A modular and enantioselective synthesis of the pleuromutilin antibiotics

See allHide authors and affiliations

Science  02 Jun 2017:
Vol. 356, Issue 6341, pp. 956-959
DOI: 10.1126/science.aan0003

A versatile synthesis of pleuromutilin

Synthetic flexibility is crucial for antibiotic development, because numerous subtle structural variations can contribute to combating resistant strains. A derivative of the fungal natural product pleuromutilin was approved a decade ago for treatment of Gram-positive bacterial skin infections; recent efforts to tune the structure for activity against Gram-negative bacteria have focused on the stereochemistry at a particular carbon center. Murphy et al. present a synthetic route to pleuromutilin that allows the configurations in that segment of the molecule to be varied, offering a distinct path for structural optimization.

Science, this issue p. 956


The tricyclic diterpene fungal metabolite (+)-pleuromutilin has served as a starting point for antibiotic development. Semisynthetic modification of its glycolic acid subunit at C14 provided the first analogs fit for human use, and derivatization at C12 led to 12-epi-pleuromutilins with extended-spectrum antibacterial activity, including activity against Gram-negative pathogens. Given the inherent limitations of semisynthesis, however, accessing derivatives of (+)-pleuromutilin with full control over their structure presents an opportunity to develop derivatives with improved antibacterial activities. Here we disclose a modular synthesis of pleuromutilins by the convergent union of an enimide with a bifunctional iodoether. We illustrate our approach through synthesis of (+)-12-epi-mutilin, (+)-11,12-di-epi-mutilin, (+)-12-epi-pleuromutilin, (+)-11,12-di-epi-pleuromutilin, and (+)-pleuromutilin itself in 17 to 20 steps.

The diterpene fungal metabolite (+)-pleuromutilin (1 in Fig. 1A) (14) inhibits the growth of predominantly Gram-positive pathogens (GPPs). Because of its unique mechanism of action, which involves binding to the highly conserved peptidyl transferase center of the bacterial ribosome, resistance to pleuromutilin is slow to develop, and it displays minimal cross-resistance with existing antibiotics (5, 6). Extensive efforts have been devoted to improving the pharmacological profile of (+)-pleuromutilin (1) by derivatization (7). Whereas the primary binding interactions of (+)-pleuromutilin (1) with the ribosome arise from its tricyclic core, the majority of semisynthetic efforts have focused on modification of its C14 side chain (6, 7). More than 3000 C14 derivatives have been prepared. These efforts culminated in the approval of retapamulin (3) in 2007 for the treatment of topical methicillin-resistant Staphylococcus aureus (MRSA) infections (8) and the development of lefamulin (currently in phase 3 clinical trials) for the treatment of community-acquired bacterial pneumonia (6).

Fig. 1 Pleuromutilins and 12-epi-pleuromutilins are promising classes of antibiotics.

(A) Structures of selected pleuromutilins and 12-epi-pleuromutilins. Pleuromutilins are active against predominantly Gram-positive pathogens, whereas 12-epi-pleuromutilins display extended spectrum activity, including activity against Gram-negative pathogens. (B) Retrosynthetic analysis of the mutilin scaffold. The synthetic fragments 7 and 8 were used in the synthesis of pleuromutilins.

Despite progress in the treatment of GPPs, the lack of antibiotics for the treatment of Gram-negative pathogens (GNPs) remains a global health concern. Various C14-modified pleuromutilins exhibit poor uptake into Gram-negative bacteria and are efficiently transported out of target cells by efflux pumps (6). Berner reported in 1986 that the C12 quaternary stereocenter of pleuromutilin could be epimerized (to a ~1:1 mixture of C12 epimers) by an unusual zinc-mediated retroallylation-allylation reaction (see 4, Fig. 1A) (9). Capitalizing on this discovery, researchers at Nabriva Therapeutics reported in 2015 that functionalization of the transposed alkene provides 12-epi-pleuromutilins with extended-spectrum activity. These compounds exhibit activity against various GNPs, including carbapenem-resistant Enterobacteriaceae (CRE) (10). Given the increasing occurrence of drug-resistant GNPs (11) and the emerging crisis of CRE (12), a flexible synthetic strategy to access 12-epi-pleuromutilins was deemed valuable. Such a strategy would overcome the inherent limitations of semisynthetic modification of (+)-pleuromutilin (1) and provide an opportunity to develop pleuromutilins with varied architecture and potentially improved antibacterial properties.

