Desmethyl Macrolides: Synthesis and Evaluation of 4,10-Didesmethyl Telithromycin

Abstract

Novel sources of antibiotics are required to keep pace with the inevitable onset of bacterial resistance. Continuing with our macrolide desmethylation strategy as a source of new antibiotics, we report the total synthesis, molecular modeling and biological evaluation of 4,10-didesmethyl telithromycin (4), a novel desmethyl analogue of the 3rd-generation drug telithromycin (2). Telithromycin is an FDA-approved ketolide antibiotic derived from erythromycin (1). We found 4,10-didesmethyl telithromycin (4) to be four times more active than previously prepared 4,8,10-tridesmethyl congener (3) in MIC assays. While less potent than telithromycin (2), the inclusion of the C-8 methyl group has improved biological activity suggesting it plays an important role in antibiotic function.

The rapid development of bacterial resistance to antibiotic drugs, which is a natural consequence of their use and abuse, has resulted in a global health crisis. To exacerbate the problem, many pharmaceutical companies have terminated their antimicrobial research programs due to economic pressures. Thus, new sources of antibiotics are critical.¹ To address this need, we recently launched a structure-based drug design program wherein des-methyl analogues (i.e., CH₃ → H) of the 3^rd-generation macrolide antibiotic telithromycin (2) are made via synthesis (Figure 1).² We have reported the total synthesis, molecular modeling and biological evaluation of 4,8,10-tridesmethyl telithromycin (3) against both wild-type and macrolide-resistant bacteria, which was found to be active against various bacterial strains (vide infra). Herein we report the total synthesis, molecular modeling and biological evaluation of 4,10-didesmethyl telithromycin (4), our second desmethyl analogue of 2. Importantly, we have installed the (R)-C-8 methyl group in a chemo- and stereoselective manner by means of substrate-controlled asymmetric synthesis,³ enabling us to test the consequences of a single methyl group on bioactivity. All macrolide antibiotics target the 50S subunit of the bacterial ribosome by reversibly binding in the peptidyl transferase center, thus blocking protein synthesis.⁴ Telithromycin (2) is a 3^rd generation semisynthetic drug used clinically since 2004 and is derived from the flagship macrolide antibiotic erythromycin (1).⁵

Figure 1.

Structures of erythromycin (1), telithromycin (2) and novel analogues 4,8,10-tridesmethyl telithromycin (3) and 4,10-didesmethyl telithromycin (4).

All macrolide antibiotics based on the erythromycin scaffold are semisynthetic (i.e., prepared from erythromycin). Our approach distinguishes itself from others in that total synthesis is employed to access each analogue, which allows significant flexibility and provides ample opportunity to explore the chemical space of this important class of safe and effective broad-spectrum antibiotics.⁶

Antibiotic resistance mechanisms fall into three major categories: (1) drug modification; (2) drug efflux; and (3) target (i.e., ribosomal) modification arising from either ribonucleotide N-methylation of residues critical for binding (e.g., A2058) or single point mutations (e.g., A2058G).⁷ Inspiration for our desmethylation strategy is derived from Steitz’s elegant structural studies of macrolide drugs (e.g., 1 and 2) co-crystallized with 50S ribosomal subunits of the archaeon Haloarcula marismortui (Hm).⁸

Unlike eubacteria that possess adenine at 2058, all archea possess a guanine (Escherichia coli numbering) and do not efficiently bind macrolides 1 and 2. However, a point mutation of guanine to adenine at position 2058 of 23S rRNA (i.e., G2058A) rendered mutants susceptible to the antibiotics, allowing the structure of telithromycin (2) bound to the HmA2058 mutant at 2.6 Å resolution to be obtained (Figure 2A). Thus, Steitz showed that A2058G mutations in bacteria confer resistance by (1) loss of hydrogen bonding to the C-2′ hydroxyl of desosamine and (2) a steric clash of the exocyclic C-2 amino group of guanine 2058 with C-4 methyl of the macrolide drug (Figure 2B). We in turn hypothesized that replacing the C-4 methyl group with hydrogen (i.e., desmethylation) should relieve the steric clash component as other residues within the ribosome (e.g., 2059) can also form hydrogen bonds with desosamine’s hydroxyl.⁹ The rationale behind removing methyls at C-8 and C-10 was to (1) simplify chemical synthesis and (2) understand their roles in antibiotic function.

