Synthesis of an α-phosphono-α,α-difluoroacetamide analogue of the diphosphoinositol pentakisphosphate 5-InsP7

A synthetic, fluorinated analogue of 5-InsP7 binds to the kinase domain of PPIP5K2, suggesting new strategies for designing diphosphoinositol phosphate mimics.


Introduction
The diphosphoinositol phosphates (PP-InsPs, "inositol pyrophosphates") are of fundamental importance to all eukaryotes, with pivotal roles in cellular and organismic metabolic homeostasis (Fig. 1). 1 The acute clustering of monophosphate and diphosphate groups around the hexahydroxycyclohexane ring of myo-inositol (Ins) endows the PP-InsPs with the most concentrated three-dimensional array of phosphate groups found in Nature. 1 The PP-InsPs are formed through the enzymatic phosphorylation of myo-inositol hexakisphosphate (InsP 6 , Fig. 1) by inositol hexakisphosphate kinases (IP6Ks) and diphosphoinositol pentakisphosphate kinases (PPIP5Ks). 2 Among the PP-InsPs, 5-diphospho-myo-inositol pentakisphosphate (5-PP-InsP 5 , also known as "5-InsP 7 ") is both the most abundant and the most intensively studied member of this signalling family.
An understanding of the molecular actions of PP-InsPs by their addition to cell-free systems can be confounded by the enzymatic and chemical instability of the diphosphate (PP) component. Therefore, we and other workers have developed more stable, synthetic versions of PP-InsPs in which the PP groups are replaced with phosphonoacetate (PA) 3,4 or methylenebisphosphonate (PCP) 4-7 groups (Fig. 1). 8 Thus, a tethered version of 5-PCP-InsP 5 was recently used to search for novel binding proteins for PP-InsPs. 9 Metabolically-stable PP-InsP analogues can also be informative for structural analysis of enzyme/substrate crystal complexes. For example, we have used 5-PA-InsP 5 to reveal a previously unidentified ligand capture site on PPIP5K2. 10 While PA analogues are relatively easy to synthesise, the carboxylic ester of the PA group could be prone to chemical hydrolysis at high pH and/or enzymatic cleavage by cellular esterases. 11 In both PA and PCP analogues, the bridging oxygen of the diphosphate group is replaced by a methylene (CH 2 ) group, making the resulting carbon-phosphorus bonds resistant to hydrolysis in comparison to the native oxygenphosphorus bonds. 12 However, the CH 2 group in methylene phosphonates is less electronegative than the bridging oxygen atom of phosphates, causing an increase in the pK a value of the methylene phosphonic acid in its second deprotonation. This can mean that a methylene phosphonate analogue is less strongly ionised than the phosphate equivalent at physiological pH, potentially leading to a decreased affinity of the analogue for protein binding sites. 12 A well-established approach to this problem, originally developed by Blackburn and co-workers, 13 involves replacing the phosphonate CH 2 group with difluoromethylene (CF 2 ). The more electronegative CF 2 group increases the acidity of the phosphonic acid and, in addition, the CF 2 group itself has greater electronic and steric similarity to the bridging oxygen atom of a phosphate than does CH 2 . 14 The difluoromethylenephosphonate (PCF 2 ) group has been particularly useful as a phosphate mimic in the development of protein tyrosine phosphatase inhibitors. 15,16 In nucleotide chemistry, the difluoromethylenebisphosphonate (PCF 2 P) motif has been used to mimic PP in stable analogues of nucleoside diphosphates and triphosphates. 17 Although the PCF 2 P group has been proposed as a potential PP mimic in stabilised analogues of PP-InsPs, this was anticipated to present a considerable synthetic challenge, 5,18 and while several PCP-InsPs have been synthesised, 4-7 no fluorinated equivalents have yet been disclosed.
