d-chiro-Inositol Ribophostin: A Highly Potent Agonist of d-myo-Inositol 1,4,5-Trisphosphate Receptors: Synthesis and Biological Activities

Abstract

Analogues of the Ca²⁺-releasing intracellular messenger d-myo-inositol 1,4,5-trisphosphate [1, Ins(1,4,5)P₃] are important synthetic targets. Replacement of the α-glucopyranosyl motif in the natural product mimic adenophostin 2 by d-chiro-inositol in d-chiro-inositol adenophostin 4 increased the potency. Similar modification of the non-nucleotide Ins(1,4,5)P₃ mimic ribophostin 6 may increase the activity. d-chiro-Inositol ribophostin 10 was synthesized by coupling as building blocks suitably protected ribose 12 with l-(+)-3-O-trifluoromethylsulfonyl-6-O-p-methoxybenzyl-1,2:4,5-di-O-isopropylidene-myo-inositol 11. Separable diastereoisomeric 3-O-camphanate esters of (±)-6-O-p-methoxy-benzyl-1,2:4,5-di-O-isopropylidene-myo-inositol allowed the preparation of 11. Selective trans-isopropylidene deprotection in coupled 13, then monobenzylation gave separable regioisomers 15 and 16. p-Methoxybenzyl group deprotection of 16, phosphitylation/oxidation, then deprotection afforded 10, which was a full agonist in Ca²⁺-release assays; its potency and binding affinity for Ins(1,4,5)P₃R were similar to those of adenophostin. Both 4 and 10 elicited a store-operated Ca²⁺ current I_CRAC in patch-clamped cells, unlike Ins(1,4,5)P₃ consistent with resistance to metabolism. d-chiro-Inositol ribophostin is the most potent small-molecule Ins(1,4,5)P₃ receptor agonist without a nucleobase yet synthesized.

Introduction

Phospholipase C hydrolyzes phosphatidylinositol 4,5-bisphosphate to release the water-soluble second messenger, d-myo-inositol 1,4,5-trisphosphate [Ins(1,4,5)P₃1, Figure 1]. Ins(1,4,5)P₃ binds to the Ins(1,4,5)P₃-binding core (IBC) of its receptor (Ins(1,4,5)P₃R),¹ a tetrameric structure with an intrinsic Ca²⁺-permeable pore that opens after Ins(1,4,5)P₃ binding.²⁻⁴ This allows Ca²⁺ to be released from intracellular stores to the cytosol to cause an increase in cytosolic Ca²⁺ concentration. Hence, Ins(1,4,5)P₃ links the many extracellular stimuli that activate phospholipase C to the Ca²⁺ signals that regulate diverse activities in animal cells. We have previously synthesized numerous modified ligands to probe their interactions with Ins(1,4,5)P₃Rs.⁵⁻⁷ Such ligands include modifications or mimics of Ins(1,4,5)P₃ and also simplified benzene polyphosphate surrogates.^6,7

Figure 1.

Structures of inositol trisphosphate 1, the adenophostins 2 and 3, and inositol adenophostin 4.

The natural product glyconucleotides adenophostin A 2 and adenophostin B 3 (Figure 1) were isolated from fungal broths⁸⁻¹⁰ and shown to be ca. 10-fold more potent than Ins(1,4,5)P₃ in evoking Ca²⁺ release through each of the three mammalian Ins(1,4,5)P₃R subtypes (Ins(1,4,5)P₃R1–3).¹¹ Their structure has inspired much synthetic work by us¹¹⁻¹³ and others¹⁴⁻¹⁷ and provided numerous ligands with modifications at the phosphate groups, nucleobase motif, and at both sugars.^11,18,19 However, the bisphosphorylated glucose moiety that is thought to mimic the key Ins(1,4,5)P₃ pharmacophore has never been replaced by a cyclitol bisphosphate structure until very recently.²⁰ Here, the glucose motif of adenophostin A was replaced with d-chiro-inositol²⁰ and thus the pyranose–furanose disaccharide linkage was replaced with a sec–sec ether to give d-chiro-inositol adenophostin 4, which is nearly 2-fold more potent than adenophostin A.²⁰

Our model for the interaction of adenophostin A with the Ins(1,4,5)P₃ binding core (IBC; residues 224–604 in Ins(1,4,5)P₃R1)²¹ is supported by mutagenesis²² and SAR analyses^{13,19,23−25} and suggests that the glucose 3″,4″-bisphosphate motif mimics the Ins(1,4,5)P₃ vicinal bisphosphate, while the 2″-hydroxyl and 2′-phosphate groups mimic the nonessential 6-hydroxyl²⁶ and the 1-phosphate of Ins(1,4,5)P₃, respectively. A recently published cryo-electron microscopy (EM) structure² of Ins(1,4,5)P₃R1 with bound adenophostin A hints at substantially different binding modes for this ligand. However, the ligand resolution in these protein structures is poor, the adenophostin A structure shown is incorrect (the pentose sugar is shown as xylose rather than a ribose derivative), and it is puzzling that the occupancy of the IBC by adenophostin A is very low in the presence of a saturating concentration of the ligand.² Until these data are validated, we continue to work with our hypothesis that the glucose 2″-hydroxyl and 3″,4″-bisphosphate triad motif of adenophostin A and related motifs in the corresponding disaccharide analogues discussed below mimic the respective 6-OH and 4,5-bisphosphate motifs of Ins(1,4,5)P₃ (Figure 1).

The simplest adenophostin A derivative designed to date is the glucopyranoside 2′,3,4-trisphosphate [Glc(2′,3,4)P₃] (5, Figure 2), in which the adenine and part of the ribose (C-1, C-4, and C-5 of ribose) were removed, leaving only an hydroxyethylphosphate moiety tethered at the α-anomeric position of glucose.^27,28 Glc(2′,3,4)P₃ was ca. 10-fold less potent than Ins(1,4,5)P₃ in releasing Ca²⁺ through Ins(1,4,5)P₃R,¹⁹ confirming the likely importance of the adenine and entire ribose sugar of adenophostin A. Others confirmed similar activity for Glc(2′,3,4)P₃ and showed that it is not degraded by either of the enzymes that metabolize Ins(1,4,5)P₃, the 5-phosphatase and 3-kinase.²⁹ The ribose sugar tethered at the anomeric center of glucose was reintroduced to give ribophostin (6), which has also a β-OMe group replacing the adenine base at position-1 of ribose, and furanophostin (7), in which the anomeric OMe group and the methylene hydroxyl (at C-5 on ribose) were removed (Figure 2). The potency of both 6 and 7 was similar to that of Ins(1,4,5)P₃.¹¹ Further pruning of furanophostin by removal of the pyranoside methylene hydroxyl group to give 8 reduced the potency to ca. 3-fold below that of Ins(1,4,5)P₃.^11,30 Hence, ribophostin and furanophostin probably provide the best minimal disaccharide-based mimics of Ins(1,4,5)P₃. Compound 9 with the same phosphorylated tetrahydrofuran motif as 8, but with β-stereochemistry at the hemiacetal, was inactive,³⁰ confirming an essential α-stereochemical requirement for ribophostin.

Figure 2.

Structures of synthetic Ins(1,4,5)P₃ mimics based on part of adenophostin, including ribophostin 6 and inositol ribophostin 10.

Since d-chiro-inositol adenophostin 4 is more potent than adenophostin A,²⁰ we speculated that replacing the glucose motif of ribophostin with a d-chiro-inositol derivative might also increase the potency of 6. We now report a synthesis of d-chiro-inositol ribophostin 10 and demonstrate that it both binds with greater affinity than ribophostin to Ins(1,4,5)P₃R and more potently evokes Ca²⁺ release, confirming the value of our approach. Moreover, the potency difference between 6 and 10 is considerably greater than that observed between 2 and 4, and unlike Ins(1,4,5)P₃, 10 is efficacious and exhibits metabolic resistance in an assay using patch-clamped whole cells.