Figure 2.

(A) Telithromycin and A2058 interactions in H. marismortui with select distances in Angstroms (Steitz et. al. PDB=1YIJ). (B) Steric consequences of A2058G mutation.

To test the consequences of adding another methyl vis-à-vis our simplest analogue 3 whilst probing the effects of removing the C-4 and C-10 methyl groups of telithromycin (2), we launched a total synthesis of 4 to realize biological evaluation (i.e., minimum inhibitory concentrations, or MICs).¹⁰

The installation of the requisite (R)-C-8 methyl group commenced with enoate 5, which was utilized in the synthesis of tridesmethyl analogue 3 (Scheme 1).² Sharpless dihydroxylation (AD mix-β) of 5 furnished γ-lactone 6 by in situ lactonization of the newly formed C-6 hydroxyl in 85% yield (er>20:1).¹¹ Protection of the C-5 alcohol as its triethylsilyl (TES) ether was accomplished with TESCl and imidazole. Treatment of 7 with lithium diisopropylamide (LDA) at −78 °C and alkylation with MeI afforded a (S)-C-8 methyl intermediate. Inversion at C-8 to obtain the (R) configuration was accomplished by (1) re-enolization with LDA and (2) quenching with trimethylacetic acid (dr = 6:1, 54% overall yield from 6).

Scheme 1.

Installation of (R)-C-8 methyl group and stereochemical confirmation.^a

^aReagents and conditions: (a) AD mix-β, 85%, er>20:1; (b) TESCl, imidazole; (c) LDA, MeI; (d) LDA, Me₃CCO₂H (dr = 6:1), 54% over three steps or LDA, Ph₃CCO₂H (dr = 14:1, 82% over three steps); (e) TBAF, THF; (f) DCC, (R)-MTPA, DMAP, 60% over two steps.

Alternatively, quenching the enolate with a bulklier acid source (e.g., triphenylacetic acid) greatly improved the selectivity of epimerization (dr = 14:1, 82% overall from 6). Confirmation of the stereochemical course of the experiment was obtained by preparing Mosher ester 9 and subsequent single crystal x-ray analysis, which was realized by removing the TES ether on the C-5 hydroxyl with TBAF and esterification with (R)-α-methoxy-α-(trifluoro-methyl)phenylacetic acid (MTPA).¹¹

Subsequent steps in the synthesis of 4,10-didesmethyl telithromycin (4) were guided by the previously reported synthesis of tridesmethyl congener 3 (Scheme 2).² Thus, lactone 8 was reduced with LiAlH₄ and the primary alcohol chemoselectively protected as its tert-butyldimethylsilyl (TBS) ether to afford 10 (78% yield over two steps). Methylation of the tertiary C-6 alcohol was accomplished with MeOTf and 2,6-di-tert-butyl-4-methylpyridine (DTBMP) to furnish 11 in 75% yield. The use of Proton Sponge, previously employed in the synthesis of 3, gave an inferior 35% yield. Hydrogenolysis of benzyl ether 11 (80% yield) afforded an intermediary alcohol that was oxidized to aldehyde 12 with the Swern method.

Scheme 2.

Synthesis of 4,10-didesmethyl telithromycin (4)^a.

^aReagents and conditions: (a) LiAlH₄, THF; (b) TBSCl, imidazole, 78% over two steps; (c) MeOTf, DTBMP, 75%; (d) H₂, Pd/C, 80%; (e) (COCl)₂, DMSO, Et₃N; (f) Bu₂BOTf, Et₃N, dr>20:1, 76% over two steps; (g) TBSOTf, 2,6-lutidine; (h) LiOOH, THF, H₂O, 88% over two steps; (i) Cl₃PhCOCl, Et₃N, DMAP, 16, 78%; (j) TBAF, AcOH, 20%; (k) DMP, NaHCO₃; (l) vinyl MgBr, THF; (m) DMP, 50% over three steps; (n) 20 mol% Grubbs’ 2^nd Generation catalyst, 60%; (o) NaBH₄, CeCl₃•7H₂O, dr=5:2; (p) TESOTf, 2,6-lutidine; (q) p-TsOH, 30% over three steps; (r) TESCl, imidazole; (s) 21, AgOTf, DTBMP, 50% over two steps; (t) HF•Et₃N, Et₃N, CH₃CN; (u) DMP, CH₂Cl₂, 60% over two steps; (v) NaH, CDI, THF/DMF; (w) 23, 45% over two steps; (x) TAS-F, DMF/H₂O, 75%; (y) NCS, Me₂S, Et₃N, 70%; (z) MeOH, 80%.

