosphates (Barker et al., 2009b; Shears et al., 2011; Saiardi, 2012) (see section IV) or the Ins(1,2,3)P₃-based inositol polyphosphates with iron-binding properties (Barker et al., 1995; Spiers et al., 1996; Irvine, 2005). In terms of the direct effects of the molecule, two or three different modes of operation for InsP₆ can be identified. First, insights from nutritionists clearly indicate its importance as a metal ion chelator, especially of iron (Zhou and Erdman, 1995). This continues to be the case for human nutrition, although a particular role in intracellular iron metabolism has now been challenged (Torres et al., 2005). Second, InsP₆ acts as an allosteric regulator of specific proteins, and third, in a related but not identical role, it also acts as a structural component of some important enzymes.

The role of InsP₆ as a chelator in iron metabolism and as an antioxidant (Graf et al., 1987) has diminished in the light of a study showing that, in the ionic environment of the cytosol, it exists as a neutral pentavalent Mg²⁺ salt that has no capacity to bind iron (Torres et al., 2005). However, it still may fulfill this role in other cellular contexts in vesicles or other subcellular organelles and more widely in nutrition, in which its abundance and chelating properties are known to cause malnutrition in developing countries (Zhou and Erdman, 1995). Furthermore, the InsP₆ breakdown product Ins(1,2,3)P₃ (Barker et al., 1995) has now been subject to the same rigorous assessment as InsP₆ (Veiga et al., 2009) but, in contrast, remains one of the best candidates in the mammalian cell for both complexing iron and preventing iron-catalyzed free radical cascades, acting therefore as a potent antioxidant. The importance of InsP₆ as an antioxidant then becomes as the reservoir from which Ins(1,2,3)P₃ is derived.

As mentioned in the introductory section, studies on yeast first exposed a fundamental role for InsP₆ in the context of the nucleus. This novel function was the export of mRNA from the nucleus to the cytoplasm (York, 2006). This was established by deletion experiments that systematically removed combinations of plc1, ipk2, and ipk1, the yeast homologs of PLC, inositol polyphosphate multikinase (IPMK or Ipk2), and Ins(1,3,4,5,6)P₅ 2-kinase (or Ipk1), respectively (Fig. 1). InsP₆ lies at the end point of the pathway controlled by these enzymes (Fig. 1) (York et al., 1999). Subsequent studies have revealed that InsP₆ specifically binds to the C-terminal of Gle1, a nuclear-pore associated protein (Folkmann et al., 2011). This interaction leads to increased activation of Dbp5, a highly-conserved ATPase, which is thought to use its ATPase activity to remodel RNA. The constitutive activity of Dbp5 is not sufficient to drive mRNA export; this apparently requires the InsP₆/Gle1 complex. More recently, the InsP₆/Gle1 activation of Dbp5 has been shown to operate during another aspect of mRNA processing, namely, in translational termination (Alcazar-Roman et al., 2010).

Continuing on this mRNA-based theme, a third aspect of the mode of operation of InsP₆ occurs in mRNA editing. During the determination of the crystal structure of the RNA editing enzyme ADAR2, InsP₆ was found to be present and tightly bound to the enzyme. Furthermore, InsP₆ was a necessary cofactor for enzyme activity (Macbeth et al., 2005). This discovery is reminiscent of an earlier study in which InsP₆ was identified as a factor required for the activity of DNA-dependent protein kinase (DNA-PK) (Hanakahi et al., 2000), although a later work (Ma and Lieber, 2002) indicated that subunits of this kinase complex (Ku70/80) were the actual binding sites of InsP₆. Of interest, there is some resemblance between the binding domain proffered by the Ku70/80 complex and ADAR2. It is possible, as other structures are resolved, that InsP₆ may arise again in interesting but unexpected places. An important process that InsP₆ mediates through its binding to the DNA-PK complex is nonhomologous end joining (Cheung et al., 2008), a process that repairs double DNA strand breaks caused by genotoxic agents. This repair process maintains genomic stability and cell viability. This last section highlights the fact that many of the functions of InsP₆ outside the realm of the pancreatic β-cell focus on nuclear-related processes.