Results and Discussion

Chemistry

Previously, we discussed why a rigidly protected myo-inositol ring was key for the synthesis of d-chiro-inositol adenophostin²⁰ and that a rigid inositol triflate l-3-O-trifluoromethylsulfonyl-6-O-p-methoxybenzyl-1,2:4,5-di-O-isopropylidene-myo-inositol 11 can undergo an S_N2 reaction with a hard nucleophilic ribose derivative as its anion. Thus, triflate 11 (vide infra for an alternative formal identification of its absolute configuration as employed here) was coupled with the known suitably protected ribose derivative 12(20,31) in tetrahydrofuran (THF)-hexamethylphosphoramide (HMPA) with NaH as base²⁰ (Scheme 1). The fully protected ribose–inositol sec–sec ether derivative 13 was formed in a reasonable yield (62%), and the more acid-sensitive trans-isopropylidene was then removed selectively and rapidly in the presence of ethylene glycol and catalytic p-toluenesulfonic acid (10–15 min) to give diol 14 in 85% yield. The presence of the ribose now attached to the d-chiro-inositol derivative via an ether seemed to affect the ease with which the trans-isopropylidene acetal was removed by acid in the presence of ethylene glycol. Possibly, the axial ribose of derivative 13 may increase the strain of the trans-acetal and could facilitate its removal. Our previous studies³² showed that a longer reaction time (40 min) and more tetraol (from removal of both diol-protecting groups) resulted from a simple 1,2:4,5-di-O-isopropylidene derivative with allyl protection at the 3- and 6-positions. Monobenzylation was then achieved by the formation of a tin acetal on diol 14 in the presence of dibutyltin oxide. The addition of cesium fluoride and excess benzyl bromide gave the separable monobenzylated products 15 and 16 in roughly equal amounts, demonstrating that no selective monobenzylation was achieved under these conditions. Compound 16 was identified by ¹H NMR spectroscopy since it showed a double triplet where the proton at C-3-H (of the d-chiro-inositol motif) flanked by two adjacent axial protons also had an extra coupling to the hydroxyl group C-3-OH that exchanged with D₂O. If the exchangeable OH was at C-2-H, next to the axial ribose–inositol ether bond, there should be a double double doublet, which upon exchange should give a double doublet. The next step required the removal of the two p-methoxybenzyl groups. Deprotection under acidic conditions was not an option as this would also remove the cis-isopropylidene moiety. Thus, 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ) in dichloromethane–water was used, giving the protected triol 17 for phosphorylation in a reasonable yield (63%) after 1 h (Scheme 1).

Scheme 1. Coupling of the Suitably Protected Inositol Triflate 11 and Ribose 12 Building Blocks, Selective Deprotection, and Benzylation To Give the Separable Regioisomers 15 and 16 and Selective Deprotection of 16 To Afford Intermediate 17 for Phosphorylation.

Phosphitylation of 17 in CDCl₃ provided trisphosphite 18 (Scheme 2). ³¹P NMR spectroscopy showed the expected AB system for the vicinal bisphosphite motif,³³ with one half centered around δ = 140.5 and the other at δ = 138.8, J = 6.0 Hz. Intermediate 18 was then oxidized with mCPBA to give the trisphosphate 19 in good yield (72%). Compound 19 was then deblocked at room temperature using palladium hydroxide and hydrogen. The best deprotection method was to leave the mixture stirring for 1 week at room temperature; the isopropylidene group was removed due to the acidic nature of the adjacent phosphates. Purification of the product, first using a gradient of triethylammonium bicarbonate (TEAB) buffer, then over RP18 resin, provided the target trisphosphate 10 in 43% yield. The product was quantified for biological assays after RP18 purification using the Briggs test.^34,35

Scheme 2. Phosphitylation of 17, Oxidation of Resulting 18 to 19, and Complete Deprotection of 19 To Give the Target Inositol Ribophostin 10.

The synthesis and separation of camphanate 20, leading to compound 22 and the more polar camphanate derivative (not shown but used to synthesize compound 11), were discussed previously.²⁰ The p-methoxybenzyl (PMB) group of 20 was first removed using DDQ in a dichloromethane–phosphate buffer. Since 22 is highly soluble in both dichloromethane and water, it would be difficult to extract and purify. The reaction was also buffered because the trans-4,5-isopropylidene group may be easily removed from 20 in its conversion to 21 in the presence of the acidic phenol DDQH₂. The camphanate derivative was then deacylated in methanolic sodium hydroxide. The less polar diastereoisomer was identified as being a derivative of d-1,2:4,5-di-O-isopropylidene-myo-inositol 22, by virtue of its rotation, and close to the literature values.³⁶⁻³⁸ The other, more polar diastereoisomeric camphanate was isolated by chromatography, then deacylated, as previously described,²⁰ giving l-6-O-p-methoxybenzyl-1,2:4,5-di-O-isopropylidene-myo-inositol, and used for the synthesis of inositol triflate 11 (Scheme 3).

Scheme 3. Determination of the Absolute Configuration of the Partially Protected d-1,2:4,5-di-O-isopropylidene-myo-inositol Diol 22 from the Separated Diastereoisomer 20.

Biology

In permeabilized avian DT40 cells expressing only rat Ins(1,4,5)P₃R1 (DT40-Ins(1,4,5)P₃R1 cells), maximally effective concentrations of Ins(1,4,5)P₃1, ribophostin 6, and d-chiro-inositol ribophostin 10 each released the same fraction (ca. 65%) of the intracellular Ca²⁺ stores. Ins(1,4,5)P₃ and ribophostin were equipotent, but d-chiro-inositol ribophostin was ca. 7-fold more potent than ribophostin (Figure 3 and Table 1). Similar results were obtained from analyses of Ca²⁺ release in permeabilized mammalian cells expressing only rat Ins(1,4,5)P₃R1 (HEK-Ins(1,4,5)P₃R1 cells), where 1, 2, 6 and 10 each released the same fraction (ca. 70%) of the intracellular stores, and d-chiro-inositol ribophostin was 3-fold more potent than ribophostin and only 2-fold less potent than adenophostin A (Figure 4 and Table 2).

Figure 3.

Effects of ligands on Ca²⁺ release from the intracellular stores of permeabilized DT40-Ins(1,4,5)P₃R1 cells. Results, expressed as a percentage of the Ca²⁺ content of the intracellular stores, show the effects of the indicated concentrations of each ligand. Mean ± SEM from three independent experiments (summarized in Table 1).

Table 1. Effects of Ligands on Ca²⁺ Release through Ins(1,4,5)P₃R1 Expressed in DT40 Cells^a.

	pEC50 (EC50) (nM)	Ca2+ release (%)	h
Ins(1,4,5)P3 (1)	6.88 ± 0.18 (133)	63 ± 5	2.0 ± 0.8
d-chiro-inositol ribophostin (10)	7.93 ± 0.06 (12)	70 ± 2	1.6 ± 0.4
ribophostin (6)	7.05 ± 0.10 (88)	64 ± 7	1.4 ± 0.4

^aResults (from Figure 3) show mean ± standard error of the mean (SEM) (pEC₅₀, Ca²⁺ release (%) and h) and means (EC₅₀) from three independent experiments. EC₅₀, half-maximal effective concentration; pEC₅₀, -logEC₅₀; h, Hill coefficient. *P < 0.05 was considered significant. None of the mean values for Ca²⁺ release or h differed significantly between ligands. There were significant differences between pEC₅₀ values for Ins(1,4,5)P₃ versus d-chiro-inositol ribophostin and ribophostin versus d-chiro-inositol ribophostin 10.

Figure 4.

Effects of ligands on Ca²⁺ release from the intracellular stores of permeabilized HEK-Ins(1,4,5)P₃R1 cells. Results show mean ± SEM from nine independent experiments, each performed in duplicate (summarized in Table 2).

Table 2. Comparison of Binding and Functional Effects of Ins(1,4,5)P₃R Ligands^a.

	Ca2+ release			binding
	pEC50 (EC50) (nM)	release (%)	h	pKd (Kd) (nM)	h	EC50/Kd (95% CI)
Ins(1,4,5)P3 (1)	6.68 ± 0.09 (209)	76 ± 2	1.0 ± 0.1	8.06 ± 0.03 (8.71)	1.1 ± 0.2	24 (11–56)
adenophostin A (2)	7.57 ± 0.11 (27)	76 ± 3	1.0 ± 0.3	8.86 ± 0.14 (1.38)	1.2 ± 0.2	19 (6–58)
d-chiro-inositol ribophostin (10)	7.28 ± 0.12 (52)	71 ± 1	1.1 ± 0.1	8.49 ± 0.03 (3.24)	1.0 ± 0.1	16 (5–49)
ribophostin (6)	6.81 ± 0.10 (155)	72 ± 2	1.0 ± 0.1	8.12 ± 0.03 (7.59)	0.9 ± 0.1	20 (8–51)

^aSummary of the effects of ligands on Ca²⁺ release from the intracellular stores of HEK-Ins(1,4,5)P₃R1 cells (from Figure 4) and on [³H]-Ins(1,4,5)P₃ binding to cerebellar membranes (from Figure 5). Results show mean ± SEM (pEC₅₀, Ca²⁺ release (%), pK_d and h) and means (EC₅₀, K_d) from nine (Ca²⁺ release) or three (binding) experiments. EC₅₀/K_d ratios show means with 95% confidence intervals (CI). P < 0.05 was considered significant. None of the mean values for Ca²⁺ release, h or EC₅₀/K_d ratios, differed significantly between ligands. There were significant differences between mean pEC₅₀ values for Ins(1,4,5)P₃ versus adenophostin A, Ins(1,4,5)P₃ versus d-chiro-inositol ribophostin, adenophostin A versus ribophostin and d-chiro-inositol ribophostin versus ribophostin; and between mean pK_d values for Ins(1,4,5)P₃ versus adenophostin A, Ins(1,4,5)P₃ versus d-chiro-inositol ribophostin, and adenophostin A versus ribophostin.

