Cystobactamid 507: Concise Synthesis, Mode of Action, and Optimization toward More Potent Antibiotics

Abstract Lack of new antibiotics and increasing antimicrobial resistance are among the main concerns of healthcare communities nowadays, and these concerns necessitate the search for novel antibacterial agents. Recently, we discovered the cystobactamids—a novel natural class of antibiotics with broad‐spectrum antibacterial activity. In this work, we describe 1) a concise total synthesis of cystobactamid 507, 2) the identification of the bioactive conformation using noncovalently bonded rigid analogues, and 3) the first structure–activity relationship (SAR) study for cystobactamid 507 leading to new analogues with high metabolic stability, superior topoisomerase IIA inhibition, antibacterial activity and, importantly, stability toward the resistant factor AlbD. Deeper insight into the mode of action revealed that the cystobactamids employ DNA minor‐groove binding as part of the drug–target interaction without showing significant intercalation. By designing a new analogue of cystobactamid 919‐2, we finally demonstrated that these findings could be further exploited to obtain more potent hexapeptides against Gram‐negative bacteria.


Introduction
The ongoing prevalence of antibiotic-resistant bacteria poses an imminent threat to humanity. [1,2] Therefore, the need to dis-cover novel antibiotics with new chemical scaffolds has been established. Nature represents ar ich repository of antibiotics; however,t he major parto ft hese natural products (NPs) have to be modifiedt oo ptimize pharmacokinetic and pharmacodynamic properties. [3,4] Recently,w er eported the discovery of the cystobactamidsan ew family of antibiotics-isolated from Cystobacter sp. (Figure 1). [5,6] Cystobactamids (1a-1f)h ave hexapeptidic structures comprised of three p-aminobenzoic acid motifs (eastern part) and two p-nitro/aminobenzoic acids (western part) connected through different linkers. The structurally simplest natural cystobactamid5 07 (2), is at ripeptide representing the eastern partofm ost of the hexapeptides (1a-f)( Figure 1). The cystobactamidsd isplay broad-spectrum antibacterial activity through inhibition of topoisomerases typeI IA, namely DNA gyrase and topoisomerase IV. [5] Here, we reports tudieso nt he optimization of 2 because it is the only compound with appropriate physicochemical properties for oral absorption according to Lipiniski. [7] Results and Discussion We used gyrase inhibition to establish the structure-activity relationship (SAR), as it is the primary target of cystobactamids in Escherichia coli. [5] The assay evaluates the effect on the DNA supercoiling activity of gyrase using ar elaxed circularp lasmid as as ubstrate. Cystobactamids have been shown to inhibit gyrase similar to quinolones throughs tabilization of the covalent enzyme-DNAc omplex with double-stranded DNA breaks resultingi na na ccumulationo fl inear DNA. [5] Previously,w ereported that the methyl homologue 3 (Figure 2a nd Ta ble 1) showedo nly as light decreasei na ctivity,s uggesting that the naturala ntibiotic 2 could be amenable to structural modification. [8] We therefore investigated the role of the substituents and the conformation of 2 for activity.R eplacement of the isopropoxy side chain in the middle ring of 2 with am ethoxy group (4)d id not impact activity,w hereass ubstitution with chlorine (5)r esulted in as ignificant loss of gyrase inhibition (Table 1).
This result highlights the importance of the alkoxys ide chains for ligand-target interaction. On the other hand, the isopropoxy of the C-terminal ring showeds tronger inhibition than the methoxy in this position( 4 vs. 3), probably duet oi ts steric and hydrophobic properties. Consequently,w ek ept it in the optimization process.
