Understanding inositol pyrophosphate metabolism and function: kinetic characterization of the DIPPs

Abstract

We illuminate inositol pyrophosphate turnover and cell-signaling activities, by showing that regulation of yeast cyclin-kinase by 1-InsP₇ is not conserved for mammalian CDK5, and by kinetically characterizing Ddp1p/DIPP-mediated dephosphorylation of 1-InsP₇, 5-InsP₇ and InsP₈. Each phosphatase exhibited similar K_m values for every substrate (range: 35-148 nM). The rank order of k_cat values (1-InsP₇ > 5-InsP₇ = InsP₈) was identical for each enzyme, although DIPP1 was 10-60 fold more active than DIPP2α/β and DIPP3α/β. We demonstrate InsP₈ dephosphorylation preferentially progresses through 1-InsP₇. Conversely, we conclude that the more metabolically and functionally significant steady-state route of InsP₈ synthesis proceeds via 5-InsP₇.

1. Introduction

Among the many members of the inositol phosphate signaling family, the diphosphoinositol polyphosphates (inositol pyrophosphates;1-InsP₇, 5-InsP₇ and InsP₈) receive particular attention. These “high-energy” signals operate at the interface of cell signaling and metabolic homeostasis [1,2], by phosphorylating proteins [3] (but see [4]), by regulating transcriptional responses to environmental stress [5-7], and by modulating PtdIns(3,4,5)P₃-signaling [8,9].

Understanding metabolic regulation of inositol pyrophosphate turnover reveals how this signaling cascade is controlled. Indeed, no metabolic pathways can be modeled accurately without knowledge of the kinetic parameters of the participating enzymes [10]. Inositol pyrophosphate metabolism is complicated by the two kinase pathways from InsP₆ to InsP₈ (I and II; ref [11] and Fig. 1), and by diphosphoinositol polyphosphate phosphohydrolases (DIPPs). Another confounding factor is that humans express five DIPP isoforms: types 1, 2α, 2β, 3α and 3β [12-15] S. cerevisiae only express one: the diadenosine and diphosphoinositol phosphohydrolase (Ddp1p) [16]. Little is known concerning the Michaelis-Menten kinetic parameters for Ddp1p/DIPPs; no such data are published for 1-InsP₇ and InsP₈. While some kinetic data have been provided for 5-InsP₇, their reliability is now being questioned. For instance, K_m values for 5-InsP₇ vary from 4 nM for DIPP1 [12] to 4 μM for DIPP3β [15]. Such variation seems inconsistent with the high conservation of DIPP's catalytic domain. Instead, recent improvements in enzymatic [17] and chemical [18,19] methods for synthesizing inositol pyrophosphates have called into question the quality of some early sources of 5-InsP₇ (see [18]). Here, using improved methods for the enzymatic synthesis and electrophoretic purification of 1-InsP₇, 5-InsP₇ and InsP₈ [17], we have kinetically characterized the DIPP family.

Figure 1. Metabolic interconversions of InsP₆, InsP₇and InsP₈.

DIPP = Diphosphoinositol polyphosphate phosphohydrolase (Ddp1p in yeast); IP6K = inositol hexakisphosphate kinase (Kcs1p in yeast); PPIP5K = Diphosphoinositol polyphosphate kinase (Vip1p in yeast). Pathways I and II are named as in [11].

2. Methods and Materials

2.1 Materials

GST-DIPP3α and GST-DIPP3β were prepared as described [15]. GST-DIPPs 1, 2α and 2β plasmids [12,13] were subcloned from pQE30 into pGEX6P-1 using BamHI and SalI. Expression plasmids were transformed into E. coli (BL21), and induced at 26°C overnight with 100 μM IPTG. Cells were harvested and sonicated in 20 mM Tris, 150 mM NaCl, 2 mM DTT, 0.1 mM EGTA, pH 7.5, 5 μg/ml leupeptin and 1 μg/ml aprotinin. GST-DIPPs were then purified using a Glutathione Sepharose-4 Fast-Flow column (Amersham Pharmacia Biotech). Protein was stored at −80°C in 10% v/v glycerol. Professor A.G. McLennan kindly provided Ddp1p.