One extra-nuclear role for InsP₆ that has emerged during the final preparation of this article is in the regulation of synaptic transmission in hippocampal neurons (Yang et al., 2012). In neurons, in contrast to pancreatic β-cells (Efanov et al., 1997; Høy et al., 2003), InsP₆ has been reported to inhibit exocytotic events, and this was suggested to occur through synaptotagmin (Llinás et al., 1994). The new data show, at the molecular level, that InsP₆ dose-dependently inhibits hippocampal autaptic excitatory postsynaptic currents. It does this by the inhibition of the synaptotagmin-1 C2B domain-mediated fusion by interacting with the synaptotagmin Ca²⁺-binding sites. Thus, the elevation of InsP₆ levels in activated neurons mediates a unique negative feedback signal that controls hippocampal excitatory transmission.

A. The Mechanism of Action of Pyrophosphates

Studies performed in the past eight years have revealed that inositol pyrophosphates may exert their influence by two modes of action, one based on classic allosteric interactions and the second by a novel phosphate transfer reaction. In both cases, aspects of the work have been challenged, although there is clearly enough data now to assert that the suggested two modes of action need to be taken seriously.

Because of the doubts now expressed on the physiologic relevance of the binding of inositol pyrophosphates to mammalian adapter protein modules (Shears, 2001; Shears et al., 2012) (see section IV.B), the first clear example of allosteric modulation exerted by inositol pyrophosphates comes from yeast (Lee et al., 2007). However, this inositol pyrophosphate is not the dominant inositol 5-PP-InsP₅ in mammalian cells (formed by the IP6Ks), but 1-PP-InsP₅, formed via PPIP5Ks and is a less abundant variant. It is important to reinforce here an earlier remark. In the studies on yeast, the novel PP-InsP₅ isomer is referred to as being one of the 4/6-PP-InsP₅ enantiomeric pair (Lee et al., 2007). This is incorrect. In this review, we use the correct structural allocation. 1-PP-InsP₅ is associated with the regulation of the Pho complex, an assembly of cyclin-related proteins consisting of a kinase, a cyclin, and an inhibitory protein, Pho81. This complex is an established regulator of yeast phosphate homeostasis.1-PP-InsP₅ strengthens the inhibitory action of Pho81. The interpretation of some of these data has been challenged, especially in the integration between 1-PP-InsP₅ concentrations and changes in phosphate levels (Onnebo and Saiardi, 2007). Nonetheless, the binding shows specificity, because the common 5-PP-InsP₅ does not significantly bind, and neither does InsP₆.

Within the past couple of years, another example of inositol pyrophosphate-mediated regulation has been proposed. It has been suggested that 5-PP-InsP₅ acts to inhibit the key regulatory protein Akt/PKB (Chakraborty et al., 2010). This interaction must be subject to careful scrutiny, because other inositol phosphates existing at much high concentrations than the pyrophosphates have also been suggested to inhibit Akt/PKB [e.g., Ins(1,3,4,5,6)P₅] (Piccolo et al., 2004). Furthermore, even the binding of 5-PP-InsP₅ to this protein has been disputed (Downes et al., 2005). However, the authors of this new work are not just proposing a direct effect on Akt/PKB, but a more subtle interference of the interaction between Akt/PKB and the regulatory protein PDK1 (Chakraborty et al., 2010). We have debated these complexities in more detail elsewhere (Barker et al., 2009b). It is possible to combine the previous negative data with the current positive data when considering the different conditions in which the binding studies were performed. In summary, the central issue is that the prebinding of PtdIns(3,4,5)P₃ to Akt/PKB prevents 5-PP-InsP₅ binding. Therefore, in studies such as those performed by the Downes laboratory (Downes et al., 2005), in which displacement of phosphatidylinositol 3,4,5-trisphosphate [PtdIns(3,4,5)P₃] was used to quantify 5-PP-InsP_5-binding, 5-PP-InsP₅ appears to bind Akt/PKB poorly. However, in the studies from the Snyder group and others, in which PtdIns(3,4,5)P₃ is not prebound, 5-PP-InsP₅ binds and inhibits Akt/PKB (Barker et al., 2009b). Therefore, we remain open to the possibility that these initial studies will be supported by future experimental investigations.

