Retrosynthetic strategy for 1–3 and 5–7We envisioned the synthesis of polycyclic 1–3 and 5–7 by employing the highly convergent strategy depicted in Fig. 2b. Retrosynthetically, we anticipated that 5–7 could be first traced back to hemiketals 1–3, respectively. The construction of the three contiguous stereocenters in the A and B rings of 1–3 would be a key issue of the synthesis. Disconnections between the C7 and R1 group (highlighted in blue), the C6′ and O at C1, and C1′ and H were considered to address the stereochemistry of C7 in the B ring and achieve desired configurations at C6′ and C1′, respectively. The double Michael addition annulation cascade reaction has proven to be a powerful tool in organic synthesis48. Thus, we envisaged using the double Michael addition reaction for the construction of consecutive three stereocenters. The strategy suggested that the α,β-unsaturated enone II is a key intermediate in the synthesis. Subsequently, disassembly of II provided red fragment 12 and tetramethylxanthenone III, which would be coupled via a Knoevenagel reaction. However, the presence of the carbonyl groups in III would undergo a competing Knoevenagel reaction, which would make the anticipated reaction challenging. III would be installed through a Rieche reaction of IV. IV would be expected to be accessible from rac-IV with the inexpensive and available 13 ( ~ 40 $/50 g) by a unique Mitsunobu-mediated chiral resolution. However, rac-IV would undergo the competitive chemo- and regioselective Mitsunobu reaction because of the presence of free phenol and the aromatic hydrogen in the precursor49. Moreover, the ee value may be affected because of the involvement of a stereocenter in this reaction50. These issues would make the anticipated Mitsunobu-mediated resolution approach challenging.Asymmetric syntheses of (+)- and (-)−1 in addition to (+)- and (-)−5We began our investigations with the synthesis of (+)- and (-)-myrtucommulone D (1), and (+)- and (-)-myrtucommulone E (5). As initial efforts to achieve the catalyzed asymmetric synthesis of the important intermediate isomyrtucommulone B (14) according to a previously reported protocol were not successful (see Supplementary Fig. 1.for details), we set out to develop a more robust route. To the best of our knowledge, chiral resolution via the diastereoisomers, which can be separated by silica gel column chromatography or recrystallization, has been widely used in asymmetric synthesis and pharmaceutical applications51,52. Thus, our synthesis of the key intermediate (+)- and (-)-14 commenced with a chiral resolution between rac-14 (prepared in two steps, see Supplementary Pages 18–20 for details) and the chiral reagents. After extensive experimentation (see Supplementary Figs. 14 and 15 for details), rewardingly, Mitsunobu-mediated chiral resolution of rac-14 was realized with the inexpensive chiral alcohol 13 at 25 °C under DEAD and PPh3 in PhMe providing the diastereoisomers, but the products were complex because of the poor regioselectivity of the reaction. To address this issue, regioselective protection of the free phenol should be considered. Given the subsequent facile unveiling of the phenol, treatment of rac-14 with BnBr followed by Mitsunobu-mediated chiral resolution (reaction condition: 13, DEAD, PPh3 in PhMe), afforded the desired compounds 15 and 16 with dr 1.3:1 in an overall 41% yield. After further optimization efforts, gratifyingly, we discovered that exposure of rac-14 to BnBr without purification, and subsequent Mitsunobu-mediated chiral resolution (reaction condition: 13, DEAD, PPh3 in PhMe/THF) resulted in excellent yield of the desired 15 and 16 with dr 1.1:1 (Fig. 3, 2.0 g scale, see Supplementary Pages 30 and 31 for details), which could be readily separated by recrystallization. At this stage, BCl3-promoted removal of the protective groups in diastereoisomers 15 and 16 readily occurred to deliver (+)- and (-)-14 in 96% yield with 99% ee and 99% ee (1.0 g scale), respectively. The absolute configurations of (+)- and (-)-14 were confirmed by X-ray diffraction analysis (see Supplementary Page 100 for details). The route described above allowed for the facile synthesis of 13 g of (+)- and (-)-14 (see Supplementary Page 32 for details), which highlights the robust nature of this chemistry. This strategy provides