At this stage, the Evans aldol reaction with propionimide 13 was employed to set the configurations at both C-2 and C-3 positions.¹² In the event, the aldol adduct was isolated in 78% yield (dr>20:1). Protection of the C-3 hydroxyl with TBSOTf to access 14 and removal of the auxiliary furnished acid 15 in 88% yield (two steps). Chemoselective Yamaguchi esterification of 15 with known diol 16 delivered ester 17 in 78% yield.²

To prepare the 14-membered macrolactone ring, we recruited the ring-closing metathesis (RCM) strategy previously employed in the simpler congener 3.² To this end, we removed the primary TBS ether in 17 with tetrabutylammonium fluoride (TBAF) buffered with AcOH (20% yield, unoptimized).¹³ The resulting alcohol was oxidized to the aldehyde with the Dess-Martin Periodinane (DMP);¹⁴ moreover, addition of vinyl MgBr and subsequent DMP oxidation afforded vinyl ketone 18 in 50% overall yield. Treatment of dienone 18 with 20 mol% Grubbs’ second-generation catalyst effected the desired RCM, affording macroketolactone 19 in 60% yield.¹⁵

Stereo- and regioselective installation of desosamine at the C-5 hydroxyl paralleled the approach taken for 3. Specifically, the C-9 ketone in 19 was reduced under Luche conditions to avoid ketalization with the C-5 hydroxyl with a dr of 5:2, and the C-12 hydroxyl was protected as its TES ether to prevent glycosylation with desosamine donor 21.² Silylation of both C-9 and C-12 hydroxyls with TESOTf and subsequent treatment with p-toluenesulfonic acid (p-TsOH) selectively deprotected the C-5 and the C-9 silyl ethers while leaving the tertiary C-12 hydroxyl TES-protected (30% yield over three steps).

Site-selective silylation of the secondary, allylic C-9 alcohol in the presence of the secondary C-5 alcohol with TESCl and imidazole furnished 20, which was subjected to glycosylation with known desosamine donor 21 (50% yield over two steps) under the agency of AgOTf and DTBMP.¹⁶ When removing both C-9 and C-12 silyl ethers, we found that HF•3Et₃N gave better selectivity compared to TBAF. DMP oxidation afforded glycosylated macroketolactone 22 in 60% yield over two steps.

Installation of the C-11/C-12 carbamate was accomplished by employing methods originally developed by Baker at Abbott and later adapted by Hoechst Marion Roussel to prepare 2.^17,5 Thus, treatment of 22 with NaH and carbonyldiimidazole (CDI) and subsequent addition of butylamine 23 effected a tandem carbamoylation/intramolecular aza-Michael sequence to stereoselectively afford oxazolidinone 24 in 45% overall yield.¹⁸ Removal of the C-3 TBS ether with tris(dimethylamino)sulfonium difluorotrimethyl-silicate (TAS-F) proceeded in 75% yield.¹⁹ Corey-Kim oxidation afforded the C-3 ketone (70% yield) and methanolysis of the methyl carbonate on the C-2′ position of desosamine delivered target molecule 4,10-didesmethyl telithromycin (4) in 80% yield.²⁰

With 4 in hand, we initiated biological evaluation by testing its activity against several bacterial strains of Escherichia coli and Staphylococcus aureus.¹⁰ Telithromycin (2) and 4,8,10-trides-methyl telithromycin (3) were used as comparators (Table 1).²

Table 1.

Minimum inhibitory concentration (MIC) values in μg/mL for 4,8,10-tridesmethyl analogue (3), 4,10-didesmethyl analogue (4) and telithromycin (2).

