Alanine Scan of [L-Dap²]Ramoplanin A2 Aglycon: Assessment of the Importance of Each Residue

Abstract

In efforts that define the importance of each residue and that identify key regions of the molecule, an alanine scan of the ramoplanin A2 aglycon, a potent antibiotic that inhibits bacterial cell wall biosynthesis, is detailed. As a consequence of both its increased stability (lactam vs lactone) and its “relative” ease of synthesis, the alanine scan was conducted on [Dap²]ramoplanin A2 aglycon which possesses antimicrobial activity equal to or slightly more potent than ramoplanin A2 or its aglycon. Thus, 14 key analogues of the ramoplanin A2 aglycon, representing a scan of residues 3–13, 15, and 17, were prepared enlisting a convergent solution-phase total synthesis that consolidated the effort to a manageable level. The antimicrobial activity of the resulting library of analogues provides insight in the importance and potential role of each residue of this complex glycopeptide antibiotic.

Introduction

Due to the overuse of broad spectrum antibiotics in medicine and agriculture, the increasing frequency of bacterial resistance to existing drugs presents a serious threat to public health.¹ Such growing concerns have increased with the recent reports of the emergence of multi-drug resistant organisms including methicillin-resistant Staphylococcus aureus (MRSA) or methicillin-resistant enterococci that exhibit a reduced vancomycin susceptibility.² As a result, it has become increasingly important to identify new antibiotics that can replace vancomycin as the antibiotic of last resort for such resistant bacterial infections.

Ramoplanin is a lipoglycodepsipeptide first isolated from the fermentation broths of Actinoplanes sp. ATCC 33076 in 1984 (Figure 1).³ The ramoplanin complex, which consists of three closely related compounds 1–3,⁴ was found to be 2–10 times more active than vancomycin against Gram-positive bacteria (500 strains), and maintains full activity against vancomycin-resistant enterococci (VRE, MIC = 0.5 μg/mL) and all known strains of MRSA.⁵ Although the mechanism of action of ramoplanin is not yet fully defined, it is a promising drug candidate for vancomycin-resistant bacterial infections since it is known to function in a unique manner.⁶ Both antibiotics inhibit bacterial cell wall biosynthesis, but vancomycin prevents cell wall cross-linking by primarily inhibiting the late stage transpeptidase-catalyzed step by binding to its substrate D-Ala–D-Ala at the terminus of peptidoglycan precursors,⁷ whereas ramoplanin inhibits the earlier stage intracellular glycosyltransferase (MurG) and the more accessible extracellular transglycosylase (PBP1b)-catalyzed steps by binding their substrates Lipid I and II, thereby precluding maturation of the bacterial cell wall.^6,8 Consequently, no cross-resistance of ramoplanin with existing antibiotics, including vancomycin, has yet been observed. Currently, ramoplanin is in phase III clinical trials for topical, nasal, and GI infections.^5,9 However, the therapeutic applications of ramoplanin are limited because of its poor pharmacokinetics: it is not orally absorbed and it is not stable in plasma due to rapid hydrolysis of the labile lactone.

Figure 1.

Structure of the ramoplanins.

Ramoplanin A1–A3 consists of a 49-membered ring composed of 17 amino acids including 12 unnatural amino acids and 7 possessing the D-configuration. The initial structure of ramoplanin was disclosed in 1989 and the three compounds that make up the ramoplanin complex differ only in the lipid side chains attached to the Asn¹ N-terminus.⁴ Originally, the two double bonds in the three different acyl groups were assigned the cis-cis stereochemistry^4a and this has since been corrected to be cis-trans.¹⁰ The C-terminal 3-chloro-4-hydroxyphenylglycine (Chp¹⁷) forms a lactone bond with the hydroxy group of β-hydroxyasparagine (HAsn²). In 1991, the structure of ramoplanose (4) was disclosed by Williams and coworkers and its composition was identical to ramoplanin A2 except for the branched mannose trisaccharide (vs mannose disaccharide) at Hpg¹¹ and the stereochemistry of the acyl side chain (cis, trans- vs cis, cis-7-methyloctadi-2,4-enoic acid).¹¹ Kruz and Guba subsequently corrected the olefin stereochemistry of ramoplanin A2 as cis-trans by 2D NMR in 1996.¹⁰ They also established the Hpg⁶ and Hpg⁷ absolute stereochemistry and found that the solution conformation consists of two antiparallel β-strands (HAsn²–D-Hpg⁷ and D-Orn¹⁰–Gly¹⁴) stabilized by six transannular H-bonds and a cluster of hydrophobic aromatic side chains (D-Hpg³, Phe⁹, and Chp¹⁷) providing a U-shape topology to the β-sheet with a reverse β-turn (aThr⁸–Phe⁹) at one end and a more flexible connecting loop (Leu¹⁵–Chp¹⁷) at the other end in the solution structure.