a route to the inaugural asymmetric synthesis of (+)- and (-)-14 by application of the unique Mitsunobu-mediated chiral resolution on an advanced tricyclic substrate containing the C7 stereocenter.Fig. 3: Syntheses of (+)−14 and (-)-ent−14 by the Mitsunobu-mediated resolution.DEAD, Diethylazodicarboxylate.With the optimized condition in hand, the generality of this strategy was explored using different substrates, including the precursors 11 and 17a for the synthesis of 2–3, to give various functionalized xanthenes. As shown in Fig. 4a, different 2-substituted and 4-substituted xanthene substrates (rac-8–9, rac-11, rac-17a–17 m) were tolerated in the optimized condition of the Mitsunobu-mediated chiral resolution to give the corresponding products in good total yields (58%–66%) with excellent ee values (92%–99%) in two steps. Hence, this method has broad generality for the asymmetric synthesis of xanthene derivatives. The absolute configuration of (-)-8 was confirmed by X-ray diffraction analysis (see Supplementary Page 100 for details). Notably, asymmetric syntheses of eight natural products (853, 929, 1154, 17a30, 17e55, 17 f56, 17i30, and 17l57) were achieved, overcoming the difficulties previously encountered in the asymmetric synthesis of 1148, 17, and 17e–17 f. Furthermore, the route described above allowed for the facile synthesis of 2.5 g of (+)-8 and (-)-8, 1.3 g of (+)-11 and (-)-11 as well as 2.6 g of (+)-17a and (-)-17a (see Supplementary Pages 23, 28, 35 for details, respectively). Encouraged by these results, this reaction will be beneficial for the asymmetric synthesis of complex natural products such as the myrtucommunins A and B58 (Fig. 4b), which is currently underway in our laboratory. Of note, there are currently no approaches available for the direct asymmetric synthesis of the 2-oxabicyclo[3.3.1]nonane scaffold, which is present in biologically interesting complex natural products such as myrtucyclitone C59. To our delight, (-)-17n, and (-)-17o, which is probably the key intermediate for the asymmetric synthesis of myrtucyclitone C, were provided by this Mitsunobu-mediated chiral resolution at an excellent 98.5% ee and 97% ee, respectively (see Supplementary Pages 73, 74, 76 and 77 for details). The absolute configurations of (-)-17n and (-)-17o were confirmed by X-ray diffraction (see Supplementary Page 100 for details). These results revealed that the unique Mitsunobu-mediated chiral resolution could be regarded as an alternative strategy for construction of synthetically difficult polycyclic chiral scaffolds.Fig. 4: Substrate scope of Mitsunobu-mediated resolution and potential applications.a Mitsunobu-mediated resolution of rac−8-9, 11 or rac−17a-17m. b The structures of myrtucommunins A-B with a xanthene scaffold. c The structures of highly optically pure (-)−17n and (-)−17o obtained through Mitsunobu-mediated resolution, along with the potential application of (-)−17o. Reaction conditions:[a]rac-7-8 (1.0 equiv.), or rac-11 (1.0 equiv.), or rac-17a-17o (1.0 equiv.), K2CO3 (1.5 equiv.), BnBr (1.05 equiv.), 60 oC, acetone (0.05 M), 10 h, then filtered and concentrated, then DEAD (1.5 equiv.), PPh3 (1.5 equiv.), 13 (1.2 equiv.), 0-25 oC,THF/PhMe (V/V = 1:1, 0.1 M), 1 h. [b]BCl3 (10.0 equiv.), CH2Cl2 (0.1 M), −50 oC, 1 h. cIsolated yields for 2 steps. [d]The ee values were determined by chiral HPLC analysis. Blue box with a background refers to natural products.Having established a reliable method for constructing (+)- and (-)-14 as well as their congeners, we then turned our focus to the asymmetric synthesis of (+)- and (-)-myrtucommulone D (1). Initially, we focused on the synthesis of Int-2 from 12 and 18 by a Knoevenagel reaction (Fig. 5a). This reaction is challenging not only because of the unfavorable regioselective Knoevenagel reaction of 18 with three carbonyl groups, but also because of the concomitant chemoselective retro-Aldol reaction (Fig. 5a, via Path II). Nevertheless, we speculated that the aldehyde group in 14 and the triketone in 12 are a strong nucleophilic acceptor and nucleophile, respectively, which possess high reactivity and would make it possible for this Knoevenagel reaction to proceed. The Rieche reaction of 14 with dichloromethyl ether in CH2Cl2 led to aldehyde 18 in 78% yield (1.2 g scale, see Supplementary Page 79 for details). On