Synthesis of Densely Phosphorylated Bis-1,5-Diphospho-myo-Inositol Tetrakisphosphate and its Enantiomer by Bidirectional P-Anhydride Formation

Abstract

The ubiquitous mammalian signaling molecule bisdiphosphoinositol tetrakisphosphate (1,5-(PP)₂-myo-InsP₄, or InsP₈) displays the most congested three-dimensional array of phosphate groups found in nature. The high charge density, the accumulation of unstable P-anhydrides and P-esters, the lack of UV absorbance, and low levels of optical rotation constitute severe obstacles to its synthesis, characterization, and purification. Herein, we describe the first procedure for the synthesis of enantiopure 1,5-(PP)₂-myo-InsP₄ and 3,5-(PP)₂-myo-InsP₄ utilizing a C₂-symmetric P-amidite for desymmetrization and concomitant phosphitylation followed by a one-pot bidirectional P-anhydride-forming reaction that combines sixteen chemical transformations with high efficiency. The configuration of these materials is unambiguously shown by subsequent X-ray analyses of both enantiomers after being individually soaked into crystals of the kinase domain of human diphosphoinositol pentakisphosphate kinase 2.

Cells propagate and integrate information by using phosphorylated second messengers based on the myo-inositol scaffold.[¹] Their polar character and relatively low abundance often necessitate tedious isolation procedures, which is especially true for the family of inositol pyrophosphates.[²] The need to have ready access to inositol pyrophosphates for biological studies motivated the development of a facile, scalable, and stereoselective synthesis of these molecules.

Inositol pyrophosphates, such as 5-PP-myo-InsP₅ (1), 1-PP-myo-InsP₅ (2, myo-InsP₇), and 1,5-(PP)₂-myo-InsP₄ (3, myo-InsP₈; Figure 1), can signal by two distinctly different processes, that is, by either binding to a specific receptor[³] or by non-enzymatic transfer of the β-phosphate to phosphorylated serine residues in proteins (transphosphorylation).[⁴] In vivo, both processes contribute to complex signaling networks that are far from being understood.[⁵]

Figure 1.

Mammalian inositol pyrophosphates 1–3 and an unnatural enantiomer 4. The dashed box indicates the generic description, with the more precise names given underneath the compounds.

Inositol pyrophosphates and their dedicated kinases (IP6Ks: inositolhexakisphosphate kinases; PPIP5Ks: diphosphoinositol pentakisphosphate kinases)[⁶] and phosphatases (DIPPs: diphosphoinositol polyphosphate phosphohydrolases)[⁷] have been linked to a range of biological phenomena, such as insulin signaling,[⁸] telomere length regulation,[⁹] pleckstrin homology (PH) domain regulation,[¹⁰] cellular energy homeostasis,[¹¹] apoptosis,[¹² and regulation of the innate immune system.[¹³] Some of these studies have been facilitated by the use of chemically synthesized myo-InsP₇ and derivatives thereof.[^4b,8b,12,13b] In contrast, the specific biological functions of myo-InsP₈ remain largely unexplored, mainly because a method to chemically synthesize these molecules in sufficient purity has not been reported. Nonetheless, it has been established that concentrations of myo-InsP₈ in mammalian cells are regulated in a stimulus-dependent fashion increasing severalfold following osmotic stress or thermal challenge, whereas bioenergetic stress significantly decreases levels of myo-InsP₈.[¹⁴]

In mammals, the structure of the myo-InsP₈ isomer has been identified as the 1,5 isomer 3.[¹⁵] Chemical synthesis of this isomer would greatly aid studies into its function. Having access to both enantiomers 3 and 4 would be even more useful for functional studies, in which biological specificity could be tested.

