Liban
Saney
a,
Kirsten E.
Christensen
a,
Miroslav
Genov
c,
Alexander
Pretsch
c,
Dagmar
Pretsch
c and
Mark G.
Moloney
*ab
aThe Department of Chemistry, Chemistry Research Laboratory, University of Oxford, 12 Mansfield Road, Oxford, OX1 3TA, UK. E-mail: mark.moloney@chem.ox.ac.uk
bOxford Suzhou Centre for Advanced Research, Building A, 388 Ruo Shui Road, Suzhou Industrial Park, Jiangsu, 215123, P.R. China
cOxford Antibiotic Group, The Oxford Science Park, Magdalen Centre, Oxford OX4 4GA, UK
First published on 14th April 2023
A general route which provides direct access to substituted bicyclic tetramates, making use of Dieckmann cyclisation of oxazolidine derivatives derived from allo-phenylserines, is reported. Of interest is the high level of diastereoselectivity observed for the N-acylation reaction of oxazolidines and the complete chemoselectivity of their ring closure in the Dieckmann cyclisation. Significantly, the sense of the chemoselectivity is different to earlier reported threo-phenylserine systems, showing the importance of steric bulk around the bicyclic ring system. The derived C7-carboxamidotetramates, but not C7-acyl systems, exhibited potent antibacterial activity against MRSA, with the most active compounds exhibiting well-defined physicochemical and structure–activity properties. This work clearly demonstrates that densely functionalised tetramates are both readily available and may exhibit high levels of antibacterial activity.
With amino esters 7a–f in hand, reaction with pivaldehyde in the presence of Et3N and petroleum ether 40:60 with heating to over 100 °C under a Dean–Stark trap for 16–24 h formed oxazolidines 10a–f in good yields – typically between 74–97% – as roughly a 5:1 mixture of diastereomers at the C2-centre, as determined from NMR analysis (Scheme 1 and Table S5, ESI†).5,7 Confirmation of their relative stereochemistry was achieved by nOe analysis. Since we had earlier found that oxazolidines derived from threo-phenylserines were not very stable to chromatographic purification,12 these products were taken directly into the next reaction sequence. Oxazolidines 10a–f were readily N-acylated by reaction with ethyl malonyl chloride under basic conditions to form malonamides 11a–f in 37% up to quantitative yields (Scheme 1). Once again, the crude products were pure by NMR analysis but could easily be chromatographically purified if required to obtain malonamides 11a–f with high levels of purity (Table S6, ESI†), being typically a 9:1 mixture of diastereomers at the C2-position. nOe studies clearly demonstrated the cis-relationship between the H2, H4 and H5-protons for malonamide 11a and this was also confirmed by a single-crystal X-ray structure (Fig. S1, ESI†).18 This stereochemical outcome was also assumed for the compound series as a whole on the basis of the similarity of their NMR spectra.
Interestingly, the N-acylation of oxazolidines 10a–f was highly diastereoselective in favour of the all cis product (Scheme 1 and Fig. 2), and much more so than the related threo-phenylserine derivatives.12 Therefore, it was apparent that the cis-C4–C5 stereochemistry had a significant effect on the diastereoselectivity of both the oxazolidine cyclisation and N-acylation reactions. While the all cis oxazolidine diastereomer 10a–f is the major one, favoured ca. 5:1 over the minor one, it is also the most reactive in the N-acylation reaction, since there is an exposed face for reactions which is away from all three ring substituents, and this gives the major N-acyl derivative 11a–f. The minor oxazolidine is not so reactive to N-acylation, since both sides of attack on the oxazolidine experience some steric hindrance from ring substituents, allowing for equilibration to the major oxazolidine and onward reaction to the major N-acyl product, leading to an overall 9:1 mixture of malonamide diastereomers 11a–f, starting from roughly a 5:1 mixture of oxazolidine diastereomers 10a–f.
When oxazolidine 10a was reacted with ethyl methylmalonyl chloride 12c (Scheme 3),19 malonamide 13a was obtained as a 1.8:1 mixture of diastereomers at the C7-position, which could be partially separated by careful chromatographic purification, and whose ring relative stereochemistry was confirmed by nOe and NOESY analysis with full assignment achievable by a single-crystal X-ray analysis (Fig. S1, ESI†).18 Similarly, oxazolidine 10a was reacted with ethyl phenylmalonyl chloride 12d (Scheme 3) and malonamide 13b was obtained as a 1.7:1 mixture of diastereomers at the C7-position in good yield, but unlike methyl derivatives 13a, were inseparable by chromatographic purification.19 The relative stereochemistry of the ring protons of malonamides 13b was confirmed as being all cis by a single-crystal X-ray structure of one of the diastereomers (Fig. S1, ESI†).18 Equilibration of malonamide diastereomers 13a and 13b in CDCl3 solution over several weeks was observed.