the basis of Romo’s work60, the Knoevenagel reaction of 12 and 18 under pyrrolidine as the base condition in CH2Cl2 proceeded smoothly to provide 14, but not Int-2 (via Path I) or 19 (Fig. 5a, via Path I, then Path III). We next turned to a variety of different bases (piperidine, Na2CO3, Et3N, and L-proline) in CH2Cl2, which forged the undesired 14 along with low yield (<10%) of the hemiketal 19 and some recovered starting material with no observed Int-2 under these conditions. To overcome this problem, we hypothesized that it may be possible to differentiate the reaction pathways resulting in 14, Int-2, and 19 by selecting suitable solvents. After an extensive investigation, with L-proline (1 equiv.) as the base and DMF as the solvent, though the desired Int-2 was not afforded, the cyclization product 19 with dr 1:1 was generated in 86% yield at 25 °C. Subsequently, we envisaged that retro-hemiketalization of 19 under strong basic conditions (i.e., LDA, and LiHMDS) provided Int-2. To our disappointment, the starting materials were recovered under basic conditions. At this stage, we proposed that retro-hemiketalization may occur. However, the above reaction was prone to form the more structurally stable 19 instead of Int-2 after quenching with solution. Thus, we expected that direct treatment of 19 with the isopropyl reagents would afford 1. We found that using a cascade reaction of hemiketal 19 to directly construct the three contiguous stereocenters at C7, C1′, and C6′ posed significant challenging. It was difficult to use hemiketal 19 to provide 1 in the presence of various isopropyl reagents [iPrMgBr, iPrLi, and (iPr)2Zn] and different solvents (THF, Et2O, CH2Cl2, and PhMe). Only some starting materials were recovered, probably because of the presence of the competing 1, 2-addition reaction. Subsequently, given the important utility of a cuprous reagent in the Michael addition reaction61,62, we initially sought to take advantage of this together with the reactivity of 19. To our disappointment, when we treated 19 and Grignard reagents or zincon with 3% to 30% of copper catalysts (i.e., CuI, CuBr, or CuBr·SMe2), the reactions were unsuccessful. At this stage, we envisioned that the complex was formed between CuI or CuII and the carbonyl group at C8 with an oxyanion at C3, which led to undesirable results63,64. Gratifyingly, treatment of 19 with iPrMgBr (3.5 equiv.) and an excess of CuI (1.1 equiv.) in THF at −50 °C afforded the desired (+)-myrtucommulone D (1) in 18% yield. Meanwhile, some starting materials were recovered. Encouraged by this result, we further explored a variety of solvents including CH2Cl2, PhMe, Et2O, 1,4-dioxane and THF/CH2Cl2. Rewardingly, using THF/CH2Cl2 instead of THF as the solvent accelerated the transformation, increasing the yield of (+)-1 to 58%. Subsequent screening of copper reagents (CuI, CuBr, CuCN, CuBr∙SMe2) revealed that CuCN provided the most excellent yield for 1. After extensive investigation, the optimal protocol was identified: when 19 was treated with iPrMgBr (3.5 equiv.) in the presence of CuCN (1.1 equiv.) in THF/CH2Cl2 at −78 °C to −50 °C, (+)-1 was obtained in 81% yield without the need for protecting groups. The structure of (+)-1 was confirmed by X-ray diffraction analysis (see Supplementary Page 99 for details). Subsequently, to further improve the efficiency for the synthesis of (+)-1, we explored a sequence of a Knoevenagel-hemiketalization annulation reaction of 12 and 18 without purification followed by iPrMgBr and CuCN to give 1 in 66% yield (120 mg scale, see Supplementary Pages 81-82 for details). Notably, a xanthene core and three contiguous stereocenters and one quaternary stereocenter were constructed with high diastereoselectivity by a two-step reaction in one pot. Encouraged by this success, a sequence of Knoevenagel-hemiketalization, and subsequent iPrMgBr and CuCN as well as elimination with p-TsOH, was exploited to afford (+)-myrtucommulone E (5) in 58% yield (120 mg scale, see Supplementary Pages 91 and 92 for details). The structure of (+)-5 was confirmed by X-ray diffraction analysis (see Supplementary Page 99 for details). Moreover, the preparation of (-)-myrtucommulone D (1) and (-)-myrtucommulone E (5) were completed from (-)-14 (Fig. 5b, see Supplementary Pages 81 and 82 and 91 and 92 for details). The 