The background to the proposition that 5-PP-InsP₅ regulates Akt/PKB signaling arose from a study of the phenotype of a mouse in which IP6K1 has been globally knocked out (Bhandari et al., 2008). This mouse exhibits a striking increase in insulin sensitivity and a resistance to obesity, which is clearly consistent with enhanced Akt/PKB signaling, but other explanations are also possible (see below) (Chakraborty et al., 2010). Moreover, it is not clear whether this regulation is general to all Akt/PKB isoforms or specific to one. Verification of this is important, especially in the pancreatic β-cell, in which different Akt/PKB isoforms exert control over distinct signaling pathways (Buzzi et al., 2010; Leibiger et al., 2010). We shall return to this in the context of the role of InsP₇ in exocytosis. Of significance, this inhibition is specific to the abundant 5-InsP₇ isoform (the 1-PP-InsP₅ isomer is ineffective) and, thus, shows stereo-selectivity, an important prerequisite for allostearic regulation. It also highlights a role for IP6Ks rather than PPIP5Ks in these events.

The second method of action of inositol pyrophosphates is through a so-called pyrophosphorylation reaction. In this case, a high-energy phosphate from the inositol pyrophosphate, rather than ATP, is used to posttranslationally modify a specific subset of proteins that are not regulated by the classic ATP-dependent phosphate donation. Initially, this was suggested to be a kinase-independent event (Saiardi et al., 2004), but now it is clear that prephosphorylation by CK2 is required (Bhandari et al., 2007). Therefore, the phosphorylation by pyrophosphates is not a classic phosphorylation but a pyrophosphorylation on an already phosphorylated serine residue (Bhandari et al., 2007). It is widely accepted that this pyrophosphorylation can occur in an in vitro setting, but its role in the intact cellular environment is more contentious (Shears et al., 2011). Hurdles that have to be overcome include the fact that InsP₆, which is generally one of the most abundant inositol polyphosphates in mammalian cells, inhibits the reaction. The current defense of this discrepancy has involved cellular compartmentalization, and although this may not be accurate, the compartmentalization of the Ins(1,3,4,5,6)P₅ 2-kinase, exclusively responsible for InsP₆ synthesis and that we have mentioned in the section on InsP₆, gives some substance to segregation (Brehm et al., 2007;Sarmah and Wente, 2009).

Since the original articles describing and refining the nature of the pyrophosphorylation (Saiardi et al., 2004; Bhandari et al., 2007), there have been a recent study that highlights the functional involvement of this novel posttranslational modification in an important cellular process and another study that questions it (Azevedo et al., 2009). On the positive side, pyrophosphorylation of the β3A subunit of the AP3 complex, when bound to kinesin, promotes a dissociation that prevents AP3-mediated HIV Gag protein trafficking (Azevedo et al., 2009). In this latter study, a back phosphorylation protocol was used to better verify the relevance of the pyrophosphorylation in intact cells. However, other interpretations are possible (Shears et al., 2011). On the negative side, another laboratory, using an admittedly indirect method, were not able to find an association between pyrophosphorylation and function (Yang et al., 2008). This study relates to the human homolog of Nsr1 or nucleolin, which is pyrophosphorylated in yeast (Saiardi et al., 2004). The authors were unable to correlate assumed pyrophosphorylation of this protein with levels of inositol pyrophosphates that might be predicted to influence it (Yang et al., 2008), although it is possible that, in the different biologic systems (yeast and mammalian), other factors come into play. For example, the authors studying the human protein used the same kind of ploy that had been exploited in the yeast system, but without being able to establish a clear linkage (Yang et al., 2008). For this novel posttranslational modification to gain broader acceptance, further independent verification would be helpful. Because data on β-cell exocytosis, discussed in the next section, seemed not to select between PP-InsP₅ isoforms (Illies et al., 2007), there was an indirect indication that pyrophosphorylation may play a role in these cells. The reason for this is that pyrophosphorylation can be performed by a variety of inositol pyrophosphates without much selectivity between them. Although we have looked carefully, we have yet to produce convincing evidence that pyrophosphorylation occurs in β-cells (C. Illies, P.-O. Berggren, and C. J. Barker, unpublished results). However, of importance, the current strategy to identify new pyrophosphorylated proteins is through a genomic rather than proteomic approaches (Werner et al., 2010). This tactic may be driven by the inability of current mass spectrometry methods to reliably identify the pyrophosphorylated motif in a protein (Shears et al., 2011).