Malonamides 11a–f were subjected to the Dieckmann cyclisation to form the desired bicyclic tetramates 14a–f (Scheme 1). Initially, malonamide 11a was reacted with KOtBu in THF at reflux overnight, but this led to a complex mixture of unidentifiable products. However, changing the base to DBU, following the procedure of Josa-Culleré et al.,20 effectively gave tetramates 14a–f (Table S7, ESI†) in excellent yields as single diastereomers (Scheme 1) and with full chemoselectivity, following path A for ring closure (Fig. 1). This corresponds to a switch in the chemoselectivity observed for threo-phenylserines12 and this is discussed in more detail below. These tetramates did not require chromatographic purification, being pure by NMR analysis, since aqueous extraction of the highly acidic tetramate salts proved to be effective. Overall, pure tetramates 14a–f could be synthesised from β-oxoesters 5a–g in 5 synthetic steps without the need for purification by chromatography at any stage of the reaction sequence. Additionally, tetramates 14a–f were stable in CDCl3 solution and as an oil for many months. Of interest is that long-range coupling between the H2 and H7-protons (coupling constant of 0.8 Hz) was observed for tetramates 14 (Fig. S2, ESI†). The C5-ethyl ester stereochemistry was later confirmed from a single-crystal X-ray structure of C7-carboxamides 23a and 23d (Fig. S1, ESI†).18 Substituted tetramates 21a and 21b were formed via the Dieckmann cyclisation of malonamides 13a and 13b respectively (Scheme 3). Tetramate 21a was obtained in 62% yield as a mixture of keto/enol tautomers, but in the case of tetramate 21b, a yield of only 10% was obtained, and the product was observed exclusively in the enolic form in CDCl3 solvent.
Of interest was both the high level of, and the switch in, chemoselectivity in the Dieckmann cyclisation of malonamides 11a–f derived from allo-phenylserines (Fig. 3) when compared with those derived from threo-phenylserine derivatives.12 In the formation of enolates 15a–f and then 16a–f from malonamides 11a–f which ultimately led to the formation of the observed bicyclic tetramates 14a–f, both the tBu and aryl groups are on the exo-face in pseudo-diequatorial locations, thus minimising steric interactions. Tetramates 17a–f which might have formed via the ring closure of enolates 18a–f would require C-5 epimerisation in the original substrate, in which both the tBu and aryl groups would be found on the exo-face of the bicyclic lactam, but this material was not observed experimentally. Of interest is that tetramates 19a–f and 20a–f were also not observed experimentally and this is most likely due to the fact that the tBu and aryl groups are on the endo-face of the bicyclic lactam in both cases, causing unfavourable steric congestion – consistent with that observed for the threo-phenylserine-derived tetramic acids, for which detailed theoretical calculations were made.12 Overall, this work demonstrates that allo-phenylserines behave in a similar manner to the reactions of L-serine, L-threonine, L-allo-threonine and L-cysteine-derived products that were condensed with pivaldehyde,5,7 reacting by path A (Fig. 1) but very differently to threo-phenylserine derivatives in Dieckmann ring closures,12 reacting by path B (Fig. 1).
