Towards New Antibiotics Targeting Bacterial Transglycosylase: Synthesis of a Lipid II Analog as Stable Transition-State Mimic Inhibitor

Abstract

Described here is the asymmetric synthesis of iminosugar 2b, a Lipid II analog, designed to mimic the transition state of transglycosylation catalyzed by the bacterial transglycosylase. The high density of functional groups, together with a rich stereochemistry, represents an extraordinary challenge for chemical synthesis. The key 2,6-anti- stereochemistry of the iminosugar ring was established through an iridium-catalyzed asymmetric allylic amination. The developed synthetic route is suitable for the synthesis of focused libraries to enable the structure-activity relationship study and late-stage modification of iminosugar scaffold with variable lipid, peptide and sugar substituents. Compound 2b showed 70% inhibition of transglycosylase from Acinetobacter baumannii, providing a basis for further improvement.

Graphical abstract

The increasingly common occurrences of infections caused by the drug-resistant bacteria represent a major threat to public health.¹ The urgent demand for novel antibacterials has led to the search for underexploited drug targets. In the past, targeting the assembly of peptidoglycan (PG), a polymer-like structure that helps maintain the integrity of bacteria cell and protects it from lysis, has proven to be a successful strategy for the discovery of antibiotics. Our group has a long-standing interest in the enzyme transglycosylase (TG) as target, which catalyzes the polymerization of Lipid II (1, Figure 1A) to generate a nascent PG before it is cross-linked by transpeptidase (Figure 1B). First reported 50 years ago,^2a TG is still viewed as a difficult,^2b albeit an attractive target.^2c–d Located on the external surface of the cytoplasmic membrane, TG is accessible to potential inhibitors. As TG does not have any mammalian counterpart, it is possible to design new antibiotics that are specific against prokaryotic pathogens. In addition, because TG recognizes an invariant carbohydrate backbone, it may be less susceptible to the traditional mechanisms of resistance development.³ Although, antibacterials that inhibit Lipid II polymerization by sequestering its substrate have been identified (e.g., vancomycin), the direct binders of TG with potent inhibitory activities and pharmacological properties suitable for clinical use have yet to be developed. One major effort in this direction has been the optimization of moenomycin structure,⁴ the only TG-specific inhibitor known to date.

Figure 1.

(A) The structure of Lipid II (1) and the target iminosugar derivatives. (B) Schematic representation for the inhibition of the donor site of TG with iminosugar 2.

Based on recent efforts towards finding the minimal required features of Lipid II/Lipid IV,^5–7 we designed structure 2 (Figure 1A) as a potential transition-state mimic of the TG-catalyzed reaction.⁵ Compound 2 consists of iminosugar ring connected to an additional ring of GlcNAc, a truncated peptide moiety with two essential methyl groups from the lactyl-alanine sequence,^5a,6a,d along with a phosphono-phosphate linked lipid chain,^6b which is necessary for the proper recognition and binding. Towards structure 2, our group has initially reported the synthesis of the truncated analog 2a (Figure 1A), which indeed showed inhibition of TG function.^6b The two drawbacks of compound 2a are of note. First, because TG is a processive enzyme,⁷ it is highly unlikely that this mono-sugar derivative 2a can reach the desired donor site of TG due to the lack of the second GlcNAc, therefore preventing enzyme to process 2a. Hence, the observed activity could be a result of 2a binding to the acceptor site only, and its designation as a transition-state analog inhibitor could not be fully realized. Second, the synthetic strategy developed for the assembly of 2a is not suitable for the preparation of the highly functionalized iminosugar 2 and its derivatives,^6b required for the detailed structure-activity relationship (SAR) study of TG inhibition. To solve the drawbacks of compound 2a, herein we report an optimized asymmetric synthesis of the iminosugar 2b as a core inhibitor of TG. Since 2b is a pseudo-disaccharide derivative of Lipid II, we envisioned that upon binding it could be processed by the enzyme and pulled into the donor site, where it can block any further transglycosylation reactions (Figure 1B). The synthetic route developed for the assembly of 2b can be further applied for the preparation of other analogs, e.g., 2c and 2d (Scheme 1).

Scheme 1.

Retrosynthetic analysis of target molecule 2.

The retrosynthetic analysis of the target molecule is presented in Scheme 1. The introduction of the sugar and lipid substituents can take place at the late stage using conventional glycosylation with GlcNAc donor (3) and a CDI-activated lipid phosphate (4). Functional groups at positions C-3 and C-4 can be installed via double bond functionalization of 6. Although the proposed synthesis leads to a mannosyl-like iminosugar, the hydroxyl group R² is well positioned for the inversion with azide to provide the NHAc substituent R³ if required. The most challenging part is the 2,6-anti configuration of the key iminosugar core, which could be accessed through a ligand-controlled asymmetric allylic amination.⁸ The stereo centers at positions C-5 and C-6 of amine 9 can be obtained from (R)-Garner’s aldehyde. The dense array of functional groups, high polarity and rich stereochemistry of the target molecule signify a substantial synthetic challenge. The most difficult steps of our synthesis, which we successfully solved, include installation of the 2,6-anti configuration of iminosugar, the C-P bond formation and pyrophosphate coupling of the complex pseudo-disaccharide to a long hydrophobic lipid chain.