1H and 13C NMR spectra of the newly synthesized 1 and 5 were identical to those of the natural products (see Supplementary Tables 2–3 and 7 for details). Of note, we have also tried to provide 1 and 5 via various other routes (see Supplementary Fig. 4–6 for details, respectively). For example, with 8 in hand, 8 and 20 underwent a Friedel-Crafts type Michael addition (K2CO3/CH2Cl2) to assemble 21 and 22 with dr 1:1.1 in 92% yield (Fig. 5c). However, unfortunately, the regioselective intramolecular cyclization followed by dehydration of 22 with different acids (i.e., p-TsOH, CSA, TFA, TiCl4, and BF3·Et2O) did not furnish 5 (see Supplementary Fig. 5 for details). Therefore, the new synthetic route reported in this article was highly efficient for the asymmetric synthesis of 1 and 5 with an angular [6-6-6-6-6] pentacyclic skeleton.Fig. 5: Asymmetric syntheses of 1, 5, and 21-22.a, b Asymmetric syntheses of (+) and (-)−1 in addition to (+)- and (-)−5. c Asymmetric syntheses of 21 and 22.Encouraged by the above experimental outcome, we next turned our attention to the two possible mechanisms responsible for the diastereoselective construction of 1 from 19, which are proposed in Fig. 6a (Path A, highlighted in red; Path B, highlighted in blue). Subsequently, to gain insight into which pathway is more probable, the density functional theory (DFT) calculations were carried out. We found that retro-hemiketalization of 19 (Path A) occurred with a lower barrier than that of the Michael addition of 19 (Path B) to give Int-5 and Int-6 (see Supplementary Fig. 80 for details), which implied that hemiketal 19 is more likely to proceed via path A to give 1. However, the diastereoselective construction of the three contiguous stereocenters remains an interesting problem. Initially, the optimization of the 3D structure of Int-3a by DFT was afforded because the anionic Int-3 could not be found to be a minimum on the potential energy surface, and the steric effects of Int-3 and Int-3a were similar. Clearly, Michael addition of Int-3 would occur on the Re-face (Fig. 6b) leading to the generation of Int-4 because of less steric hindrance. Subsequently, the thermodynamic process of Int-4 to 1 was suggested by the following experiments: a. treatment of 1 with NIS and NaOH in THF at −40 °C to 25 °C resulted in the sprio-1a (Fig. 6c, see Supplementary Fig. 75 for details), which indicated that retro-hemiketalization of 1 indeed occurred (most likely via Int-9); b. exposure of 1 to NaOH and THF without NIS at the same temperature range could only recover the starting material. Then, the subsequent DFT-minimized structures of Int-7 and Int-8 were obtained, which showed that an intramolecular oxa-Michael addition of Int-4 would afford Int-7 due to the more stable envelope-like conformation in the A ring, with a free energy that was 6.0 kcal/mol lower than that of Int-8 (Fig. 6a, d). Similarly, comparison of the conformations and free energies of 1 and 1′-epi-1 also showed good diastereoselectivity toward the formation of 1 would be gave (Fig. 6a, d). Thus, the mechanism of stereoselective formation of 1 was explained. Based on the above results, we hypothesized that the structural configurations of (-)-callistenone D (2)32 and (-)-rhodomyrtosone R (3) are probably identical, and the structural assignment of (-)-3 probably needs to be revised. To validate our proposal, we would plan to use the above strategy to achieve (-)−2 and (-)-3.Fig. 6: Mechanistic investigation of the diastereoselective synthesis of 1.a The proposed and computational studies to understand mechanism of diastereoselectivity of the cascade reaction. b Depiction of the attack onto the two faces of the Int-3a. c Investigation of retro-hemiketalization. d 3D structures of Int-7-8, 1′-epi−1 and (+)−1 with their A ring highlighted in four different colored circles. NIS N-Iodosuccinimide.Asymmetric syntheses of (+)- and (-)-2–3 as well as (+)- and (-)−6–7Having achieved the asymmetric syntheses of (+)- and (-)-1 as well as (+)- and (-)-5, and gained an understanding of the mechanism of the cascade reaction, we then set out to the construction of (+)- and (-)-2–3 as well as (+)- and (-)-6–7 as shown in Fig. 7a-c. The preparation of (+)- and (-)-2–3 and (+)- and (-)-6–7 was obtained from (+)-11, (-)-11, (+)-17a, and (-)-17a, respectively, through a