Entry	Strain	Bacteria	wt/mutant	Previous work		This work
				MIC trides (3)a	MIC TELI (2)a	MIC dides (4)b	MIC TELI (2)b
1	SQ171/2058G	E. coli	A2058G	>512	>512	>512	>512
2	DK/pKK3535	E. coli	wt	32	0.5	8	0.5
3	DK/2058G	E. coli	A2058G	64	1	16	1
4	UCN14	S. aureus	A2058T	32	>256	>256	>128
5	ATCC33591	S. aureus	ermA	>128	>128	>128	>128

^aPrevious experiment carried out in EtOH.

^bCurrent experiment carried out in DMSO.

Although both resistant strains (entries 1 and 5) were not susceptible to any of the macrolides, both E. coli wild-type and A2058G mutant (entries 2 and 3) were inhibited by tridesmethyl analogue 3, didesmethyl analogue 4 and telithromycin (2). In these strains, the didesmethyl congener was fourfold more potent than tridesmethyl variant. Compared to telithromycin (2), the novel analogue was sixteenfold less potent. While the current evaluation was carried out in DMSO and not EtOH, the MIC values for telithromycin (2) were found to be identical in both thus suggesting the potency of didesmethyl analogue 4 vis-à-vis 3 is not derived from solvent change. Thus, these data illustrate the consequences of deleting methyl groups from the erythronolide scaffold. Interestingly, 4 did not inhibit the growth of the UCN14 strain that carries an A2058T mutation as compared to 3.

To facilitate interpretation of the MICs, we employed the Conformationally Sampled Pharmacophore (CSP) approach using Hamiltonian Replica Exchange Dynamics (HREX MD), from which conformational properties of the 2, 3 and 4 were obtained.²¹ Probability distributions of distances between select points on telithromycin (2, black), 4,8,10-tridesmethyl (3, blue), and 4,10-didesmethyl (4, red) telithromycin are shown in Figure 3(A–D). The distributions for both desmethyl analogues overlap well with that of telithromycin, consistent with both analogues binding to the ribosome. The additional conformational flexibility of 3 and 4 as seen in panels A, B and C of Figure 3 are suggested to contribute to the decrease in their MICs vs. 2. Consistent with the additional methyl in didesmethyl 4 vs. tridesmethyl 3, 4 has decreased conformational flexibility, which is supported by the improved MIC values. Altogether, these results indicate a model where removal of the methyl groups lead to increased conformational flexibility, which has a negative impact on the biological activity, a consideration that must be taken into account when removing the C-4 methyl group to avoid resistance associated with the A2058G mutation.

Figure 3.

CSP probability distributions for telithromycin (2, black); 4,8,10-tridesmethyl telithromycin (3, blue); and 4,10-didesmethyl telithromycin (4, red). The vertical line corresponds to the crystallographic distances from PDB 1YIJ. Atom pairs represented in A through D are shown in the inset figure.

In conclusion, 4,10-didesmethyl telithromycin (4) was prepared via total synthesis using a RCM approach in 44 steps overall (32 steps in the longest linear sequence) from commercially available starting materials. The novel analogue was evaluated for biological activity and found to inhibit the growth of two bacterial strains, including an A2058G mutant. Significantly, the addition of an extra methyl group at C-8 resulted in a fourfold increase in activity as compared to tridesmethyl congener suggesting it plays an important role in antibiotic function. We are currently working toward the synthesis of other desmethyl telithromycin analogues. Those results will be reported in due course.

ABBREVIATIONS

TBS

tert-Butyldimethylsilyl

TES

triethylsilyl

LDA

lithium diisopropylamide

DTBMP

2,6-di-tert-butyl-4-methylpyridine

RCM

ring-closing metathesis

MIC

minimum inhibitory concentration

CSP

Conformationally Sampled Pharmacophore

DMP

Dess-Martin Periodinane

Tf

trifluoromethanesulfonyl

TBAF

tetrabutylammonium fluoride

HDREX MD

Hamiltonian Replica Exchange Dynamics (HREX MD)

TELI

telithromycin

CDI

carbonyldiimidazole

TAS-F

tris(dimethylamino)sulfonium difluorotrimethylsilicate