B. Inositol Pyrophosphates as Regulators of Exocytosis

The observation of the higher levels of 5-PP-InsP₅ in insulin-secreting cells, mentioned in the introduction to Section IV, was also confirmed in primary islets and β-cells (Illies et al., 2007), which paved the way for a detailed investigation of a potential role of this pyrophosphate in the β-cell. When we consider a role for inositol pyrophosphates in pancreatic β-cells, we turn to exocytosis. There has been an indication from the work of Snyder and colleagues that one of the kinases involved in 5-PP-InsP₅ production, IP6K1, is involved in neuronal exocytosis (Luo et al., 2001). However, this was independent of any catalytic activity and, thus, not related to the generation of inositol pyrophosphates. Earlier studies have suggested the involvement of InsP₇ in vesicle trafficking, related to its high affinity for subunits of the AP2 and AP180 adaptor proteins associated with the formation of clathrin coated vesicles, but as we have discussed above, it is now thought to be unlikely that this has a physiologic context (Shears, 2001). Deletion of IP6Ks in yeast led to disruption of intracellular membrane compartments. However, in mammalian cells, inositol pyrophosphates had no established track record in influencing vesicle trafficking or the resultant exocytotic events.

It was established, through electrophysiological experiments measuring membrane capacitance, that 5-PP-InsP₅ was important for the filling of the RRP of vesicles and, thus, controlled the exocytotic potential of the β-cell (Illies et al., 2007). The effect of InsP₇ was dose-dependent in the physiologic concentration range of the pyrophosphate found in mammalian cells. Although overexpression of either of the IP6Ks (IP6K1 and 2) found in β-cells increased the RRP, only the siRNA knockdown of IP6K1 had any effect on exocytosis. These data hint at a spatial segregation in 5-PP-InsP₅ production, which, as we have indicated earlier, has several precedents in β-cell stimulus-secretion coupling. It has previously been shown that InsP₆ can promote exocytosis in pancreatic β-cells (Efanov et al., 1997; Høy et al., 2003), but the effect was not restricted to the RRP-like 5-PP-InsP₅ (Illies et al., 2007). Thus, a clear delineation can be made between the effects of InsP₆ and 5-PP-InsP₅ on exocytosis, and this suggests that they act through independent mechanisms.

From a physiologic perspective, the RRP has been suggested to reflect the first phase of insulin secretion after glucose stimulation (Rorsman and Renstrom, 2003). A caveat needs to be added about the RRP pool, because it is an empirical definition based on insights from electrophysiology. Recent insights using alternative techniques may give a different picture. The most prominent new approach is based on the total internal reflection microscopy technique, which exploits an evanescent wave to selectively follow events at or just below the plasma membrane. This procedure has suggested that the electrophysiological RRP, based on direct membrane depolarization, causes the fusion of a pool of vesicles that hardly overlaps with the pool of vesicles that glucose stimulation alone can drive to fusion (Seino et al., 2011). In the case of the data published in the article in Science on the effect of 5-PP-InsP on the RRP (Illies et al., 2007), both electrophysiological and indirect measurements of insulin secretion were used to draw the conclusions. Furthermore, we have recently developed a direct dynamic insulin release assay, and silencing IP6K1 in this system reduces the first phase of insulin release (G. C. Gaboardi, P. O. Berggren, and C. J. Barker, unpublished data). Because this phase is thought to reflect the liberation of the RRP, we can conclude that 5-PP-InsP₅ also regulates the glucose-induced RRP.

The report that 5-PP-InsP₅ and, thus, IP6K1 is important for insulin secretion is backed by two further pieces of evidence. First, a genetic linkage study in a Japanese family with T2D found a putative disruption of the IP6K1 gene (Kamimura et al., 2004), and second, in an article by Snyder and colleagues, in which they made a general knockout of IP6K1 in a mouse, showed that it had decreased insulin in the blood, consistent with our studies on β-cell insulin secretion (Bhandari et al., 2008). The IP6K1 knockout mouse did, however, expose a paradox that may also have a bearing on pancreatic β-cell regulation by pyrophosphates.