The starting amine 9 was prepared on a gram-scale through a 5-step sequence from the (R)-Garner’s aldehyde requiring only one purification step (Scheme 2). Carbonate 8a was obtained in a high yield via a 3-step sequence. Compounds 8b and 8c were synthesized in a similar manner according to the known procedures.⁹

Scheme 2.

Synthesis of the amine 9 and carbonate 8a.

Using chiral amines 9 and 10, and different carbonates 8a-c, we tested the conditions for the asymmetric allylic amination reaction. The results are summarized in Table 1. During the initial screen using 10 (or its HCl salt) and carbonate 8a (entries 1-2, Table 1),^8d we observed a trace amount of the desired product 13 together with unreacted starting material. Because the free hydroxyl groups might influence reactivity, we then switched to the protected amine 9, which provided 13 in an excellent yield and selectivity. Since it is very difficult to introduce the phosphonate group into the highly functionalized iminosugar at the late stage,¹⁰ we decided to test esters 8b-c under the optimized reaction conditions. Unfortunately, using 8b we could only observe the product of elimination (14). The reaction with 8c, where the acidic protons are replaced with fluorine atoms,¹¹ gave only a linear diene (15). Nevertheless, the availability of intermediate 13 allowed us to pursue the synthesis of our target molecule.

Table 1.

Screen of substrates for the Ir-catalized allylic amination reaction.^a

entry	amine	8	additive	resultb
1	10·HCl	8a	NaH2PO4	13 (trace)
2	10	8a	–	13 (trace)
3	9	8a	–	13 (>95%)c
4	9	8b	–	14
5	9	8b	NaH2PO4	14
6	9	8c	–	15

^aStandard reaction conditions: 8 (1.1 equiv.), 9 or 10 (1.0 equiv.), [Ir(dbcot)Cl]₂ (2 mol%), 12 (4 mol%), n-BuNH₂ (4 mol%), DMSO, 50 °C, 12 h.

^bDetermined by LCMS.

^cIsolated yield.

Synthesis of amine 13 was scaled up without any impact on the stereoselectivity or yield, and afforded gram-quantities of material. Upon switching to inert atmosphere, the loading of iridium catalyst can be lowered to 1 mol% (Scheme 3). With the key intermediate 13, we aimed to close the ring and to introduce substituents at positions C-3 and C-4. For the optimization of ring-closing metathesis conditions, we screened different salts of amine 13, which however failed to provide the cyclized product. Trifluoroacetamide protection of amine (16),¹² however, allowed a high-yield preparation of ene-17 using the second-generation Grubbs catalyst. Crystallization of the intermediate 17 resulted in compound 18 (Scheme 3),^13a a crystal structure of which confirmed the 2,6-anti selectivity of the asymmetric allylic amination step. Next, we performed an osmium-catalyzed dihydroxylation of ene-17, which gave diol 19a. To verify the stereochemistry of the last step, we carried out a global deprotection to obtain 19b, which matched the NMR data reported for the iminosugar scaffold previously described in the literature (Scheme 3).^13b

Scheme 3.

Synthesis of the iminosugar core and determination of the absolute stereochemistry.

The remaining sequence of steps is presented in Scheme 4. The TFAc protection of amine 19a was replaced by the base-compatible Cbz-protection giving the diol-20 in a high overall yield. Next, protection of the diol with the benzyl group (21), removal of PMB and iodination of 22 afforded intermediate 23. One of the most challenging steps in our synthesis was the installation of C-P bond. After screening a variety of different substrates and reagents, the reaction of iodo-derivative 23 with triethyl phosphite was identified as the only effective method, which provided phosphonate 24 in moderate yield. Selective opening of the benzylidene ring afforded acceptor 25,¹⁴ which was glycosylated with donor 3 to give the desired pseudo-disaccharide 26.¹⁵ Subsequent transformation of NHTroc to NHAc, hydrolysis of phosphonate ester¹⁶ and global removal of protecting groups yielded the key intermediate 28. Coupling of 28 with CDI-activated geranylgeranyl phosphate 4 afforded target molecule 2b.

Scheme 4.

Diversification of iminosugar scaffold and synthesis of target molecule 2b.

The inhibition of TG by 2b was determined using HPLC-based assay.¹⁷ At 50 μM, 2b showed 70% inhibition of TG from Gram-negative A. baumannii; and the observed activity of 2b is comparable with di- and monosaccharide mimics of moenomycin that bind to the donor site of TG.^4a This result validates our strategy for the design of TG inhibitors and suggests that the presence of a peptide moiety may be required to improve the potency of 2b. The related analogs are being investigated in our laboratory, and the results will be reported in a due course.

In conclusion, we have developed an efficient route towards the Lipid II analog 2b from the commercially available (R)-Garner’s aldehyde. The key step, installation of the 2,6-anti-stereochemistry of iminosugar was achieved using the iridium-catalyzed asymmetric allylic amination procedure, which was optimized to the gram-scale process. The developed route could be used to access other Lipid II mimics, particularly 2c and 2d, which are expected to have better binding affinities towards TG, than 2b; these structures will serve as a template for further SAR and structural studies, hence accelerating the development of new antibiotics.