In the current report, we explore an alternative approach to a fluorinated isopolar analogue of a PP-InsP, building on our earlier work with phosphonoacetic acid (PA) esters. Although replacing the CH 2 group in the β-ketophosphonate fragment of a PA ester with CF 2 is possible, the resulting difluoroacetate ester would be very labile to hydrolysis. Therefore, we chose to replace the ester of PA with a stable amide linkage, allowing the inclusion of the electron-withdrawing CF 2 group in place of CH 2 . In the resulting analogue, 5-PCF 2 Am-InsP 5 (1, Fig. 1), the terminal phosphonate group should more closely resemble the corresponding β-phosphate group of 5-PP-InsP 5 in its electronic properties than in either the PA or the PCP equivalents. The α-phosphono-α,αdifluoroacetamide (PCF 2 Am) unit lacks a close equivalent to the α-phosphate of PP, but the amide carbonyl retains the potential to accept H-bonds and the rigidity of the amide structure itself may confer advantages at some binding sites. To the best of our knowledge, the PCF 2 Am motif 19 has not previously been explored as a diphosphate isostere, although it has been used successfully in analogues of 1,3bisphosphoglyceric acid as inhibitors of phosphoglycerate kinase 20 and in the design of inhibitors of aspartate carbamoyltransferase 21 and protein tyrosine phosphatases. 16 Results and discussion Synthesis of 5-PCF 2 -Am-InsP 5 (1) To construct compound 1, we needed to synthesise an appropriately protected 5-deoxy-5-amino-myo-inositol intermediate (Scheme 1). We have previously shown 22 that regioselective sulfonylation of butanediacetal (BDA) protected myo-inositol 2 (ref. 3, 22 and 23) using triflic anhydride, followed by solvolysis using wet dimethylacetamide, gives inversion of configuration at C-5, to produce the neo-inositol acetate derivative 3 via an iminium ion intermediate. 22 In the present work, we used a second triflate esterification of the free OH group in 3 followed by treatment with sodium azide in DMF to give a second configurational inversion, returning us to the myo-inositol configuration in the 5-deoxy-5-azido-myoinositol derivative 4. ‡  Compound 4, with its combination of acid-and base-labile protecting groups, is a potentially versatile intermediate in itself. However, to simplify the current synthesis, we sequentially replaced these labile protecting groups with benzyl ethers. Thus, replacement of the 2-O-acetate ester with a 2-O-benzyl ether gave 5, a precursor for the synthesis of analogues of 5-PP-InsP 4 . Next, the BDA groups were removed and replaced with benzyl ethers to give pentabenzyl 5-deoxy-5-azido-inositol 6. Re-duction of the azide group in 6 using LiAlH 4 now gave the 5-deoxy-5-amino-inositol 7 in quantitative yield. The next step was to introduce the phosphonodifluoroacetamide unit (Scheme 2). Thus, reaction of 7 with diethyl phosphonodifluoroacetic acid 24 gave phosphonodifluoroacetamide 8. The benzyl protecting groups in 8 were removed by catalytic hydrogenolysis over PdĲOH) 2 /C to give pentaol 9. Phosphitylation using bisbenzyloxydiisopropylaminophosphine activated with 5-phenyl-1H-tetrazole then gave the intermediate pentakisphosphite, which was observed by 31 P NMR, but not isolated. Oxidation of phosphites with mCPBA yielded the fullyprotected pentakisphosphate 10. Finally, the benzyl and ethyl protecting groups in 10 were cleanly removed using TMSBr in dichloromethane followed by MeOH.
Compound 1 was isolated as the triethylammonium salt, containing six TEA + ions per molecule of 1. The nonfluorinated phosphonoacetamide (PCH 2 Am) equivalent of 1 was also initially synthesised to explore synthetic methods (see ESI † for details). During this synthesis, we observed facile exchange of the CH 2 protons with deuterium from D 2 O during NMR spectroscopy of the PCH 2 Am-containing pentaol intermediate analogous to compound 9. The deuterium was retained after the subsequent phosphorylation and deprotection steps. Attempts to exchange deuterium back to hydrogen in the final product were unsuccessful, possibly because ionisation of the methylenephosphonate group disfavours the required enolisation (see ESI † for further details). This may suggest a strategy for developing tritiated versions of the non-fluorinated equivalent of 1 and related analogues of PP-InsPs.  The 31 P NMR spectrum of 1 in D 2 O showed the signal corresponding to the PCF 2 Am phosphorus atom at δ 0.64, shifted approximately 13.5 ppm up-field relative to the equivalent signal in the PCH 2 Am equivalent (ESI † Fig. S1). This reflects the profound effect of the CF 2 group on the electronic properties of the phosphonate group in 1. Compound 1 was very stable in aqueous solution; the NMR sample showed no sign of decomposition after >1 year in solution in D 2 O.