In equilibrium-competition binding assays using [³H]-Ins(1,4,5)P₃ and cerebellar membranes, which express large amounts of Ins(1,4,5)P₃R1s, the affinity of d-chiro-inositol ribophostin for Ins(1,4,5)P₃Rs (defined by the equilibrium dissociation constant, K_d) was 2.3-fold greater than that of ribophostin and only 2-fold lower than that of adenophostin A (Figure 5 and Table 2). As reported previously,³⁹ a comparison of the ratio of the concentrations of ligand required to evoke half-maximal Ca²⁺ release (EC₅₀) and occupy 50% of Ins(1,4,5)P₃Rs (K_d), the EC₅₀/K_d ratio, can be used to report efficacy, that is, the ability of a ligand to activate Ins(1,4,5)P₃R once it has bound. The ratios were similar for all four ligands (Table 2), and since Ins(1,4,5)P₃ and adenophostin A are full agonists,³⁹ we conclude that ribophostin and d-chiro-inositol ribophostin are also full agonists.

Figure 5.

Equilibrium-competition binding to cerebellar membranes using [³H]-Ins(1,4,5)P₃ and the indicated concentrations of competing ligands. Results show mean ± SEM from three independent experiments (summarized in Table 2).

These results demonstrate that replacing the glucose moiety of ribophostin with an inositol substantially increases the affinity of the resulting ligand, d-chiro-inositol ribophostin, for Ins(1,4,5)P₃Rs without compromising efficacy. d-chiro-Inositol ribophostin is the most potent ribophostin-type analogue synthesized to date.

Structure–Activity Relationships (SAR)

Structure–activity relationship (SAR) parameters are summarized in Figure 6, and those for Ins(1,4,5)P₃ have been well rehearsed^5,7 (Figure 6a). Adenophostin A data (Figure 6b) are also more relevant for considering the activity of 10 since it possesses the core disaccharide motif of ribophostin (Figure 6c). Figure 6 demonstrates that apparently similar structural modifications do not necessarily correlate across the three ligand classes. Since compound 10 is a hybrid of Ins(1,4,5)P₃ and ribophostin and approaching the potency of adenophostin for Ca²⁺ release, a summary of the various strategic functional groups and the effect of various positional modifications within these molecules is given in Figure 6.

Figure 6.

Comparative summary of structure–activity parameters for analogues of (a) Ins(1,4,5)P₃, (b) adenophostin A, and (c) ribophostin illustrating major differences between the three ligand classes.

Exploration of a potential “supra-optimal” phosphate group in adenophostin A led to the synthesis of phosphorylated disaccharides⁴⁰ with conformational restriction, for example, sucrose 3,4,3′-trisphosphate (23), α,α′-trehalose 2,4,3′,4′-tetrakisphosphate (24), and α,α′-trehalose 3,4,3′,4′-tetrakisphosphate (25) (Figure 7). Comparing the three structures for optimal positioning of the single phosphate to that of the 2-phosphate on the ribose of ribophostin, compound 23 is 25-fold less potent for Ca²⁺ release, and therefore with the phosphate not in an optimal position. Compound 24 is only 6-fold weaker for Ca²⁺ release than ribophostin,⁴⁰ and compound 25 is 11.5-fold weaker for Ca²⁺ release and less potent than 24.

Figure 7.

Disaccharide polyphosphate agonists to explore conformational restriction.

The Ins(1,4,5)P₃R binding²¹ model suggested mimicry of the key structural components of Ins(1,4,5)P₃, with binding enhanced by the adenine base of adenophostin A interacting with Arg504. This is supported by mutagenesis data.¹² The N6 position can be modified with little effect on activity²¹ and, while substantial modifications to the adenine base are tolerated,^16,17,41 a bicyclic ring system is preferred. The two sugar hydroxymethyl groups are not important for potent activity, and ribose 2′-phosphate, while not critical, mimics the 1-phosphate of Ins(1,4,5)P₃.¹¹ The idea that this phosphate may facilitate binding in a supra-optimal fashion was discounted in favor of synergy between the adenine base and the phosphate group.¹² The vicinal glucose bisphosphate is critical for potent activity, and removing either the 3′-phosphate or 4″-phosphate massively decreases activity, but the 4′′-phosphate is less critical.^11,13,22 Acyclophostin 26, an analogue that can be viewed as either an “opened” ribose or similar to 5 but with an additional adenine motif, is only 14–17-fold less potent than adenophostin A⁴⁵ and a pH-dependent partial agonist. Transposing the 3″-phosphate of adenophostin to the glucose 2′′-hydroxyl group gave a 2,2′,4-trisphosphate 27 (Figure 8) that is surprisingly only 12-fold less potent than Ins(1,4,5)P₃ and was perhaps the most surprising of all analogues to date.⁴⁵ However, with an understanding of the structure of the IBC, this could subsequently be rationalized without disturbing the original binding model.^13,21

Figure 8.

Structures of the ribose ring-opened adenophostin analogue acyclophostin 26 and the active regioisomeric 2,2′,4-trisphosphate adenophostin analogue 27.

Figure 6c illustrates the SAR for ribophostin for which modifications have been synthesized. The skeletal tetrahydrofuran derivative such as 8 (Figure 2) exhibits a potency close to Ins(1,4,5)P₃ and ribophostin.¹¹ There is little effect when the primary hydroxymethyl groups are removed either individually or together.¹⁹ If 3,4-bisphosphate is replaced with bis-phosphorothioate,¹⁹ EC₅₀ drops 5-fold.

Compound 28 (Figure 9) is a diastereoisomer of compound 8 (Figure 2). 28 has a very low potency in Ca²⁺-release assays,^19,30 and 29, with β-stereochemistry, releases Ca²⁺ with only ∼19-fold lesser potency than Ins(1,4,5)P₃.^19,30 The reasons for this are unclear. Xylopyranoside 2′,3,4-trisphosphate 30 has 17-fold lower affinity than Ins(1,4,5)P₃, but 2-deoxy-xylopyranoside 2′,3,4-trisphosphate 31 has 2000-fold lower affinity, indicating the importance of the pyranose 2-hydroxyl group (equivalent to the C-6-OH of Ins(1,4,5)P₃).⁴⁶

Figure 9.

Structures of minimal xylopyranoside-based agonists.

Thus, all three ligands exhibit diverse SAR features and, while these structural modifications on activity individually have been extensively reviewed,^5,47 they have not before been juxtaposed between analogue classes as in Figure 6.

Two sterically constrained phosphorylated epimers (R)-32 and (S)-33 and the “spirophostin” compounds (R)-34 and (S)-35 were also synthesized (Figure 10).^48,4932 and 33 were, respectively, 18- and 14-fold less potent than Ins(1,4,5)P₃, and 34 and 35 were 6-fold and ca. 12-fold less potent for Ca²⁺-release. No compound (28−35) is as potent as ribophostin, which behaves similarly to Ins(1,4,5)P₃.⁴⁰

Figure 10.

Synthetic sterically constrained epimers to explore the conformational restriction at the potentially supraoptimal nonvicinal phosphate group.

d-chiro-Inositol adenophostin 4 was the first adenophostin A analogue of its kind to possess a glucose-to-inositol replacement, offering increased possibilities for structural diversity and engineering by virtue of its extra hydroxyl group replacing the pyranoside oxygen. It was more active than adenophostin A itself, albeit by a relatively modest margin.²⁰ The simplest interpretation is that the d-chiro-inositol trisphosphate moiety of 4 better fits into the receptor than the glucose-based mimic of adenophostin A, consistent with our original binding model.²¹ Similarly, d-chiro-inositol ribophostin 10 is the first ribophostin analogue of its kind where the glucose 3,4-bisphosphate moiety is replaced with a d-chiro-inositol 3,4-bisphosphate unit. Different chemistry is used to form the hemiacetal of ribophostin 6 and the sec–sec ether of 10. The main difference between 6 and 10 is that the 5- and 6-positions of d-chiro-inositol (mimicking the 3- and 2-positions of Ins(1,4,5)P₃ in our model²⁰) replace the 5-hydroxymethyl group and the pyranoside oxygen. Figure 6a–c shows how individual positions on Ins(1,4,5)P₃, ribophostin, and adenophostin influence activity. It might be expected from this that there would be little gain in activity by replacing a pyranoside oxygen by a C2–OH group. Similarly, the loss of a hydroxymethyl group would not be expected to greatly affect activity, and its replacement by an inositol hydroxyl group should not have any marked effect.

