Design, Synthesis and Biological Evaluation of Conjugates of 3-O-Descladinose-azithromycin and Nucleobases against rRNA A2058G- or A2059G-Mutated Strains

Structurally unrelated antibiotics MLSB (macrolide-lincosamide-streptogramin B) compromised with clinically resistant pathogens because of the cross-resistance resulting from the structural modification of rRNA A2058. The structure–activity relationships of a novel 3-O-descladinose azithromycin chemotype conjugating with nucleobases were fully explored with the aid of engineered E. coli SQ110DTC and SQ110LPTD. The conjugates of macrolides with nucleobases, especially adenine, displayed antibacterial superiority over telithromycin, azithromycin and clindamycin against rRNA A2058/2059-mutated engineered E. coli strains at the cost of lowering permeability and increasing vulnerability to efflux proteins against clinical isolates.


Introduction
Public health has been threatened due to the dramatically surging incidence of resistance in clinical isolates. More than half of the marketed antibacterial drugs serve as protein synthesis inhibitors by targeting 30S or 50S subunits of bacterial ribosomes [1]. Among them, macrolides are widely prescribed to treat community-or hospital-acquired bacterial pneumonia.
Recently, high-resolution crystal structures of 70S ribosome complexed with erythromycin derivatives revealed the convincible model for a prevailing resistance against structurally unrelated MLS B (macrolide-lincosamide-streptogramin B) antibiotics [2]. The methylated A2058 or A2058 mutation would disrupt the water-mediated hydrogen bonding to the 5-O-desosamine of macrolides, and the greatly decreasing affinity could not maintain the tight binding of the antibiotics when nascent peptides egress, leading to high-level resistance.
The second-generation semi-synthetic macrolides, clarithromycin and 15-membered azithromycin (Figure 1), failed to afford additional contacts with the ribosome and thus showed no efficacy against erythromycin-resistant bacteria. The third-generation ketolide (replacing 3-O-cladinose with 3-keto), telithromycin (Figure 1), succeeded in partially restoring the activities by supplying a secondary π-π interaction between the A752-A2609 base pair and the newly introduced imidazolyl pyridyl group [3,4]. The extended side chain of another ketolide, solithromycin, interacts with the same base pair [5]. However, the affinities of telithromycin or solithromycin to the A2058-methylated or mutated ribosome are too low to allow the usage of these drugs for treating high-level MLS B -resistant infections [2]. To strengthen the affinities to the modified ribosome, a variety of mono/bi/fusedheteroaromatic groups were installed onto the end of the linkers derivatized from the 6-OH, 9-keto, 11-OH and 4′′-OH of macrolides [6,7]. It was reported that the antibacterial activities of the novel macrolides that were constructed by a total synthesis route still heavily depend on the pendant side chains capable of binding the ribosome [8]. Carbamolides (3-O-carbamoyl), acylides (3-O-acyl) and alkylides (3-O-alkyl) showed a new mode of action, namely π-π interacting with the base pair C2610-G2505 [9][10][11][12]. In 2019, we attempted to introduce some heteroaromatic groups to 3-OH clarithromycin bridged by three-to eight-atom-long spacers. The molecular docking suggested that the most potent compound possessing a three-atom-long spacer shared the same mode of action as that of the known macrolides [13].
Apart from the chemotypes mentioned above, some conjugates have been investigated to achieve dual modes of action [14][15][16]. Quinolones linking to 4′′-OH of macrolides have been extensively investigated since 2010, which was well summarized in a recent review [7]. The potent hybrids proved to be non-inhibitors in gyrase-based supercoiling assays [17,18]. In 2020, we reported that quinolones attaching to 6-OH of azithromycin could show moderate activities against E. coli DNA gyrase [19].
