Concise Synthesis of Capuramycin

Abstract

A concise total synthesis of capuramycin (1), a promising preclinical TB drug lead, is achieved by high-yield formations of the cyanohydrin 5a and 4″,5″-glycal derivative 12. Capuramycin can be synthesized in eight steps from the uridine building block 5a with >30% overall yield. The synthetic intermediates reported here are useful for generation of analogs to improve pharmacokinetic properties of capuramycin.

Graphical Abstract

Mycobacterium tuberculosis (Mtb) causes tuberculosis (TB) and is responsible for nearly two million deaths annually.¹ In particular, people who are HIV-AIDS patients are susceptible to TB infection. There are significant problems associated with treatment of AIDS and Mtb coinfected patients. Rifampicin and isoniazid (a key component of the DOTS therapy²) induce the cytochrome P450 3A4 enzyme, one of the enzymes responsible for drug metabolism, in liver which shows significant interactions with protease inhibitors for HIV infections.³ In addition, rifampicin strongly interacts with non-nucleoside reverse transcriptase inhibitors for HIV. Thus, clinicians avoid starting highly active antiretroviral therapy (HAART), which consists of three or more highly potent reverse transcriptase inhibitors and protease inhibitors, until the TB infection has been cleared.⁴ Moreover, the emergence of multidrug-resistant (MDR) strains of Mtb seriously threatens TB control and prevention efforts. Thus, there is an urgent need and significant interest in developing new TB drugs.⁵

Since peptidoglycan (PG) is an essential bacterial cell-wall polymer, the machinery for PG biosynthesis provides a unique and selective target for antibiotic action.⁶ However, only a few enzymes in PG biosynthesis such as the penicillin binding proteins (PBPs), which are inhibited by the β-lactams and glycopeptides, are extensively studied. Thus, the enzymes associated with the PG biosynthesis (i.e., MurA, B, C, D, E, and F, MraY, and MurG) are still considered to be a source of unexploited drug targets.⁷ Our interest in targets related to PG biosynthesis is MraY, which catalyzes the transformation of UDP-N-acylmuramyl-l-alanyl-γ-d-glutamyl-meso-diaminopimelyl-d-alanyl-d-alanine (Park's nucleotide) to prenylpyrophosphoryl-N-acylmuramyl-l-Ala-γ-d-glu-meso-DAP-d-Ala-d-Ala (lipid I). MraY is inhibited by nucleoside-based complex natural products such as muraymycin, liposidomycin, caprazamycin, and capuramycin.⁸ Capuramycin (1) and its analogs exhibited significant mycobacterial growth inhibitory activities in vitro and in vivo (Scheme 1) and very low toxicity in mice.⁹ Moreover, a capuramycin analog killed M. tuberculosis much faster than other first-line TB drugs (>90% of the bacilli were killed within 48 h) and thus could dramatically reduce the time frame for effective anti-TB chemotherapy.¹⁰ Therefore, capuramycin and its congeners have been considered important lead molecules for the development of a new drug for MDR-TB infections. However, extensive SAR studies of capuramycins to improve pharmacokinetic properties have been limited due to difficult modifications of the desired position(s) of the complex natural product with biologically interesting functional groups. Consequently, it is essential to establish a concise and convergent synthesis of capuramycin that is amenable to synthesis of analogs for SAR studies. To date, Knapp and Nadan have reported the only total synthesis of capuramycin.¹¹ Their synthesis requires 22 linear steps from diisopropylidene-d-glucofuranose, and relatively lengthy synthesis of the manno-pyranuroate glycosyl donor. We now report a concise total synthesis of capuramycin that is amenable to performing comprehensive medicinal chemistry studies based on the core structure of capuramycin.

Scheme 1.