In practice, however, the effects are more marked than expected and exceed those observed in the case of 4. When the potency of adenophostin A 2 is compared to that of its counterpart inositol adenophostin 4, 4 is some 1.6-fold more potent in Ca²⁺-release assays.²⁰ Similarly, replacing the glucose unit of ribophostin with an inositol motif enhances potency, but more appreciably. Thus, in Ca²⁺-release assays in two systems, inositol ribophostin 10 was ca. 7- and 3-fold more potent than ribophostin 6 and 11-fold and 4-fold more potent than Ins(1,4,5)P₃, approaching adenophostin A in its activity. 10 also had a greater affinity than Ins(1,4,5)P₃ for Ins(1,4,5)P₃R1. Thus, glucopyranoside replacement can further increase the biological activity, presumably through a better fit of the new inositol bisphosphate moiety with the receptor than the original glucose bisphosphate mimic derived from adenophostin A. Perhaps also the lack of a base motif as in 4 allows more conformational flexibility for 6 in its search for an optimal binding conformation. It is conceivable, therefore, from our binding model²¹ and related work implicating a specific role for the adenine base of adenophostin^{12,13,18,22,23} that conformational movement of adenophostin 2 (and therefore perhaps also of inositol adenophostin 4, if assumed to bind similarly in the nucleotide region) may be more restricted in its extended binding mode, influencing options available for the glucose or inositol bisphosphate components. By contrast, the simpler 4 could be more flexible and more able to optimize its fit in the likely more important region of Ins(1,4,5)P₃R. Note also that when comparing ribophostin 6 with Ins(1,4,5)P₃1 in activity, the former appears to be ca. 1.3- to 1.5-fold more potent. The precise reasons why a ligand without an adenine base, but now based on an inositol–sugar polyphosphate hybrid rather than a sugar–sugar counterpart, should significantly exceed Ins(1,4,5)P₃ in potency need further investigation. Since the conformation of disaccharides and a polyphosphate derivative such as 6 is mainly determined by the populations of rotamers around the glycosidic linkage, it could be pertinent to consider how changing this linkage to a sec–sec ether in 10 might change this. However, 10 appears to be the most potent simple (i.e., nondimeric and non-adenophostin A-based) small-molecule Ca²⁺-mobilizing analogue yet synthesized. Also, its possession of an axial Ins(1,4,5)P₃ 2-hydroxyl mimic, in place of a pyranoside oxygen as in 4 affords the same extra synthetic flexibility for further structural elaboration that might allow access to the cleft of the Ins(1,4,5)P₃R clam by a suitably tailored pendant group.

We also compared the relative abilities of 10 and its nucleotide parent 4 in a functional sense to activate I_CRAC in rat basophilic leukemia RBL-2H3 cells in comparison to Ins(1,4,5)P₃ and adenophostin A. We selected a concentration of 200 nM for Ins(1,4,5)P₃ because this dose is just below the threshold for I_CRAC activation.⁵⁰ Dialysis with 200 nM adenophostin A evoked a large I_CRAC after a short delay (Figure 11b), as reported previously.^50,51 The current–voltage relationship, taken at steady state, is shown on the right and reveals the strong inward rectification and positive reversal potential that are hallmarks of I_CRAC. Current amplitude from several cells is shown in the plot in Figure 11e. Inositol adenophostin 4 activated I_CRAC with similar kinetics to adenophostin A, and the amplitude of the current (at −80 mV) was similar to that evoked by adenophostin A (Figure 11a). Inositol ribophostin 4 also activated I_CRAC but after a slightly longer delay, and the current typically attained a smaller amplitude (Figure 11d). We thus demonstrate for the first time that inositol adenophostin 4 is able to elicit a functional store-operated Ca²⁺ current I_CRAC in patch-clamped cells, similar to adenophostin A 2 (Figure 11b), an activity downstream of the Ins(1,4,5)P₃R, whereas Ins(1,4,5)P₃ itself (Figure 11d) is inactive. Inositol ribophostin 10 (Figure 11c) is also able to stimulate this current. As is apparent from Figure 11e, inositol adenophostin is thus functionally similar to adenophostin and both are slightly more potent than inositol ribophostin. All three are considerably more potent than Ins(1,4,5)P₃ under these conditions with its lack of activity presumably the result of extensive metabolism. This confirms not only the high activity of 10 but suggests it to be substantially hydrolysis-resistant in whole cells, potentially enhancing its value along with 4 as an investigative tool for chemical biology.

Figure 11.

Activation of the store-operated Ca²⁺ current I_CRAC in RBL-2H3 cells following whole-cell dialysis with the Ins(1,4,5)P₃R ligands (a) d-chiro-inositol adenophostin 4, (b) adenophostin A 2, (c) d-chiro-inositol ribophostin 10, and (d) Ins(1,4,5)P₃1. Representative time courses (measured at ÷ 80 mV) and current–voltage relationships are shown. 200 nM Ins(1,4,5)P₃ routinely failed to evoke a detectable whole-cell current. Data are summarized in (e) as mean ± SEM, three to five cells for each condition.

Conclusions

In summary, a concise first synthesis of d-chiro-inositol ribophostin 10, an inositol-based counterpart of the glyconucleotide natural product adenophostin analogue ribophostin has been achieved. Key features include coupling of a chiral-protected myo-inositol diol derivative activated as its triflate 11 to the chiral-protected ribose derivative 12. The chiral diol d-(−)-1,2:4,5-di-O-isopropylidene-myo-inositol 22,³⁶⁻³⁸ derived from the separable diastereoisomeric 3-O-camphanate derivative of its 6-O-p-methyoxybenzyl ether, was used to define the absolute configuration of 11 by comparison with the literature values. Subsequent elaboration of the coupled product after deprotection of the labile isopropylidene group and monobenzylation of the stannylene-mediated vicinal hydroxyl group with separation and identification of regioisomers afforded a suitably protected triol 17. After phosphitylation and oxidation, the removal of all protecting groups afforded the target 10 that was evaluated as an agonist for intracellular Ca²⁺ release through the Ins(1,4,5)P₃R in permeabilized cells stably expressing Ins(1,4,5)P₃R1. d-chiro-Inositol ribophostin was more potent than Ins(1,4,5)P₃ and, more particularly, its disaccharide parent ribophostin surprisingly approached adenophostin A in activity, being only ca. 2-fold weaker. This was also confirmed using equilibrium-competition binding studies using displacement of [³H]-Ins(1,4,5)P₃. Thus, glucopyranoside-inositol replacement can further optimize biological activity, presumably by facilitating a better fit to Ins(1,4,5)P₃R than a pyranoside-based mimic directly based on the natural product adenophostin A. 10 is the most potent simple analogue agonist of Ins(1,4,5)P₃R so far identified that does not possess the adenine base motif of adenophostin A thought to enhance receptor binding. Moreover, using the patch-clamp methodology, 10 was shown to elicit a downstream store-operated Ca²⁺ current like adenophostin A, demonstrating both its efficacy and metabolic stability. Ins(1,4,5)P₃ under the same conditions showed no response and is presumably heavily metabolized. This further underlines the potential of such ligands as chemical biology tools. Structure–activity implications are discussed in the context of Ins(1,4,5)P₃, adenophostin A, and ribophostin. d-chiro-Inositol ribophostin thus represents a new class of simple but highly potent Ins(1,4,5)P₃ analogue as a phosphorylated inositol–sugar conjugate that may be useful for further development and chemical biology applications.

Experimental Section

Ca²⁺-Release Assays

The methods used to measure Ca²⁺ release from the intracellular stores of permeabilized HEK or DT40 cells stably expressing only Ins(1,4,5)P₃R1 were exactly as reported previously.^11,20 Briefly, stable cell lines expressing rat Ins(1,4,5)P₃R1 were established from cells in which all endogenous Ins(1,4,5)P₃R genes had been disrupted. The intracellular stores were loaded with a low-affinity fluorescent Ca²⁺ indicator (Mag-fluo4), the plasma membrane of the cells was then permeabilized by incubation with saponin, and the intracellular stores were loaded with Ca²⁺ by incubating cells in 96-well plates in a cytosol-like medium containing ATP. A FlexStation III plate-reader was used to record Mag-fluo4 fluorescence at 1.44 s intervals during Ca²⁺ loading and then in the presence of cyclopiazonic acid, to inhibit further Ca²⁺ uptake, added either with the Ins(1,4,5)P₃R ligands (DT40 cells) or 60 s before addition of the ligands (HEK cells). Ca²⁺ release was calculated as a percentage of the fluorescence signal from fully loaded stores (F_full) minus the signal from empty stores (F_full – F_empty).

[³H]-Ins(1,4,5)P₃ Binding Assays

Equilibrium competition binding to membranes prepared from rat cerebellum was performed in a medium comprising 50 mM Tris, 1 mM ethylenediaminetetraacetic acid (EDTA), pH 8.3 with [³H]-Ins(1,4,5)P₃ (19.3 Ci/mmol, 1.5 nM), and the competing ligand and evaluated exactly as reported previously.³⁹

Data Analysis

Concentration–effect relationships and equilibrium competition-binding results were fitted to Hill equations (GraphPad Prism, version 5), and EC₅₀/K_d ratios were processed as previously described.³⁹ Statistical analysis used analysis of variance (ANOVA) followed by Bonferroni’s multiple comparison test (GraphPad Prism, version 5). P < 0.05 was considered significant.