As an ext step, we performed in silico conformational analyses of compounds 2-4 using molecular dynamics (MD) to identify the potential conformationso ft he molecules and their relative energies. Generally,t he compounds adopt linear conformations with abackbone curvature [9] of about 1588 ( Figure S31 and S32). They can adopt two constrained conformations (anti and syn)w ith respect to the alkoxy side chains that are controlledb yt he hydroxy group at the middle ring. The lowest energy conformationi st he anti form ( Figure 3A), which is stabilizedb yt hree intramolecular hydrogen bonds (IMHBs). An IMHB between C4-NH and C3-alkoxy group restricts rotationo f the C-terminal ring aroundt he Ar-NH axis. Another IMBH between C1'-CO and C2'-OH restricts rotationo ft he middle ring aroundt he Ar-CO axis. The thirdI MHB between C4'-NH and C3'-alkoxyg roup restricts rotation of the middle ring around Ar-NH axis. The syn conformer is also stabilized by three IMHBs similar to the anti form, except that C2'-OH switches from HB donor to HB acceptora nd forms as ix-membered ring with C4-NH (restricting rotation of the middle ring aroundt he Ar-CO axis) ( Figure 3B). The energy difference (DE)b etween the anti and syn formsi s0 .4-0.7 kcal mol À1 .S uch am inor energy difference could allow the interconversion between both conformations at ambient temperature.
We then investigated the preferred conformation for 2-4, their esters and nitro precursors experimentally in solution by aN OESY study in ab iomimetic solvent (20 %H 2 O/ [D 6 ]DMSO). [10] In agreement with MD calculations, all compounds show as trong cross-peak between C4-NH and C6'-H, and aw eak or no cross-peak with C2'-OH, indicating that these compounds predominantly exist in the anti conformation (Figure 4a nd S8). Using standard NMRs olvents, the same results were obtained (FigureS1-S10).
Subsequently,w ev erified whether the syn conformer could exist under physiological conditions by a 1 HNMR experiment for the cystobactamid 507 isopropyl ester (26)a t2 0a nd 37 8C. We found that the chemical shift of the C2'-OH proton moved upfield from 12.40 ppm at 20 8C( mainly HB donor form, anti)   to 12.35 ppm at 37 8C( pushing the conformational equilibrium to HB acceptorf orm, syn)( Figure S29). In addition, we observed the syn conformationi nt he solids tate by determining the crystal structure for the dipeptidep recursor of 3 (Figure S37E), whereas in solution there was ap revalence of the anti conformation ( Figure S11). These resultsd emonstrate that these compoundsc an easily interconvert from anti to syn at physiological temperature. Since the anti form was the predominant conformation in solutionf or 2 and all derivatives so far,w ed esigned compoundst hat preferentially adopt the syn form by blockingt he ambiguous hydroxy motif by methylation, thusc onverting it into aH Ba cceptorg roup only (compound 6). MD calculations indicated that 6 typically adopts an IMHB-stabilized syn conformation with al arge DE compared to the anti form (3.8 kcal mol À1 ). NOESY studies showed ac ross-peak between C4-NH and C2'-OMe, and no cross-peak with C6'-H, indicating that the syn conformation is predominant ( Figure S15). Compound 6 showedathreefold highera ctivity than 2 ( Table 1), indicating that the hydroxy group is not essential for activity.
To clarify whether this enhancement was due to the induction of the syn conformation or due to an additional hydrophobic interaction, we first designedc ompound 7,b earing an isopropoxyg roup in place of the methoxy( 6). This modification resultedi nasix-foldi mprovement in activity compared to 2,i ndicating that, besides restricting the conformation to the syn form, alkoxy groups at position2 of the middle ring also contributetotarget interactions.