Enduracidin A and B are additional members of ramoplanin family that have been employed as feed additives and are known to inhibit Gram-positive bacterial cell wall biosynthesis¹² by the same mechanism as the ramoplanins^12f (Figure 2). The enduracidins and ramoplanins share a high degree of structural similarity including two-antiparallel β-strands and a conserved D-Hpg³–aThr⁸ region thought to be important for Lipid I and II recognition and binding. Moreover, many of the remaining residues in the enduracidins and ramoplanins are identical (Hpg¹¹, Gly¹⁴, D-Ala¹⁶) or represent conservative structural departures (D-Ser¹² vs D-aThr¹², Dpg¹³ vs Hpg¹³, Hpg¹⁷ vs Chp¹⁷, and Thr² vs HAsn²). Even some of the significant departures (Cit⁹ vs Phe⁹ and D-End¹⁰ vs D-Orn¹⁰) represent changes that maintain the stereochemical and potential functional features (D-End¹⁰ vs D-Orn¹⁰) of the ramoplanins. Recently, Marazzi and coworkers disclosed a solution phase conformation of the enduracidins determined by NMR exhibiting only subtle structural differences between the enduracidins and ramoplanins.¹³ The enduracidins do not contain a di- or trisaccharide at Hpg¹¹ and the lipid side chains are longer than those found in the ramoplanins. Enduracidin includes an additional basic residue at End¹⁵ (vs Leu¹⁵) and an acidic residue at Asp¹ (vs Asn¹) that are proximal and engaged in a transannular salt bridge, as well as a flexible side chain at the Cit⁹, which is exposed to the solvent (H₂O–DMSO-d₆, 4:1), whereas ramoplanin incorporates a hydrophobic side chain at Phe⁹ forming a well-packed hydrophobic core along with other aromatic side chains (D-Hpg³, Chp¹⁷) and the lipid side chain.¹⁰ The net result being that the characteristic ramoplanin hydrophobic core is disrupted within the enduracidins. The significance of this difference is yet to be defined and it is not reflected in different transglycosylase inhibition kinetics.^12f Although less well characterized, janiemycin represents an additional member of ramoplanin family.¹⁴

Figure 2.

Structures of the enduracidins.

In 2002, we reported the first total synthesis of the ramoplanin A2 and ramoplanose aglycon (5) confirming the assigned structure.¹⁵ Three key subunits composed of D-Hpg³–Phe⁹ (subunit A), Leu¹⁵–Asn¹ (subunit B), and D-Orn¹⁰–Gly¹⁴ (subunit C) were synthesized, sequentially coupled, and cyclized in a solution phase approach to the 49-membered macrocyclic core of ramoplanin. Two macrocyclization sites, Phe⁹–D-Orn¹⁰ and Gly¹⁴–Leu¹⁵, were examined that maximize the convergency of the synthesis, minimize the use of protecting groups, prevent late stage opportunities for racemization of carboxylate-activated phenylglycine-derived residues, and benefit from a β-sheet preorganization of an acyclic substrate for ring closure.¹⁶ Macrocyclization at the Phe⁹–D-Orn¹⁰ site additionally benefits from closure at the corner of a β-turn incorporating a D-amine¹⁷ while the alternative closure at the Gly¹⁴–Leu¹⁵ site represents a non-hindered glycine site incapable of racemization. Additional keys to the success of the approach were the choice of a SES protecting group for Orn⁴/Orn¹⁰ and Fmoc protection for Asn¹ furnishing orthogonal protecting groups to Boc, Cbz, and benzyl ester deprotections yet capable of sequential and selective removal in the presence of the unstable depsipeptide ester bond. Deliberate final stage incorporation of the Asn¹ lipid side chain provided late stage access to all three ramoplanins as well as side chain analogues of the aglycons. As such, the total synthesis of the two minor components of the ramoplanin complex (A1 and A3) was achieved confirming a reassigned cis-trans stereochemistry for the lipid side chains.¹⁸