Electrophysiology

Patch clamp experiments were conducted on rat basophilic leukemia RBL-2H3 cells (ATCC CRL-2256) in the tight seal whole-cell configuration at room temperature (20–24 °C) as previously described.⁵⁰ Pipettes were pulled from borosilicate glass and then Sylgard-coated and fire-polished. They had resistances of 4–6 MΩ when filled with an internal solution containing (mM) 145 Cs glutamate, 8 NaCl, 1 MgCl₂, 10 HEPES, 0.35 EGTA, 2 Mg-ATP, pH 7.2 (CsOH). The internal solution was supplemented with 200 nM Ins(1,4,5)P₃, adenophostin A, inositol adenophostin, or inositol ribophostin, as indicated. The external solution contained (mM) 155 NaCl, 10 CaCl₂, 10 CsCl, 2.8 KCl, 2 MgCl₂, 10 HEPES, 10 glucose, pH 7.4 (NaOH). Whole-cell currents were filtered using an eight-pole Bessel filter at 2.5 kHz and digitized at 100 μs. The store-operated Ca²⁺ current I_CRAC was measured at −80 mV from voltage ramps spanning −100 to +100 mV (50 ms duration, 0.5 Hz, 0 mV holding potential). Whole-cell currents were normalized to cell capacitance. Capacitative currents were automatically compensated for before each ramp by the EPC 9–2 amplifier (HEKA, Lambrecht/Pfalz, Germany). Leak currents were removed by subtracting the first one to five ramps obtained immediately after break-in.

Chemistry

Chemicals were purchased from Acros Organics, Alfa Aesar, Fisher Scientific, and Sigma-Aldrich. thin-layer chromatography (TLC) was carried out on Merck TLC aluminum sheets coated with silica 60F₂₅₄. The products were developed by dipping a TLC plate in an ethanolic solution of phosphomolybdic acid, and then heated at a high temperature. Flash chromatography was performed using silica 60 A (Fisher Scientific). Hexamethylphosphoramide (HMPA) was dried over calcium hydride and distilled under pressure. Organic solutions of compounds were dried over dry MgSO₄. All compounds were characterized by spectroscopic methods (NMR and MS), and the final compound was purified using ion-exchange chromatography with a LKB-Pharmacia Medium Pressure Ion Exchange Chromatograph using Q-Sepharose Fast Flow and a 0–2.0 M gradient of triethylammonium bicarbonate (TEAB) followed by purification over RP18 resin. Fractions containing the final compound were identified by the Briggs test,³⁴ and compounds were then quantified by this assay, using a standard curve. Briefly, column fractions containing a polyphosphate were assayed for phosphate by a modification of the assay as follows:³⁵ a molybdate solution (12.5 g of ammonium molybdate dissolved in 250 mL of water and 35 mL of concentrated H₂SO₄), a hydroquinone solution (0.5 g of hydroquinone dissolved in 100 mL of water and a drop of concentrated H₂SO₄), and a sulfite solution (20% w/v sodium sulfite in water) were prepared. Aliquots (250 μL) of the fractions were transferred into test tubes, concentrated H₂SO₄ (three drops) was added, and the samples were heated at 150 °C for 1 h. After cooling, water (500 μL), molybdate solution (500 μL), hydroquinone solution (250 μL), and sulfite solution (250 μL) were added and the samples were boiled for 7 s and allowed to cool. Phosphate-containing fractions were identified by their blue color. For quantitative analysis, samples containing known amounts of potassium dihydrogen phosphate were co-assayed with samples of unknown phosphate contents. After being processed as above, fractions were transferred to 10 mL volumetric flasks, water was added to give 10 mL of solution, and the UV absorbance at 340 nm was recorded using 3 mL quartz cells. The concentration of the unknown samples was calculated from a standard curve derived from the absorbances of the reference samples. Final compounds were used in biological assays after such quantification as their triethylammonium salts. All compounds were of purity >95% by NMR. ¹H NMR spectra were recorded at 400 MHz and 500 MHz, ¹³C NMR spectra were recorded at 100 MHz and 125 MHz, and ³¹P spectra at 202.4 MHz on a Varian Mercury VX400 NMR machine and a 500 MHz Bruker Advance III. Mass spectra were recorded on a Bruker MicrOTOF electrospray electrospray ionization time-of-flight (ESI-TOF) spectrometer with sodium formate as the standard. Melting points were recorded on a Systems Optimelt automated melting point system. Optical rotations were measured in an Optical Activity Ltd AA-10 polarimeter.

(3aS,4S,4aS,7aR,8R,8aS)-4-(((3R,4R,5R)-2-((Benzyloxy)methyl)-5-methoxy-4-((4-methoxybenzyl)oxy)tetrahydrofuran-3-yl)oxy)-8-((4-methoxybenzyl)oxy)-2,2,6,6-tetramethylhexahydrobenzo[1,2-d:4,5-d′]bis([1,3]dioxole) (13)

Dry HMPA (2 mL) was added to methyl 5-O-benzyl-2-O-p-methoxybenzyl-β-d-ribofuranoside (12, 200 mg, 0.53 mmol) and NaH (25 mg, 95%, 1.0 mmol), and the mixture was stirred under argon for 30 min and cooled with ice. l-(+)-1,2:4,5-di-O-isopropylidene-6-O-p-methoxybenzyl-3-O-trifluoromethanesulfonyl-myo-inositol (11, 555 mg, 1.07 mmol) was added as a solid, and the remainder in the weighing vessel was washed with dry THF (2 mL). The reaction mixture turned orange/yellow, and the reaction was allowed to warm to room temperature as the ice melted. The reaction was allowed to slowly warm to lab temperature after the addition of the triflate, and the mixture was stirred overnight, for 17 h in total.

The THF was evaporated and the solution of HMPA was poured into a saturated solution of ammonium chloride (25 mL), and the solution was left to separate. The liquid part was decanted off, and this contained mainly the HMPA with a trace amount of product and starting material. The remaining solid products were extracted with ether (4 × 25 mL) after a further addition of saturated ammonium chloride (25 mL). Purification by chromatography 40–60 °C petroleum ether/EtOAc (2:1) provides the product (13, 241 mg, 62%). [α]_D = +33.33°, 20 °C, (c = 1.74, CHCl₃).

¹H NMR (500 MHz, CDCl₃) 1.24, 1.31, 1.33, 1.42 (12 H, 4 s, 4 × CH₃), 3.29 (3 H, s, ribose-OMe), 3.57–3.63 (3 H, m, CH₂OBn, 1 H, H-4-Ins), 3.73 (1 H, dd, J = 2.6, 10.0 Hz, H-2-Ins), 3.76 (3 H, s, CH₂PhOMe), 3.80 (3 H, s, CH₂PhOMe), 3.88 (1 H, d, J = 4.5 Hz, H-2-ribose), 4.02 (1 H, t, J = 10.0 Hz, H-3-Ins), 4.19–4.25 (2 H, m, H-4-ribose, H-4-Ins), 4.32–4.35 (2 H, m, H-1-Ins, H-6-Ins), 4.39 (1 H, dd, J = 4.5, 7.2 Hz, H-3-ribose), 4.57–4.75 (6 H, m, CH₂Ph, CH₂PhOMe), 4.81 (1 H, s, H-1-ribose), 6.84–6.88 (4 H, m, 2 × CH₂PhOMe), 7.25–7.36 (9 H, m, CH₂Ph, CH₂PhOMe).

¹³C NMR (125 MHz, CDCl₃) 25.73, 26.79, 27.20, 27.70 (4 q, 2 × C(CH₃)₂), 54.89, 56.24, 56.25 (3 q, 2 × CH₂PhOMe, ribose-OMe), 70.88, 71.46, 72.19, 73.37 (4 t, CH₂Ph, CH₂PhOMe, CH₂OBn), 72.80, 75.20, 76.69, 77.90, 79.78, 80.12, 80.37, 80.51, 81.76, 106.20 (10 × -C-H ring carbons for inositol and ribose), 109.25, 111.19 (C_q, 2 × C(Me)₂), 113.53, 113.81 (C-H, 2 × CH₂PhOMe), 127.53, 128.32, 129.27, 129.50 (d, C-H, CH₂Ph, CH₂PhOMe), 130.22, 130.59, 138.07 (C_q, s, CH₂Ph, CH₂PhOMe), 159.01, 159.21 (C_q, s, CH₂PhOMe). M/z C₄₁H₅₃O₁₂ [M + H]⁺ expected 737.3532; found 737.3532: C₄₁H₅₂NaO₁₂ [M + Na]⁺ expected 759.3351; found 759.3340.

(3aS,4R,5R,6R,7S,7aR)-4-(((3R,4R,5R)-2-((Benzyloxy)methyl)-5-methoxy-4-((4-methoxybenzyl)oxy)tetrahydrofuran-3-yl)oxy)-7-((4-methoxybenzyl)oxy)-2,2-dimethylhexahydrobenzo[d][1,3]dioxole-5,6-diol (14)

The fully blocked compound (13, 381 mg, 517 μmoles) was dissolved in methylene chloride (25 mL) followed by the addition of a catalytic amount of PTSA (5 mg, 0.025 mmol) and ethane 1,2-diol (38 mg, 0.612 mmol), 1.18 equiv. The mixture was stirred at room temperature, and the reaction was complete within 10–15 min and monitored by TLC [EtOAc/petroleum ether (2:1)], R_f = 0.44 for the diastereoisomer 14. The product was purified by flash chromatography using a combiflash and gradient of EtOAc/petroleum ether 2:1. Yield 306 mg, 85%. [α]_d 23 °C = +63.2°, (c = 2.5, CHCl₃).