In continuation of our work, we report here a series of 3-O-descladinose azithromycin derivatives substituting 3-O-cladinose by seven-to seventeen-atom-length spacers anchored with various nucleobases (adenine, guanine, cytosine, thymine and uracil). There are two reasons to further this work. First, thymine and adenine were linked to 9-or 11positions of some 3-O-cladinose-containing erythromycin derivatives, but the resulting activities, for unknown reasons, are poor [20,21]. The 3-O-derivatives of macrolides are not expected to activate the expression of inducible resistance genes, as the presence of 3-O-cladinose is required for efficient induction [22], but they failed to afford potency against A2058/2059-methylated or mutated strains. As purines and pyrimidines might form Watson-Crick interactions with rRNA bases, it would be interesting to know whether nucleobases attached at the C3-position of the macrolactone ring of macrolides could restore the activities against the ribosome-mutated strains. Secondly, the modes of action of ribosome-targeting inhibitors are less investigated in the early stage of new macrolides discovery [23] because it is difficult to isolate resistance mutants with alterations in the drug target sites because of the redundancy of rRNA genes in bacteria. To determine the biological function of nucleobases when attached to macrolides, we mutated ribosomal RNA A2058 or A2059 of a public strain E. coli SQ110 that has been engineered with only one chromosomal rrn allele [24].

Results and Discussion
The derivatization of the five nucleobases followed a modified procedure [25]. The nucleobases 3, 4, 8, 12 and 14 were obtained through three steps: a substitution reaction with methyl bromoacetate, protection and saponification, which are outlined in Schemes To strengthen the affinities to the modified ribosome, a variety of mono/bi/fusedheteroaromatic groups were installed onto the end of the linkers derivatized from the 6-OH, 9-keto, 11-OH and 4"-OH of macrolides [6,7]. It was reported that the antibacterial activities of the novel macrolides that were constructed by a total synthesis route still heavily depend on the pendant side chains capable of binding the ribosome [8]. Carbamolides (3-O-carbamoyl), acylides (3-O-acyl) and alkylides (3-O-alkyl) showed a new mode of action, namely π-π interacting with the base pair C2610-G2505 [9][10][11][12]. In 2019, we attempted to introduce some heteroaromatic groups to 3-OH clarithromycin bridged by three-to eight-atom-long spacers. The molecular docking suggested that the most potent compound possessing a three-atom-long spacer shared the same mode of action as that of the known macrolides [13].
Apart from the chemotypes mentioned above, some conjugates have been investigated to achieve dual modes of action [14][15][16]. Quinolones linking to 4 -OH of macrolides have been extensively investigated since 2010, which was well summarized in a recent review [7]. The potent hybrids proved to be non-inhibitors in gyrase-based supercoiling assays [17,18]. In 2020, we reported that quinolones attaching to 6-OH of azithromycin could show moderate activities against E. coli DNA gyrase [19].
In continuation of our work, we report here a series of 3-O-descladinose azithromycin derivatives substituting 3-O-cladinose by seven-to seventeen-atom-length spacers anchored with various nucleobases (adenine, guanine, cytosine, thymine and uracil). There are two reasons to further this work. First, thymine and adenine were linked to 9-or 11-positions of some 3-O-cladinose-containing erythromycin derivatives, but the resulting activities, for unknown reasons, are poor [20,21]. The 3-O-derivatives of macrolides are not expected to activate the expression of inducible resistance genes, as the presence of 3-O-cladinose is required for efficient induction [22], but they failed to afford potency against A2058/2059-methylated or mutated strains. As purines and pyrimidines might form Watson-Crick interactions with rRNA bases, it would be interesting to know whether nucleobases attached at the C3-position of the macrolactone ring of macrolides could restore the activities against the ribosome-mutated strains. Secondly, the modes of action of ribosome-targeting inhibitors are less investigated in the early stage of new macrolides discovery [23] because it is difficult to isolate resistance mutants with alterations in the drug target sites because of the redundancy of rRNA genes in bacteria. To determine the biological function of nucleobases when attached to macrolides, we mutated ribosomal RNA A2058 or A2059 of a public strain E. coli SQ110 that has been engineered with only one chromosomal rrn allele [24].
