Total Synthesis of Dynobactin A

Abstract

The first total synthesis of the potent antimicrobial agent dynobactin A is disclosed. This synthesis enlists a singular aziridine ring opening strategy to access the two disparate β-aryl branched amino acids present within this complex decapeptide. Featuring a number of unique maneuvers to navigate inherently sensitive and epimerizable functional groups, this convergent approach proceeds in only 16 steps (LLS) from commercial materials and should facilitate the synthesis of numerous analogues for medicinal chemistry studies.

Graphical Abstract

Darobactin A¹ (1, Figure 1) and Dynobactin A^2–3 (2, Figure 1) are recently discovered natural products that show promising potent antibacterial properties against gram-negative pathogens, as well as low cytotoxicity. Adding further intrigue and promise to this class of antibiotics is its apparent ability to evade bacterial resistance through mutation.⁴ Both 1 and 2 target BamA, a bacterial insertase within the β-barrel assembly machinery (BAM) complex^5–6 that facilitates folding and insertion of outer membrane proteins in gram-negative bacteria.^7–13 However, 2, a decapeptide featuring two linked macrocycles, stands structurally distinct from its BamA-targeting counterpart, heptapeptide 1. The unfused bicyclic structure possibly endows dynobactin A with flexibility, and contains additional ionizable sites, resulting in a water solubility that exceeds that of 1 by more than 20-fold.² The synthetic challenge posed by 1 was addressed concurrently last year by two separate groups.^14–15 2 consists of several unique structural elements, including a C–C bond formed between C6 of the tryptophan indole and the β-carbon of the asparagine (C21), as well as a rare N-C linkage bridging the imidazole of histidine and β-carbon (C39) of the tyrosine unit. The intriguing bioactivity and structural complexities render 2 an exciting subject for synthetic pursuit. Yet, the intricate structure of 2 along with its natural scarcity pose a challenge to conventional (semi)synthetic approaches. In this Communication, the first total synthesis of 2 is disclosed, leveraging the recognition of a hidden strategic symmetry wherein parallel aziridine openings forge the challenging β-aryl disubstituted amino acids prior to convergent assembly.

Figure 1.

Darobactin A (1) and Dynobactin A (2) Retrosynthetic strategy

Retrosynthetically, 2 (Figure 1) was envisioned to arise from a C-terminal Arg-Phe dipeptide unit and appendage of the two macrolactam subunits to the central valine residue, resulting in two fragments comprising a 17-membered (3, western fragment) and 12-membered (4, eastern fragment) macrolactam, respectively. Our lab recently published a modular, atroposelective synthetic route to 1.¹⁴ While distinct in overall structure, the shared features of 1 and the western macrocycle of 2 suggested that the Larock macrocyclization strategy employed in that synthesis might also be applicable herein to obtain 3. In order to generate the western fragment via indole synthesis, the framework would be constructed from an asparagine unit, serine unit, alkyne fragment 5, and the requisite aniline in the form of a β-aryl asparagine. 4 was proposed to be constructed via macrocyclization along the His-Ser peptide bond. The overall synthetic strategy was centered around the construction of the β-branched amino acid (AA) units which are characteristic of the biosynthetic origin of 2 as a ribosomally synthesized and post-translationally modified peptide (RiPP) natural product.² Key to the execution of this plan was that branched AAs 6 and 8 could both be derived using the same nucleophilic ring opening strategy upon aziridines 7 and 9, respectively (vide infra).

Extensive route scouting was conducted to determine feasible strategies for assembly of the requisite branched AAs (Figure 2 with further detail in SI). Initially, the synthesis of the β-branched Asn fragment of the western fragment 3 was pursed via an epoxide opening reaction with enolate-type nucleophiles (Figure 2a, strategy A). Although the primary alcohol obtained proved competent for construction of the western fragment, all attempts for its late-stage conversion to the corresponding primary amide were hampered by extensive epimerization of the benzylic α-carbonyl stereocenter. For construction of the β-branched tyrosine moiety of the eastern fragment 4, a broad range of strategies were evaluated (Figure 2, strategies B–D). Linkages of this type are extremely rare and notoriously difficult to obtain and no successful strategy has been uncovered to date.^16–17 Attempted nucleophilic substitution reactions led consistently to either formation of the corresponding unsaturated dehydro-amino acid or decomposition of the substrates (Figure 2b, strategy B). A potentially biomimetic intramolecular cyclization by addition of the histidine imidazole to the corresponding tyrosine-derived quinone methide was considered (Figure 2b, strategy C). However, this approach led to either decomposition or tautomerization to the corresponding dehydro-tyrosine.