Despite the lower level of insulin secretion, the IP6K1 knockout mouse was not diabetic and responded normally to a glucose tolerance test (Bhandari et al., 2008). The explanation for this, as suggested above, was exposed after additional studies (Chakraborty et al., 2010) that revealed that Akt/PKB signaling was enhanced, and this leads naturally to an increased insulin sensitivity. This improved sensitivity was more than able to compensate for the β-cell defect. The mouse is also resistant to the effects of a high-fat diet and is thus a supermouse; however, the actual molecular phenotype at the level of the 5-PP-InsP₅/Akt/PKB interaction needs careful consideration. This stems from the fact that there is at least one alternative molecular explanation for the enhanced Akt/PKB signaling resulting from the elimination of IP6K and, second, that there is doubt in the literature regarding the 5-PP-InsP₅/Akt binding, both of which we have discussed above.

The first caveat arises from a still unexplained phenomenon that is manifested when silencing IP6K in yeast (Saiardi et al., 2000) or in pancreatic β-cells (Barker et al., 2009b; C. J. Barker, C. Illies, and P.-O. Berggren, unpublished observations). This silencing also reduces the level of Ins(1,3,4,5,6)P₅. The effect in yeast is more pronounced, but this may reflect the fact that there is only one IP6K in this organism. Of course, it is possible that it is not Ins(1,3,4,5,6)P₅ that is being reduced, but a yet unidentified InsP₄-based pyrophosphate with the same elution properties. Snyder and colleagues report that this reduced level of the Ins(1,3,4,5,6)P₅ peak is not observed in either brain or embryonic fibroblasts of IP6K1 knockout mice; indeed, there is an increase of Ins(1,3,4,5,6)P₅ in IP6K1 knockout brain (A. Chakraborty and S.H. Snyder, unpublished observations). This suggests that lowering of Ins(1,3,4,5,6)P₅ concentrations after IP6K knockout may be a phenomenon restricted to β-cells. We raise this issue, because Ins(1,3,4,5,6)P₅ is also a known inhibitor of Akt/PKB (Piccolo et al., 2004) (see above), and loss of Ins(1,3,4,5,6)P₅ may produce a phenotype similar as that produced by loss of PP-InsP₅. Another concern with this mouse model is the reported regulation of 5-PP-InsP₅ by insulin. We have not seen this, at least not in short-term experiments. The effects that are reported are from studies on dividing cells that have been serum starved. Under these conditions, cells can lock in mid G1, and addition of insulin probably releases them from this lock. Because it has also been demonstrated that PP-InsP₅ concentrations increase in the cell cycle during the same time frame (Barker et al., 2004), the changes may reflect cell cycle changes rather than the kind of regulation that would be observed in the mature β-cell, with its low rate of cell division. In fact, if cells are held in a serum-starved state, they can rapidly enter cell cycle again, after serum is re-administered; we have observed (C. J. Barker, unpublished data) that, under these conditions, PP-InsP₅ increases fairly rapidly. However, unpublished data from A. Chakraborty and S. H. Snyder (personal communication) suggest that even short (2 hours) serum starvation protocols can reduce PP-InsP₅ levels. Subsequent addition of growth factors seems to then be able to increase PP-InsP₅. With such a short starvation protocol, changes in 5-PP-InsP₅ may not be linked to specific cell cycle–related changes.

Whatever the final outcome of these deliberations, in terms of β-cell regulation, the possibility that IP6Ks, directly or indirectly, are also modulating Akt/PKB, is an important one. Additional studies by the Snyder and colleagues (Prasad et al., 2011) have indicated an enhanced Akt/PKB signaling in neutrophils from IP6K1 knockout mice, reinforcing a likely in vivo role for this 5-PP-InsP₅/Akt interaction.