Interaction of 5-PCF 2 -Am-InsP 5 with PPIP5K2
To determine whether 1 could act as a mimic of 5-PP-InsP 5 , we examined its interaction with the kinase domain of human PPIP5K2. We used a "reverse-kinase" assay that records ATP generated from 0.1 mM ADP during the dephosphorylation of 100 nM 1,5-[PP] 2 -InsP 4 . This approach avoids the need to use radiolabelled material and slow-throughput HPLC analysis. Furthermore, the assay of ATP production is inherently more sensitive compared to measuring ATP consumption. Compound 1 inhibited 1,5-[PP] 2 -InsP 4 metabolism by PPIP5K2 with an IC 50 of 375 nM ( Fig. 2A). We then obtained an X-ray structure of 5-PCF 2 Am-InsP 5 (1) in complex with PPIP5K2 kinase domain and the stable ATP analogue AMPPNP. The structure shows that 5-PCF 2 Am-InsP 5 binds to the catalytic site of PPIP5K2 in a similar orientation to 5-PP-InsP 5 (Fig. 2B-D). While crystallography previously showed that 5-PA-InsP 5 binds to two sites in PPIP5K2 (the catalytic site and a surface-mounted ligand capture site), 4,10 electron density for 1 was seen only in the catalytic site (Fig. 2B). This is similar to the complex structures of PPIP5K2 with 5-PCP-InsP 5 (ref. 7) or with 5-PP-InsP 5 itself. 25 The electron density for 1 (Fig. 2C) clearly shows that the planar amide unit adopts the Z conformation, and with the amide NH antiperiplanar to H-5 of myo-inositol. The carbonyl oxygen of the amide is orientated similarly to one nonbridging oxygen atom of the α-phosphate of bound 5-PP-InsP 5 ( Fig. 2E and F). Additionally, one of the fluorine atoms in the CF 2 group stretches its position to that of the other non-bridging O-atom of an α-phosphate, which is expected to bring the negative charge distribution in this region closer to that of 5-PP-InsP 5 than occurs with the PA analogue. The terminal phosphonate group in bound 1 is held in an extended conformation (Fig. 2E) although its orientation is different to that of the corresponding β-phosphate group in 5-PP-InsP 5 bound to PPIP5K2 (Fig. 2F).

Conclusions
We have described the design and synthesis of a novel type of stable PP-InsP analogue containing a phosphonodifluoroacetamide (PCF 2 Am) structure, intended to mimic the 5-diphosphate (PP) group of 5-PP-InsP 5 . As proof of principle, we examined the interaction of the new analogue, 5-PCF 2 Am-InsP 5 (1), with the catalytic domain of the diphosphoinositol pentakisphosphate kinase PPIP5K2. A crystallographic study showed that 1 binds to the PP-InsPbinding site of the PPIP5K2 kinase domain in a similar orien-tation to that of the natural substrate 5-PP-InsP 5 and that the PCF 2 Am structure can mimic several functionally important aspects of the diphosphate group (PP) in 5-PP-InsP 5 .
Recent years have seen a rapid increase in reported functions of PP-InsPs, the identification of their target proteins, and the characterisation of metabolising enzymes. Stabilised analogues of PP-InsPs have often played valuable roles in this work. No single analogue of a PP-InsP can perfectly mimic its steric and electronic features in all environments, and a range of compounds with complementary properties therefore offers a more versatile analytical approach. It is in this context that we propose that 5-PCF 2 Am-InsP 5 (1) brings new opportunities. For example, the presence of the CF 2 group in 1 means that its terminal phosphonate group should be more acidic than the equivalent group in 5-PA-InsP 5 or 5-PCP-InsP 5 , making it more similar in this respect to the β-phosphate of 5-PP-InsP 5 and potentially allowing stronger interactions with protein binding sites. In addition, the rigidity of the amide unit in 1 means that the PCF 2 Am unit contains fewer rotatable bonds than the equivalent PP, PA or PCP structures. A more rigid ligand is less likely to lose entropy on binding than a more flexible one, and thus may bind more tightly to some sites. Note also that the 5-diphosphate in 5-PP-InsP 5 participates in complexing a hydrated Mg 2+ ion, a property that is mimicked in the complex of 5-PCP-InsP 5 with PPIP5K2. 7 This Mg 2+ ion is not present in the complex with 1 (Fig. 2E). Thus, 5-PCF 2 Am-InsP 5 may be particularly useful for studying ligand/protein interactions that do not involve Mg 2+ . 9 To the best of our knowledge, the PCF 2 Am group has not been previously employed as a diphosphate mimic; given the large number of biomolecules that contain diphosphate or polyphosphate motifs, it may find wider applications than those outlined here.