Initially, the MICs of compounds 26-32 were determined against a panel of clinical isolates. The data in Table 1 suggest that the influence of the bases on the activities followed the order of A > G > U > C (29A, 29G, 29U, 29C; 30A, 30G, 30U, 30C), and the activity increased when prolonging the length of the linker (27U-32U). Scheme 6. Synthesis of compounds 37U and 37A. Reagents and conditions: a. CH 2 Cl 2 , compound 5 or 10, ZnCl 2 , NaBH(OAc) 3 ; b. MeOH, 60 • C; c. EtOH, HCl.  Unexpectedly, the activities of compounds 26-32 were inferior to azithromycin against susceptible and efflux-encoded strains. To further determine whether conjugates 26-32 were vulnerable to outer membrane barriers and efflux pumps, we used the engineered E. coli stains SQ110DTC and SQ110LPTD to examine the MICs of compounds 30U, 30A, 30G, 30C, 32U and 32A (Table 2). SQ110 was rendered hypersusceptible to macrolides by deleting the gene encoding the multidrug efflux transporter TolC (strain SQ110DTC) or mutating the lptD gene, which renders the outer membrane more permeable (strain SQ110LPTD) [24]. The decreasing MICs against the SQ110 strains in comparison with the reference E. coli strain ATCC25922 indicated that the low activities of compounds 26-32 could be attributed to their poor permeability and vulnerability to efflux. Among them, compounds 30A and 32A, which have adenine groups, were the most active, which was in agreement with the SARs against the clinical isolates, as shown in Table 1. The presence of the uracil base was inferior to the adenine (30U vs. 30A, 32U vs. 32A). None of the compounds were active against the A2058/A2059-mutated SQ110 stains (Table 2). 30A, 30G, 30C, 32U and 32A (Table 2). SQ110 was rendered hypersusceptible to macrolides by deleting the gene encoding the multidrug efflux transporter TolC (strain SQ110DTC) or mutating the lptD gene, which renders the outer membrane more permeable (strain SQ110LPTD) [24]. The decreasing MICs against the SQ110 strains in comparison with the reference E. coli strain ATCC25922 indicated that the low activities of compounds 26-32 could be attributed to their poor permeability and vulnerability to efflux. Among them, compounds 30A and 32A, which have adenine groups, were the most active, which was in agreement with the SARs against the clinical isolates, as shown in Table  1. The presence of the uracil base was inferior to the adenine (30U vs. 30A, 32U vs. 32A).
None of the compounds were active against the A2058/A2059-mutated SQ110 stains (Table 2). It was reported that menadione (VK3) could inhibit the efflux pumps and alter the fluidity of the bacterial membrane of S. aureus [34]. Thus, compound 32A, in combination with the usage of menadione (VK3) at concentrations of MIC/4 or MIC/8, exhibited better activities against S. aureus than 3-O-descladinose-3-OH azithromycin while much lower activities of 32A than azithromycin (Table 3), which suggested that the introduction of nucleobases is detrimental to the resulting activities of the conjugates against the clinical isolates. It was reported that menadione (VK 3 ) could inhibit the efflux pumps and alter the fluidity of the bacterial membrane of S. aureus [34]. Thus, compound 32A, in combination with the usage of menadione (VK 3 ) at concentrations of MIC/4 or MIC/8, exhibited better activities against S. aureus than 3-O-descladinose-3-OH azithromycin while much lower activities of 32A than azithromycin (Table 3), which suggested that the introduction of nucleobases is detrimental to the resulting activities of the conjugates against the clinical isolates.  With the most potent compound, 32A, in hand, t screened, as shown in Table 4. The varying lengths of the that the homolog of 33A was more potent than 32A and G/U (32A vs. 32G; 37A vs. 37U) and removing cyclic ca 35A) were not favorable for activities. The conversion of a ity) to a flexible one (amine) resulted in two-to four-fold With the most potent compound, 32A, in hand, the analogs of 32A were then screened, as shown in Table 4. The varying lengths of the linker (32A, 33A, 34A) revealed that the homolog of 33A was more potent than 32A and 34A. Replacing the base A with G/U (32A vs. 32G; 37A vs. 37U) and removing cyclic carbonation at 11,12-OH (32A vs. 35A) were not favorable for activities. The conversion of a rigid linker (amide functionality) to a flexible one (amine) resulted in two-to four-fold higher activities (32A vs. 37A).