synthetic route similar to that used to generate 1 and 5 (Fig. 7a, see Supplementary Pages 81 and 91 for details). The structure of (+)-2 was confirmed by X-ray diffraction analysis (see Supplementary Page 99 for details). The 1H and 13C NMR spectra of the newly synthesized 2–3 and 6–7 were identical to those of the natural products (see Supplementary Tables 4–6, 8–10 for details, respectively). Notably, among them, the absolute configurations of natural 3 and 7 have not been reported previously29,31, and the signs of the optical rotation of natural 3 and 7 were identical to those of the newly synthesized (-)-3 and (-)-7. Therefore, the structure of natural 3 was revised to be (1′S, 6′R, 7 R, 7″S) (Fig. 7b), and the absolute configuration of natural 7 was determined to be (7 S, 7″S) (Fig. 7c). These outcomes demonstrate that this strategy has the huge potential for the preparation of more complex compounds with a xanthene core.Fig. 7: Asymmetric syntheses of 2-3, 6-7 and stereochemical assignment of 3 and 7.a Asymmetric syntheses of (+)- and (-)−2–3 as well as (+)- and (-)−6–7. b, c The structural revisions of natural (-)−3 and (-)−7.Antibacterial activities of synthetic compounds in vitro and in vivoNext, we evaluated the antibacterial activity of the 66 compounds prepared in the present study against seven Gram-positive and three Gram-negative bacteria. The results showed that there was no significant difference in the antibacterial activity between the enantiomers. In addition, the hemiketal moiety plays a critical role in their antibacterial activity, based on the comparison of the MICs of 1-3 and 5-7. Moreover, 22 exhibited the most potent antibacterial activity against MRSA (MIC 0.5 μg/mL), compared to other synthetic compounds (see Supplementary Table 11 for details). Therefore, we evaluated the antibacterial activity of 22 in vivo. As expected, the treatment of 22 (3.75 mg/kg) resulted in a reduction of wound area and the prevention of skin ulcer formation in a mouse skin infection model (P < 0.0001). This effect was comparable to that of vancomycin (see Supplementary Fig. 254 for details). In addition, we compared the time of emergence of resistance in S. aureus induced by 22 and norfloxacin. The results showed that the time of induction of drug resistance by 22 was significantly longer than that by norfloxacin (see Supplementary Fig. 255 for details).Investigation on the mechanism of actionTo elucidate the mode of action of 22, we successfully obtained four strains (SA22-SR-1 to SA22-SR-4) that were spontaneously resistant to 22 through serial passaging (see Supplementary Table 12 for details). We initially confirmed that the susceptibility of SA22-SR remained unchanged relative to other antibacterial agents, including vancomycin, ofloxacin, linezolid, kanamycin, meropenem, and tigecycline (see Supplementary Table 13 for details). Then the genomic differences between the wild-type (WT) parental strain SA29213 and the spontaneously resistant strains were analyzed. Nine genomic differences, including seven single nucleotide (gyrA, walK, valS, plsY, and three hypothetical genes) and two indels (glpK and a hypothetical gene), were identified in all SA22-SR strains (see Supplementary Table 14 for details). We selected one of the strains with the least number of mutations (SA22-SR-1) for subsequent experiments, so as to exclude the interference of other non-shared mutations (see Supplementary Table 14 for details) as much as possible. To the best of our knowledge, WalK histidine kinase is essential and specific to low G + C Gram-positive bacteria such as S. aureus, and has been recognized as promising targets65. Given the antibiogram of 22, which is mainly effective against Gram-positive bacteria, WalK could be considered as a reasonable target. Thus, we focused on examining the effects of 22 on the activity of WalK.WalK is a transmembrane histidine kinase that works in conjunction with the response regulator WalR to regulate the expression of autolysins, as well as virulencefactors such as hemolysis genes hla, hlb and hlgABC, and genes responsible for biofilm formation in S. aureus. In this study, we observed increased transcription levels of WalKR-regulated genes (atlA, lytM, hla, ssaA, sceD, SA0710, SA2097, SA2353, and isaA), increased sensitivity to lysostaphin-induced lysis