To investigate whethert he syn conformation is ad ecisive factor,w et hen designed two rigid cystobactamid 507 analogues, 8 and 9,w ith ap yridine scaffold adopting only one conformation, anti or syn,r espectively ( Figure 2). Rigidity was achieved via ab ifurcated IMHB between C4-NH and oxygen atom of C3-isopropoxyaswell as the nitrogen atom of the pyridine ring. We confirmed the stability of 8 and 9 by MD calculations,w hich showedo nly the desired conformation in an energy window DE of 7.0 kcal mol À1 .I nN OESY experiments, no cross-peak was observed between C4-NH andp yridine C3-H at 27 8Ca nd at higher temperatures up to 67 8C( Figure S21-S27). Moreover,X -ray crystal structures of the compoundsw ere as expected ( Figure 5, S37C and S37D). Results revealed that 9 is fourfold and sevenfold more potent than 8 and 2,r espectively ( Table 1), indicating that the syn form is indeedt he more active conformation of cystobactamid5 07. Notably,t he slightly bettera ctivity of 8 than 2 suggestst hat the introduced pyridine ring might contribute to the interaction with gyrase and the inhibitory effect.
The newly designed compound 9 is clearly advantageous compared to the natural compound 2.I ntroduction of ap olar pyridine ring in lieu of the benzene enhances water solubility and ligand-lipophilicity efficiency [11] (Sol pH 7.4 :5 9.4 mmol mL À1 , LLE:3 .54 for 9 vs. 9.6 mmol mL À1 ,1 .75 for 2,r espectively). In addition, removal of the hydroxy group, while keeping the right conformation,i ncreases ligand efficiency [11] (LE:0 .17 for 9 vs. 0.13 for 2). Moreover,t he enhancement of the binding affinity of the rigid ligand 9 is at least in part due to the reduction of the unfavorable entropic contribution to the Gibbs free energy of binding. [12] Another important achievementc ould be that hopping of the cystobactamid5 07 scaffold to the novel pyridine-based chemotype (compound 9)m ight circumvent the cystobactamids' cleavage through AlbD, ak nown resistance protein inactivatingt he structurally relateda ntibiotics albicidins. [13] The AlbD endopeptidase hydrolyses the amide bond between the middle and the N-terminal ring of the eastern parto fa lbicidins/cystobactamidsa nd the tripeptidic derivatives. [13] By incubation of compound 9 with AlbD under the reported conditions, [13] our hypothesis turned out to be correct;n oc leavage of 9 in the presence of AlbD was observed ( Figure 6).
For furtherS AR exploration, compound 2 was simplified by omitting either the isopropoxy or the hydroxy group from the centralr ing in 10 and 11,r espectively ( Table 1). Removal of the former resulted in al oss of activity,e mphasizing the importance of alkoxy substituents for activity.I nterestingly,r emoving the hydroxy group increased thea ctivity twofold compared to 2.T his confirms that the hydroxy group is not essential for activity.I ts removal permits free rotation of the middle ring, as   Figure S28). Introduction of an isopropoxyorhydroxy group at the N-terminal ring of 10 and 11,t om aintain the beneficial free rotation, resulted in restoration of activity with af ourfold enhancement compared to 2 (13 and 12,r espectively)( Ta ble 1). These results demonstrate the usefulness of the alkoxy motifs (2 and 13 vs. 10), and indicate that varying the distance between them can be tolerated because of the flexibility of the new analogues.
Esterificationo f4 and the most potent compounds 6, 7, 9 and 11-13 decreaseda ctivity strongly,revealing that the terminal carboxyl group is important for activity (Table S2). Replacing the amino moiety of 12 and 13 with nitro groups, as in the hexapeptidic cystobactamids, resulted in similar inhibitory activities (14 and 15,r espectively) ( Table 1). This indicatest hat substitution in the 4''-position of cystobactamid 507 analogues provides only little contribution to activity.
