Crystal structure of human diphosphoinositol phosphatase 1

Phosphorylated variants of inositol play diverse and crucial roles in a wide range of cellular processes. The fully phosphorylated species, IP6 or phytic acid, was for long thought to be the endpoint of cellular inositol phosphorylation but some 15 years ago two laboratories independently discovered mono- and bis-diphosphorylated inositols PP-IP5 (IP7) and [PP]2-IP4 (IP8).1, 2 Since then, a large number of functions have been ascribed to these diphosphorylated inositols, for example, for telomere maintenance, DNA repair, protein phosphorylation, apoptosis, and even for effective exocytosis in pancreatic β-cells.3-5 The cellular turnover of inositol diphosphates is immense with a large part of the IP6 pool being converted to and from IP7 every hour.1, 3, 4 At any given time point, however, the cellular concentrations of IP7 and IP8 are very low (0.1–2 μM).3, 4 In mammalian cells, IP7 has for some time been known to be of the 5-[PP]-IP5 isoform.4 However, a second IP7 isoform was recently discovered and the new IP7 isoform was proposed to be either 1-[PP]-IP5 and 3-[PP]-IP5.6 Consequently, the IP8 isoform(s) present in mammalian cells are most likely either 1,5-[PP]2-IP4 and/or 3,5-[PP]2-IP4. The ambiguity of the IP7 and IP8 isoforms arises from the lack of stereo-selectivity in the techniques used so far. Early on, it was observed that cells incubated with fluoride display very high levels of inositol diphosphates and it was therefore presumed that the diphosphatase that breaks down inositol diphosphates to IP6 again is fluoride sensitive.1 Indeed, a fluoride sensitive diphosphoinositol diphosphatase (DIPP), was identified in 1998 by Shears and coworkers.7 From sequence similarity, it was found to belong to the Nudix family of hydrolases that commonly have nucleotide and nucleotide derivative substrates.7, 8 Mutagenesis in the Nudix motif, GX5EX7REUXEEXGU, of DIPP rendered the enzyme largely inactive.7, 9 Later, one could define a DIPP-family consisting of five highly homologous members that are all capable of selective inositol diphosphate degradation.10 Here, we describe the crystal structure of the first described DIPP, the human DIPP1, as determined by a combination of sulfur and phosphor SAD. Furthermore, we describe a cocrystal structure of DIPP1 with the product, IP6, and a magnesium fluoride cluster bound. On the basis of the cocrystal structure, we propose a catalytic mechanism for DIPPs. DNA encoding residues 1–148 of the DIPP1 gene (gi: 14043478) was cloned by ligation-independent cloning into a pET-28 based expression vector incorporating an N-terminal hexa-His-tag fusion (pNIC-Bsa4; gi:EF198106). After transformation and liquid culture growth using standard methods,11 recombinant expression of human DIPP1 in Escherichia coli strain BL21(DE3) was induced at 291 K by addition of 0.5 mM IPTG to Terrific Broth media. Induction was maintained for 18 h before harvesting. DIPP1 was purified using IMAC12 on a 1 mL HiTrap chelating HP column followed by gel filtration on a 120 mL Superdex 75 column. DIPP1 was essentially pure as judged by SDS-PAGE analysis and denaturing electrospray-ionization mass spectrometry verified protein integrity. All columns used were from GE-Healthcare, Uppsala, Sweden. Crystals of DIPP1 were grown by sitting-drop vapor diffusion at 277 K. DIPP1 (11 mg/mL) in 20 mM Hepes pH 7.5, 300 mM NaCl, 10% glycerol, 2 mM TCEP, and 5mM myo-inositol hexakisphosphate (IP6) were mixed with an equal amount (100 nL) of reservoir solution (30% PEG 8000, 200 mM lithium sulfate, and 100 mM sodium acetate, pH 4.5) using a Phoenix crystallization robot (Art Robbins Instruments, Sunnyvale, CA). Crystals grew as 0.05 mm × 0.05 mm × 0.1 mm rods after 3 weeks. Crystals were flash-frozen in liquid-nitrogen directly from the crystallization solution. A single-wavelength anomalous dispersion dataset was collected with Cu-Kα radiation on an X8 PROTEUM system equipped with a four-circle Kappa goniostat and a PLATINUM-135 CCD (all from Bruker AXS, Delft, Netherlands). SAINT and SADABS (Bruker AXS) were used to process the data (Table I). XPREP (Bruker AXS) was used to prepare FA-values, up to a resolution of 2.4 Å, for substructure solution in SHELXD13 that found five anomalous scatterer sites. Identified sites were used for phase calculation followed by density modification in SHELXE.13 The resulting map could be auto-traced using ARP/wARP.15 At this stage, it was apparent that two of the five sites used for phasing were phosphors