General chemistry methods
All reagents and solvents were of commercial quality and were used without further purification. Petroleum ether used for chromatography and crystallisations was of fractions 40-60°C. Alcohol 3 was synthesised as previously reported. 22 Diethyl phosphonodifluoroacetic acid was synthesised according to a literature procedure. 24 Thin-layer chromatography (TLC) was performed on pre-coated aluminium plates (Merck, silica gel 60, F254). Chromatograms were visualised under UV light and by dipping plates into either phosphomolybdic acid in EtOH, vanillin in acidic EtOH, or alkaline aqueous KMnO 4 , followed by heating. Flash column chromatography was performed on an ISCO CombiFlash Rf automated flash chromatography system using RediSep Rf disposable flash columns. NMR spectra were recorded on Bruker Avance III 400 and 500 MHz NMR spectrometers. Proton chemical shifts are reported in ppm (δ) relative to internal tetramethylsilane (TMS, 0.0 ppm) or with the solvent resonance relative to TMS employed as the internal standard (D 2 O, 4.79 ppm). The following abbreviations are used to describe resonances: br, broad; s, singlet; d, doublet; dd, double doublet; q, quartet; m, multiplet; t, triplet. 13 C chemical shifts are reported relative to internal TMS (TMS, 0.0 ppm) or with the solvent resonance relative to TMS employed as the internal standard. The assignments of the proton and carbon atoms are based on 2D-NMR experiments ( 1 H-1 H-COSY, HSQC). 31 P chemical shifts are reported in ppm (δ) relative to an 85% H 3 PO 4 external standard (H 3 PO 4 , 0.0 ppm). 19 F chemical shifts are reported in ppm (δ) relative to a CFCl 3 external standard (CFCl 3 , 0.0 ppm). Melting points were determined using a Stanford Research Systems Optimelt MPA100 automated melting point system and are uncorrected. High resolution time-of-flight mass spectra were obtained on a Bruker Daltonics micrOTOF mass spectrometer using electrospray ionisation (ESI †).

1,2,3,4,6-Penta-O-benzyl-5-deoxy-5-azido-myo-inositol (6)
To a stirred solution of 5 (507 mg, 0.968 mmol) in dichloromethane (5 mL) was added 95% aqueous TFA (5 mL). A yellow colour (butanedione) appeared within 30 s. After 30 min, the solution was concentrated to leave the crude tetraol as a white solid (325 mg). This solid was taken up in dry DMF (10 mL) and the solution was cooled to 0°C, before sodium hydride (60% suspension in mineral oil, 240 mg, 6.0 mmole) was added. The stirred suspension was allowed to warm to room temperature and then cooled again to 0°C before benzyl bromide (0.60 mL, 5.1 mmol) was added, dropwise. The cooling bath was removed and the mixture was stirred overnight (16 h) at room temperature. Excess NaH was destroyed by careful addition of water and the solution was concentrated. The residue was taken up in dichloromethane (50 mL) and washed with water and brine (50 ml each), dried (MgSO 4 ) and concentrated to give an oily residue. Purification by flash chromatography on silica (ethyl acetate in petroleum ether, 0 to 50%) gave 6 as a colourless oil, which slowly crystallised (576 mg, 0.878 mmol, 91% over two steps); R f 0.