The molecular docking studies of the ligands (37A and 32A) against the G2099A mutant Haloarcula Marismortui 50S ribosomal subunit (PDB ID: 1YHQ [35]) were performed utilizing Autodock VINA in YASARA (YASARA Biosciences GmbH, Vienna, VIE, AUT) [36,37]. The suggested binding modes of 37A and 32A ( Figure 2) indicated that hydrogen bonds were formed between the adenines of the conjugates and the surrounded ribosomal nucleobases (Tables S1 and S2), which accounted for the better activities of 37A and 32A than azithromycin against A2058/2059-mutated SQ110 strains. Meanwhile, higher binding affinities were determined for 37A than 32A (Table 5), which was in agreement with the higher activities of 37A than those of 32A (Table 4). With the most potent compound, 32A, in hand, the analogs of 32A were then screened, as shown in Table 4. The varying lengths of the linker (32A, 33A, 34A) revealed that the homolog of 33A was more potent than 32A and 34A. Replacing the base A with G/U (32A vs. 32G; 37A vs. 37U) and removing cyclic carbonation at 11,12-OH (32A vs. 35A) were not favorable for activities. The conversion of a rigid linker (amide functionality) to a flexible one (amine) resulted in two-to four-fold higher activities (32A vs. 37A). The molecular docking studies of the ligands (37A and 32A) against the G2099A mutant Haloarcula Marismortui 50S ribosomal subunit (PDB ID: 1YHQ [35]) were performed utilizing Autodock VINA in YASARA (YASARA Biosciences GmbH, Vienna, VIE, AUT) [36,37]. The suggested binding modes of 37A and 32A ( Figure 2) indicated that hydrogen bonds were formed between the adenines of the conjugates and the surrounded ribosomal nucleobases (Tables S1 and S2), which accounted for the better activities of 37A and 32A than azithromycin against A2058/2059-mutated SQ110 strains. Meanwhile, higher binding affinities were determined for 37A than 32A (Table 5), which was in agreement with the higher activities of 37A than those of 32A (Table 4). The adenine, 3′-N(CH3)2 of the desosamine, the carbamoyl group in 37A formed six conventional hydrogen bonds with both sugar and phosphate groups in the RNA backbone linked to A2096/ U2539/ G2540/ U2541 (Table S1). The adenine and 2′-OH of the desosamine in 37A formed three conventional hydrogen bonds with the 2-NH2 of G2611 and O6 of G2102. The adenine and 3′-N(CH3)2 of the desosamine in 32A formed two conventional hydrogen bonds with the sugar group in the RNA backbone linked to A2096 and G2540. The adenine in 32A formed two conventional hydrogen bonds with the O2 of U2607 and U2539 (Table S2). Therefore, it revealed that the 37A bound tightly to the backbone of the G2099A large ribosomal subunit. The reliability of the models constructed  The adenine, 3 -N(CH 3 ) 2 of the desosamine, the carbamoyl group in 37A formed six conventional hydrogen bonds with both sugar and phosphate groups in the RNA backbone linked to A2096/ U2539/ G2540/ U2541 (Table S1). The adenine and 2 -OH of the desosamine in 37A formed three conventional hydrogen bonds with the 2-NH 2 of G2611 and O6 of G2102. The adenine and 3 -N(CH 3 ) 2 of the desosamine in 32A formed two conventional hydrogen bonds with the sugar group in the RNA backbone linked to A2096 and G2540. The adenine in 32A formed two conventional hydrogen bonds with the O2 of U2607 and U2539 (Table S2). Therefore, it revealed that the 37A bound tightly to the backbone of the G2099A large ribosomal subunit. The reliability of the models constructed from protein-ligand docking was verified by using molecular dynamics (MD) simulations. Both methods yielded similar complex structures. Quantitatively, the