This approach has since been extended to the synthesis of two key analogues of ramoplanin containing an amide linkage in place of the labile ester between HAsn² and Chp¹⁷ in which HAsn² was replaced with L-2,3-diaminopropionic acid (Dap) or L-2,4-diaminobutyric acid (Dab).¹⁹ The two derivatives are both much more stable and synthetically more accessible than the natural ramoplanin aglycon. In a subsequent mechanistic analysis of these analogues, both amide linkage substitutions as well as removal of the lipid chain did not affect Lipid II binding indicating that the residue 2 modifications and the acyl side chain do not play an important role in substrate recognition and binding.²⁰ However, the antimicrobial activities of the Dab² analogue and compounds containing truncated side chains were not comparable to those of ramoplanin or the Dap² analogue. The ring expanded 50-membered Dab² analogue, but not the 49-membered Dap² analogue, was found to aggregate extensively in aqueous buffer where its increased conformational flexibility perhaps permits the β-strands in the molecule to associate in an intermolecular manner (aggregation),²⁰ whereas the lipid side chain presumably helps ramoplanin localize to the bacterial cell membrane.²⁰

In 1998, Brötz et al. reported the first direct evidence that ramoplanin binds to a substrate involved in peptidoglycan biosynthesis.²¹ In 2000, Walker and coworkers showed that ramoplanin inhibits the transglycosylation step of peptidoglycan biosynthesis by binding to Lipid II.^6d Shortly following that, insights into the interaction between ramoplanin A2 and a peptidoglycan precursor (Park’s nucleotide) were disclosed by McCafferty using NMR studies suggesting that the octapeptide sequence (D-Hpg³–D-Orn¹⁰), ^8a which is highly conserved among ramoplanins and enduracidins, constitutes the substrate recognition domain. Contemporary with these studies, inhibition kinetic studies^6g and NMR titration experiments^6f performed by Walker and coworkers suggest that the inhibitory species binds with a stoichiometry of 2:1 ramoplanin:Lipid II and in a second NMR study, Lo et al. observed that ramoplanin A2 exists as a mixture of monomer and dimer in methanol.^6e This latter series of studies led Walker to propose that Lipid II may bind in a cleft formed by the dimerization of two ramoplanin molecules and defined the dimer interface region as D-Orn¹⁰–Hpg¹³.

Herein, we report an alanine scan of ramoplanin A2 aglycon and the resulting antimicrobial properties of the derivatives in efforts to define the importance of the individual residues and to identify key regions of the molecule.²² As a consequence of its “relative” ease of synthesis and its resulting stability (lactam vs lactone), the alanine scan was conducted on the [Dap²]ramoplanin A2 aglycon (6) which possesses antimicrobial activity equal to or slightly more potent than the authentic ramoplanin A2 aglycon itself (Figure 3).¹⁹ Notably, the efforts constitute a scan of residues 3–13, 15, and 17 conducted implementing a convergent solution-phase total synthesis of the 14 key analogues.

Figure 3.

Structure of [Dap²]ramoplanin A2 aglycon (6) and the three key subunits A–C.

Chemistry

Following the completion of the total synthesis of the natural product aglycons^15,18 and the identification of an active and stable amide (vs depsipeptide ester) template, [Dap²]ramoplanin A2 aglycon,¹⁹ we initiated and herein report an alanine scan of the ramoplanin A2 aglycon enlisting [Dap²]ramoplanin A2 as the template. These efforts include a full alanine scan of the subunit A (D-Hpg³–Phe⁹) and C (D-Orn¹⁰–Gly¹⁴) which contain the putative substrate recognition domain (D-Hpg³–D-Orn¹⁰) proposed by McCafferty,^8a the dimer interface region (D-Orn¹⁰–Hpg¹³) defined by Walker,^6e and D-Orn⁴ and D-Orn¹⁰ which have been established to be important to the biological properties of ramoplanin.^6f,8a,15b,19 Selective semisynthetic modifications of D-Orn⁴ or D-Orn¹⁰ are notorious for their low yields or poor selectivity. Consequently, the identification of the modified site is difficult and the results derived from their examination have not been entirely conclusive.^4c–e Therefore, the site specific incorporation of D-Ala⁴ and D-Ala¹⁰ by the synthetic methods detailed herein unambiguously address these issues. In addition and to more clearly elucidate the role of the chlorine substitution at Chp¹⁷, a Hpg¹⁷ as well as Ala¹⁷ analogue was prepared.

The synthesis and assembly of three key subunits were facilitated by the convergent nature of the solution-phase approach (Figure 3). For example, each change within the A subunit utilized the common [Dap²]B and C subunits for completion of the synthesis and this expedited the parallel synthesis of the library. Similarly, each of the subunits A, B, and C was also assembled in a convergent fashion (e.g., Ala³ through Ala⁵ were prepared by coupling the modified tripeptide to a common Hpg⁶–Phe⁹ tetrapeptide), and this consolidated the work to a manageable level reducing the complexity of the syntheses and number of new intermediates. Thus, each residue analogue required at most five coupling reactions to assemble the linear peptide, a macrocyclization, the side chain introduction (2 steps), and a final global deprotection (9 operations total) enlisting unmodified subunits used in the synthesis of [Dap²]ramoplanin A2 aglycon.