For the synthesis of 2 and its analogues, retrosynthetic analysis revealed three units of either p-aminobenzoic acid or 5aminopicolinic acid derivatives linked by amide bonds. Accordingly,t he individual middle, C-and N-terminal rings were prepared followed by the coupling of the constituents. We established brief and efficient synthetic pathways for novel as well as reported amino acids, which are also precursors of other NPs [14][15][16][17] (Scheme S1). Compared to previous methods, [6,8,18] the middle ring of 2 (20)w as prepared in only four steps using catechola sastarting material. Nitration of the catechol to the 3-nitro derivative 17 followed by regioselective isopropylation at the 2-hydroxy positionu sing as toichiometric amount of 2bromopropanep rovided 18. ortho-Formylationo f18 with paraformaldehyde in MgCl 2 /TEA/MeCN mixture under strictly anhydrous conditions produced the p-nitrobenzaldehyde 19.T he latter was finally oxidized using AgNO 3 under basic conditions to yield the corresponding acid 20 (Scheme1), the structure of which was confirmedb yX -ray crystallography ( Figure S37A). The C-terminal ring 22 was prepared by isopropylationo f3 -hydroxy-4-nitrobenzoic acid followed by reduction of the resulting nitro derivative 21 via heating with iron in ethanol (Scheme1).
For amide coupling, three obstacles were encountered:T he free OH as an interfering group, the nonreactivec arboxylic acids, and the weakly reactive aromatic amines. [6,8,16,18] We developedastraightforward strategy that could overcome these difficulties and bypasst he activation of carboxylic acids and phenolic OH protection/deprotection. [6,8,16,18] The total synthesis of 2,f or example, wasa ccomplished in overall 11 steps instead of 21 and 13, respectively. [8,18] Coupling of the central ring 20 to the C-protected C-terminal ring 22 wasa chieved by heatingt he mixture with dichlorotriphenylphosphorane in anhydrous chloroform to afford dipeptide 23 (Scheme 1). After reduction,t he corresponding amine 24 was coupled with p-nitrobenzoic acid by using the same procedure as described above.R eduction of the nitro substituted tripeptide 25 followed by C-deprotection via ester saponification yielded the amino acid 2 (Scheme 1). This synthetic procedure was broadly applicable ande nabled us to preparealarge series of peptidomimetics in short time with good to excellent yields (Scheme S2-S4).
We evaluated the inhibitory activity of 2 and the most potent gyrase inhibitors 7, 9,a nd 12 on the second bacterial target of cystobactamids( topoisomerase IV) using ar elaxation assay,w hich assessest he effect on the conversion of as upercoiled plasmidi ntoarelaxedf orm. [5] Results indicated that 2 was not active up to 500 mm,w hereas the new analogues displayed moderate inhibitory activities (Table 1). This demonstrates that the modifications applied to improvethe gyrase inhibitory activity are also valid for topoisomerase IV,w hich is not surprisingd ue to the high homology between both enzymes. [19] Moreover,t he same activity trend of the compounds on both targets suggestsasimilar mode of action. The primary binding site of the cystobactamids is probably located at the gyrase-DNA interface overlapping that of the quinolone antibiotics. [5] To gain adeeper insightinto the mode of action,weinvestigated whether and eventually how cystobactamids and their analogues bind to the DNA part of the target complex. There are two main binding modeso fs mall molecules to DNA: minor groove binding or intercalation. [20] Intercalationi so fp articularc oncern, as compounds that adopt this binding mode may trigger genotoxic effects in eukaryotes. [21] We carriedo ut displacementt itratione xperiments using fluorescent dyes that show increased fluorescence upon DNA binding:H oechst 33342 for DNA minor-groove binding and ethidiumb romide (EtBr) for intercalation. Titration of calf thymusD NA bound Hoechst 33342 with cystobactamids 1a, 1b,a nd 2-15 induced ac oncentration-dependent loss of fluorescence( Figure 7A and S39). No compound-induced fluorescenceq uenching was observed in the absence of DNA. In the presence of EtBr,noc ompound showed significant reduction in fluorescence ( Figure 7B and S39). These resultsi ndicate that the cystobactamids are able to bind to DNA utilizing the minor groove withouts ignificant intercalation. Moreover,t hey reveal that the cystobactamids and their analogues are an ew chemical frame for minor groove recognitionb esides the known family of five-membered fused and non-fused heterocyclic polyamides. [22] Evaluation of the antibacterial activity against ap anel of Gram-positivea nd Gram-negative bacteria revealed that the new compounds show up to 8-to 16-fold enhanced activities againstG ram-positive strains compared to the parenta ntibiotic 2 ( Table 2). Ag ood correlation between the antibacterial effects and topoisomerases IIA inhibitory activities was observed. In contrast to 1a and 1b, [5,6] compound 2 and its analogues did not show activity against E. coli wild-type, but they were active against the efflux-deficient E. coli tolC3 mutant. This implies that efflux is responsible for the inactivity toward E. coli wild-type. Some compounds, like 7 and 12,s uffered from penetration issues throught he Gram-negative outer membrane, as indicated by the enhanced MIC values (eightfold) in the presence of ap ermeabilitye nhancer.