of SA29213, biofilm formation and hemolysis on sheep blood agar after 22 treatment (Fig. 8a–c, see Supplementary Fig. 256 for details), indicating the activation activity of 22 on WalK. In contrast, compared to SA29213, SA22-SR-1 exhibited decreased transcription levels of WalKR-regulated genes (atlA, hla, SA0620, ssaA, sdrD, ebpS, sceD, SA2097, isaA), biofilm formation, hemolysis on sheep blood agar and inhibited lysostaphin-induced lysis (Fig. 8a–c, see Supplementary Fig. 256 for details), indicating that the decreased activity of WalK in SA22-SR-1 which probably due to the R86C mutation in WalK. As expected, all the changes associated with WalKR function were not affected by 22 in the SA22-SR-1 cells (Fig. 8a–c). These results suggested that the WalKR pathway was activated by 22.Fig. 8: Effects of 22 on WalK activity.a Transcription levels of genes regulated by WalKR in the presence or absence of 22 (0.1 μg/mL). b Lysostaphin-induced lysis process in the presence or absence of 22 (0.025 μg/mL). c Biofilm formation of different strains in the presence or absence of 22 (0.03, 0.06 μg/mL). Three biologically independent experiments at least were performed in a–c. The mean is shown, and error bars represent the s.d. P values were determined using a nonparametric one-way ANOVA. d SPR analysis demonstrated the binding of 22 to WalK. e 22 docked with the erWalK structure (PDB ID: 5IS1). The space-filling model shows the binding of 22 in the β-sheet pocket. f 22 docked with the erWalKR86C structure (PDB ID: 7DUD). The protein residues, hydrogen bonds and hydrophobic bonds are shown in a solid-ribbon and in 2D view. Pi-Donor hydrogen bonds, conventional hydrogen bonds and hydrophobic bonds are colored orange, green and purple, respectively. Source data are provided as a Source Data file.We then employed CRISPR-Cas9 and successfully recreated the R86C mutation [SAWalK(R86C)] in the SA29213 background (see Methods section for details). The MIC of 22 against SAWalK(R86C) was 2 μg/mL, indicating that the level of resistance did not reach that observed in SA22-SR strains. The transcription levels of lytM, hla, ssaA, sdrD, ebpS, sceD, SA0710, SA2097 and SA2353 were decreased in SAWalK(R86C), indicating that the function of WalK was down-regulated (see Supplementary Fig. 257 for details). The MIC of vancomycin against SAWalK(R86C) was 2 μg/mL, suggesting that other point mutations in SA22-SR strains contribute to restoring vancomycin susceptibility. Our results, along with previous reports that have generated isogenic mutants66,67,68, suggest that mutations at different positions of WalK contribute differently to vancomycin susceptibility. However, we found that the R86C mutation of WalK is highly unstable and usually spontaneous reverses to the wild state after 2-3 passages of culture. We deduced that some of other point mutations in SA22-SR strains may act as necessary compensation for the walK mutation. Therefore, we constructed SA22-SR-1:walK by the episomal expression of wild-type walK in SA22-SR-1. Our results showed that transcription level of genes which are positively regulated by WalKR and the sensitivity to lysostaphin-induced lysis in the presence of 22 were restored in SA22-SR-1:walK cells, suggesting that 22 may act on wild type WalK (see Supplementary Fig. 258 for details).For more evidence, the effect of 22 on bacterial autolysis was investigated by examining the membrane integrity. As expected, we observed that the use of 22 significantly increased the uptake of propidium iodide (PI) (see Supplementary Fig. 259a for details) and increased numbers of dead bacteria as the cell membranes burst (see Supplementary Fig. 259b for details). Moreover, using the fluorescence agent DiSC3(5), we observed a significant decrease in fluorescence intensity after treatment of 22, suggesting that this compound caused hyperpolarization of the bacterial membrane (see Supplementary Fig. 259c for details).To verify this speculation, we turned attention to exploring the binding between 22 and WalK. Since the full-length protein is prone to degradation, we purified the extracellular part of the receptor WalK (erWalK, residues 35-182 of uniprot ID: Q6GKS6) (see Supplementary Fig. 260-262 for details) and determined the binding kinetics of 22 to erWalK using surface plasmon resonance (SPR) analysis. The