Figure 2.

Synthetic strategies towards β-branched amino acids

While addition to aziridines has previously been demonstrated as a useful strategy for the synthesis of β-branched amino acids,¹⁸ very few examples exist for comparable additions to α-carbonyl, β-aryl aziridines as was necessitated in the case of 2. Early stage attempts to generate the eastern fragment via addition of the histidine imidazole to an acyl aziridine moiety were hampered by the electronic nature of the acyclic twisted amide obtained (Figure 2b, strategy D).¹⁹ In general, addition to aziridines requires the use of electron-withdrawing N-activators (such as tosyl or nosyl protecting groups), which could potentially introduce further synthetic issues. Specifically, cleavage of N-tosyl is incompatible with use of an aryl bromide,^20–21 preliminary experiments indicated difficulties in controlling the reactivity for N-nosyl,²² and rearrangement is a known issue for carbamate-based protecting groups.^23–24 For these reasons, a more efficient approach was developed based on direct addition into free-NH aziridines. A breakthrough was reached upon recognition that the β-branched amino acid motifs for both 3 and 4 could, in principle, be derived from such an aziridine opening. The β-branched motif of 4 could thus be traced back to addition of histidine into chiral aziridine 9 to forge the N-C39 bond of 2. To mitigate the risk of imidazole eliminating from tyrosine to form the corresponding para-quinone methide (QM) (Figure 1, highlighted in red), bromide was chosen as a stable surrogate for the hydroxyl group of tyrosine with the intention to introduce the requisite phenol moiety at a late stage. The synthesis of the western fragment 3 branched tryptophan-asparagine motif (Figure 1, highlighted in green) was envisioned in the same fashion by addition of 2-methyl furan into the corresponding aziridine 7. In this case, furan acts as an epimerization-proof synthetic precursor to the primary asparagine amide, to be unveiled towards the end of the synthesis. With a strategy in hand to address the challenge of synthesizing the β-branched AAs 6 and 7, our attention next turned to construction of the two macrocycles and final assembly of 2.

The synthesis of 2 is outlined in Scheme 1a. To construct the eastern fragment 4, enantioenriched 9 (93:7 er) was prepared in high yield on multigram-scale in two steps, leveraging the multicomponent asymmetric aziridination developed by the Wulff group (see SI for experimental detail).²⁵ With free aziridine 9 in hand, a stereo- and regio-selective aziridine opening reaction was conducted using histidine derivative 11 as a nucleophile under acidic conditions. This reaction was subjected to extensive optimization (see inset Table 1 and SI for further detail) and showed a pronounced solvent dependence (Table 1 entries 1–3), wherein the use of trifluoroethanol (TFE) was found to be critical.^26–28 p-Toluenesulfonic acid and triflic acid performed comparably (Table 1 entries 3 and 4). The major side reaction pathway under optimized conditions was identified as dimer formation by competing nucleophilic attack of the aziridine. The ratio between the desired product and dimer is critically influenced by the protonation equilibrium of the aziridine and nucleophile (see SI for a comparison of nucleophiles with varying pK_a). The reaction between His 10 and aziridine 9 (Table 1 entry 4) was hampered by extensive dimer formation due to an insufficient pK_a difference. In comparison, utilization of less basic 5-Br-His 11 led to a subtle shift in protonation equilibrium resulting in reduced dimer formation to give the desired aziridine addition in 41–65% yield depending upon the equivalents of His 11 utilized (Table 1 entries 5 and 6). Subsequent reductive cleavage of the bromide under acidic conditions using Zn/AcOH proceeded with excellent selectivity for the bromo-imidazole and enabled access to the key β-branched amino acid coupled product 8. Subsequently, amide coupling with a protected serine residue furnished cyclization precursor 12 in 65% yield, which was subjected to TBAF, resulting in simultaneous Fmoc and (trimethylsilyl)ethyl (TSE) deprotection. After screening a variety of activation reagents, the combination of PyAOP with HOAt was found to be the most effective conditions under which to forge the macrocycle to yield 4 in 54% yield over two steps.

Scheme 1.

Total synthesis of dynobactin A^a

^aFor details on reagents and conditions, see SI.