Different isomers and analogues of myo-InsP₇ have been accessed by chemical synthesis, but it is still a challenge to obtain reasonably pure material.[¹⁶] The method must be mild enough to protect the delicate P-anhydride, while enforcing the generation of a highly congested structure. The more intense molecular crowding in 1,5-myo-InsP₈ presents even more complex synthetic challenges. In fact, the chemical synthesis of only one myo-InsP₈ isomer is reported, namely the unnatural and symmetric meso 2,5-isomer; its published purity of approximately 10% underscores the technical demands in synthesizing these molecules.[¹⁷]

Synthetic access to enantiopure nonsymmetric myo-InsP₇ can be achieved by using a C₂-symmetric P-amidite through the desymmetrization of a protected analogue of the meso compound myo-inositol.[¹⁸] The efficiency of this approach is based on the fact that the chiral auxiliary also serves as an orthogonal protecting group that can be cleaved under very mild conditions. As all positions in myo-InsP₈ are phosphorylated, desymmetrization by phosphorylation[^{18a, 19}] significantly decreases the required number of steps. Consequently, extension of this approach to synthesize myo-InsP₈ derivatives 3 and 4 could potentially provide efficient access to these molecules.

The synthesis commenced with myo-inositol (5) that was converted[^18a] (Scheme 1) into benzylidene-protected intermediate 6 with a free 5-OH group by the method of Holmes et al.[²⁰] The 5-axial hydroxy group in alcohol 6 was phosphitylated with a bis-fluorenyl-protected P-amidite 7,[²¹] oxidized, and the benzylidene acetal was selectively cleaved yielding meso-diol 8.

Scheme 1.

Synthesis of the unsymmetric hexakisphosphates 13a and 13b by phosphitylation with C₂-symmetric P-amidite 9, followed by the gram-scale separation of the diastereomers 10a,b. The chiral auxiliary (R^χ) serves as an orthogonal protecting group throughout the synthesis. DCM = dichloromethane, PMB = para-methoxybenzyl, Fm = fluorenylmethyl, tBuOOH = tert-butylhydroperoxide, pTsOH = para-toluenesulfonic acid, mCPBA = meta-chloroperbenzoic acid, TFA = trifluoroacetic acid, FC = flash chromatography, XE = o-xylylene.

Diol 8 was subjected to phosphitylation with C₂-symmetric P-amidite 9, followed by oxidation of the P^III intermediate. The obtained mixture of diastereomers 10a/b was then separated by repeated flash chromatography (FC) until an enrichment of either diastereomer, 10a or 10b, to a purity of 90% to 95%, was achieved. Note that with this approach, a total of 2.7 g of the diastereomeric mixture was separated, providing access to more than 1 g of each of the purified diastereomers (diastereomeric ratio, d.r. >9:1). After cleavage of the para-methoxybenzyl (PMB) protecting groups under acidic conditions and crystallization, both diastereomers 11a and 11b were obtained in highly enriched form (d.r. ≥ 98:2), as evidenced by mixing of the separated compounds and ³¹PNMR spectroscopic analysis (see the Supporting Information). Consequently, enrichment of the diastereomers 10a or 10b prior to cleavage of the PMB groups is sufficient to obtain essentially diastereopure material in the next step. The diastereopure tetraols 11a and 11b were then phosphitylated with an o-xylylene-derived phosphoramidite 12[^16g] and oxidized in situ. The obtained unsymmetric hexakisphosphates 13a and 13b were again analyzed for diastereomeric purity by mixing experiments (d.r. ≥ 98:2; Supporting Information).

In the key step, the hexakisphosphates 13a and 13b were subjected to a bidirectional “telescoping” P-anhydride-forming reaction (Scheme 2). This step involved the introduction of both protected pyrophosphates in the 1 or 3 and 5 positions in a single, one-pot reaction without workup of the reaction product between steps. As this process combined sixteen discrete chemical reactions, precise control over every single step was required. Overall, four protecting groups (Fm and chiral auxiliaries R^χ) were removed by replacement with more labile trimethylsilyl groups (TMS) under mild, basic conditions in the presence of the TMS donor N,O-bis(trimethylsilyl) trifluoroacetamide (BSTFA). Subsequently, all four TMS groups were cleaved by methanolysis and the nucleophilic dibasic phosphates formed were stabilized by mono-protonation with trifluoroacetic acid (TFA). The monobasic phosphates obtained were then phosphitylated with bis-benzyl P-amidite 14, oxidized in situ, and the generated bis-P-anhydrides 15a and 15b were isolated in almost pure form by simple precipitation. It was mandatory to obtain the bis-P-anhydrides 15a and 15b in high quality, as they were not stable enough for further purification.