Synthesis of the A subunit Ala derivatives: heptapeptide D-Hpg³–Phe⁹

The seven synthetic targets of the alanine-substituted subunit A (8–14) and details of the preparation of the [Ala⁵] analogue (10) are summarized in Scheme 1.²³ Following the approach defined for synthesis of authentic subunit A (7), heptapeptides 8–14, which contains all but Orn¹⁰ of the putative Hpg³–Orn¹⁰ recognition domain, were assembled from the D-Hpg³–D-aThr⁵ tripeptide and Hpg⁶–Phe⁹ tetrapeptide.¹⁵ In turn, the tripeptide for 10 was obtained by the coupling of Boc-D-Hpg³-D-Orn⁴(SES)-OH and H-D-Ala-OBn (72%). Benzyl ester hydrogenolysis followed by coupling with H-Hpg⁶–D-Hpg⁷–aThr⁸–Phe⁹-OBn gave the heptapeptide 10 (48%). In similar fashion, the remainder of the alanine-substituted A subunits were prepared (Supporting Information).

Scheme 1.

²³ Alanine-substituted subunit A analogues and synthesis of [Ala⁵] subunit A (10).

Synthesis of the C subunit Ala derivatives: pentapeptide D-Orn¹⁰–Gly¹⁴

The preparation of the pentapeptide 15,¹⁵ which contains the entire dimer interface domain,^6e and the synthetic alanine-substituted C subunits (16–19) are summarized in Scheme 2 along with details of the key [Ala¹⁰] derivative. Coupling of HCl·H-L-Hpg¹¹-D-aThr¹²-OBn with Boc-D-Ala-OH provided the tripeptide in 99% yield. Benzyl ester deprotection and coupling with H-L-Hpg¹³-Gly¹⁴-OBn gave pentapeptide 16 (90% yield). The synthesis of remaining alanine-substituted C subunits proceeded similarly (Supporting Information). Since residue 14 is glycine, its replacement with an alanine was not conducted.

Scheme 2.

²³ Alanine-substituted subunit C analogues and synthesis of [Ala¹⁰] subunit C (16).

Synthesis of the B subunit derivatives: pentapeptide [L-Dap²]B

Within the B subunit, residue 2 was fixed as the simplified and stable Dap² where the side chain linkage is attached to the coupled and similarly unmodified Asn¹.¹⁹ Additionally, residue 16 already constitutes a D-Ala and it was not further modified. Rather, our interest focused on Leu¹⁵ as well as Chp¹⁷ which was replaced with Hpg¹⁷ to directly assess the role of the unusual chloride substituent as well as Ala¹⁷ (Scheme 3). The modified pentapeptide subunit 23, incorporating the Hpg¹⁷ (or Ala¹⁷ for 22) residue and the Dap² amide, was obtained from an intermediate tripeptide which in turn was prepared by coupling Boc-Leu¹⁵–D-Ala¹⁶-OH with HCl·H-Hpg¹⁷-OBn (EDCI, HOAt, NaHCO₃, 20% DMF–CH₂Cl₂, 0 °C, 20 h) followed by deprotection of the benzyl ester (H₂, 10% Pd/C, MeOH, 25 °C, 20 min, 100%). Fmoc-Dap²(NH₂·HCl)-OBn obtained from Fmoc-Dap²(Boc)-OH (HCl–EtOAc) was coupled with this tripeptide using DEPBT (3-(diethoxyphosphoryloxy)-1,2,3-benzotriazin-4(3H)-one)²³ (NaHCO₃, DMF, 0 °C, 1 h, then 25 °C, 52 h, 100%) to give a single diastereomer of the corresponding tetrapeptide with no detectable racemization of the sensitive Hpg¹⁷ residue. Fmoc removal (Bu₄NF, i-PrOH, DMF, 25 °C, 1 h, sonication), coupling the free amine with Fmoc-Asn(Trt)-OH (EDCI, HOAt, DMF, 0 °C, 48 h, 61% for 2 steps), and benzyl ester hydrogenolysis (H₂, 10% Pd/C, EtOH, 25 °C, 45 min, 100%) provided 23a.¹⁹ The original pentapeptide subunit B and the simplified [Dap²]B subunit 20 were found to be sensitive to the benzyl ester deprotection step potentially generating mixtures with a contaminate dechlorinated side product if not closely monitored. In an effort to take advantage of this observation and to reduce the number of synthetic steps for the preparation of 23a, direct dechlorination of the [Dap²]B subunit 20 or [Dap²]ramoplanin A2 aglycon itself under various conditions were examined. However, the rate of dechlorination was too slow and the approach proved to be impractical.