To maximize the scope of accessible derivatives, we conceived a modular approach to the 12-epi-mutilin scaffold. Our design involved late-stage construction of the macrocycle by using a conjunctive reagent that could be easily modified at positions 11 to 13 (see 5 and 6, Fig. 1B). After considerable experimentation, the neopentyl iodide 8 and the enimide 7 were developed as the conjunctive reagent and electrophile, respectively. Three remarkable total syntheses of pleuromutilin (27 to 34 steps) have been reported (1315), and many synthetic approaches to the molecule have been disclosed (1623). Our synthesis is modular in nature, which we expect will facilitate analog production. We illustrate our approach through the enantioselective synthesis of (+)-12-epi-mutilin (4), (+)-11,12-di-epi-mutilin (26), (+)-12-epi-pleuromutilin (29), (+)-11,12-di-epi-pleuromutilin (32), and (+)-pleuromutilin (1) itself in 17 to 20 steps.

To begin our synthesis of the enimide 7, we developed a method for asymmetric conjugate addition–C-acylation, which relies on the unique reactivity of zincate enolates (Fig. 2A) (24). This reaction entails copper-catalyzed enantioselective 1,4-addition of dimethylzinc to cyclohex-2-ene-1-one (9), in situ activation of the resulting zinc enolate with methyllithium, and C-acylation with methylcyanoformate. The β-ketoester product undergoes diastereoselective α-methylation (25) to provide 10 with 71% yield (two steps), 97:3 enantiomeric ratio (er), and >20:1 diastereomeric ratio (dr). Deprotonation of 10 and trapping of the resulting enolate with N-phenyltriflimide provided the vinyl triflate 11 (88%), which was converted to the dienone 12 (83%) via a palladium-catalyzed carbonylative coupling (26). Copper-catalyzed Nazarov cyclization (27) then formed the hydrindanone 13 as a single double-bond regioisomer (88%). In agreement with earlier studies (19), extensive experimentation was required to productively functionalize the hydrindanone 13. Conjugate addition of diethylaluminum cyanide (28) to 13 proceeded with 3:1 selectivity at C9 (pleuromutilin numbering), but these epimers were difficult to separate on preparative scales. Fortunately, we found that the undesired minor epimer could be selectively reduced in situ with di-iso-butylaluminum hydride. Treatment of the unpurified product mixture with dilute sodium hydroxide then effected quantitative inversion of the C4 stereocenter to yield the desired cis-hydrindanone 15, which was isolated as a single diastereomer (65% yield from 13). The relative stereochemistry of 15 was confirmed by x-ray crystallography. Protection of the ketone as the ethylene glycol ketal then provided 16 (84%) (29). Selective functionalization of the nitrile substituent in 16 was challenging because of its steric congestion and the similar reactivity profile of the ester substituent. Ultimately, we found that the addition of excess methyllithium to 16, followed by di-tert-butyl dicarbonate, provided the cyclic enimide 7 (80%). This cascade reaction likely comprises the addition of methyllithium to the nitrile, intramolecular cyclization of the resulting anion 17, deprotonation to form 18, and N-acylation. In support of this mechanistic scenario, the intermediacy of 18 was confirmed by N-protonation in the absence of di-tert-butyl dicarbonate and isolation. This cascade accomplishes both C–C bond formation at C10 and activation of the C14 methyl ester for fragment coupling. The conjunctive reagent, iodoether 8, was prepared by site- and stereoselective α-alkylation of the chiral tigloyl imide 19 with para-methoxybenzyl chloromethyl ether (60%, 7:1 dr) (30), imide reduction (71%), and deoxyiodination (74%) (Fig. 2B).

Fig. 2 Stereoselective syntheses of the coupling fragments.

(A) Synthesis of the enimide 7. Reagents and conditions: (1) Zn(CH3)2, copper(II) triflate [Cu(OTf)2] (0.5 mol %), L* (1.0 mol %), toluene, 0°C, then CH3Li, –78°C, then methylcyanoformate, –78°C; (2) iodomethane, sodium tert-butoxide (t-BuONa), CH3OH, 0°C, 71%; (3) potassium bis(trimethylsilyl)amide (KHMDS), N-phenyl triflimide, tetrahydrofuran (THF), –78°C; (4) CO (1 atm), tetravinyltin, LiCl, tetrakis(triphenylphosphine)palladium(0) [Pd(PPh3)4] (5 mol %), dimethylformamide (DMF), 40°C; (5) Cu(OTf)2 (5 mol %), (CH2Cl)2, 70°C; (6) diethylaluminum cyanide (Et2AlCN), THF, 0°C, then di-iso-butylaluminum hydride (DIBALH), –78°C, then 0.01 M NaOH, CH3OH–H2O (5:1), 0°C; (7) trimethylsilyl trifluoromethanesulfonate (TMSOTf), bis(trimethylsilyl)ethylene glycol, CH2Cl2, 30°C; and (8) CH3Li, toluene, 0°C, then di-tert-butyl dicarbonate, 0°C. (B) Synthesis of the bifunctional iodoether 8. Reagents and conditions: (1) sodium bis(trimethylsilyl)amide (NaHMDS), para-methoxybenzyl chloromethyl ether (PMBOCH2Cl), THF, –78°C to 20°C; (2) LiAlH4, Et2O, 0°C; and (3) PPh3, I2, imidazole, THF, 70°C.

