Synthesis of macrocyclic nucleoside antibacterials and their interactions with MraY

The development of new antibacterial drugs with different mechanisms of action is urgently needed to address antimicrobial resistance. MraY is an essential membrane enzyme required for bacterial cell wall synthesis. Sphaerimicins are naturally occurring macrocyclic nucleoside inhibitors of MraY and are considered a promising target in antibacterial discovery. However, developing sphaerimicins as antibacterials has been challenging due to their complex macrocyclic structures. In this study, we construct their characteristic macrocyclic skeleton via two key reactions. Having then determined the structure of a sphaerimicin analogue bound to MraY, we use a structure-guided approach to design simplified sphaerimicin analogues. These analogues retain potency against MraY and exhibit potent antibacterial activity against Gram-positive bacteria, including clinically isolated drug resistant strains of S. aureus and E. faecium. Our study combines synthetic chemistry, structural biology, and microbiology to provide a platform for the development of MraY inhibitors as antibacterials against drug-resistant bacteria.


pKa prediction
Sphaerimicin has three ionizable functional groups, a carboxyl group, a secondary amine attached at the 3'''-position, and a tertiary amine in the piperidine ring. The carboxylic acid seems to be deprotonated in an aqueous solution, on the other hand, it is unclear which amine could be protonated. To anticipate its ionic state, the acidity of DMSO used in this study was -268.34 kcal/mol. 4 The predicted pKa values of protonated secondary amines and piperidines were 9.4-13.9 and 13.0-16.6 respectively, and the pKa values of the piperidines were higher than secondary amines within the eight diastereomers (Supplementary Table 1

Conformation analysis
Next, we focused on the stereochemistry of the piperidine ring of the spaherimicins.
As mentioned above, piperidine ring has three stereocenters. Therefore, the conformation of the possible eight diastereomers were investigated by the MM calculation performed by Macromodel suite of program using the MCMM method, 5 followed by PRCG minimization 6 with the OPLS3e force field 7  between the cationic 3′′′-nitrogen atom and lone pair of the piperidine nitrogen, and a hydrogen bond discussed above stabilize the conformations. Although the piperidine rings in a twist-boat, boat and half-chair were found in some diastereomers (SSS, SRS, SRR, RSR, and RRR), their relative energies seemed to be too high to exist in a solution phase.
Supplementary Figure 5. Conformational analysis of model compounds.

Construction of docking models
With the aid of the calculation, we perfumed docking study of MraY. The low-energy conformers of the SRS and RSR (Supplementary Figure 6), whose stereochemistries are identical to the designed SPM-1 and SPM-2, were docked with the crystal structure of MraY, which derived from the crystal structure of MraY from Aquifex aeolicus bound to carbacaprazamycin (PDB code: 6OYH). As a result, 3 rd stable conformer of SRS and global minimum of RSR were docked well and following embrace minimization by using Macromodel program afforded the docking model.

Docking model of SPM-3 bound to MraY
The docking model of SPM-3 was constructed by the modification of the crystal structure of MraY bound to SPM-1. The structure of SPM-1 in the crystal structure was manually modified to that of SPM-3 to give an initial complex. The initial complex was

Preparation of compounds General
All reactions except that carried out in aqueous phase were performed under argon atmosphere, unless otherwise noted. Isolated yields were calculated by weighing products. Supplementary Figure 9. Synthesis of cyclopentene units rac-3.
The reaction mixture was partitioned between AcOEt and sat. aq. NH4Cl. The organic phase was washed with H2O and brine, dried (Na2SO4),

Confirmation of the structure of S9.
The newly formed stereochemistry of carbon in S9, which is connected to the nitrogen of the aminoribose moiety, was independently confirmed as illustrated in Scheme S4. Namely, a chiral cyclopentenol (+)-S12 was reacted with 2 by Mitsunobu reaction to give S9, during the course of which the Boc group at the uridine moiety was partially removed. Subsequent treatment of the mixture with acetic acid in MeOH provided S14 as a single diastereomer. The Boc group of S9 obtained by asymmetric allylic alkylation was removed to afford S14. 1 H NMR data of S14 obtained by this route are identical to those obtained by the asymmetric allylic alkylation. Moreover, the know stereochemical outcome of the Trost ligand is entirely consistent with our results. 12 Supplementary Figure 14. Confirmation of the stereochemistry of S9.
The residue was purified by Hi-Flash silica gel column chromatography (20-50% acetone/hexane) to afford S14 (236 mg, 230 mol, 58% over 2 steps) as a white foam. All the data (see below) were identical to those obtained from S9 prepared via allylic alkylation.

Evaluation of antibacterial activity
MICs were determined by a microdilution broth method as recommended by the CLSI with cation-adjusted Mueller-Hinton broth (MHB). Serial twofold dilutions of each compound were made in appropriate broth, and the strains were inoculated with 5 × 10 5 cfu/mL in 96-well plates (each 0.1 mL/well). The plates were incubated at 37 °C for 18 h and then MICs were determined.

Data collection and structure determination
Supplementary