The CDK5RAP1 open-reading frame was PCR-amplified from IMAGE clone 4418659, subcloned into pENTR/TEV/D-TOPO, and further subcloned into pDest606 in E. coli (DH5α). The CDK5RAP1-pDest606 construct was transformed into E. coli DH10Bac cells, yielding CDK5RAP1/bMON14272 bacmid; this was amplified by successive infections of Sf9 cells, then used to infect Sf9 cells for protein expression. After 72 hr., cells were lysed in I-Per (Pierce) plus protease inhibitor cocktail (Roche). CDK5RAP1 was purified using Ni-NTA resin and then an amylose resin column. Protein was stored at −80°C.

InsP₆ was obtained from Calbiochem. Enzymatically-prepared inositol pyrophosphates were electrophoretically purified [17,20]. Samples from “blank” gels (i.e. no inositol pyrophosphates), processed in parallel, provided vehicle controls. Purity (by HPLC) of 5-InsP₇, 1-InsP₇, and InsP₈ was, respectively, 86%, 95% and 94%. Such slight decomposition of the pyrophosphate groups does not significantly affect assays of DIPP activity [19]. Assays were corrected for the “zero-time” products.

2.2 Assays

Phosphatase activity was determined at 30°C (Ddp1p) or 37°C (DIPPs) in 20 μl of buffer containing 50 mM KCl, 50 mM HEPES (pH 7.2 with KOH), 4 mM CHAPS, 0.05 mg/ml BSA, 1 mM Na₂EDTA, 2 mM MgSO₄. [³H]labeled and non-radiolabeled InsP₇ or InsP₈ [20] were added as indicated (see Figures). After 2-25 min (corresponding to approx. 20% substrate metabolism), reactions were quenched and neutralized [14], loaded onto a HiChrom 4.6 × 125 mm Partisphere SAX column, and eluted (1 ml/min) with one of two protocols generated by mixing Buffer A (1 mM Na₂EDTA) with buffer B [A + 1.3 M (NH₄)₂HPO₄, pH 3.85 with H₃PO₄ ]: protocol 1), 0–5 min, B=0%; 5–10 min, B=0–45%; 10–55 min, B=45-100% (1ml fractions); protocol 2), 0–5 min, B=0%; 5–10 min, B=0–50%; 10–55 min, B=50-100% (0.5ml fractions). [³H] was assayed by liquid scintillation spectrometry. K_m and k_cat values were obtained by non-linear regression.

CDK5 activity was measured after preincubating (30 min) 50 ng CDK5/p35 (Sigma) and 20 ng CDK5RAP1 in 45 μl reactions containing 20 mM MOPS pH 7.4, 1 mM EGTA, 20 mM MgCl₂, 1 mM dithiothreitol, 100 μM (γ-³²P)ATP (0.05 μCi) and the appropriate inositol phosphate. Reactions were initiated with 5 μl of 1 mg/ml peptide substrate (BioMol International). After 30 min, 40 μl of each reaction was quenched by spotting onto P81 phospho-cellulose filters (Millipore). These were washed (4×; 0.75% H₃PO₄) and acetone-rinsed. [³²P] was assayed by Cerenkov counting.

3. Results and Discussion

3.1 Ddp1p kinetics

By HPLC analysis of the dephosphorylation of [³H]-radiolabeled substrates, we have kinetically characterized recombinant Ddp1p from S. cerevisiae. Ddp1p hydrolyzed 1-InsP₇ and 5-InsP₇ to InsP₆ (Fig 2A,B). InsP₈ was dephosphorylated to InsP₆, with relatively little accumulation of InsP₇ (Fig. 2C, and see below). Substrate saturation plots (Fig. 2) showed that K_m values for each substrate were very similar (93nM, 105 nM, 148 nM; Table 1), but 1-InsP₇ was hydrolyzed at a 5 to 6-fold faster rate than 5-InsP₇ and InsP₈ (Table 1).

Figure 2. Analysis of the catalytic activities of Ddp1p.

HPLC was used to obtain substrate saturation plots for Ddp1p against either A, 1-InsP₇, (0-1024 nM incubated with 2 ng Ddp1p for 0-25 min); B, 5-InsP₇ (0-512 nM incubated with 8 ng Ddp1p for 0-25 minutes); or C, InsP₈, (0-512 nM incubated with 8 ng Ddp1p for 0-25 min). Insets provide illustrative HPLC analyses using protocol 1 (Section 2.2): A, 10 nM 1-InsP₇, 20 ng Ddp1p, 10 min (not used in the substrate saturation plot). B 10 nM 5-InsP₇, 10 ng Ddp1p, 6 min (not used in the substrate saturation plot). C, 16nM InsP₈, 8ng Ddp1p, 4 min.