¹H NMR (500 MHz, Me₂SO) 1.20, 1.32 (6 H, 2 s, C(CH₃)₂), 3.21 (3 H, s, ribose-OMe), 3.26 (1 H, dd, J = 7.2, 8.6 Hz, H-Ins), 3.46–3.58 (3 H, m, CH₂OBn, H-5-ribose, H-2 or H-3-Ins), 3.66 (1 H, dt, J = 2.2, 4.7 Hz, H-2 or H-3-Ins, D₂O exch.), 3.73, 3,73 (6 H, 2 s, CH₂PhOMe), 3.77 (1 H, dd, J = 2.1, 5.1 Hz, H-Ins), 3.87 (1 H, d, J = 4.7 Hz, H-2-ribose), 4.02 (1 H, m, H-4-ribose), 4.13 (1 H, dd, J = 6.8, 8.5 Hz, H-Ins), 4.20–4.25 (2 H, m, H-Ins and H-ribose), 4.51–4.67 (6 H, m, CH₂Ph, 2 × CH₂PhOMe), 4.85 (1 H, s, H-1-ribose), 4.97 (1 H, d, J = 4.7 Hz, D₂O exch., Ins-OH), 5.15 (1 H, d, J = 5.1 Hz, D₂O exch., Ins-OH), 6.87–6.89 (4 H, m, 2 × CH₂PhOMe), 7.24–7.32 (9 H, m, CH₂Ph, 2 × CH₂PhOMe).

¹³C NMR (125 MHz, Me₂SO) 25.47, 27.71 (2 q, C(CH₃)₂), 54.28, 55.02, 55.02 (3 q, 2 × CH₂PhOMe, ribose-OMe), 71.14, 71.16, 71.71, 72.19 (4 t, CH₂Ph, CH₂PhOMe, CH₂OBn), 73.08, 74.04, 76.74, 77.85 (2 carbon signals), 78.38, 79.07, 80.04, 83.52, 105.74 (10 × -C-H ring carbons for inositol and ribose), 108.31 (C_q, C(Me)₂), 113.37, 113.59 (C-H, 2 × CH₂PhOMe), 127.28 (2 carbon signals), 127.32, 128.17, 129.00, 129.72 (d, C-H, CH₂Ph, CH₂PhOMe), 129.95, 130.98, 138.39 (C_q, s, CH₂Ph, CH₂PhOMe), 158.47, 158.47 (C_q, s, CH₂PhOMe).

M/z C₃₈H₄₈NaO₁₂ [M + Na]⁺ expected 719.3038; found 719.3009.

(3aS,4R,5R,6R,7S,7aR)-6-(Benzyloxy)-4-(((3R,4R,5R)-2-((Benzyloxy)methyl)-5-methoxy-4-((4-methoxybenzyl)oxy)tetrahydrofuran-3-yl)oxy)-7-((4-methoxybenzyl)oxy)-2,2-dimethylhexahydrobenzo[d][1,3]dioxol-5-ol (15) and (3aR,4S,5R,6R,7R,7aS)-6-(Benzyloxy)-7-(((3R,4R,5R)-2-((Benzyloxy)methyl)-5-methoxy-4-((4-methoxybenzyl)oxy)tetrahydrofuran-3-yl)oxy)-4-((4-methoxybenzyl)oxy)-2,2-dimethylhexahydrobenzo[d][1,3]dioxol-5-ol (16)

The diol (14, 419 mg, 0.60 mmol) and dibutyltin oxide (180 mg, 0.72 mmol) in toluene (100 mL) were heated at reflux temperature overnight for 18 h with the Dean–Stark apparatus with the removal of water. The reaction mixture was cooled and the solvents were evaporated off to give a glassy residue, which was dried overnight. Cesium fluoride (327 mg, 2.16 mmol), dry dimethylformamide (DMF) (10 mL), and benzyl bromide (240 μL, 2.0 mmol) were added to the mixture, which was stirred overnight (18 h). TLC (EtOAc/petrol, 1:1) showed two products R_f = 0.50 and R_f = 0.42. The solvents were evaporated under reduced pressure, the residue was partitioned between EtOAc and water (20 mL of each), and the organic phase was stirred with sodium hydrogen carbonate solution (20 mL) for 30 min. The resulting milky organic layer was filtered over a bed of celite, dried, and the solvent was evaporated. The resulting residue was purified by flash chromatography (EtOAc/petroleum ether, 2:3) to give the title compound as a glass (122 mg, 30.5%) (of consumed compound), (upper spot) together with the lower R_f product (124 mg, 31%) and 66 mg of starting material, thus 353 mg is used up. Yields were based upon the benzylated compound.

3-O-Benzyl Diastereoisomer (15)

¹H NMR (400 MHz, CD₃CN) 1.35, 1.47 (6 H, 2 s, C(CH₃)₂), 3.28 (3 H, s, ribose-OMe), 3.49–3.55 (2 H, m, H-Ins), 3.60–3.71 (3 H, m, -CH₂OBn, H-Ins), 3,79, 3.84 (6 H, 2 s, 2 × CH₂PhOMe), 3.87–3.92 (2 H, m, 2 × H-Ins), 4.00 (1 H, dd, J = 1.3, 4.9 Hz, H-2-ribose), 4.15 (1 H, ddd, J = 3.9, 5.8, 6.8 Hz, H-4-ribose), 4.22 (1 H, m, H-Ins), 4.28 (1 H, dd, J = 4.8, 6.7 Hz, H-3-ribose), 4.33 (1 H, dd, J = 4.7, 6.5 Hz, H-Ins), 4.58–4.78 (8 H, CH₂Ph, CH₂PhOMe), 5.05 (1 H, d, J = 1.2 Hz, H-1-ribose), 6.92 (2 H, d, J = 8.9 Hz, CH₂PhOMe), 6.93 (2 H, d, J = 8.9 Hz, CH₂PhOMe), 7.32 (2 H, d, J = 8.9 Hz, CH₂PhOMe), 7.34–7.43 (12 H, m, 2 × CH₂Ph, CH₂PhOMe).

¹³C NMR (100 MHz, CD₃CN) 25.96, 28.17 (2 q, C(CH₃)₂), 55.39 (2 carbon signals), 55.83 (3 q, 2 × CH₂PhOMe, ribose-OMe), 72.41, 72.46, 73.73, 73.77, 74.11 (5t, CH₂Ph, CH₂PhOMe, CH₂OBn), 72.23, 76.61, 79.44, 79.68, 79.90, 80.57, 81.73, 82.78, 83.14, 106.95 (10 × -C-H ring carbons for inositol and ribose), 110.08 (C_q, C(Me)₂), 114.44, 114.70 (C-H, 2 × CH₂PhOMe), 128.37, 128.47, 128.58, 128.81, 129.15, 129.30, 130.40, 131.05 (d, C-H, CH₂Ph, CH₂PhOMe), 130.50, 132.08, 139.63, 140.00 (C_q, s, CH₂Ph, CH₂PhOMe), 160.09, 160.48 (C_q, s, CH₂PhOMe). M/z C₄₅H₅₄NaO₁₂ [M + Na]⁺ expected 809.3507; found 809.3534. 3-O-Benzyl diastereoisomer: [α]_d 20 °C = +61.8°, (c = 2.48, CHCl₃).

2-O-Benzyl Derivative (16)

¹H NMR (500 MHz, CDCl₃) 1.28, 1.43 (6 H, 2 s, C(CH₃)₂), 3.28 (3 H, s, ribose-OMe), 3.40 (1 H, dd, J = 6.9, 8.9 Hz, H-4-Ins), 3.56 (2 H, d, J = 5.0 Hz, -CH₂OBn), 3.62 (1 H, dd, J = 2.6, 8.0 Hz, H-2-Ins), 3.73, 3.79 (6 H, 2 s, 2 × CH₂PhOMe), 3.85 (1 H, dd, J = 1.3, 8.3 Hz, H-2-ribose), 3.98 (1 H, dt, J = 1.8, 8.3 Hz, H-3-Ins), 4.08 (1 H, t, J = 3.2 Hz, H-1-Ins), 4.19–4.22 (1 H, m, H-4-ribose), 4.24–4.31 (3 H, m, H-3-ribose, H-6-Ins, H-5-Ins), 4.44–4.85 (8 H, m, 2 × CH₂Ph, 2 × CH₂PhOMe), 4.77 (1 H, d, J = 1.3 Hz), 6.76 (2 H, d, J = 8.7 Hz, CH₂PhOMe), 6.86 (2 H, d, J = 8.7 Hz, CH₂PhOMe), 7.15 (2 H, d, J = 8.9 Hz, CH₂PhOMe), 7.25–7.32 (12 H, m, 2 × CH₂Ph, CH₂PhOMe).