Scheme 3.

²³ Modified subunit B and its analogues.

Synthesis of [Dap², Ala^m]ramoplanin A2 Aglycon

The convergent strategy developed for the total synthesis of the ramoplanin and [Dap²]ramoplanin A2 aglycons was utilized for the preparation of the alanine-substituted [Dap²]ramoplanin A2 aglycon analogues.¹⁹ For example, assemblage of the alaninesubstituted subunit A analogues (8–14) required coupling with the common unmodified B and C subunits and maximized the convergent nature of the synthesis. Representative of the final stages of the analogue preparations, the synthesis of [Dap², D-Ala³]ramoplanin A2 aglycon begins with the treatment of 8 with 4 N HCl–dioxane followed by coupling of resulting amine with the B subunit carboxylic acid 20a to provide 8b without competitive β-elimination (Scheme 4). An excess of 20a was employed to consume all 8a which proved difficult to remove from the product enlisting either an acid wash or chromatography. Boc removal of 8b under mild conditions (B-bromocatecholborane ≡ BCB, CH₃CN, 0 °C, 3 h) which do not affect the Asn¹ trityl group and successive washing of reaction mixture with aqueous NaHCO₃, aqueous 1 N HCl, and water gave the corresponding amine.¹⁹ The resulting amine was then coupled with the unmodified subunit C (15a) to yield 8c. Successive Boc removal, benzyl ester hydrogenolysis and macrocyclization provided the cyclic peptide core 8d in superb yield attributable the β-sheet preorganization of the cyclization substrate as well as closure at a D-amine terminus.¹⁷ Fmoc removal under specially developed conditions (8 equiv of Bu₄NF, 10 equiv of i-PrOH, DMF, 25 °C, 1 h), acyl side chain introduction and final global deprotection (HF, anisole, 0 °C, 90 min) yielded [Dap², D-Ala³]ramoplanin A2 aglycon (8g).¹⁵ In similar manner, each of the ramoplanin A2 aglycon analogues was synthesized and details of their synthesis and their intermediates are provided in the Supporting Information. The final products were purified by reverse-phase HPLC to provide homogeneous materials that were used for characterization and the biological evaluation of the final products. However, it is notable that the final products were obtained in purities ranging from 46–91% (avg. = 73%, see Supporting Information) even before this final purification. The ¹H NMR spectroscopic properties of each of the analogues did not reveal a loss of conformational rigidity that might contribute to a change or loss of antimicrobial activity and only Ala⁸, Ala¹⁰, Ala¹¹, and most notably Ala¹⁷ exhibited slightly less crisp spectra that might be attributable to a conformational heterogeneity (line broadening).

Scheme 4.

²³ Synthesis of [Dap², Ala³]ramoplanin A2 aglycon (8g).

A 48-membered ramoplanin A2 aglycon analogue

One of the more successful structural modifications of the ramoplanin aglycon has been the replacement of L-HAsn² with L-Dap² simplifying the synthesis and removing the hydrolysis labile depsipeptide ester. Surprisingly, the analogous replacement with L-Dab², which results in a ring expansion to a 50-membered ring, resulted in a >100-fold loss in antimicrobial activity and provided an analogue with a propensity for aggregation in aqueous buffer. As a consequence, we now examined the ring contracted 48-membered analogue 25g incorporating a D-Orn² in place of L-HAsn² where the side chain also incorporates an amine for lipid side chain attachment removing the need for Asn¹ (Scheme 5). The simplified B subunit 25a was prepared and sequentially coupled with the authentic A (7a) and C (15a) subunits (Scheme 5). To our disappointment, all efforts to promote the macrocyclization of the linear peptide 25c following C- and N-terminus deprotection failed to provide the macrocyclic peptide 25d. Presumably, the decreased flexibility of 25c resulting from deleting the one methylene is sufficient to preclude macrocyclization. These observations along with the behavior of the 50-membered [Dap²]ramoplanin A2 aglycon underscores the special role residue 2 plays in conveying antimicrobial activity to the compounds and highlights how remarkable the stable L-Dap² for L-HAsn² substitution is.

Scheme 5.

Attempted synthesis of the 48-membered [D-Orn²]ramoplanin A2 aglycon (25g).