Table 1. Ddp1p/DIPP kinetic data.

	1-InsP7		5-InsP7		InsP8
	Km (nM)	102 × kcat (s-1)	Km (nM)	102 × kcat (s-1)	Km (nM)	102 × kcat (s-1)
Ddp1p	105	2.4 ± 0.05	93	0.4 ± 0.1	148	0.5 ± 0.1
DIPP1	42	110 ± 30	52	13 ± 3	85	10 ± 2
DIPP2α	60	5 ± 2	35	0.7 ± 0.2	55	0.24 ± 0.08
DIPP2β	70	1.7 ± 0.5	40	0.3 ± 0.09	42	0.16 ± 0.02
DIPP3α	104	23 ± 2	146	4 ± 2	126	2.2 ± 0.4
DIPP3β	73	8 ± 2	63	0.88 ± 0.07	78	0.37 ± 0.06

Kinetic data were obtained as described in Section 2. Average (n=3-5) substrate affinities were first compiled as -log K_m (SEMs were <3% of the mean) and then transformed to K_m for the Table.

No previous studies have described Michaelis-Menten kinetic parameters for Ddp1p activity towards either 1-InsP₇ or InsP₈. There is one earlier study [21] that incubated Ddp1p with 250 μM of either 1-InsP₇ or 5-InsP₇; in an electrophoretic analysis, only 1-InsP₇ was found to be dephosphorylated. That inability to detect 5-InsP₇ metabolism may have reflected its assay concentration exceeding the K_m value by 3 orders of magnitude (see Table 1). Also, SDS-PAGE mass analysis is less sensitive than is HPLC analysis of [³H]-labeled substrates. So although the preference of Ddp1p for 1-InsP₇ over 5-InsP₇ was previously known [21], our new kinetic data indicate the relative reaction rates are closer than previously suggested.

3.2 Human DIPP kinetics

We obtained substrate saturation plots (not shown) to derive K_m and k_cat values for all five human DIPPs (Table 1). DIPP1 exhibited the highest k_cat, irrespective of substrate (Table 1). Thus, for example, the relative levels of cellular expression of DIPP1, versus the 20 to 60-fold less active DIPP2 isoforms, could dictate the rapidity of inositol pyrophosphate turnover on a cell-to-cell basis. This in turn could influence the sensitivity with which levels of inositol pyrophosphates respond to appropriate stimuli. This is analogous to how differential expression of cAMP-phosphodiesterases isoforms influence cell-to-cell differences in the sensitivity of agonist-mediated cAMP signaling [22].

All human DIPPs behaved similarly to Ddp1p in exhibiting higher k_cat values for 1-InsP₇ compared to 5-InsP₇ and InsP₈ (Table 1). The K_m data only vary over a 4-fold range (35 to 148 nM), across every enzyme and every substrate. The demonstration that rates of InsP₇ dephosphorylation equal or exceed those of InsP₈ metabolism (Table 1) explains why so little InsP₇ accumulated when InsP₈ was incubated with DIPPs (Fig. 3); the InsP₇ was rapidly converted to InsP₆.

Figure 3. HPLC analyses of the hydrolysis of InsP₈ by DIPP2β and DIPP3β.

HPLC analyses (protocol 1; Section 2.2) of the metabolism (filled circles) at 37°C of 16 nM InsP₈ incubated with (A) 10 ng DIPP2β for 6 min, or (B) 22.5 ng DIPP3β for 2 min. Open circles depict zero-time controls.

Our new kinetic data yield further conclusions: since cellular levels of 5-InsP₇ are about 1-2 μM [15,23], these can now be considered sufficient (see Table 1) to be at a saturating concentration for all DIPPs. The k_cat values for 5-InsP₇ dephosphorylation are 5 to 9-fold lower than those for 1-InsP₇ (Table 1). However, the affinities of the two InsP₇ isomers for DIPPs are similar (Table 1), so they will compete for the DIPP active sites in direct proportion to their cellular concentration ratios. In mammalian cells, steady-state levels of 1-InsP₇ are10 to15-fold lower than those of 5-InsP₇ [24]. Thus, in vivo it is unlikely that the steady-state rate of 1-InsP₇ dephosphorylation will be significantly greater than that for 5-InsP₇. The latter conclusion - which can only be made now that kinetic parameters are available (Table 1) - counters a recent proposal [21] that, in vivo, a preferential DIPP activity towards 1-InsP₇ “masks” its rate of kinase-mediated synthesis relative to that for 5-InsP₇.