¹³C NMR (125 MHz, CDCl₃) 25.92, 27.93 (2 q, C(CH₃)₂), 55.13, 55.20, 55.25 (3 q, 2 × CH₂PhOMe, ribose-OMe), 71.06, 72.12, 72.75, 72.20, 73.32 (5t, CH₂Ph, CH₂PhOMe, CH₂OBn), 71.47, 76.05, 76.56, 79.02, 79.94, 80.27, 80.45, 80.91, 83.41, 106.47 (10 × -C-H ring carbons for inositol and ribose), 109.39 (C_q, C(Me)₂), 113.68, 113.74 (C-H, 2 × CH₂PhOMe), 127.54, 127.55, 127.66, 127.86, 128.34, 128.35, 129.53, 129.55 (d, C-H, CH₂Ph, CH₂PhOMe), 130.11, 130.75, 138.10, 138.16 (C_q, s, CH₂Ph, CH₂PhOMe), 159.12, 159.15 (C_q, s, CH₂PhOMe). M/z C₄₅H₅₄NaO₁₂ [M + Na]⁺ expected 809.3507; found 809.3526. 2-O-Benzyl diastereoisomer: [α]_d 20 °C = +19.1°, (c = 2.44, CHCl₃).

(3aR,4S,5S,6R,7R,7aS)-6-(Benzyloxy)-7-(((3S,4R,5R)-2-((benzyloxy)methyl)-4-hydroxy-5-methoxytetrahydrofuran-3-yl)oxy)-2,2-dimethylhexahydrobenzo[d][1,3]dioxole-4,5-diol (17)

A mixture of the 2-O-benzyl derivative (16, 184 mg, 0.234 mmol) and DDQ (188 mg, 0.825 mmol) was stirred in dichloromethane (9.5 mL) and water (0.5 mL) for 1 h after which a salmon-colored precipitate was observed. The reaction was diluted with dichloromethane (40 mL), and the precipitate was filtered off and washed with 10% sodium sulfite solution (2 × 20 mL), then washed with sodium hydrogen carbonate solution (20 mL) and water (20 mL), dried, filtered, and the dichloromethane was evaporated to give an oil. The oily residue was chromatographed on silica gel using a gradient of petrol to 70% ethyl acetate to give the title compound 17; yield (80 mg, 63%); R_f = 0.42 (ethyl acetate).

¹H NMR (500 MHz, CDCl₃) 1.25 1.43 (6 H, 2 s, C(CH₃)₂), 2.66 (1 H, d, J = 3.4 Hz, D₂O, exch. Ins-3-OH), 2.88 (1 H, d, J = 4.5 Hz, D₂O, exch. Ins-4-OH), 3.32 (3 H, s, ribose-OMe), 3.51–3.63 (4 H, m, CH₂OBn, H-5-ribose, H-2 or H-3-Ins), 3.86 (1 H, t, J = 2.9 Hz, H-1 or H-6-Ins), 3.89 (1 H, dt, J = 3.1, 8.8 Hz, D₂O exch. J = 8.8 Hz, H-3-Ins), 4.02–4.04 (2 H, m, Ins and/or H-ribose), 4.10–4.16 (3 H, H-Ins and/or H-ribose, ribose-2-OH, D₂O exch.), 4.18 (1 H, dd, J = 2.8, 5.4 Hz, H-Ins or H-ribose), 7.26–7.38 (10 H, m, 2 × CH₂Ph).

¹³C NMR (125 MHz, CDCl₃) 25.72, 27.83 (2 q, C(CH₃)₂), 54.92 (ribose-OMe), 71.81, 73.45, 74.10 (3t, CH₂Ph, CH₂OBn), 71.81, 72.51, 74.37, 75.77, 76.11, 78.39 (2 carbon signals), 79.20, 81.35, 108.01 (10 × -C-H ring carbons for inositol and ribose), 110.89 (C_q, C(Me)₂), 127.61, 127.72, 128.42, 128.71 (d, C-H, CH₂Ph), 136.83, 137.90 (C_q, s, CH₂Ph). M/z C₂₉H₃₈NaO₁₀ [M + Na]⁺ expected 569.2357; found 569.2366. [α]_d 22 °C = + 23.5°, (c = 1.1, CHCl₃).

Tetrabenzyl ((3aS,4R,6S,7R,7aS)-6-(Benzyloxy)-7-(((3R,4R,5R)-2-((benzyloxy)methyl)-4-((bis(benzyloxy)phosphoryl)oxy)-5-methoxytetrahydrofuran-3-yl)oxy)-2,2-dimethylhexahydrobenzo[d][1,3]dioxole-4,5-diyl) Bis(phosphate) (19)

A mixture of bis(benzyloxy)diisopropylaminophosphine (0.345 g, 1 mmol) and 1H-tetrazole (0.21 g, 3 mmol) in dry CDCl₃ (3 mL) was stirred for 15 min at room temperature. The phosphitylating complex was added to the triol (17, 65 mg, 0.119 mmol) in CDCl₃ (2 mL) and stirred for 30 min to give the P(III) intermediate (18). Solid mCPBA (345 mg, 2.0 mmol) was added to the cooled solution using a dry ice–acetone bath, and the P(III) intermediate was oxidized and stirred for a further 30 min. The solution was diluted with EtOAc (20 mL) and washed with 10% aqueous sodium sulfite solution (2 × 20 mL). The organic layer was dried (MgSO₄), and the product was purified by flash chromatography using a gradient of petroleum ether ethyl acetate 100% petroleum ether to (EtOAc/petroleum ether, 2:1), to give the trisphosphate (19, 114 mg, 72%).

¹H NMR (500 MHz, CD₃CN) 1.25 1.43 (6 H, 2 s, C(CH₃)₂), 3.55–3.69 (2 H, ABX, J = 10.7, 3.5, 5.4 Hz, H-5-ribose), 4.10–4.13 (1 H, m, obscured, H-4-ribose), 4.14 (1 H, dd, J = 2.4, 6.5 Hz, H-Ins), 4.24 (1 H, t, J = 2.2 Hz, H-1-Ins), 4.38 (1 H, dd, J = 6.9, 9.35 Hz, H-Ins), 4.46 (1 H, t, J = 6.5 Hz, H-Ins or H-ribose), 4.46 (1 H, dd, J = 2.1, 7.3 Hz, H-Ins or H-ribose), 4.53–4.69 (7 H, m, H-2-ribose, H-3-Ins, H-4-Ins, 2 × CH₂Ph), 4.83 (1 H, s, H-1-ribose), 4.96–5.10 (12 H, 3 × P(O)O(CH₂Ph)₂), 7.23–7.37 (40 H, 3 × P(O)O(CH₂Ph)₂, 2 × CH₂Ph). ³¹P NMR (202.4 MHz, CD₃CN) −1.43, −1.60, −1.77.

¹³C NMR (125.7 MHz, CD₃CN) 25.72, 27.93 (2 q, C(CH₃)₂), 55.44 (ribose-OMe), 69.91 [(P-coupled, J = 5.6), 70.17 (P-coupled, J = 5.8 Hz), 70.29 (P-coupled J = 5.7), 5.37 (P-coupled, J = 5.7 Hz), 70.49 (P-coupled, J = 5.8 Hz, 2 carbon signals), (3 × P(O)(OCH₂Ph)₂)], 71.90, 73.60, 73.80 (t, 2 × CH₂Ph, CH₂OBn), 76.43, 77.77, 78.25, 78.33 [(P-coupled, J = 4.6 Hz), 78.61 (P-coupled, J = 5.2 Hz), 79.27 (P-coupled, J = 5.3 Hz), 80.98, 81.07, 81.82 (P-coupled, J = 5.8, 7.6 Hz)], 107.04 (P-coupled, J = 4.6 Hz), 111.47 (C_q, C(Me)₂), 128.46, 128.48, 128.51, 128.52, 128.84, 129.01, 129.03, 129.15, 129.20, 129.25, 129.30, 129.39, 129.42, 129.45, 129.47, 129.49, 129.52, 129.56 (d, C-H, CH₂Ph), 137.02, 137.04, 137.07, 137.09, 137.13, 137.26, 137.33, 137.48, 137.54, 139.13, 139.53 (C_q, s, CH₂Ph). M/z C₇₁H₇₇NaO₁₉P₃ [M + Na]⁺ expected 1349.4164; found 1349.4121. [α]_d 23 °C = + 20.7°, (c = 1.9, CHCl₃).

(2R,3S,4R,5S,6S)-3,4,6-Trihydroxy-5-(((3R,4R,5R)-2-(hydroxymethyl)-5-methoxy-4-(phosphonatooxy)tetrahydrofuran-3-yl)oxy)cyclohexane-1,2-diyl bis(phosphate) (10) [d-chiro-Inositol Ribophostin (10)]

The phosphorylated compound (19, 90 mg, 67.8 μmoles) was dissolved in MeOH (4 mL), and water (1 mL) was added. Palladium hydroxide (10, 15 mg) was stirred vigorously in the presence of a balloon of hydrogen for 19.5 h. About one-third (by volume) was removed and filtered, then two drops of triethylamine were added to form the salt and ¹H NMR was conducted. At this stage, the removal of the isopropylidene was incomplete (about a third was removed). One-third of the sample was evaporated to dryness and acidified using 80% acetic acid in D₂O (5.0 mL in total), then stirred for 3 h. The NMR results indicated that it was incomplete and was stirred in 80% acetic acid for a further 6 h and then for another 2 h. The remaining compound was then left under stirring for about 1 week. After this time, methanol was removed and triethylamine was added. NMR results indicated that the isopropylidene group had been removed. The remaining compound was purified over Q-Sepharose ion-exchange resin using a gradient of 0–2 M TEAB buffer. The compound eluted between 1.1 and 1.4 M buffer and then on RP18 resin. Yield (28.9 μmoles, 42.6%).