Antimicrobial Activity

The results of the antimicrobial assay of the analogues against S. aureus alongside the ramoplanin complex (MIC = 0.19 μg/mL) and [Dap²]ramoplanin A2 aglycon (6, MIC = 0.07 μg/mL) are summarized in Figure 4. Analogous to observations first reported for [Dap²]ramoplanin A2 aglycon (6) upon its initial disclosure,¹⁹ it proved to be 2–3 fold more potent than the natural ramoplanin complex and slightly more active than the ramoplanin A2 aglycon.^19,20 Perhaps the most important of the comparisons to highlight first are the activities of the Ala⁴ and Ala¹⁰ analogues probing the importance of Orn⁴ and Orn¹⁰. Semisynthetic modifications of these sites have indicated that both contribute significantly to ramoplanin’s activity, but their relative importance and potential role remained unclear.^8a,6f,15b,19 Consistent with Walker’s observations with such semisynthetic derivatives,^6f the Ala¹⁰ analogue 16g (MIC = 38 μg/mL) proved to be 540-fold less active than 6, whereas the Ala⁴ analogue 9g (MIC = 3.1 μg/mL) was 45-fold less potent. This clearly indicates that Orn¹⁰ is essential to the activity of ramoplanin and plays a much more fundamental role than Orn⁴, although both Ala¹⁰ and Ala⁴ are among the least effective of the Ala analogues examined.

Figure 4.

Antimicrobial minimum inhibitory concentrations (MIC).

Within the remainder of the putative Lipid II binding domain proposed by McCafferty (Hpg⁴–Orn¹⁰),^8a the additional largest impacts on activity were observed with D-Ala³ (vs D-Hpg³, MIC = 5.2 μg/mL), D-Ala⁷ (vs D-Hpg⁷, MIC = 3.7 μg/mL), and Ala⁸ (vs aThr⁸, MIC = 2.5 μg/mL) representing 75-fold, 50-fold, and 35-fold losses in antimicrobial activity, respectively. Thus, both D-Hpg³ and D-Hpg⁷ proved more significant than D-Orn⁴, and aThr⁸ approaches this level of importance. Notably, the importance of aThr⁸ is unique among the three aThr sites in ramoplanin, and each of the remaining D-Ala/D-aThr⁸ substitutions (residues 5 and 12) resulted in ≤10-fold reductions in activity. Surprisingly, Ala⁶ (vs Hpg⁶, MIC = 0.9 μg/mL) as well as D-Ala⁵ (vs D-aThr⁵, MIC = 0.16 μg/mL) central to this region had the least impact on ramoplanins properties resulting in only 10-fold and 2-fold reductions in antimicrobial activity. In fact, Hpg⁶ proved to be the least important of the phenylglycines imbedded in the ramoplanin structure and each of the Ala/Hpg substitutions elsewhere resulted in >20-fold reductions in activity, and the Ala⁵ analogue [Dap², Ala⁵]ramoplanin A2 aglycon still proved to be slightly more active than the natural ramoplanin complex despite its 2-fold reduction in activity versus 6.

The Ala⁹ analogue (MIC = 0.6 μg/mL) with modification of Phe⁹ resulted in a modest 8–9 fold reduction in antimicrobial activity despite its role in capping the hydrophobic core and important location at the corner of a β-turn in ramoplanins structure adjacent to D-Orn¹⁰.¹⁰ Given that the Orn¹⁰ importance almost certainly arises as a consequence of a key stabilizing binding interaction with the diphosphate central to the structure of Lipid I and II,^6,8 it is surprising that the adjacent residue Phe⁹, unlike Hpg¹¹, would have such a modest effect. Nonetheless, it is consistent with the departure in structure observed with the enduracidins which possess a solvent accessible Cit⁹ residue and suggests this may represent a useful modification site for further functionalization or conjugation efforts.

The Ala analogues spanning the dimerization domain defined by Walker and observed in MeOH (Hpg¹¹–Gly¹⁴),^6e which resides adjacent to the key Orn¹⁰ and transannular to the putative Lipid II recognition domain defined by McCafferty,^8a proved especially interesting. Each of the Ala substitutions resulted in a ≥10-fold reduction in antimicrobial activity with that of Ala¹¹ (vs Hpg¹¹, MIC = 2.5 μg/mL) being the greatest (35-fold), followed by Ala¹³ (vs Hpg¹³, MIC = 1.4 μg/mL, 20-fold) and D-Ala¹² (vs D-aThr¹², MIC = 0.7 μg/mL, 10-fold). In addition to illustrating a significant role that residues in this potential dimerization domain may play in ramoplanin’s interaction with its biological target(s), it also defines D-aThr¹² as the residue at which to probe such roles. In this regard, it is interesting to note that the enduracidins incorporate a D-Ser¹² in place of D-aThr¹² and that Walker’s ramoplanin dimer structure established by NMR places its residue 12 hydroxyls within 3.87 Å of one another on the same face of the dimerization interface. Clearly, this represents a unique residue whose functionalization, deliberate dimerization, or conjugation may be exploited to probe mechanistic questions without perturbing the intrinsic structure of the ramoplanin monomer or dimer.