RMSD values of the structures from the MD and docking are small ( Figure S2), which indicates that the structural variation is minor. Interestingly, we found that after the molecular dynamic simulation, the 6-NH 2 of compound 37A's adenine formed two hydrogen bonds with the sugar in the RNA backbone linked to G2609. The -NH-of the linker chains formed two hydrogen bonds with the phosphate groups of G2540, and the -C=O-of the linker chains and the backbone of macrolides formed two hydrogen bonds with 6-NH 2 of A2099 ( Figure 3A, Table S3). The 6-OH of compound 32A formed a hydrogen bond with the sugar in the RNA backbone linked to U2541. The 2 -OH formed two hydrogen bonds with the 4-NH 2 and N3 of C2104 and a hydrogen bond with 1-NH of G2102. Moreover, the -N-CH 3 and O6 of the backbone of macrolides formed a hydrogen bond with the O2 of C2104 and 2-NH 2 of G2102, respectively. The 3 -N(CH 3 ) 2 of the desosamine, 6-NH 2 and N7 of adenine formed a hydrogen bond with the phosphate groups, O1 and the sugar in the RNA backbone linked to G2539, respectively. The -C=O-of the backbone of macrolides formed a hydrogen bond with 3-NH of U2620, and the N3 of adenine formed a hydrogen bond with the 2-NH 2 of G2611 ( Figure 3B, Table S4).
Molecules 2023, 28, x FOR PEER REVIEW 11 of 28 from protein-ligand docking was verified by using molecular dynamics (MD) simulations. Both methods yielded similar complex structures. Quantitatively, the RMSD values of the structures from the MD and docking are small ( Figure S2), which indicates that the structural variation is minor. Interestingly, we found that after the molecular dynamic simulation, the 6-NH2 of compound 37A's adenine formed two hydrogen bonds with the sugar in the RNA backbone linked to G2609. The -NH-of the linker chains formed two hydrogen bonds with the phosphate groups of G2540, and the -C=O-of the linker chains and the backbone of macrolides formed two hydrogen bonds with 6-NH2 of A2099 ( Figure  3A, Table S3). The 6-OH of compound 32A formed a hydrogen bond with the sugar in the RNA backbone linked to U2541. The 2′-OH formed two hydrogen bonds with the 4-NH2 and N3 of C2104 and a hydrogen bond with 1-NH of G2102. Moreover, the -N-CH3 and O6 of the backbone of macrolides formed a hydrogen bond with the O2 of C2104 and 2-NH2 of G2102, respectively. The 3′-N(CH3)2 of the desosamine, 6-NH2 and N7 of adenine formed a hydrogen bond with the phosphate groups, O1 and the sugar in the RNA backbone linked to G2539, respectively. The -C=O-of the backbone of macrolides formed a hydrogen bond with 3-NH of U2620, and the N3 of adenine formed a hydrogen bond with the 2-NH2 of G2611 ( Figure 3B, Table S4).

Synthetic Procedures
All of the solvents and reagents were obtained from commercial sources (Innochem, Beijing, China; Bidepharm, Shanghai, China; J&K scientific, Beijing, China) and used without further purification unless otherwise noted. The reactions were monitored by using thin-layer chromatography (TLC, Qingdao Ocean Chemical Co., Ltd., Qingdao, China) with silica gel HSGF254 precoated plates (0.2 mm), and the compounds were visualized under UV light (λ = 254 or 365 nm) and/or stained with iodine. Column chromatography was performed on silica gel (100-200 mesh). 1 H and 13 C spectra were taken in CDCl 3 , MeOH-d 4 or DMSO-d 6 on Bruker Ascend 400 MHz or Ascend 700 MHz spectrometers (Bruker, Massachusetts, Germany) with tetramethylsilane (TMS) as an internal standard. High-resolution mass spectra (HRMS) were obtained with an Agilent Q-TOF 6520 (Agilent Technologies Inc., Santa Clara, CA, USA). The spectra of compounds are shown in Table S5.