belonging to a bound IP6 molecule. A higher resolution dataset was collected for refinement purposes on beam line I911-3 at Max-Lab (Lund, Sweden) (Table I). The model from ARP/wARP was completed and improved through several rounds of model building and refinement in COOT16 and SHELXL-9717 (Table I). For obtaining an IP6-Mg-F structure, it was necessary to gradually transfer the crystals from the original crystallization solution to 30% PEG 8000 with 200 mM LiCl, 5 mM MgCl2, 20 mM NaF, and 5 mM IP6. These crystals were flash-frozen directly from the soaking solution and a 1.65 Å dataset was collected on the X8 PROTEUM system described earlier (Table I). It was apparent after a few rounds of refinement that the IP6 molecule bound in two conformations. The occupancy of each conformation was refined in SHELXL-97.17 The resulting occupancies were used in a final refinement using REFMAC5.18 The atomic coordinates and structure factors have been deposited at the RCSB under PDB code 2FVV (IP6) and 2Q9P (IP6-Mg-F). DIPP1 adopts the canonical Nudix fold with two β-sheets—one mixed and one anti-parallel—flanked by short helices [Fig. 1(a)]. A PDB structural similarity search using EBI-SSM19 lists the T. thermophilus AP6A hydrolase (1VC8) as the closest structural homologue with an RMSD on C-α of 1.8 Å and sequence identity over the structural alignment of 26%. The Nudix motif (GX5EX7REUXEEXGU) that normally adopts a loop-helix-loop fold display a strand-loop-helix motif (β3-loop-α1) in DIPP1 due to the insertion of an extra residue in between the first glycine and first glutamate in the motif and a tight association of the first three residues of the Nudix motif to a neighboring β-strand (β1). The cocrystallized IP6 molecule binds in a cavity lined with sequence-conserved positively charged residues (data not shown). In our initial DIPP1 structure (Crystal I; 2FVV), the IP6 molecule binds distant to the canonical Nudix motif in a catalytically irrelevant mode, presumably due to a sulfate ion bound in the extended, positively charged, active-site region. Therefore, we devised a scheme to remove the lithium sulfate from the initial crystallization by gradual transfer to a solution with lithium chloride, magnesium, and fluoride to be able to obtain a catalytically relevant cocrystal structure. This was obtained [Crystal II; Fig. 1(a,b)] and this is the crystal form that will be discussed hereafter. In Crystal II, IP6 binds in two conformations—one with an α-phosphate at the inositol C4 (P4) and one with an α-phosphate at the inositol C5 (P5)—close to the Nudix motif catalytic site [Fig. 1(c)]. They are denoted the P4 and the P5 conformation hereafter. For simplicity, only the P4 conformation will be used in the description of the active site and the mechanism discussion. (a) Overall structure of Hs-DIPP1 (light blue). Residues belonging to the Nudix motif are in sticks and colored beige. Magnesium ions are in grey, fluoride ions in green, and waters molecules in red. (b) Close-up of the active site. Same coloring scheme as in (A). (c) Stereo image showing the two conformations in which IP6 is bound in 2Q9P. The P4 conformation is shown in green and the P5 conformation is shown in violet. Map shown is a Fo-Fc synthesis calculated before IP6 was added to the model. In human DIPP1, the Nudix motif coordinates a cluster of magnesium and fluorides ions flanking IP6 [Fig. 1(b)]. There are four magnesium ions bound of which three (MgA, MgB, and MgC) are proposed below to be catalytically relevant. The fourth magnesium ion, MgD coordinates octahedrally a fluoride and IP6-P4 as axial ligands with a set of two fluorides and two waters as planar ligands. This constellation, which is reminiscent of the pentacovalent intermediate found in associative phosphoryl transfer reactions, is situated at the main conserved site of bond breakage in the Nudix family [Fig. 2(a)]. The IP6-Mg-F cocrystal structure allows us to propose a catalytic mechanism for DIPP1 [Fig. 2(b)]. In the proposed mechanism, Glu66, bridges a binuclear complex of MgB and MgC that in turn helps Glu69 to activate a nucleophilic water species for an in-line attack of Pβ. Coordinated by Glu70 and the backbone carboxylate of Gly50, MgA promotes the scissile bond breakage by first stabilizing the transient pentacovalent intermediate and later by binding to the negatively charged leaving Pα. Here, Arg20, that forms a bi-dental bond to the Pα from below [Fig. 2(c,d)], also helps to stabilize the developing negative charge on the leaving group. This is analogous