In the key fragment-coupling step, an organolithium reagent derived from 8 by lithium-halogen exchange was added to the enimide 7 (Fig. 3). In situ hydrolysis of the resulting lithio-enamine provided the methyl ketone 21 (48%). Essential to the success of this twofold neopentylic fragment coupling was the electronic activation of the C14 carbonyl group and the minimization of nearby nonbonded interactions via construction of the cyclic enimide functional group. Before this achievement, >15 different fragment-coupling strategies were examined by using reactants derived from compounds 10 to 15, but each strategy was far inferior to the enimide-organolithium coupling described here. For example, imide N-substituents that were less electron-withdrawing or more sterically hindered than the tert-butoxy carbonyl group diminished the electrophilicity of the C14 carbonyl, which led to little or no detectable coupling products. In addition, carbonyl electrophiles that lacked the cyclic enimide functional group, such as those derived from compounds 10 to 15, could not be successfully elaborated to related fragment-coupling products. Next, formal dehydration of the methyl ketone 21 via base-induced elimination of a transient vinyl triflate provided the alkyne 22 (81%). Removal of the para-methoxybenzyl group followed by oxidation of the resulting alcohol with the Dess-Martin periodinane (31) generated the alkynyl aldehyde 23 (83%, two steps).

Fig. 3 Synthesis of (+)-12-epi-mutilin (4) and (+)-11,12-di-epi-mutilin (26) from 7 and 8.

Reagents and conditions: (1) t-BuLi, 8, Et2O, –45°C, then 7, –45°C, then HCl, THF, 0°C; (2) KHMDS, N-(5-chloro-2-pyridyl)bis(trifluoromethanesulfonimide) (Comins’ reagent), THF, –78°C; (3) 2,3-dichloro-5,6-dicyano-1,4-benzoquinone (DDQ), CH2Cl2, pH 7 buffer, 20°C; (4) Dess-Martin periodinane (DMP), CH2Cl2, 20°C; (5) bis(cyclooctadiene)nickel(0) [Ni(cod)2] (30 mol %), 1,3-bis-(2,6-di-iso-propylphenyl)imidazol-2-ylidene (IPr, 30 mol %), triethylsilane, THF, 20°C, then tetra-n-butylammonium fluoride (TBAF), 20°C; (6) DMP, CH2Cl2, 20°C; (7) SmI2, THF, CH3OH, 20°C; (8) Na, EtOH, 20°C; (9) HCl, H2O, CH3OH, THF, 20°C; (10) (i) SmI2, THF, H2O, 20°C; (ii) lithium triethylborohydride (LiEt3BH), THF, 20°C; (11) Na, EtOH, 20°C, then HCl, H2O, CH3OH, THF, 20°C.

We envisioned using the alkynyl aldehyde 23 in a reductive cyclization reaction to forge the macrocycle of the targets. Although aldehyde- and alkyne-reductive cyclizations have enabled the synthesis of 5, 6, and >10-membered rings (3235), the synthesis of medium rings using this method is conspicuously absent. This is likely due to unfavorable reaction kinetics—for example, C–O bond-forming ring closures to make five-membered cyclic ethers are ~105 times as fast as those to make eight-membered cyclic ethers (36). In the reductive cyclization of 23, we envisioned that the limited number of rotatable bonds along the nascent macrocycle would lower the entropic penalty of ring closure while enhancing regio- and stereocontrol. Furthermore, the presence of sp2-hybridized carbon atoms at locations C10 and C14 of the allylic alcohol product 24 would alleviate transannular nonbonded interactions in the eight-membered ring. After extensive experimentation with iridium, rhodium, ruthenium, titanium, and nickel-based catalysts, we found that treatment of 23 with bis(1,5-cyclooctadiene)nickel (30 mol %), 1,3-bis-(2,6-di-iso-propylphenyl)imidazol-2-ylidene (IPr, 30 mol %), and triethylsilane (3 equivalents) at 20°C resulted in smooth reductive cyclization (35). A single diastereomer of the allylic silyl ether product was formed, which was desilylated to provide allylic alcohol 24 [60% yield, >20:1 dr, >20:1 regiomeric ratio (rr)].