The aziridine-based strategy outlined above could be leveraged again to forge the tryptophan-asparagine linkage in the western fragment, as shown in Scheme 1b. Thus, 3 was prepared via ring opening of aziridine 7 using 2-methyl furan as a nucleophile. The use of 2-methylfuran as a masked primary amide provided several advantages, notably avoiding undesired side reactions including epimerization of the β-carbon and imide formation during carboxylic acid activation. The resulting ring-opened product 6 was enlisted directly in a one-pot amide coupling with compound 13. A subsequent Fmoc deprotection and amide coupling with alkyne 5 yielded the cyclization precursor 14 in 56% yield over three steps. Borrowing from the strategy previously developed to access 1,¹⁴ a Larock macrocyclization was employed to form the 17-membered ring, providing 15 in 49% yield using a catalytic quantity of Pd. Due to the increased ring size of 15 relative to 1, no atropisomerism was observed for this macrocycle.²⁹ The protected primary amide functionality was then unveiled via furan oxidation using a Ru catalyst and subsequent aminolysis to afford the completed western fragment 16 in 41% yield. A number of different permutations were tested for the sequence of valine introduction, east-west-fragment coupling, and furan oxidation (an overview is given in inset Table 2 and in the SI). Approaches to undertake the oxidative cleavage of the furan late-stage after coupling together the 8-mer or 10-mer peptide (with/without the Arg-Phe side chain, Table 2 entries 1 and 2) led only to decomposition of the substrate, presumably due to interference with the electron-rich imidazole moiety. Coupling of the valine residue to the western fragment 15 followed by oxidative cleavage of the furan/aminolysis, while successful on small scale Table 2 entry 3), resulted in long reaction times (5 days) and diminished yields (< 20%) upon scaleup. In contrast, oxidative cleavage of the furan directly on intermediate 15 proceeded faster (12–24 h) and in higher yield (Table 2 entry 4) and was thus chosen as the optimal order of operations for this sequence. The ethyl ester moiety of 16 was then hydrolyzed using trimethyltin hydroxide³⁰ to afford 3 and set the stage for the final assembly of the decapeptide (see inset Table 3 and Scheme 1c).

Following Boc deprotection with formic acid, eastern fragment 4 was coupled to the central valine residue to generate 17 in 47% yield. Trityl deprotection of 17 followed by coupling with western fragment 16 gave 18 in 51% yield over two steps. Remarkably, C-terminal fragment coupling of the branched asparagine proceeded with no substantial aspartimide-related side reactions (e.g., deamination, dehydration, isomerization).³¹ This fragment coupling strategy, conjoining both fragments as equally sized tetrapeptides, is the most convergent successful approach to the fragment coupling of the iterations evaluated (Table 3 entry 4). As mentioned above, fragment coupling attempts prior to furan oxidation were non-viable, given the sensitivity of late-stage intermediates to the oxidation conditions (Table 3 entries 1 and 2). Appendage of the valine unit to the western macrocycle followed by coupling of the eastern macrocycle was comparably successful (Table 3 entry 3). Assembly of the decapeptide featuring introduction of the dipeptide Arg-Phe to the eastern fragment before east-west-fragment coupling was not successful (Table 3 entry 5). As such, 18 was subjected to ester hydrolysis and subsequent coupling with 19 to arrive at the decapeptide 20 in 66% yield over two steps. At this point, all that remained was conversion of the bromide to the requisite hydroxyl group, followed by global deprotection to complete the synthesis of 2. The final transformations were accomplished in a one-pot procedure consisting of: (1) Miyaura borylation of the bromide and concomitant deprotection of acid-labile protecting groups,³² (2) oxidation of the resulting boronic acid to the phenol, (3) hydrogenolysis of the benzyl ester, and (4) removal of the final acetyl protecting group on the indole moiety. This sequence provided 2 in 55% yield from 20, with 16 steps LLS from commercially available starting materials. Synthetic 2 was found to be spectroscopically and chromatographically identical to an authentic sample of the natural product (see SI).

To summarize, a convergent total synthesis of 2 has been accomplished by leveraging hidden symmetry of the seemingly distinct eastern and western halves of the molecule. A unique (and perhaps general) strategy for the synthesis of β-aryl branched amino acids that enlists simple N-H aziridine building blocks in either stereo- and regioselective C–C (western fragment) or N–C (eastern fragment) ring opening events enabled this plan. The strategic use of a furan and aryl bromide to shield the most sensitive parts of the final natural product are also noteworthy tactics. The modular approach to 2 should enable an extensive study of the SAR of this fascinating family of antibiotics.

Supplementary Material

Experimental procedures, analytical data (¹H and ¹³C NMR, MS) for all new compounds as well as optimization tables. This material is available free of charge via the Internet at http://pubs.acs.org.