C. Inositol Hexakisphosphate Kinase as a Metabolic Sensor with a Role in T2D

Glucose-stimulated insulin secretion is biphasic in mammals, including humans. One of the key early defects in T2D is a blunted first phase of secretion (Ahren, 2005). As we have discussed above, the RRP is the major contributor to the first phase of insulin release and 5-PP-InsP₅ is a quantitatively significant regulator of the RRP. These fundamental relationships raise the question of whether 5-PP-InsP₅ and/or IP6K1, the kinase that produces it, have any role in the disease phenotype and its progression. One indication that this might be the case is to consider possible genetic evidence linking IP6K1 to T2D. As we have mentioned above, there was a single report describing a Japanese family in which a putative disruption in the IP6K1 gene was proposed to be linked to the disease (Kamimura et al., 2004). However, T2D is a polygenetic disease with many potential dysfunctional genes, and thus, this single case, although indicating a proof of principle, is not one in which a wider involvement of IP6K1 in T2D can be built. Perhaps a more insightful question is how the 5-PP-InsP₅-generated RRP will be affected in a diabetic milieu. This question leads to the discussion of two fundamental properties of IP6K1 that may make it vulnerable in T2D and that are shown in Fig. 6 and discussed below.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

IP6K1 as a metabolic sensor in physiology and under diabetic conditions. The figure illustrates the central position occupied by IP6K1 in the regulation of the RRP. Because IP6Ks have a millimolar K_m for InsP₆ (Voglmaier et al., 1996; Saiardi et al., 2000), they are potentially sensitive to cellular ATP levels. Thus, reductions in cellular ATP:ADP ratio can lead to significant reductions in PP-InsP₅ (Nagel et al., 2010). The IP6Ks are also sensitive to inhibition by oxidation resulting from a sensitive cysteine in the catalytic site (Onnebo and Saiardi, 2009). Therefore, in the β cell, we propose, along with others (Shears, 2009), that IP6K1 is a metabolic sensor. It is likely to be inhibited by the increased oxidative stress and disrupted energy metabolism that are associated with T2D. Because the RRP is responsible for the first phase of insulin secretion and this is an early sign of T2D, it is possible that loss of first phase may be, at least in part, attributable to the compromise of IP6K1 function.

To set these properties in context, it is important to re-emphasize the tight connection between glucose metabolism, changes in ATP:ADP ratios, and the drive to stimulate insulin release that occurs in pancreatic β-cells. In summary, in the β-cell, increases in the ATP:ADP ratio, instituted by glucose metabolism, leads to closure of K_ATP channels, depolarization of the β-cell and influx of Ca²⁺ through voltage-dependent Ca²⁺ channels, which then drives the final stages of insulin release. Thus, changes in ATP:ADP are integrated into insulin secretion, via the K_ATP channel (Ashcroft and Rorsman, 1990). There are, however, other examples in which changes in the ATP:ADP ratio are coupled to exocytosis. One of these is phosphatidylinositol 4 kinase, which can sense changes in the ATP:ADP ratio (Olsen et al., 2003). An increase in the ATP:ADP ratio after glucose stimulation leads to its activation, a concomitant increase in the production of PtdIns(4,5)P₂, and the promotion of secretion. With these precedents in mind, it is possible that IP6Ks could also be sensitive to ATP:ADP increases. This possibility is likely, because IP6Ks, including IP6K1, have an unusually high (mM) K_m for ATP (Voglmaier et al., 1996; Saiardi et al., 2000). Therefore, the enzymes could be subject to regulation by prevailing millimolar cellular ATP concentrations. Shears, also noting the kinetic properties of IP6Ks with regard to ATP, has already proposed that they may be metabolic sensors (Shears, 2009). The next important question is, can we connect this ATP:ADP sensitivity of the IP6K to diabetes? Disturbances of ATP and energy production in pancreatic β-cells is an established phenomenon in diabetic models (Ma et al., 2012), and as we have stated in the introduction, the initiation of glucose-induced insulin secretion is dependent on changes in the ATP:ADP ratio. 5-PP-InsP₅ levels seem to be exquisitely sensitive to reductions in cellular ATP levels, even more so than the lipids (Nagel et al., 2010). Thus, there could be a direct coupling of disrupted energy metabolism to a reduced activity of IP6K1. The consequence of this would be a reduction in the RRP and a subsequent diminishing of the first phase of insulin secretion.