3.3 Inositol pyrophosphates do not regulate the human homologue of the yeast Pho80/81/85 cyclin kinase complex

The debate over the relative functional significance of pathway I and II (see above) has been influenced by the observation [5] that, in yeast, 1-InsP₇ reversibly empowers Pho81 to inhibit the Pho80/85 cyclin kinase complex. This remains the only demonstration to date that 1-InsP₇ serves a unique biological function. The mammalian homologs of Pho85 (the kinase), Pho80 (the kinase activator) and Pho81 (the kinase inhibitor) are, respectively, CDK5, p35 and CDK5RAP1 (originally: C42) [25,26]. However, unlike Pho81, there is no evidence recombinant CDK5RAP1 requires ancillary factors for inhibiting cyclin kinase activity [27].

In earlier experiments [6], 1-InsP₇ augmented inhibition of Pho85 by Pho81 when the latter was added at a molar ratio with Pho85 that was 20-30 fold lower that which, by itself, inhibited Pho85. In our CDK5 assays, we increased assay sensitivity by having CDK5RAP1 at the threshold level at which its own inhibitory effects can be detected (Fig. 4). We then added 4 μM 1-InsP₇, a concentration that is >20-fold higher than is physiological (see above); CDK5 activity was unaffected (Fig. 4B); InsP₆, 5-InsP₇ and InsP₈ were similarly ineffective (i.e. p>0.05 vs controls) (Fig. 4B). These data provide the first evidence that specific regulation of yeast cyclin kinase activity by 1-InsP₇ [5] is not conserved in mammals.

Figure 4. Inositol pyrophosphates do not regulate CDK5 activity.

Panel A describes the concentration-dependent inhibition of CDK5 activity by CDK5RAP1, measured as described in Section 2. In panel B the molar ratio of CDK5RAP1:CDK5 was 0.4:1. None of the added inositol phosphates (4 μM) affected CDK5 activity (p>0.05 vs vehicle control; analysis by ANOVA with a Dunnett's post hoc test, n=5-8). Further experiments (not shown) did not uncover “order-of-addition” effects upon CDK5 activity.

3.4 The positional specificity of DIPPs towards InsP₈

The positional specificity of DIPPs towards the 1- and/or 5-diphosphate groups on InsP₈ also influences relative fluxes through the inositol pyrophosphate pathways (Fig. 1). Positional specificity is difficult to quantify, as so little InsP₇ accumulates when InsP₈ is dephosphorylated by DIPPs (e.g. Fig 3). Furthermore,1-InsP₇ and 5-InsP₇ are not generally resolved by strong anion-exchange HPLC. However, the columns used in the current study yielded a partial separation, particularly when we collected smaller fractions from a shallower gradient (Fig. 5A).

Figure 5. HPLC analyses of InsP₈ dephosphorylation by Ddp1p, DIPP1, DIPP2α and DIPP3α.

Representative HPLC runs (protocol 2; Section 2.2) are shown for reactions containing either no enzyme (A; “zero”), or B, 16 ng Ddp1p, C, 2 ng DIPP1, D, 20ng DIPP2α and E, 10ng DIPP3α, all incubated for 8 min with 16 nM InsP₈. The right hand panels in each pair amplify the InsP₇ region of the chromatograph (open circles; InsP₇ is quantified as percentage of total); filled symbols show elution of 1-InsP₇ and 5-InsP₇ standards (determined individually).