Scheme 2.

One-pot synthesis of protected 1,5- or 3,5-(PP)₂-myo-InsP₄ (15a/15b) followed by high-pressure hydrogenation to yield enantiopure myo-InsP₈ (3 or 4) as octasodium salts. Dotted Box: PAGE analysis of synthetic myo-InsP₈ with synthetic standards (5-myo-InsP₇, 1-myo-InsP₇) and commercial standards (polyphosphate and myo-InsP₆). Stained with toluidine blue. BSTFA = N,O-bis(trimethylsilyl) trifluoroacetamide, DBU = 1,8-diazabicyclo[5.4.0]undecene, TFA = trifluoroacetic acid, mCPBA = meta-chloroperbenzoic acid, Bn = benzyl.

Rapid deprotection by high-pressure hydrogenation of 15a or 15b led to 3 or 4, respectively (Scheme 2), and precipitation from solution gave material of suitable purity for biological studies. The presence of trace impurities (10%) of myo-InsP₇ (e.g. 1 and 2) was detected by ³¹P NMR spectroscopy and PAGE analysis[²²] (Scheme 2, dotted box; Supporting Information). Even though α-unprotected P-anhydrides[²³] are less prone to hydrolysis compared to P-anhydrides that are fully protected,[^{16g, 17}] complete preservation of the moiety could not be achieved.

It is difficult to assign the absolute configuration of inositolphosphates exclusively by measurement and comparison of optical rotations, as these values are usually very small and strongly depend on the pH value.[¹⁹] Optical rotation values of myo-InsP₈ have not been reported. However, the naturally occurring isomer of myo-InsP₈, 3, has previously been identified in crystal complexes of the kinase domain of human PPIP5K2 (PPIP5K2^KD).[¹⁵] New crystal complexes of PPIP5K2^KD were now prepared, into which either myo-InsP₈ enantiomer 3 or 4 were individually soaked (Figure 2). The electron-density maps allowed for an unequivocal assignment of the absolute configuration of each of the myo-InsP₈ enantiomers. It is interesting to note that the unnatural enantiomer 4 is also accommodated in the catalytically active site of the enzyme. In addition to the active site of PPIP5K2^KD accommodating the natural (3) and unnatural (4) enantiomers, both were also substrates for the phosphatase enzyme DIPP1 (3: V_max = 124.9 ± 5.5 nmolmg⁻¹ min⁻¹; 4: V_max = 239.9 ± 9.3 nmolmg⁻¹ min⁻¹, n = 3; measured by inorganic phosphate (P_i) release as described previously (see Ref. [24] and the Supporting Information).

Figure 2.

Refined 2 F_o– F_c maps contoured at 2.0σ (panels A, B) or 1.0σ (panels C, D) and shown in blue mesh. Myo-InsP₈ molecules are shown in a stick model. The kinase domain of PPIP5K2 is shown in a transparent representation. The phosphate groups around the inositol ring are numbered. Panels A, B: PPIP5K2^KD/ADP complexed with myo-InsP₈ (1,5-(PP)₂-InsP₄ 3), showing two orientations. Panels C, D: PPIP5K2^KD/AMP-PNP complexed with the enantiomer 3,5-(PP)₂-InsP₄ (4). (PDB codes: 4Q4C and 4Q4D).

Overall, this study unveils a new strategy to synthesize 1,5- and 3,5-myo-InsP₈ 3 and 4 in scalable quantities, based on desymmetrization by phosphitylation on a gram-scale, followed by a bidirectional P-anhydride synthesis in a one-pot telescoping procedure. The absolute configuration was deduced by cocrystallization with PPIP5K2^KD, allowing for the subsequent unambiguous assignment of each enantiomer. This study provides novel strategies to construct the most congested and highly charged three-dimensional assemblies of phosphate groups occurring in nature. These strategies can be applied to target other regioisomers and functionalized analogues and will help to unravel the enigmas associated with the inositol pyrophosphate signaling network.