Finally, the examination of several additional analogues within the less well defined Leu¹⁵–Asn¹ segment, which adopts a more flexible loop at one end of the ramoplanin structure and contains depsipeptide ester as well as the lipid side chain, has shed additional insights into its importance. Previously, we have shown that the lipid side chain is essential for antimicrobial activity (200–800 fold reductions),^19,20 but does not impact Lipid II binding presumably serving to deliver or anchor ramoplanin to the bacterial cell wall.²⁰ Additionally, L-Dap² (49-membered ring), but not L-Dab² (50- membered ring), effectively replaces HAsn² (49-membered ring) providing a stable amide replacement for the labile depsipeptide ester.^19,20 Herein, the extentions of this work to an attempted replacement with D-Orn² resulting in a ring contracted 48-membered macrocyclic amide revealed that the macrocyclization ring closure failed to provide the core structure. The ease of 49-membered ring macrocyclization, the failure of the corresponding (more rigid) 48-membered ring closure, and the aggregation of the 50-membered ring system,²⁰ suggests a significant structural role for this corner of ramoplanin structure and highlights how special the properties of [Dap²]ramoplanin A2 aglycon are. Additionally, we probed the importance of the adjacent Chp¹⁷ residue in this region. Its replacement with Hpg¹⁷ representing the removal of the aromatic chlorine substituent had virtually no impact on the antimicrobial properties (MIC = 0.09 μg/mL).²⁴ Consistent with this lack of functional role for the chlorine substituent, the enduracidins incorporate Hpg¹⁷ in their structure at this site. More significantly, the Ala¹⁷ replacement for Chp¹⁷ with 22g (MIC = 0.3 μg/mL) had a similar lack of effect in the antimicrobial activity resulting in only a modest 4-fold reduction in antimicrobial potency. Despite the apparent significance of the structural change, this Ala modification at residue 17 was among those that had the least effect of any of the Ala substitutions although it was among the residues including Phe⁹ that perturb the rigid solution conformation characteristic of ramoplanin. We have interpreted this behavior to suggest that it is only the stereochemistry at this center that might be important for its antimicrobial activity, and that residues 16–1, constituting a more flexible loop including a D-Ala¹⁶ at one end of the antiparallel β-sheet, do not directly interact with the biological target Lipid I or II and do not play a direct functional role. However, Phe⁹ and Chp¹⁷ do stabilize the intrinsic solution structure of ramoplanin by forming a hydrophobic core buried within the U-shaped conformation of the natural product and may indirectly contribute to its properties. Significantly, the results with Ala¹⁷ indicate that this residue, and perhaps the adjoining residue Ala¹⁶, represent prime sites for modification or conjugation in efforts to probe ramoplanin’s mechanism of action. Finally, it is similarly interesting and surprising that the conservative Ala¹⁵ replacement for Leu¹⁵ at the end of this region of the molecule resulted in a 20-fold loss in antimicrobial activity.

Discussion and Conclusions

A full alanine scan of [Dap²]ramoplanin aglycon (6), a hydrolytically stable and slightly more potent analogue of the ramoplanin aglycon (5), was conducted providing insights into the importance and potential role of each residue. By far the most important residue in ramoplanin is D-Orn¹⁰. Its replacement with Ala¹⁰ resulted in a >500-fold reduction in antimicrobial activity consistent with its proposed integral role in Lipid II diphosphate binding. In contrast and more surprising to us, the conserved Orn⁴ was found to be substantially less important than Orn¹⁰ suggesting its role in binding Lipid I or II is not as critical. Both these residues lie in the putative recognition and binding domain proposed by McCafferty.^8a Two of the remaining residues in this region (residues 3–10) exhibit a larger impact than Orn⁴ (D-Hpg³ and D-Hpg⁷), a third is comparable (aThr⁸), and three appear much less important. Significantly, two of these (D-aThr⁵, Hpg⁶) lie central to this putative Lipid II recognition domain and it is difficult to rationalize such a marginal impact central to a contiguous binding interface. An alternative explanation that might account for this behavior is that Orn⁴ may not be involved in Lipid I or II substrate binding, but rather that it may interact with the membrane phosphates and collaborate with the lipid side chain to deliver and anchor ramoplanin to the bacterial cell wall. In the monomer solution structure of ramoplanin, Leu¹⁵, Ala¹⁶, the lipid side chain and perhaps Hpg¹³/Gly¹⁴ form a hydrophobic face on one side of the molecule adjacent to Orn⁴ that could serve as the membrane binding domain, Figure 5. As such, the enduracidin’s incorporation of End¹⁵ for ramoplanin’s Leu¹⁵ would represent incorporation of an additional proximal membrane phosphate binding residue spatially bracketing this hydrophobic face of ramoplanin. The cluster of the remaining most prominent residues including Orn¹⁰ (red, >100-fold reduction) and D-Hpg³, D-Hpg⁷, aThr⁸, and Hpg¹¹ (orange, 100–25 fold reduction) traverses the opposite end of ramoplanin at one corner of its reverse β-turn at aThr⁸–Phe⁹ potentially representing an alternative recognition domain for Lipid II. In the Walker’s dimer structure, a potential and similar discontiguous recognition motif is defined by the red and orange residues and Orn⁴ is found proximal to the same hydrophobic face and lipid side chain. Although now speculative, such roles will become clearer upon examination of the derivatives herein for Lipid I or II binding in a transglycosylation inhibition assay that can dissect such roles.²⁰