The second intrinsic property of IP6K1 pertaining to diabetes is that it is sensitive to oxidants and that oxidant stress leads to its depletion (Onnebo and Saiardi, 2009). The sensitivity to oxidants is a direct one based on a vulnerable cysteine in the active site of the enzyme. Oxidative stress is a consequence of the hyperglycemia associated with both types of diabetes. In T2D, in which the β-cell still survives but malfunctions, oxidative stress continues to play an important role in the death of the cells (Rosen et al., 2001).

In summary, IP6K1 may represent a novel link between known metabolic defects in β-cell stimulus-secretion coupling and the diabetic phenotype exhibited by the cell. Therefore, the enhancement of IP6K1 activity under glucose-stimulating conditions could represent an important new concept in diabetes treatment, because it pertains to the preservation of β-cell function.

D. Insights of Inositol Pyrophosphate Function from Other Cell Systems

Although we have covered many of the important aspects of inositol pyrophosphates in the preceding sections, some topics, principally because there is no extant data coming from β-cells, have been ignored. Therefore, in this section, we focus on some of these neglected areas, because we believe that they are likely to be important in our future understanding of the physiology and pathophysiology of the β-cell, especially during the stress associated with diabetes. The first theme is the role of inositol pyrophosphates in apoptosis. In this context, we need to consider the IP6K2 isoform.

As we have suggested above, IP6K2 is involved in apoptosis, both because of its catalytic activity and, thus, its ability to generate inositol pyrophosphates, presumably 5-PP-InsP₅, and its noncatalytic activity (Barker et al., 2009b). This area is confused by the fact that two separate groups have described very different redistributions of IP6K2 during apoptosis. The explanation for these discrepancies may lie in the different pathways by which apoptosis is being induced or other cell type–specific phenomena. Essentially, the initial reports suggested that IP6K2 translocated to the nucleus in IFN-β- (and later γ-radiation) induced apoptosis (Morrison et al., 2001, 2002), whereas studies by another group suggest a normal nuclear localization, followed by a translocation to bax-positive mitochondria after pro-apoptotic stimuli, such as cytotoxic stress (Nagata et al., 2005). In trying to sort out the discrepancy between the work of these two groups, the first thing to note is that IP6K2 has a nuclear localization signal and, thus, has the intrinsic property to localize to the nucleus. This would thus support the idea that IP6K2 can travel to the nucleus. However, it does not exclude the possibility of the second case: that is, after it is in the nucleus, it could leave under the appropriate stimulation. Therefore, it is possible that both sets of data are correct but represent different scenarios occurring in different cell types in which IP6K2 is or is not nuclear resident. One possible mechanism by which a protein with a nuclear localization signal remains resident in the cytoplasm is if it is complexed to another protein that masks its localization signal or in some other way restricts its access to the nucleus. In this regard, it is interesting that IP6K2 has been shown to bind to the heat shock protein HSP90 (Chakraborty et al., 2008). This association occurs in the cytosol and masks the catalytic activity of the IP6K2. Dissociation of IP6K2 from HSP90 occurs during cytotoxic insults, including those generated by anticancer drugs (Chakraborty et al., 2008). It is possible that cytosolic IP6K2 is invisible to antibodies until it is released from its partner, which would be one way of combining the two sets of disparate data. Very recently, insights from studying IP6K2 in lymphoblast cells from patients with Huntingtons’s disease suggest that IP6K2 is in the nucleus under normal conditions, but in the cytosol during the disease state (Nagata et al., 2011). We return to this disease model later, but these new data allow us to speculate that the location of an active IP6K2 in the cytosol, and not the nucleus, may reflect a pathologic phenotype of the ovarian cancer cells used in the original studies (Morrison et al., 2001, 2002) and may be another explanation for the discrepancy between the data from the two groups.