To a suspension of thymine (1.000 g, 7.93 mmol) and K 2 CO 3 (1.100 g, 7.93 mmol) in dry DMF, methyl bromoacetate was added (0.85 mL, 7.93 mmol), and the mixture was stirred overnight. Then, the mixture was filtered, the solid residue was cooled to 0 • C and then treated with water and 4M HCl. The precipitate was collected by filtration and washed with water to afford compound 1 (0.700 g, 3 (3) Compound 1 (0.700 g, 3.54 mmol) was treated with water (3 mL) and 2M NaOH (3 mL). The mixture was boiled for 10 min, and then cooled to 0 • C and treated with 4M HCl to adjust the pH to 3. After stirring for 30 min, the solid was filtered and washed with water to afford compound 3 (0.  (4) To a suspension of uracil (1.000 g, 8.92 mmol) and K 2 CO 3 (1.356 g, 8.92 mmol) in dry DMF, methyl bromoacetate was added (0.96 mL, 8.92 mmol), and the mixture was stirred overnight. The mixture was filtered, and the solid residue was cooled to 0 ºC and treated with water and 4M HCl. The precipitate was collected by filtration and washed with water. The solid was treated with water (3 mL) and 2M NaOH (3 mL), and then the mixture was boiled for 10 min. The mixture was cooled to 0 • C and treated with 4M HCl to adjust the pH to 3. After stirring for 30 min, compound 4 (0.287 g, 1.69 mmol, 18 Compound 2 (2.313 g, 12.57 mmol) was dissolved in dry CH 2 Cl 2 , and then DMAP (0.566 g, 12.57 mmol) and di-tert-butyl dicarbonate (1.000 g, 15.08 mmol) were added to the solution over a period of 15 min at 0 • C. The cooling was discontinued, and the solution was stirred for 8 h. Subsequently, the mixture was treated with saturated NH 4 Cl repeatedly. After stirring for 30 min, the phases were separated, and the organic phase was washed with water and brine. The organic layer was concentrated and purified through column chromatography (CH 2 Cl 2 /EtOH = 10:0.1). Then, the resulting compound (0.400 g, 1.41 mmol) was dissolved in dry THF/MeOH (10:1), and NaBH 4 (0.106 g, 2.82 mmol) was added to the solution. After stirring for 1 h at room temperature, the mixture was washed with water and brine, and the organic layer was concentrated. The resulting compound (0.218 g, 0.85 mmol) and Dess-Martin periodinane (0.396 g, 0.93 mmol) were dissolved in dry CH 2 Cl 2 . The solution was stirred for 1 h at room temperature and then washed with water and brine. The organic layer was concentrated and purified through column chromatography (CH 2 Cl 2 /EtOH = 10:0.3) to afford compound 5 (0.161 g, 0.64 mmol, 75.3%). 1  3.1.5. Methyl Adenin-9-ylacetate (6) To a suspension of adenine (1.000 g, 7.40 mmol) and K 2 CO 3 (1.227 g, 8.88 mmol) in dry DMF, methyl bromoacetate was added (1.60 mL, 14.80 mmol). After stirring at room temperature for 5 h, the mixture was filtered, and the solid residue was treated with ethyl acetate, resulting in recrystallization. The solid was filtered off and followed by recrystallization using MeOH to yield 0.807 g (4.18 mmol, 56.5%) of white product. 3.1.6. Methyl (N 6 -(di-tert-butyloxy carbonyl)adenin-9-yl)acetate (7) Compound 6 (0.807 g, 4.18 mmol) was dissolved in dry CH 2 Cl 2 , and then DMAP (1.532 g, 12.54 mmol) and di-tert-butyl dicarbonate (2.737 g, 12.54 mmol) were added to the solution over a period of 15 min at 0 • C. The cooling was discontinued, and the solution was stirred overnight. Subsequently, the mixture was treated with saturated NH 4 Cl repeatedly.