Figure 5.

(A) Monomer and (B) dimer solution structures of ramoplanin (red: >100, orange: 100–25, white: <25-fold reductions in antimicrobial activity by alanine scan).

Within the residue 11–14 domain, all residues exhibit a significant effect even though they lie outside McCafferty’s putative Lipid II recognition domain. The magnitude of their impact (10–40 fold) suggest a prominent role in establishing ramoplanin’s activity. Although there may be many explanations for this behavior, this would be consistent with their stabilization of the ramoplanin dimerization interface observed by Walker.^6e It is noteworthy that the most significant of these residues is Hpg¹¹ which also lies adjacent to the critical Orn¹⁰ and is the site of glycosylation. Given the impact of Hpg¹¹, it is possible that it plays a larger role in binding Lipid II than present models suggest.

One of the most interesting regions of the molecule spans residues 15–2. Attached to residue 2 is Asn¹ which is external to the cyclic ring system and which is acylated with the unsaturated lipid side chain. We have suggested that this serves to anchor the antibiotic in the bacterial cell wall ensuring its localization at its site of action.²⁰ Consistent with this, removal of the side chain or its replacement with a minimal acetyl group results in 200–800 fold reductions in antimicrobial activity, but has no effect on Lipid II binding or transglycosylase inhibition.²⁰ This suggests that Leu¹⁵–HAsn² resides adjacent to a membrane anchoring center where this region constitutes a “relatively” flexible loop at one end of the otherwise rigid U-shaped antiparallel β-sheet. As mentioned above, it is also possible that the proximal Orn⁴ collaborates with this lipid side chain by capping this membrane binding domain with a positively charged amine that binds membrane phosphates. We have also shown and continue to highlight herein, that HAsn² may be effectively replaced with Dap² providing a hydrolytically stable 49-membered ring system in which an amide replaces the labile depsipeptide ester. The HAsn² β-carboxamide clearly does not contribute to ramoplanin’s interaction with its biological target (but may sterically hinder ester hydrolysis), and the Dap² rigid secondary trans amide bond assuredly mimics the analogous trans ester conformation observed in the solution conformations of ramoplanin and enduracidin. Although flexible, efforts to expand^19,20 or contract the macrocycle by a single carbon atom at this site have failed or resulted in nonfunctional compounds suggesting this region plays important conformationally-related roles helping to confine ramoplanin to productive conformations. With the possible exception of Leu¹⁵, the side chains of the remainder of this “flexible” loop, Leu¹⁵–D-Ala¹⁶–Chp¹⁷, do not appear to directly contribute to ramoplanin’s antimicrobial activity. The surprisingly small impact of replacing Chp¹⁷ with Ala¹⁷ (4-fold), the disparate Leu¹⁵ versus solubilizing End¹⁵ residues found in the ramoplanins versus enduracidins, and the conserved minimal side chain at D-Ala¹⁶ not only suggest this “connecting” region of the molecule may represent a membrane interacting region of the molecule, but that it may be a good site for modification or conjugation studies.

Finally, the impact of nearly each residue is so significant that it is difficult to imagine deriving a simplified ramoplanin by excision of a substantial portion of its structure (i.e., cyclization or alternative presentations of only the putative recognition domain),^8b the studies serve to define residues and regions amenable to further funtionalizations for detailed mechanistic studies in progress, and further insights into the individual roles of each residue may be forthcoming from their examination in assays establishing Lipid II binding or transglycosylation inhibition.²⁰

Supplementary Material

1si20061129_03. Supporting Information Available.

Full experimental details and compound characterizations are provided. This material is available free of charge via the Internet at http://pubs.acs.org.