Norbornene-based anion receptors as D -alanine binders

Vancomycin is currently used as last-line therapy against many Gram-positive bacterial pathogens. Herein, we report a series of peptidomimetic norbornene-based anion receptors that were designed as simple vancomycin mimics New hosts were evaluated for their affinity to both acetate and acetyl D-Ala by 1H NMR titration. Modest binding to both anions was observed in DMSO-d 6 (Log Ka 1–2 for TBA Acetyl D-Alanine) in the anticipated 1:1 mode of binding.


Introduction
An increasing incidence of bacterial resistance to clinically used antibiotic agents (such as methicillin-resistant Staphylococcus and vancomycin-resistant enterococci) combined with decreased investment in antibacterial agents by the pharmaceutical industry has led to an antibiotic crisis (1).
Vancomycin 1 is a naturally occurring antibiotic and, in a supramolecular chemistry context, is an elegant example of anion recognition by means of both a preorganised (achieved largely by the linked aromatic amino acid side chains) and complementary (matched H-bond donors/acceptors) host ( Figure 1) (2). The 'guest' is the terminal D-alanine-D-alaninate (D-Ala-D-Ala, 2) portion of Lipid II, a key substrate in the late-stage enzymatic cross-linking to form the bacterial peptidoglycan outer membrane (2,3).
Vancomycin binds D-Ala-D-Ala through five hydrogen bonds (2,3) with three of these combining to target the terminal carboxylate. The strength of binding (Log K a of 5.6 in an aqueous environment) of this interaction has been measured using a segment of the peptide portion of Lipid II, namely diacetyl-L-Lys-D-Ala-D-Ala (4). In the presence of vancomycin, the transpeptidase enzyme responsible for cross-linking the cell wall is unable to access the D-Ala-D-Ala, thus, construction of the cell wall is incomplete and the bacteria are rendered susceptible to lysis and cell death (5).
As vancomycin is an excellent example of anion recognition, supramolecular chemists have sought to mimic the binding, and in turn, the antibacterial properties of vancomycin by designing preorganised hosts specific for D-Ala-D-Ala (6,7). Examples include that of Schumuck who produced functionalised pyrroles that bind D-Ala-D-Ala (2,3,8). Bremner and co-workers synthesised functionalised binols with cationic amino acid sequences and the resultant peptidomimetics were active against vancomycinsusceptible S. aureus and vancomycin-resistant S. aureus. (9) Cohen et al. developed a vancomycin mimetic based on the calix [4]arene scaffold and studied its binding to D-alanine using diffusion NMR (10).
The structure of vancomycin can be viewed as a rigidified aryl ether backbone supporting, and preorganising, the peptide chain responsible for the hydrogen bonding interactions. Our work in the field of alicyclic frameworks (11) led us to believe that a similarly preorganised peptide (such as 2, Figure 1) might be constructed through the use of a rigid [n]polynorbornane backbone. Herein, we report our initial efforts to replicate the carboxylate-binding portion of Vancomycin in which a series of peptidefunctionalised norbornenes were designed and synthesised as carboxylate receptors (Figures 1,(3)(4)(5)(6)(7)(8)(9).
The design of these anion receptors allows for rapid synthesis with potential to incorporate a broad range of structural modifications. It is envisioned that the fragments discussed here will be elaborated in future through extension of both the peptide chain (using standard peptide coupling reagents) and also the norbornane scaffold (using 1,3-dipolar cycloaddition chemistry) to provide a macrocyclic peptidoframework such as [3]polynorbornane 2. (11,12) If the complete peptidoframework was to be successful it was envisioned that sections of this molecule, such as receptors 3-9, could bind to the carboxylate of D-Ala-D-Ala as vancomycin does (2,3,8).
The design of this fragment library ( Figure 1) includes functionalisation for variation of (i) urea/ thiourea moiety (X) to mediate the pK a ; (ii) electronwithdrawing group (Y), for further tuning of the urea/ thiourea pK a ; (iii) chiral aliphatic chain (R), for hydrophobic interaction with the D-Ala side chain and (iv) spacer between the norbornene framework (n), which may prove crucial in future as it can provide steric relief from the [n]polynorbornane scaffold.
Next, receptors 3-9 were evaluated for their ability to bind both acetate and acetyl D-alaninate as model anions representing the carboxyl portion of the bacterial cell wall precursor. Using 1 H NMR titration, all receptors were tested against both anions and the results are presented in Table 1. Anions were used as their tetrabutylammonium salts and the titrations were conducted in DMSO-d 6 using an initial host concentration of 2.5 mM. The strength of binding (Log K a ) was calculated using WinEQNMR2 (13) (see ESI for titration isotherms, modified Job plots and fit plots).
All receptors (3)(4)(5)(6)(7)(8)(9) were found to bind to both acetate and acetyl D-alanine in a 1:1 Host:Guest (H:G) stoichiometry. Although the modified Job plots used in this study indicated that the H:G stoichiometry was 1:2, we (14) and others (15) have shown previously that Job Plots can provide misleading results (in particular the modified methodology) and a sensible 1:2 H:G stoichiometry could not be envisioned. 1 Nevertheless, both 1:1 and 1:2 fitting protocols were used to analyse the data. The results of the 1:2 fitting provided large asymptotic errors (20-500%) and nonsensical K a values; the errors for the 1:1 fitting were less than 15%. Given, the quality of fit of the data to the 1:1 model, all receptors were deemed to bind acetate in a 1:1 stoichiometry with an average Log K a of 2.6 which is typical for (thio)ureas in DMSO (16).
The R=H receptor 9 provided an example where the receptor amide N-H was also able to contribute to binding of the anion (as indicated by the small downfield 1 H NMR chemical shift change, Δδ = 0.16), but for receptors such as 5, with an isopropyl side chain, the bulky group likely restricted the ability of the receptor to adopt the desired tridentate binding conformation. Unfortunately, only modest binding to acetyl-D-alanine was observed across this series (average Log K a = 1.9), with no hydrogen bonding identified from the acetyl amide. Nevertheless, receptor 8, the most abundantly available, was submitted for a disc diffusion assay against S. aureus, and unfortunately no activity was noted.
It was pleasing to see that in this initial investigation into norbornene receptors for D-alanine, the binding stoichiometry was 1:1 as predicted. While the strength of association was modest, the receptors lack the same number of H-bond donors and the same level of preorganisation present in vancomycin. Future efforts will involve using a larger, more preorganised, fused polynorbornane framework (such as 2) that will be functionalised with a more complete array of H-bond donors and acceptors.