results showed an association rate constant ka of 9.32 × 102 M−1s−1, a dissociation rate constant kd of 8.45 × 10−3 s−1, and an equilibrium dissociation constant KD of 1.11 × 10−5 M (Fig. 8d), indicating that 22 can directly interact with WalK. As expected, the SPR result of 22 and erWalKR86C (erWalK with R86C mutation) showed an equilibrium dissociation constant KD of 7.38 × 10−5 M, indicating a marked decrease in the affinity of erWalKR86C with 22 compared to erWalK (see Supplementary Fig. 263 for details).To further investigate this interaction, we docked 22 into the wild-type structure of WalK using Autodock Vina. The ligand was found to interact with WalK in the α1 and β2 regions where the mutant structure differs most from the wild-type structure. Specifically, 22 forms three hydrogen bonds within the β-sheet pocket formed by β1-5 and α1 (Fig. 8e, see Supplementary Fig. 264 for details). The calculated affinity of 22 for erWalK using AutoDock Tools is 2.01 × 10−5 M, which is consistent with the results obtained using SPR. The three residues forming hydrogen bonds with 22 are Arg93, Asp139 and Asn109, and the hydrocarbon potion of Lys106, and residues Leu112, Ile136, and Tyr140 interacts with 22 by hydrophobic interactions (Fig. 8e, Supplementary Table 17 for details). In contrast, 22 docked poorly a predicted affinity of 6.56 × 10−5 M to the same pocket in erWalKR86C (PDB ID: 7DUD, Supplementary Table 16 for details), which was obtained in our study. The local structural rearrangements in R86C diminished most of the interactions, leaving only Lys138 for pi-cation interaction and Leu112 and Try165 for hydrophobic interactions (Fig. 8f, Supplementary Table 18 for details). Taken together, these results indicated that 22 activates the WalK function. To our knowledge, WalK inhibitor has been reported69, whereas its activators have not been reported. The present results confirm that the WalKR pathway is a vulnerable site in Staphylococcus, and 22 represents the inaugural WalK activator with antibacterial activity.Of note, due to the differences in structure between the erWalK of Streptococcus pneumoniae (S. pneumoniae; extra cellular PAS domain only has one transmembrane helix) and S. aureus, as well as the MIC for S. pneumoniae with compound 22 is only double that of S. aureus, it would suggest that 22 may not be specific for erWalK and could potentially have a second target. The variance in resistance levels between SAWalK(R86C) and SA22-SR strains to 22 further supports this hypothesis. Therefore, we must point out that although this study identified the interaction between 22 and WalK, it cannot exclude the possibility of other targets. Other essential gene mutations in SA22-SR, such as plsY, which are worthy of further investigation.In this work, we present a concise approach for the collective asymmetric total syntheses of (+) and (-)-myrtucommulone D (1), (+) and (-)-callistenone D (2), (+) and (-)-rhodomyrtosone R (3), (+) and (-)-myrtucommulone E (5), (+) and (-)-myrtucomvalone D (6), and (+) and (-)-callistenone C (7) in six steps70,71,72. The key to the synthesis is the use of a Mitsunobu-mediated chiral resolution with broad substrate applicability and high optical purity (92%–99% ee). Additionally, the establishment of a unique Knoevenagel/hemiketalization annulation followed by a retro-hemiketalization/double Michael addition cascade reaction not only constructs the [6-6-6-6-6] pentacyclic system, but also stereoselectively introduces three consecutive stereocenters and a quaternary stereocenter. The mechanism of this stereoselective transformation is illustrated by Quantum mechanical calculations. Based on the total synthesis, the absolute configurations of rhodomyrtosone R and callistenone C are determined. Notably, this synthesis provides 66 compounds for further study of their antibacterial activity. Among them, compound 22 shows the most effective activity against MRSA in vitro and in vivo. Genetic and biochemical studies revealed that 22 is an antibacterial agent to affect autolysis by activating WalK, making it a promising lead compound with an unusual mechanism that is unlikely to lead to drug resistance73. This work lays the foundation for the asymmetric synthesis of various polycyclic xanthenes and will accelerate the development of antibacterial agents.