Additionally,w ei nvestigated compounds 2, 7, 9,a nd 12 for their phase Ia nd II biotransformation using human liver S9 fraction.C ystobactamid5 07 and all synthetic compounds displayed an extraordinarily high metabolic stability( t 1/2 > 240 min, FigureS42). Surprisingly,s tability was observed for both amide groups against peptidases as wella sf or the hydroxy groups of 2 and 12 toward conjugating enzymes. Moreover,c onformational modification anti to syn,i .e., 2 to 7 and 9 maintained the outstanding metabolic stability of the natural compound.
Finally,t od emonstrate that structure optimizationo f2 can be translated into potent hexapeptides, we picked an analogue with ah ydroxy group at the middle ring to keep the general features of isolated cystobactamids. [6] Accordingly, compound 4 was connected to the western part of the natural compounds through l-asparagine as as implified linker to form the cystobactamid 919-2 analogue 16 (Scheme 2). Compound 16 showed potent gyrasei nhibition and antibacterial activity on the Gram-negative E. coli wild type (Table 1a nd 2). This indicates that modification of the natural cystobactamid 919-2 using the tripeptidic cystobactamid 507 analoguesi sa n appropriate strategy to improve activity.M oreover,r emoval of the methoxy group at the linker was tolerated, offering an ew space forfurtheroptimization.
The hexapeptide 16 was preparedv ia convergent synthesis including ac oupling of two tripeptide fragments;n amely,t he eastern and the western part with al inker. [23] The eastern part was prepared as described above as C-Boc protected amino acid 34 (Scheme 2). Synthesis of the western fragment started

Conclusions
We described ab rief synthetic route for the natural antibiotic cystobactamid 507 (2)a nd the design of new derivatives with improved topoisomerases IIA inhibition, and antibacterial activity.S AR studies revealed the importance of the alkoxys ide chains and the irrelevance of the hydroxy group forg yrase inhibition.T he terminal carboxy and amino/nitro moieties were found to be necessary for activity.T he molecular conformation in solution as well as in the solid state was investigated to deduce the conformation-activity relationship. We pioneered the design of noncovalently bonded rigid structures through IMHBs to disclose the bioactive conformation of 2 as the syn form without using the classical methods of cyclization or introduction of sterically demanding moieties. [24] Increase of chemicald iversity by modification of the cystobactamid 507 scaffold into ap yridine-baseds tructure (compound 9)t urned out to make the molecule stable toward the albicidin resistance factor AlbD. It was shown that the cystobactamids' mode of action could be, at least in part, mediated by DNA minorgroove bindinga nd not intercalation. An important advantage of 2 andthe novel analogues is the high metabolic stability.Ultimately,w ed emonstrated that optimization of the tripeptides could be translated into hexapeptidic cystobactamids with improvedg yrase inhibition and antibacterial activity against Gram-negative strains. It is worth mentioning that compounds could be designedw ith better pharmacokinetic properties compared to the natural compounds. Thus, this work would be useful for the development of cystobactamidsa nd similar natural compounds [14][15][16] towardb etter antibiotics.

Experimental Section
Experimental procedures for the synthesis of the final compounds and intermediates as well as their characterization, NMR spectra, computational work, biological experiments and X-ray crystallographic data are described in detail in the Supporting Information.