¹H NMR (400 MHz, MeOD) 3.41 (s, ribose-OMe), 3.63 (1 H, dd, J = 6 Hz, 12 Hz, H-5a-ribose), 3.75 (1 H, dd, J = 3.6 Hz, 11.6 Hz, H-5b-ribose), 3.90 (1 H, dd, J = 3.2 Hz, 9.6 Hz, H-5-Ins), 3.98–4.05 (2 H, m, H-2-Ins, H-4-ribose), 4.09 (1 H, t, J = 3.6 Hz, H-1 or H-6-Ins), 4.17–4.20 (br m, H-3-ribose), 4.31 (1 H, app q, J = 8.8 Hz, J = 9.2 Hz, H-3-Ins), 4.48 (1 H, app q, J = 8.8 Hz, J = 9.2 Hz, H-4-Ins), 4.65 (1 H, dd, J = 4.4 Hz, H-2-ribose), 5.14 (s, H-1-ribose); 13 C NMR (125.7 MHz, MeOD) 52.25 (ribose-OMe), 64.73 (HOCH₂-ribose), 70.68 (C-6-Ins), 72.10 (C-2-Ins), 72.85 (C-5-Ins), 77.36 (C-2-ribose), 78.24 (C-4-Ins), 78.59 (C-3-Ins), 81.75 (C-3-Ribose), 82.37 (C-1-Ins), 83.13 (C-4-ribose), 108.69 (C-1-ribose); ³¹P NMR (202 MHz, MeOD) +2.56, +1.93, +0.82; M/z C₁₂H₂₄O₁₉P₃ [M – H]⁻ expected 565.0130; found 565.0136.

d-3-O-1S-(−)-Camphanoyl-1,2:4,5-di-O-isopropylidene-myo-inositol (21)

The upper diastereoisomer (20, 560.6 mg, 1.0 mmol) and DDQ (454 mg, 2 mmol) in dichloromethane 0.5 M phosphate buffer pH 7.4 (20 mL, 19:1) were stirred at room temperature overnight to give a deep green solution. The solution was diluted with dichloromethane (100 mL), and the pink-red precipitate that formed overnight was filtered off. The organic solution was then washed with a 10% solution of sodium hydrogen carbonate (2 × 100 mL) and water (100 mL). The organic layer was dried, filtered, and evaporated to give the crude product with R_f = 0.24 (ethyl acetate/petroleum ether, 1:1) and R_f of starting material = 0.54. The crude product was purified by flash chromatography (ethyl acetate/petroleum ether, 1:1 to ethyl acetate) to give the alcohol as a solid in 63% yield (279 mg). The compound was recrystallized from t-butylmethyl ether–hexane. [α]_D + 24° at temp = 22 °C, c = 0.50 (CHCl₃), mp = 223–226 °C. ¹H NMR (400 MHz, CDCl₃) 0.98 (3 H, s, CH₃- camphanate), 1.05 (3 H, s, CH₃- camphanate), 1.11 (3 H, s, CH₃-, camphanate), 1.28, 1.43 (6 H, 2 s, C-(CH₃)₂, isopropylidene), 1.45, 1.49 (6 H, 2 s, C-(CH₃)₂, isopropylidene), 1.65–1.72 (1 H, m, -CH- of -CH₂-, camphanate), 1.88–1.95 (1 H, m, -CH- of -CH₂-, camphanate), 2.04–2.10 (1 H, m, -CH- of -CH₂-, camphanate), 2.44–2.51 (1 H, m, -CH- of -CH₂-, camphanate), 2.53 (1 H, d, J = 2.6 Hz, D₂O exch. Ins-OH), 3.40 (1 H, dd, J = 9.5, 10.7 Hz, H-5), 3.88–3.93 (1 H, br ddd, 2.3, 6.5, 9.1 Hz, D₂O exch. J = 6.5, 10.7 Hz, H-6), 4.05 (1 H, dd, J = 9.5 Hz, 10.4 Hz, H-4), 4.08 (1 H, t, J = 5.8 Hz, H-1), 4.63 (1 H, t, J = 4.8 Hz, H-2), 5.20 (1 H, dd, J = 4.4, 10.5 Hz, H-3).

¹³C NMR (100 MHz, CDCl₃) 9.74, 16.49, 16.65 (3 q, -CH₃, camphanate), 25.74, 26.85, 26.94, 28.10 (4 q, 2 × C(CH₃)₂), 28.98, 30.58 (2 t, 2 × -CH₂-, camphanate), 54.71, 54.90 (2 s, C_q, camphanate), 72.10, 74.43, 74.49, 74.63, 78.14, 81.90 (6 d, 6 × inositol ring carbons), 91.02 (s, C_q, camphanate), 110.33, 112.99 (2 s, C_q, C(CH₃)₂), 167.14 (s, C_q, (O-C=O), camphanate), 178.11 (s, C_q, C=O, camphanate). C₂₂H₃₂O₉, Expected: C = 59.99, H = 7.32; Found: C = 60.0, H = 7.42. M/z C₂₂H₃₂NaO₉, Expected: 463.1939; Found 463.1950.

d-(−)-1,2:4,5-Di-O-isopropylidene-myo-inositol (22)

A mixture of d-3-O-1S-(−)-Camphanoyl-1,2:4,5-di-O-isopropylidene-myo-inositol (21, 237 mg, 0.538 mmol) and MeOH (50 mL) was heated under reflux in the presence of NaOH (73 mg, 1.82 mmol) for 30 min. The solvent was further diluted with MeOH (50 mL), and the solution was neutralized by bubbling carbon dioxide through it. Water (10 mL) was added, and the solvents were evaporated off to give a solid. Extraction of the product (CH₂Cl₂) gave the (EtOAc-CH₂Cl₂, R_f = 0.20) purified product as a white solid (115 mg, 82%) from EtOAc. [α]_D = −23°, temperature = 21 °C, c = 0.50 in CH₃CN, Lit. + 22° (c = 1.08, MeCN)³⁶ (for opposite enantiomer). Lit. −22° (c = 1, MeCN)³⁷ Lit. −21.7° (c = 0.46, MeCN);³⁸ mp 157–159 °C; Lit.³⁶ mp = 159–161 °C.

¹H NMR (400 MHz, CDCl₃) 1.36, 1.44, 1.47, 1.52 (12 H, 2 × C(CH₃)₂, isopropylidene), 2.56 (1 H, d, J = 8.6 Hz, Ins-3-OH, D₂O, exch.), 2.80 (1 H, d, J = 3.0 Hz, Ins-6-OH, D₂O, exch.), 3.31 (1 H, dd, J = 9.4, 10.7 Hz, H-5), 3.82 (1 H, t, J = 9.6 Hz, H-4), 3.88 (1 H, ddd, J = 3.0, 6.5, 10.7 Hz, D₂O exch. dd, J = 6.4, 10.7 Hz, H-6), 4.02 (1 H, ddd, 4.5, 8.6, 10.6 Hz, D₂O exch. dd, J = 4.5, 10.1 Hz, H-3), 4.06 (1 H, dd, J = 5.8, 6.4 Hz, H-1), 4.46 (1 H, t, J = 4.8 Hz, H-2).

¹³C NMR (100 MHz, CDCl₃) 25.86, 26.89, 26.90, 28.08 (4 q, 2 × C(CH₃)₂, isopropylidene), 69.80, 74.82, 77.59, 77.98, 78.03, 81.84 (6 d, inositol ring carbons), 110.29, 112,72 (2 s, C_q, 2 × C(CH₃)₂, isopropylidene).

Glossary

Abbreviations

DDQ

2,3-dichloro-5,6-dicyano-1,4-benzoquinone

HMPA

hexamethyl-phosphoramide

Ins(1,4,5)P₃

d-myo-inositol trisphosphate

Ins(1,4,5)P₃R

d-myo-inositol trisphosphate receptor

THF

tetrahydrofuran

HEK

human embryonic kidney

SEM

standard error of mean

IBC

inositol binding core

RBL

rat basophilic leukemic

TEAB

triethylammonium bicarbonate

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jmedchem.9b01986.

• Spectral data including ¹H NMR, ³¹P NMR, and ¹³C NMR for compounds, including final compound (PDF)

• Molecular formula strings (CSV)

Author Contributions

The manuscript was written by S.J.M. and B.V.L.P. with contributions from all authors, who approved the final version.

The authors declare no competing financial interest.