3.1.11. (2-Amino-6-(benzyloxy)purin-9-yl)acetic acid (12) NaH (0.730 g, 18.25 mmol) was dissolved in benzyl alcohol (10 mL) for 2 h at 0 • C, a solution of compound 11 (0.880 g, 3.65 mmol) in DMF was slowly added, and the resultant suspension was stirred overnight at room temperature. Then, 1M NaOH was added, and the solution was washed with ethyl acetate. The water phase was acidified to pH 3 with 4M HCl and extracted with ethyl acetate repeatedly. The combined organic phases were washed with brine and concentrated. The residue was recrystallized from EtOH (yield 74.8%). 1 (13) Cbz-Cl (2.54 mL, 18.0 mmol) was added dropwise over a suspension of cytosine (1.000 g, 9.0 mmol) in pyridine (10 mL) at 0 • C. The mixture was stirred overnight. Water was added, and the pH was adjusted to 1 with 4M HCl. The resultant white precipitate was filtered off, washed with water, and dried to afford 13 (0.510 g, 2.08 mmol, 22%). 1 (14) To a suspension of compound 13 (0.510 g, 2.08 mmol) and K 2 CO 3 (0.288 g, 2.08 mmol) in dry DMF, methyl bromoacetate was added (0.23 mL, 2.08 mmol), and the mixture was stirred overnight. The mixture was filtered, and the solid residue was treated with water and 4M HCl at 0 • C. The precipitate was collected through filtration and treated with water (3 mL) and 2M NaOH (3 mL). After stirring for 30 min, the solution was cooled to 0 ºC and treated with 4M HCl to adjust the pH to 3. The mixture was filtered and washed with water to obtain a white solid, 14 (0.

3-O-descladinosyl-3-OH azithromycin (15)
To a solution of azithromycin (20.000 g, 26.7 mmol) in ethanol (35 mL), 1 M hydrochloric acid was slowly dripped into the reaction system. After stirring for 1 h at 40 • C, the pH was adjusted to 10 with NH 3 ·H 2 O, and then CH 2 Cl 2 was added for extraction. The organic phase was washed with water and brine and dried to afford compound 15 (15.
The resulting compound (15.900 g, 25.1 mmol) was dissolved in dry CH 2 Cl 2 (10 mL), then 4-dimethylaminopyridine (6.131 g, 50.2 mmol) and N, N-carbonyl diimidazole (12.176 g, 75.3 mmol) were added to this solution under Ar. The reaction was stirred at room temperature for 12 h and was washed with saturated NH 4 Cl solution, water and brine. The organic phase was evaporated under reduced pressure, and then an appropriate amount of ethyl acetate was added. The mixture was stirred in an ice bath for 2 h and filtered to afford a white solid, 16 (

General Procedures for the Synthesis of Intermediates 17-25
Compound 16 (1 eq) was dissolved in DMF, then a diamine (5 eq) was added to this solution. The reaction was stirred at room temperature for 2 h. Water was added, and the resulting mixture was extracted with ethyl acetate. The organic layer was washed with water and brine. The organic layer was concentrated and purified through column chromatography (CH 2 Cl 2 /MeOH/NH 3 ·H 2 O = 10:0.4:0.05) to obtain intermediates 17-25.