When we considered IP6K-silencing experiments with respect to exocytosis, it was clear that only the IP6K1 had the endogenous capacity to mediate the enhancement of 5-PP-InsP₅ of the RRP, although flooding of the cell with this pyrophosphate by overexpression of any IP6K can also mediate this effect (Illies et al., 2007). This thus raised the issue of compartmentalization. Of interest, when we consider experiments that involve the IP6K2 catalytic activity in driving apoptosis, a similar phenomenon is apparent. That is, overexpression of any IP6K can promote apoptosis, but only silencing of IP6K2 can prevent apoptosis, at least that driven by cytotoxic stress (Nagata et al., 2005). Thus, here too, there is a suggestion of apparent compartmentalization. Another strong indication of the compartmentalization of IP6Ks comes from the phenotypes of the knockout mice in which IP6K1 and IP6K2, respectively, are deleted. We consider these phenotypes in more detail now, because they give an important understanding to the broader role of IP6K2 in cell and whole animal physiology (Bhandari et al., 2008; Morrison et al., 2009).

As we have described in Section IV.B, the phenotype of the IP6K1 knockout mouse is a reduced plasma insulin level (Bhandari et al., 2008), which probably reflects reduced insulin secretion (Illies et al., 2007). It also has a reduced body mass, and the males are infertile. In contrast to the IP6K1 knockout mouse, the IP6K2-knockout mouse has no reduction in plasma insulin levels or body mass; moreover, it is fertile (Morrison et al., 2009). Although these findings reinforce the unique contribution of location of the individual IP6Ks, we need to bear in mind that both kinases also exert effects independently of their catalytic activity. Nonetheless, it is apparent that the generation of inositol pyrophosphates is important in both apoptosis and exocytosis and that this can be independent of any specific protein-protein interactions (Nagata et al., 2005; Illies et al., 2007). Focusing on the unique phenotype of the IP6K2 knockout mouse, the authors were able to demonstrate an increased generation of oral- and esophageal-based squamous cell carcinomas when animals were given a carcinogen in their drinking water (Morrison et al., 2009). Furthermore, fibroblasts from the knockout animals displayed insensitivity to the antiproliferative effect of interferon β, indicating that the phenotypes observed in the previous experiments based on cultured cells can be observed in vivo. Although there is no obvious aspect of the IP6K2 knockout phenotype that suggests an important role of this enzyme in β-cells (i.e., unlike IP6K1, it does not contribute noticeably to insulin secretion), its role in apoptosis should not be neglected. This is because apoptosis in the pancreatic β-cell underscores both T1D and T2D (Thomas et al., 2009).

As we have mentioned above, a new study advocates a correlation between cellular distribution of IP6K2 in lymphoblasts and the HD phenotype (Nagata et al., 2011). In fact, the suggestion is that the residence of IP6K2 in the cytosol, which leads to an increase in basal 5-PP-InsP₅, drives the autophagy phenotype seen in this disease. This is an important observation; however, it raises an issue that may also be significant in diabetes. An early article by the same group had shown that IP6K2 overexpression promotes autophagy (Nagata et al., 2010), and this was based on the ability to generate inositol pyrophosphates. Autophagy is an important catabolic process maintaining proper organelle homeostasis in cells (Szumiel, 2011). Recent work has shown that autophagy is important in the maintenance of proper growth and function of β-cells, indicated by the fact that islets from mice in which autophagy has been disrupted in the β-cells, show reduced β-cell mass and compromised insulin secretion (Hur et al., 2010). Because autophagy is also upregulated in insulin resistance, its induction may also represent a defensive strategy against T2D. IP6K2 is likely to be similarly vulnerable to disruption by ATP/ADP changes and oxidative stress as IP6K1, therefore returning to the model that we proposed in section IV.B; it is obvious that a compromised 5-PP-InsP₅ generation—this time, as a consequence of IP6K2 inactivation—could mediate a reduced autophagy in pancreatic β-cells. This indicates another important process in which compromised inositol pyrophosphate generation could act to promote β-cell dysregulation in T2D. It is possible that the elevated levels of 5-PP-InsP₅ in β-cells do not just underscore the maintenance of the exocytotic potential, but also may help sustain the basal autophagy required for effective β-cell function. One caveat that needs to be raised is that IP6K2 also mediates apoptosis; thus, there are both positive and negative aspects to its activation. Under normal circumstances, the level of 5-PP-InsP₅ that is generated may not be sufficient to be proapoptotic but can still mediate a positive effect on autophagy, thus protecting the β-cell; however, when the cell experiences more severe stress, the proapoptotic dimension may come to the fore.