During InsP₈ dephosphorylation by Ddp1p, DIPP1, DIPP2α and DIPP3α, the accumulation of new InsP₇ product was 45-70% above the level of the (predominantly) 1-InsP₇ that was present in the no-enzyme assays (Fig. 5). Both the increase in InsP₇ peak height, and the larger degree of peak spreading, were sufficient to indicate accumulation of both 1-InsP₇ and 5-InsP₇. However, the k_cat data in Table 1 allow us to conclude that, in these experiments (Fig. 5), 1-InsP₇ that is formed from InsP₈ will be degraded faster than the 5-InsP₇ that is produced. That is, the rate of the 5-phosphate removal from InsP₈ (yielding 1-InsP₇) is underestimated relative to the rate of 1-phosphate removal from InsP₈ (producing 5-InsP₇). This conclusion is confirmed by data in Fig. 6, which show DIPP1 metabolizes 1-InsP₇ faster than 5-InsP₇ from a 1:1 mixture. Thus, where the data allow us to determine positional specificity (Fig. 5), we conclude Ddp1p/DIPPs preferentially (but not exclusively) remove the 5-phosphate from InsP₈. Positional specificity for DIPPs 2β/3β was less clear (Figs 3), but based on sequence conservation and kinetic similarities, it seems likely they also show positional selectivity for the 5-phosphate on InsP₈. Thus, the preferential route of InsP₈ dephosphorylation is metabolically distinct from the main pathway for InsP₈ synthesis. This contrasts with the old idea [28] that DIPPs remove from InsP₈ the diphosphate group that is added after 5-InsP₇ (now known to be in the 1-position [29]). The latter conclusion arose from a different experimental approach in which commercially-prepared substrates were incubated with liver homogenates. It is unclear why the earlier experiments led to a different conclusion - current concern over the quality of the substrates used (see Section 1) could be relevant here - but the possibility of a separate InsP₈ phosphatase is another option that bears some consideration.

Figure 6. Metabolism of a 1:1 mixture of 1-InsP₇ and 5-InsP₇ by DIPP1.

1-InsP₇ plus 5-InsP₇ (10 nM each) were together incubated for 10 min with 0.3 ng DIPP1 plus approx 1000 D.P.M. of either 1-[³H]InsP₇ (left hand bar) or 5-[³H]InsP₇ (right-hand bar). HPLC analysis determined the metabolism of each InsP₇. *p=0.02 (t-test; n=3).

3.5 Concluding Comments

Our study increases insight into inositol pyrophosphate turnover and function in vivo by providing the first kinetic characterization of Ddp1p/DIPP-mediated hydrolysis of 1-InsP₇, 5-InsP₇ and InsP₈. Our results lead us to argue against previous suggestions [11,21] (made in the absence of the relevant kinetic data) that DIPPs conceal the importance of pathway II (see Fig. 1) by masking the rate of PPIP5K-driven phosphorylation of InsP₆ to1-InsP₇. Furthermore, much of the attention given to the functional significance of pathway II arises from evidence that 1-InsP₇ inhibits the yeast Pho85 cyclin kinase [5,6]. Our data (Fig. 4) indicate that this mechanism is not conserved in mammals. Indeed, in mammals the 5-InsP₇ that is produced by pathway I has functional significance, by competing with PtdIns(3,4,5)P₃ for PH domains [8,9]; 1-InsP₇ is much less potent [8].

The DIPP/Ddp1 family can also hydrolyze inorganic polyphosphates [21], 5-phosphoribosyl 1-pyrophosphate [30] and nucleotide dimers [16]. However, these particular catalytic activities are unlikely to affect the conclusions reached in this study, as they have alkaline pH optima and are orders of magnitude slower than the rates of inositol pyrophosphate hydrolysis.

Although mammalian cells contain only submicromolar levels of 1-InsP₇ (Section 3.2), its concentration in S. cerevisiae has been argued to reach 30 μM in phosphate-starved cells [5]. While others have disputed that claim [21], it has nevertheless been reported that, in ddp1Δ yeast, levels of (presumably) 1-InsP₇ were elevated 5-fold upon inhibition of Kcs1p by TNP, or by kcs1 deletion [11]. Those experiments were proposed to reveal the true extent of steady-state 1-InsP₇ synthesis, again arguing that Pathway II is underappreciated. However, perhaps in those experiments, Vip1p's phosphorylation of InsP₆ was artificially enhanced merely because of the absence of the kinase's preferred (and competing) substrate, 5-InsP₇ (see [20,29]).

Highlights.

• Understanding inositol pyrophosphate turnover through kinetic study of Ddp1p/DIPPs

• k_cat values for 1-InsP₇ are 5-20 fold higher than those for 5-InsP₇ and InsP₈

• InsP₇ does not regulate CDK5, the human homologue of the yeast Pho85 cyclin kinase

• Kinetics imply differential expression of DIPP isoforms sets signaling sensitivity

• Metabolically and functionally separate routes for InsP₈ synthesis and hydrolysis