1,2,3,4,6-Penta-O-benzyl-5-deoxy-5-amino-myo-inositol (7)
To dry THF (3 mL) under argon was added LiAlH 4 (1.0 mL of a 1.0 mol dm −3 solution in THF, 1.0 mmol). The solution was stirred at 0°C under argon and a solution of the azide 6 (420 mg, 0.640 mmol) in dry THF (4 mL) was injected over 10 min. After 30 min, the cooling bath was removed and the solution was stirred at room temperature for a further 1 h. The solution was then cooled to 0°C again and excess LiAlH 4 was destroyed by addition of ethyl acetate (2 mL). The solution was allowed to warm to room temperature and then diluted with water (20 mL). The suspension was extracted with diethyl ether (3 × 20 mL). The aqueous layer was re-extracted with dichloromethane (20 mL) and the combined organic extracts were then washed with brine (50 mL), dried (MgSO 4 ) and concentrated to give amine 7 as a white solid (402 mg, 0.638 mmol, 100%); 1  To a stirred suspension of pentaol 9 (30 mg, 0.076 mmol) and 5-phenyl-1H-tetrazole (83 mg 0.57 mmol) in dry dichloromethane (3 mL) under N 2 at room temperature was added bisĲbenzyloxy)diisopropylaminophosphine (0.20 mL, 0.60 mmol). The mixture was stirred under N 2 at room temperature for 3 h and then cooled to −78°C, before mCPBA (70%, 187 mg, 0.760 mmol) was added. The mixture was allowed to warm to room temperature and then diluted with EtOAc (30 mL). The clear, colourless solution was washed with 10% aq. Na 2 SO 3 solution (3 × 25 mL), dried over MgSO 4 and concentrated. The residue was purified by flash chromatography (EtOAc in petroleum ether, 0 to 100%) to give 10 as a colourless oil (85 mg, 0.050 mmole, 66%); TLC (EtOAc : petroleum ether, 4 :  (1) A stirred solution of 10 (68 mg, 40 μmole) in dry dichloromethane (2 mL) was cooled to 0°C under N 2 and trimethylsilyl bromide (1 mL) was added dropwise over 5 min. The solution was allowed to warm gradually to room temperature, and stirring was continued for 48 h. The solution was concentrated and methanol (5 mL) was added to the residue. The resulting colourless solution was stirred at room temperature for a further 1 h, and then concentrated to give a white gum. The gum was washed with diethyl ether (3 × 2 mL), then taken up in aqueous triethylammonium bicarbonate (1.0 mol dm −3 , pH 7.6, 5 mL). This solution was then washed with diethyl ether (3 × 5 mL) and concentrated. The residue was re-dissolved in MilliQ water and lyophilised to give the triethylammonium salt of the title compound 1 as a colourless solid (47 mg, 34 μmole, 85%); 1 H NMR (500 MHz,

Protein expression, purification, crystallisation and structure determination
The kinase domain of human PPIP5K2 (residues 41-366) was sub-cloned, expressed and purified as before. 25 PPIP5K2 kinase domain was crystallised by hanging drop vapour diffusion against a well buffer of 12% (w/v) PEG 3350, 20 mM MgCl 2 , 0.1 M HEPES, pH 7.0, 2 mM CdCl 2 , 1 mM AMPPNP at 4°C. The crystals were transferred to a stabilising buffer containing 22% (w/v) PEG 3350, 10 mM MgCl 2 , 0.1 M sodium acetate, pH 5.2 at 4°C and the crystals were then soaked under the above stabilising buffer for three days with 2 mM compound 1. Cryosolvent was prepared by adding 33% ethylene glycol into the soaking buffer. Diffraction data were collected using APS beamlines 22-ID. All data were processed with the program HKL2000. The structure was determined using rigid body and direct Fourier synthesis, and refined with the equivalent and expanded test sets. The structure was further manually rebuilt with COOT and refined with REFMAC from the CCP4 package. Ligand topology and parameter files were prepared using the PRODRG server. The molecular graphics representations were prepared with the program PyMol (Schrödinger, LLC). Atomic coordinates and structure factors have been deposited with the Protein Data Bank with accession code 6N5C.

Conflicts of interest
There are no conflicts to declare.