Conclusion
A series of low-molecular weight norbornane-based vancomycin mimics were synthesised and evaluated for binding of acetyl-D-alaninate. Consistent 1:1 binding of these substrates with the desired target was noted and the compounds serve as an entry point for a larger study involving highly functionalised fused [n]polynorbornanebased vancomycin mimics.

General experimental
All reagents were obtained commercially and used without purification; all solvents used were analytical reagent grade. Petroleum spirits refers to the fraction boiling between 40 and 60°C. NMR spectra were collected on either a JEOL JNM-Ex 270 MHz, a Varian Unity Plus 300 MHz, an Eclipse JNM-ECP 400 MHz FT-NMR or a Bruker AVANCE III 500 MHz FT-NMR spectrometer as indicated. Samples were dissolved in CDCl 3 , DMSO-d 6 or CD 3 CN (~0.5 mL) and reported relative to TMS = 0.00 ppm. 1 H NMR spectra are reported as chemical shift (ppm) (integral, multiplicity (s = singlet, bs = broad singlet, d = doublet, t = triplet, sept = septet, m = multiplet), J coupling (Hz), assignment). Highresolution mass spectral data were collected on either an Agilent Technologies LC/MSD TOF mass spectrometer or an LC Agilent 1200 MS 6520 QTOF with dual-electrospray ionisation source. Samples were dissolved in acetonitrile or MeOH at a concentration of less than 0.1 mg/mL. Thin-layer chromatography was performed on Merck TLC silica gel 60 F plates, and visualised using UV light (λ = 254 and 365 nm) and/or potassium permanganate (KMnO 4 , H 2 O, K 2 CO 3 ) as an oxidising dip. Column chromatography was performed with silica gel 60 (230-400 mesh). All microwave reactions were performed using a CEM Discover S-class microwave reactor. IR spectra were collected on a Bruker AlphaP ATR-FTIR spectrometer. Compounds were named according to the IUPAC guidelines with the Von Baeyer system for polycyclic compounds and the carbohydrate α/β system for ring substituents (17).