Preliminary Studies on Glycosylation of 5a and the Revised Synthetic Strategy for Capuramycin (1)

In our preliminary synthetic studies of the advanced intermediate ii (Scheme 1), glycosylation of the cyanohydrin 5a with β-d-manno-pyranuronate imidate i provided the desired α-linked mannuronic acid derivative with very low yield (5–15%) even after extensive optimization efforts. Moreover, E2 elimination to form the 4″-enopyranosiduronic acid derivative ii did not give satisfactory results (35–45% with DBU).¹² Based on the preliminary studies summarized in Scheme 1, we revised the synthetic route for capuramycin (1) in which we envisioned performing glycosylation of the cyanohydrin 5a with the tetraacetyl thio-α-D-mannopyranoside 6, followed by one-pot oxidation of the C-6″ alcohol and elimination of the acetate at the C-4″ position to form the α,β-unsaturated aldehyde 12. The revised synthetic route for capuramycin illustrated in Scheme 1 would allow for the synthesis of the intact molecule from readily accessible building blocks with a minimum number of protecting group manipulations.

Scheme 2 illustrates our synthetic route for capuramycin (1). The partially protected uridine 2¹³ was converted to the 2-O-acetyl-3-O-methyl-uridine derivative 3 through monomethylation using ⁿBu₂SnO, acetylation, and detritylation reactions. The primary alcohol of 3 was oxidized under Pfitzner-Moffatt conditions (DCC, Cl₂CHCO₂H, DMSO)¹⁴ to provide the corresponding aldehyde 4, which was utilized after passing through a SiO₂ plug. The aldehyde 4 was unstable against Brønsted and Lewis bases (i.e., Et₃N, DABCO, cinchona alkaloids, triphenylphosphine, and triphenylphosphine oxide). Because of the instability of 4 when exposed to bases, we explored cyano addition reactions of carbonyl molecules promoted by Lewis acids¹⁵ or under neutral conditions. Several Lewis acid promoted trimethylsilylcyanations of 4 were examined. In all cases, cyanohydrin synthesis furnished a mixture of 5a and 5b in favor of undesired 5b with moderate yields.¹⁶ For example, the cyanosilyation of 4 with ZnI₂ (100 mol %) gave rise to a mixture of the cyanohydrins in 60% yield with a 5a/5b ratio of ~1:2 after desilylation. The same reaction with the (S)-or (R)-BINOL-Ti(OⁱPr)₄ complex resulted in a 5a/5b selectivity of ~1:1.5 regardless of the configuration of BINOL. On the contrary, the addition of TMSCN to 4 with 10 mol% of Ti(OⁱPr)₄ in CH₂Cl₂–H₂O (1%) provided a mixture of 5a and 5b with the 5a/5b ratio of 2:1 in 90% yield from 3. Because addition of H₂O accelerated the cyanosilylation reaction of 4, we speculated that a Ti-oxo species is responsible for this transformation.¹⁷ Although the cyanosilylation reaction catalyzed by Ti(OⁱPr)₄ in CH₂Cl₂–H₂O exhibited moderate 5a/5b selectivity, a ten-gram quantity of 5a could easily be synthesized in 60% yield via the convenient achiral Ti reagent.¹⁸ In addition, the undesired cyanohydrin 5b could be converted to the desired 5a via a modified Mitsunobu reaction (DIAD, TPP, ClCH₂CO₂H, pyridine (1:1:1:1))¹⁹ and subsequent deacylation with thiourea²⁰ in 90% overall yield.

Scheme 2.

Synthsis of Capuramycin (1)