in function to the Arg79 in the mechanism of E. coli ADPRase.20 Facing inwards from the protein surface, Glu54 fix the position of Arg65 so that it can potentiate Glu69 in activating the attacking water species. Thus, it is apparent that the Nudix motif constitutes to the full extent the catalytic machinery at work in DIPP1 since the three residues outlined earlier (Glu54, Arg65, and Glu69) are all part of the conserved Nudix motif. The mechanism proposed here for DIPP1 is similar but not identical to the mechanism proposed for E. coli ADPRase20 in that the proposed catalytic base in DIPP1 (Glu69) is part of the Nudix motif, whereas the proposed catalytic base in E. coli ADPRase (Glu162) is not. Moreover, Glu69 in DIPP1 approaches the attacking water species from another direction than Glu162 and the positively charged Arg65 in DIPP1 that is available to potentiate Glu69 has no counterpart in the E. coli ADPRase mechanism.20 Furthermore, the DIPP1 counterpart (Glu70) of the catalytic base proposed for Mycobacterium tuberculosis ADPRase,21 Glu142, cannot be at work in DIPP1 since MgB lies between Glu70 and the water to be activated [Fig. 1(b)]. So, the mechanism of DIPP1 is analogous to the mechanism of the ADPRase Nudix enzymes20, 21 but utilize a different catalytic base. (a) Superimposition of Hs-DIPP1 magnesium-fluoride cluster with the structure of the de-capping Nudix hydrolase X29 with bound P-1-7-methylguanosine-P-3-adenosine-5′,5′-triphosphate (m7GpppA) (PDB ID: 2A8T).24 The spatially shared point of bond cleavage is denoted with an arrow. Manganese ions bound in 2A8T are colored in lilac. Residue numbering in beige and grey for DIPP1 and X29, respectively. (b) Proposed mechanism of human DIPP1. (c) Molecular modeling of 3,5-[PP]2-IP4 (yellow) binding. (d) Molecular modeling of 1,5-[PP]2-IP4 (pink) binding into the active site of human DIPP1 (2Q9P). The pyrophosphates on the IP8 isomers are denoted with P1, P3, and P5, respectively. Ionic hydrogen bonds shorter than 3.4 Å between IP8 and active-site side chains are shown with green dashed bonds. Magnesium ions bound in 2Q9P are shown as grey spheres. When modeling the two most likely IP8 isoforms, 1,5-[PP]2-IP4 and 3,5-[PP]2-IP4, into the active site of human DIPP1, it is apparent that the 3,5-[PP]2-IP4 isomer is able to form several strong interactions with active site residues not available to the 1,5-[PP]2-IP4 isomer [Fig. 2(c,d)]. Using a 3.4 Å cut-off for ionic hydrogen bonds, 3,5-[PP]2-IP4 and 1,5-[PP]2-IP4 can form 12 and 8 ionic hydrogen bonds, respectively. Thus, from this modeling, 3,5-[PP]2-IP4 should be the preferred IP8 isomer for DIPP1 catalyzed breakdown. Most notably, the C5-β phosphate of 3,5-[PP]2-IP4 interacts with the Nε2 of His91 and both the NH2 and Nε of Arg89 in our model. These interactions appear unavailable to 1,5-[PP]2-IP4. Interestingly, both His91 and Arg89 have previously been shown to be important for IP8 catalysis. Specifically, a His91Leu mutant in DIPP1 reduced activity toward IP8 to 4% of wild-type while activity toward IP7 remained at 75% of wild-type activity.22 Less drastic, but still significant, the DIPP3α isoform that has a Pro89 instead of an Arg89 (90 in DIPP3 numbering) has 2.5 times smaller kcat/Km for IP8 than the DIPP3β isoform.23 It should be noted, however, that both 1,5-[PP]2-IP4 and 3,5-[PP]2-IP4 should be able to bind in the DIPP1 active site and that structural rearrangements in and close to the active site may change binding characteristics significantly from the modeling presented here. In closing, the structure of human DIPP1 provides a structural basis for further biochemical work in the important DIPP family. Future challenges include cocrystal structures of DIPPs with relevant diphosphorylated inositol substrates. B.M.H. is a research fellow of the Swedish Research Council. The authors thank Thomas Ursby (MAX-lab, Lund, Sweden) for support during data collection. The Structural Genomics Consortium is a registered charity (number 1097737) that receives funds from the Canadian Institutes for Health Research, the Canadian Foundation for Innovation, Genome Canada through the Ontario Genomics Institute, GlaxoSmithKline, Karolinska Institutet, the Knut and Alice Wallenberg Foundation, the Ontario Innovation Trust, the Ontario Ministry for Research and Innovation, Merck & Co., Inc., the Novartis Research Foundation, the Swedish Agency for Innovation Systems, the Swedish Foundation for Strategic Research, and the Wellcome Trust.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.