Efficient tin-mediated synthesis of lysophospholipid conjugates of a TLR7/8-active imidazoquinoline

The immune stimulating ability of certain antiviral/antitumor 1H-imidazo[4,5-c]quinolines¹ such as imiquimod (1, R-837; Fig. 1), which is marketed in the USA as Aldara® and Zyclara®, has been attributed primarily to the activation of Toll-like receptor (TLR) 7 in plasmacytoid dentritic cells (DCs) and B cells and the induction of type I interferons (IFN-α/β) and IFN-regulated cytokines.² The structurally related imidazoquinoline resiquimod (2, R-848) is also a ligand for TLR7 on human cells, but—in marked contrast to imiquimod—potently activates TLR8 on myeloid and monocyte-derived DCs, leading to production of proinflammatory cytokines. Since TLR7 and TLR8 are broadly expressed in DCs and other antigen presenting cells, TLR7/8 agonists may be especially useful as adjuvants in human vaccines.^3,4 However, both oral and topical preparations of imiquimod (1) and resiquimod (2) and other small-molecule TLR7/8 agonists can exhibit serious side effects.⁵ Further, since TLR7 and TLR8 receptors are located in endosomal/lysosomal compartments,⁶ cellular uptake is prerequisite for cellular activation by TLR7/8 ligands. Thus, strategies that would increase the penetration of the TLR7/8 ligand into DCs and reduce toxicity are of great interest. Lipid conjugation of nucleoside drugs⁷ is one useful strategy for membrane targeting and intracellular delivery, and is known to decrease toxic side effects, as well as facilitate incorporation of the nucleolipid into liposomes and other biodegradable nanoparticle formulations.

To our knowledge, the effect of conjugating small-molecule TLR7/8 agonists to lysophospholipids on immune cell activation has not been investigated. We were particularly interested in the synthesis of lysophospholipid conjugates of TLR7/8-active imidazoquinolines possessing different acyl groups to evaluate the effect of fatty acid structure on immunostimulatory activity. Varying the nature of the acyl chain in this phospholipid class should permit optimization of a particular formulation and route of administration.

The synthesis of lysophospholipids often involves multiple synthetic and chromatographic steps due to the introduction of the fatty acyl group onto the glycerol scaffold early in the synthesis and the potential for intramolecular acyl group migration. While acyl migration leading to regioisomers and other by-products has been mitigated by orthogonal protection of the sn-2- and sn-3-glycerol positions,^8–10 these multi-step approaches preclude the use of a common advanced intermediate (CAI) and the introduction of different acyl groups near the end of the synthesis. Herein, we describe an efficient synthesis of lysophosphatidyl derivatives of imidazoquinoline 3 via regioselective tin-mediated mono-acylation of the sn-1-position of a late-stage phosphoglycerol CAI with different acid chlorides in the penultimate step. Imidazoquinoline 3 was selected as a test compound for initial phospholipidation studies due to its known TLR7/8 activity and simplified structure relative to resiquimod 2.

To demonstrate that lysophosphatidyl derivatives of imidazoquinoline 3 are isolable and stable, we first carried out the synthesis of lysophospholipids 5 and 6 using a known orthogonal protection strategy starting from 3-p-toluenesulfonyl-sn-glycerol (7)¹⁰ and employing a one-pot, two-step phosphitylation reaction recently developed in our laboratory to install the lysophosphatidyl group (Scheme 1). Accordingly, monoacyl glycerols 8 and 9, prepared in 3 steps from 7, were treated with 2-cyanoethyl N,N,N',N'-tetraisopropylphosphordiamidite in the presence of tetrazole to give the corresponding glycerol phosphoramidites (not isolated), which were reacted in situ with imidazoquinoline 3 in the presence of imidazolium triflate (Im-OTf) to afford the desired phosphite intermediates 10 and 11 in 97% and 73% yield, respectively after chromatographic purification. Subsequent oxidation of phosphites 10 and 11 to the corresponding phosphates with t-butyl peroxide and sequential removal of the cyanoethyl and tetrahydropyranyl (THP) protecting groups provided the desired lysonucleolipids 5 and 6 in 66% and 52% yield, respectively after chromatography. Reversing the order in which the protecting groups are removed leads to the regeneration of starting material 3, presumably via initial intramolecular attack of the sn-2-hydroxyl group on the phosphotriester moiety. In this way, lysophospholipids 5 and 6 could be prepared in 8 steps and 19–33% overall yield from known tosyl glycerol 7 using the linear synthetic route shown in Scheme 1.