α-Selective mannosylation of 5a with the disarmed glycosyl donor 6 required to investigate an appropriate promoter. The standard NIS/TfOH²¹-promoted mannosylation of 5a with 6 furnished the orthoester 7²² (20%) together with a 2:1 mixture of 8a and 8b (30%) (Table 1). Under the NIS/TfOH conditions the orthoester 7 was not completely rearranged to the O-glycosides even after 16 h at 0 °C. The α/β-selectivity and isolated yield of the desired 8a were not dramatically improved by replacing with other counterion sources (entries 1–3 in Table 1).²³ On the contrary, the NIS/AgBF₄ promoted mannosylation of 5a provided the orthoester 7 in quantitative yield within 15 min, which underwent the rearrangement within 16 h to afford 8a exclusively (90% isolated yield, entry 4 in Table 1). A shorter reaction time resulted in a mixture of 7 and 8a (entry 5).²⁴ The strong acid salts (i.e., ⁿBu₄NClO₄, ⁿBu₄NOTf, LiClO₄, and ⁿBu₄NBF₄), known to promote glycosylations with thioglycoside and NXS (X = I or Br),²⁵ and the other silver salts²⁶ (i.e., AgClO₄ and AgSbF₆) were not effective in mannosylation of 5a with 6 (entries 6 and 7 in Table 1). Mechanistically, tetrafluoroborate (BF₄⁻) appears to be an innocent anion, and the intermediate iii may be stabilized by the formation of an intimate ion pair with BF₄⁻.²⁷ Fluoroboric acid (HBF₄) generated in situ in the glycosylations using NIS/AgBF₄ facilitates the rearrangement of 7 to 8a through iii.²⁸

Table 1.

α-Selective Mannosylation of the Cyanohydrin 5a with 6

entry	promotor	time (h)	yield (%)a	8a/8b
1	TfOH	16	30	2/1
2	TMSOTf	16	35	2/1
3	AgOTf	16	40	2.5/1
4	AgBF4	16	90	1/0
5	AgBF4	8	55	1/0
6	Bu4NBF4	16	<5	–
7	AgClO4	16	<5	–

^aCombined yield of 8a and 8b.

We have demonstrated the synthesis of a gram quantity of the α-mannosylated cyanohydrin 8a (Scheme 2). The cyano group of 8a was efficiently hydrated using 5 mol % of the Parkins’ [Pt] complex,²⁹ furnishing the corresponding primary amide 9 in quantitative yield. Although the BOM group of the uridine moiety could be deprotected in the late stage of capuramycin synthesis using FeCl₃ in CH₂Cl₂ at 0 °C,³⁰ the best overall yield from 8a to 1 was achieved when the BOM group was removed prior to the oxidation of the primary alcohol (11→12). Selective deacetylation of the primary acetate of 9 was accomplished by using [^tBu₂SnCl(OH)]₂.³¹ Treatment of the pentaacetate 9 with 1 mol % of the [Sn] reagent in MeOH furnished the C6”-free alcohol 10 in 70% yield (90% yield based on recovering 9). Hydrogenolytic removal of the BOM group of 10 gave 11 in an almost quantitative yield. A mild oxidation-elimination reaction of 11 using modified Parikh-Doering conditions (SO₃•pyridine in a biphasic solvent system (DMSO/Et₃N = 3/1)³² provided the α,β-unsaturated aldehyde 12 in quantitative yield (determined by ¹H NMR analysis). After all volatiles were removed, the aldehyde 12 was oxidized to the corresponding carboxylic acid 13 by using NaClO₂. The resulting crude mixture was coupled with (2S)-aminocaprolactam using a standard peptide-forming reaction condition (HOAt, EDCI, and NMM) to yield 14 in 82% overall yield from 11. Saponification of 14 by using LiOH in aq THF provided capuramycin (1) in quantitative yield. The synthetic material was characterized by ¹H NMR, ¹³C NMR, and MS.^9a,33 In addition, the biological activities of capuramycin synthesized here were evaluated in vitro (IC₅₀) against Mtb MraY and in mycobacterial growth inhibitory assays (MIC).³³ These in vitro assay data showed good agreement with those reported in the literature.⁹

In conclusion, we have developed a concise synthesis of capuramycin (1) in which the versatile aldehyde 12 for the preparation of capuramycin analogs was synthesized in seven steps from 3. Each step summarized in Scheme 2 is operationally very simple and a high-yielding conversion. Moreover, the building blocks 3 and 6 can readily be synthesized in four steps from the commercially available uridine derivative and 2 steps from D-mannose. We will synthesize capuramycin analogs by (1) modifying the structure of 3, (2) reductive amination of 12 with a variety of primary amines, and (3) coupling of 13 with pharmacologically interesting primary and secondary amines. A detailed structure antibacterial activity relationship studies of new capuramycin analogs will be reported elsewhere.