To complete the synthesis, the allylic alcohol 24 was subjected to a high-yielding two-step isomerization comprising oxidation with the Dess-Martin periodinane followed by site- and stereoselective reduction of the resulting enone with samarium diiodide (98% yield over two steps, >20:1 dr at C10) (31). The boat-chair conformation and absolute stereochemistry of 25 were confirmed by x-ray crystallography (see inset image, Fig. 3). Single-electron reduction of the diketone 25 with an excess of sodium proceeded with thermodynamic selectivity (>20:1 equatorial-axial at C14, 3:1 equatorial-axial at C11) to provide, after ketal hydrolysis, (+)-12-epi-mutilin (4) and (+)-11,12-di-epi-mutilin (26). Stepwise methods for reduction of diketone 25 were also investigated. Reduction of 25 with samarium diiodide in tetrahydrofuran and water provided a 1.3:1 mixture of the C11 epimers 28 and 27 (37). Alternatively, lithium triethylborohydride reduction of 25 provided the C11 axial alcohol 27 with 1:>20 diastereoselectivity. Reduction of the remaining C14 ketone in 27 or 28 with sodium in ethanol (13) followed by ketal hydrolysis then formed 4 and 26, respectively, both with >20:1 dr at C14 (92% and 74% for 4 and 26, respectively).

(+)-12-Epi-mutilin (4) and 11,12-di-epi-mutilin-ketal (31) were easily elaborated to pleuromutilins (Fig. 4). Adapting Procter’s protocol (15), (+)-12-epi-mutilin (4) was converted to (+)-12-epi-pleuromutilin (29) via stepwise acylation of the C11 and C14 alcohols with trifluoroacetylimidazole and O-trifluoroacetylglycolic acid, respectively, followed by in situ methanolysis of the trifluoroacetyl esters (59%, two steps, 1.1% overall from 9). Related conditions provided access to O-trityl-12-epi-pleuromutilin (30, 64% two steps). The C12 quaternary stereocenter of 30 was epimerized via the retroallylation-allylation reaction developed by Berner (9) to afford, after in situ removal of the trityl protecting group, a separable 1:1.7 mixture of (+)-pleuromutilin (1, 33%, 0.4% overall from 9) and (+)-12-epi-pleuromutilin (29, 56%). Finally, the C14-alcohol of 11,12-di-epi-mutilin-ketal (31) was selectively acylated with O-trityl-glycolic acid and then deprotected to afford (+)-11,12-di-epi-pleuromutilin 32 (66%, two steps, 2.3% overall from 9).

Fig. 4 Synthesis of (+)-12-epi-pleuromutiln (29), (+)-pleuromutilin (1), and (+)-11,12-di-epi-pleuromutilin (32).

Reagents and conditions: (1) 1-(trifluoroacetyl)imidazole, ethyl acetate (EtOAc), –78°C; (2) O-trifluoroacetylglycolic acid, 1-ethyl-3-(dimethylaminopropyl)carbodiimide (EDC), 4-dimethylaminopyridine (DMAP), 20°C, then CH3OH, NaHCO3, 20°C; (3) 1-(trifluoroacetyl)imidazole, EtOAc, –78°C; (4) O-tritylglycolic acid, EDC, DMAP, 20°C, then CH3OH, NaHCO3, 20°C; (5) Et2Zn, DMF, 100°C, then HCl, 20°C; (6) O-tritylglycolic acid, EDC, DMAP, 20°C; (7) HCl, 20°C.

We can envision adapting the strategy outlined herein to create additional derivatives with optimized structures (ring sizes, atomic substitution, and/or substitution patterns) that may further address constraints inherent to semisynthetic approaches and may be accessible by shorter, higher-yielding sequences. Recent successes in the development of convergent routes to tetracycline (38) and macrolide (39) antibiotics underscore the potential to generate new clinical antibacterial candidates by total synthesis (40).


Materials and Methods

Figs. S1 to S60

Tables S1 and S2

References (4151)


Acknowledgments: Financial support from NIH (R01GM110506), the National Sciences and Engineering Research Council of Canada (postdoctoral fellowship to S.K.M.), and Yale University is gratefully acknowledged. We thank O. Goethe and R. Holmes for assistance in preparing compounds 19, 20, and 8. We thank B. Mercado for x-ray crystallographic analysis. The structural parameters for 15 and 25 are available free of charge from the Cambridge Crystallographic Data Centre under reference numbers CCDC-1533223 and CCDC-1533222, respectively. Experimental procedures, nuclear magnetic resonance data, and high-performance liquid chromatography data are available in the supplementary materials. S.B.H., S.K.M., and M.Z. are inventors on patent application U.S. 62/453,330 submitted by Yale University that covers the synthesis of pleuromutilin derivatives.

Stay Connected to Science

Navigate This Article