Docking Methods
The molecular docking studies of the ligands (37A and 32A) against target G2099A mutant 50S ribosomal subunit (PDB ID: 1YHQ) were performed utilizing Autodock VINA in YASARA (YASARA Biosciences GmbH, Vienna, VIE, AUT) [36,37]. To remove bumps and ascertain the covalent geometry of the ligands, the structures were all energy-minimized with the NOVA force field [39]. The dockings were undertaken by defining the simulation cell box of size 30 Å × 30 Å × 30 Å for the G2099A mutant 50S ribosomal subunit. Then, the docking studies on the aptamer were carried out through the built-in docking simulation macro 'dock_run.mcr' using an AMBER03 force field with 25 poses and 9 clusters for each situation [40].

Molecular Dynamics
To justify whether the models constructed from protein-ligand docking were reliable, molecular dynamics (MD) simulations of the bindings between G2099A mutant 50S ribosomal subunit with the ligands (37A and 32A) were performed. The AMBER14 force field was utilized for the MD simulation, as implemented in the YASARA program. For the treatment of long-range coulomb forces beyond an 8 Å cutoff, the MD simulation used periodic boundary conditions and the particle-mesh Ewald method. NaCl was used at a concentration of 0.9%, and the density of HOH was 0.997 g/mL in the MD cell. No restraints were applied during the MD simulation using the settings employed in the second equilibration dynamics. Energies and coordinates every 100 ps were saved with a total simulation length of 40-50 ns under a constant temperature (298 K) and uncontrolled pressure in an NVT ensemble. The structural stability of the protein-ligand complex was examined by analyzing the average values of the potential energy with root mean square deviation (RMSD) throughout the trajectory. The RMSD profiles of all MD structures ( Figure S1) show that the variation in the RMSD values tends to be stable (<1 Å) after 40-50 ns, indicating that equilibrium was reached, and the last MD structure was chosen for the analysis.

Conclusions
In summary, macrolides conjugating with nucleobases exhibited considerably low activities against clinical isolates due to poor accumulation in the cell, which was confirmed in the presence of menadione (VK 3 ) that could inhibit the efflux pumps and alter the fluidity of the bacterial membrane of S. aureus. Thus, we turned to exploit an engineered E. coli SQ110 in combination with its rRNA-mutated strains to investigate the SARs of the ribosome-targeting inhibitors. Notably, 34A was the most potent against A2059G-mutated strains (34A 16 µg/mL vs. telithromycin 128 µg/mL, azithromycin > 256 µg/mL and clindamycin > 128 µg/mL). Compounds 37A and 37U showed weak activities against the A2058G-mutated strain (128 µg/mL vs. telithromycin > 128 µg/mL, azithromycin > 256 µg/mL and clindamycin > 128 µg/mL). Molecular docking suggested additional hydrogen bonding formation between the adenine of the conjugates and the surrounding ribosomal nucleobases.
The conjugates had superior profiles over a family of macrolide-lincosamidestreptogramin B against A2058/2059-mutated strains, which suggested that a suitable extended alkyl-aryl arm derived from the C-3 of macrolactone, perhaps supplied extra opportunities to combat high-level A2058 mutated resistance. Further research on novel conjugates to address the inefficiency of this work against clinical isolates is underway.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/molecules28031327/s1, Figure S1: The RMSD profiles of MD simulations of the G2099A mutant 50S ribosomal subunit with the ligands (37A and 32A). Figure S2: Superposition of two complexes including the last structures obtained from the MD and the complexes obtained from docking. Table S1: The interaction of hydrogen bonds between G2099A Large Ribosomal Subunits (PDB ID: 1YHQ) and 37A. Table S2: The interaction of hydrogen bonds between G2099A mutant 50S ribosomal subunit (PDB ID: 1YHQ) and 32A. Table S3: The interaction of hydrogen bonds between G2099A mutant 50S ribosomal subunit (PDB ID: 1YHQ) and 37A after molecular dynamic simulation. Table S4: The interaction of hydrogen bonds between G2099A mutant 50S ribosomal subunit (PDB ID: 1YHQ) and 32A after molecular dynamic simulation. Table S5: HRMS and NMR spectra checklist for the intermediates and target compounds (1-16 and 26-37). 1 H and 13 C NMR Spectra for the representative compounds.