Next, we turned our attention to carrying out the convergent synthesis of homologous lysonucleolipids 12 and 13 via mono-acylation of a phosphoglycerol intermediate towards the end of the synthesis (Scheme 2). While regioselective monoacylation of the sn-1-position of late-stage phosphoglycerol intermediates using carbodiimide-mediated or mixed anhydride acylation protocols has typically led to low yields of lysophosphatidylated products, the regioselective derivatization of vicinal diols using reactive stannylene acetal intermediates has been widely applied in organic synthesis.¹¹ Thus, we envisioned that the differentially acylated nucleolipids 12 and 13 could be assembled by first, constructing the protected phosphoglycerol 15 through tandem phosphitylation of alcohols 3 and 14, and then—subsequent to protecting group manipulation—selective tin-mediated mono-acylation¹² of the sn-1-position of the phosphoglycerol 17.

Accordingly, commercially available (S)-1,2-isopropylidene glycerol (14) was treated with 2-cyanoethyl N,N,N',N'-tetraisopropylphosphordiamidite in the presence of tetrazole to give the corresponding glycerol phosphoramidite, which was reacted in situ with imidazoquinoline 3 in the presence of imidazolium triflate to afford the desired phosphite intermediate in 87% yield after chromatographic purification. Subsequent oxidation of the intermediate phosphite led to phosphate 15 in quantitative yield after chromatography. Due to the incompatibility of the primary aromatic amino group of 15 with stannylene and other O-acylation protocols as well as the susceptibility of the cyanoethyl group of 15 to β-elimination in the presence of the unprotected amino group,¹³ the amine was protected as the benzyl carbamate 16 in 80% yield using 1-carbobenzoxy-3-methylimidazolium trifuoromethanesulfonate (Rapoport’s reagent).¹⁴ Sequential removal of the cyanoethyl and acetonide groups of 16 with triethylamine and trimethylsilyl triflate (TMSOTf)¹⁵ then gave the phosphoglycerol common advanced intermediate 17¹⁶ in excellent yield. Reversing the order of protecting group removal or simultaneous deprotection of the 2-cyanoethyl and acetonide groups with TMSOTf / N,N-diisopropylethylamine (DIPEA)¹⁷ resulted in significantly lower yields of 17 and the formation of imidazoquinoline 18 as a significant by-product.

Consistent with literature reports with other phosphoglycerols, attempts to selectively acylate the sn-1-hydroxy group of key intermediate 17 under carbodiimide-mediated conditions with lauric or palmitic acid or with the corresponding acid anhydrides in the presence of base led to low yields (<35%) of monoacylated products 21 and 22 accompanied by significant amounts of di-acylated and other by-products, which were difficult to separate chromatographically. Not unexpectedly, phosphoglycerol 17 could be readily diacylated under standard conditions (fatty acid/ DCC, DMAP) in high yield (data not shown).

Since dibutyltin oxide (DBTO) was recently employed to selectively mono-acylate the sn-1-position of a glycerophosphoryl choline,¹² we attempted to prepare 21 via acylation of dibutylstannylene acetal 19 with lauroyl chloride. However, under the conditions of the reaction (1 equiv DBTO, refluxing i-PrOH; then lauroyl chloride, Et₃N) the intermediate stannylene 19 was converted to imidazoquinoline 18 in nearly quantitative yield, presumably via intramolecular attack of the reactive stannylene on the phosphodiester group. Similarly, treatment of 17 with catalytic amounts of dimethyltin dichloride reagent¹⁸ in the presence of potassium carbonate in anhydrous tetrahydrofuran (THF) at room temperature to form 20 also led to the formation of 18 as the major product, despite the milder reaction conditions. Gratifyingly, by employing an organic base (Et₃N) in place of potassium carbonate as demonstrated by Onomura¹⁹ in the mono-silylation of diols via dimethylstannylene intermediates, the CAI 17 was converted to the desired mono-acyl products 21 and 22 in 59% and 80% overall yield,²⁰ respectively, after treating the in situ-prepared stannylene 20 with the requisite acyl chlorides in dichloromethane at room temperature. Hydrogenolysis of the N-Cbz group then provided the desired lysophospholipids 12 and 13 in 59% and 65% yield,²¹ respectively, after chromatographic purification.

In summary, a short and convergent synthesis of lysophospholipid conjugates of TLR7/8-active imidazoquinoline 3 was developed utilizing a common advanced intermediate strategy and a reactive tin acetal to incorporate different acyl residues in the penultimate step. The synthesis of nucleolipids 12 and 13 in 2 steps and in 35–52% overall yield from CAI 17 by this method compares very favorably to the linear preparation of the homologous lysophospholipids 5 and 6 in 8 steps and 19–33% overall yield from tosylate 7. The biological activity of these lysophosphatidylated imidazoquinolines will be reported elsewhere.