The chemoselective formation of tetramates 14a–f from malonamides 11a–f led to unfunctionalised C7-tetramates. In order to incorporate functionality at this position, tetramates 14a–f were reacted with butyl chloroformate under basic conditions, following the reaction conditions of Jeong et al.,21 to form C7-butyl ester tetramates, to allow differentiation of C5-ethyl esters 22a–f (Scheme 1 and Table S8, ESI†). The yields for these conversions were poor – typically between 10–27%, with a considerable amount of starting material being present in the reaction mixture after 24 h, probably reflecting the low reactivity of the intermediate tetramate enolate. Moreover, the C7-butyl ester tetramates 22a–f were extremely unstable in CDCl3 solution – analogous to C7-ethyl ester tetramates of type 3 (Fig. 1) reported earlier,12 needing to be both chromatographically purified and reacted on very quickly; this instability could also explain the presence of starting material in the initial reaction mixture. These compounds also picked up trace metal ions during chromatographic purification, a phenomenon we had observed earlier, but which could be readily removed with a 10% citric acid (aq.) workup. The importance of metal chelation in these systems has recently become increasingly apparent and has been reviewed.22,23
These compounds could be converted to C7-carboxamidotetramic acids 23a–f by reaction with adamantylamine in yields that ranged from 26–60% (Scheme 1 and Table S9, ESI†) and which were suitable for chromatographic purification as well as being stable in both the pure form and in CDCl3 solution for many months. The aminolysis reaction was fully chemoselective for the C7-butyl ester, with no exchange at the more hindered C5-ethyl ester. Single-crystal X-ray structures of compounds 23a and 23d confirmed their relative stereochemistry, and that no epimerisation had occurred at any of the ring stereogenic centres during the aminolysis reactions (Fig. S1, ESI†).18 Compounds 23a–f, after chromatographic purification on silica gel, picked up trace metal ions to form metal-chelates across the tri-carbonyl unit; this was immediately evident in the 1H NMR spectrum, where broad peaks were observed after chromatographic purification, a phenomenon which has been seen previously.4,21 In the case of compound 23a, the H2 and H4-signals were relatively sharp – indicating these protons were relatively far away from the metal-chelation site – whereas the adamantyl and aromatic peaks were relatively broad – indicating these groups were relatively close to the metal-chelation site (Fig. S3a, ESI†). The acid-washed tetramic acids 23a–f existed as approximately a 50:50 mixture of AB:CD tautomers in CDCl3 solution (Fig. 4), as has been observed in related systems,21 but this differed from C7-carboxamides 24 that lacked a C5-ethyl ester (i.e. those derived from threo-phenylserine), whose ratio was typically an 80:20 AB:CD tautomeric mixture in CDCl3 solution (Fig. 4).12 It appears that the allo-derived 23a binds metal ions with less affinity than threo-derived carboxamidotetramate 24a,12 since the NMR peaks were far sharper for tetramate 23a (Fig. S3, ESI†) than for tetramate 24a. Since the only difference between these two compounds was a missing C5-ethyl ester in the case of tetramate 24a, it seems that the C5-ethyl esters might play an important (steric) role in the reduction of metal-chelation across the proximal tricarbonyl core. The NMR peaks of tetramates 23a–f could be further sharpened by washing metal-chelated tetramic acids with 10% citric acid solution (aq.) (Fig. S3b, ESI†). Interestingly, the peaks for both the adamantyl and aromatic groups were sharpened by a simple acid wash, and changes in the chemical shift values for the tBu, adamantyl and H4-groups before and after acid wash (Fig. S3b, ESI†) were observed.
Inspired by an earlier report of the direct acylation of tetramates using carboxylic acids,24 tetramate 14a was coupled with aromatic and aliphatic carboxylic acids using DCC and stoichiometric amounts of DMAP (1.3 eq.) to form C7-acyltetramates 25a–s in acceptable yields (Scheme 4 and Table S10, ESI†). In this process, substituents were tolerated at the para-position of the aromatic ring system, but unsurprisingly less so at the ortho-position. Moreover, more hindered acids such as 1-naphthoic acid, 2,6-dichlorobenzoic acid, 2-chlorobenzoic acid and 9-anthracenecarboxylic acid gave no products. While the 2-thienyl substituent was obtained in poor yield, aliphatic acids were obtained in moderate to good yields. NOESY studies suggested no epimerisation had occurred at any of the stereogenic centres after DCC/DMAP coupling. Once again, purification of these tetramic acids on silica gel led to the introduction of trace metal ions, evidenced by very broad resonances in their 1H NMR spectra (Fig. S4, ESI†), and which were much broader than metal-chelated C7-carboxamides 23a described above (Fig. S3, ESI†). For example, in the case of carboxamidotetramate 23a, while discernible peaks are visible, for acyltetramate 25d, no assignable peaks were initially visible but sharp peaks arose after a simple acid wash with 10% citric acid solution (aq.).
Additionally, the products in all cases were obtained as the CD tautomers, with the exception of 25o–q, which were observed as a mixture of tautomers but in which the CD tautomers were the major ones. Tan et al. had earlier demonstrated that the major tautomers for C7-acyltetramates 26 was the CD pair, in which the C6-carbonyl peaks from the AB tautomers were more downfield due to deshielding from hydrogen bond effects (Fig. 5),21,24–26 and the spectra for 25a–s were assigned by analogy.
Although the above described synthesis of allo-phenylserines 7a–f from β-oxoesters 5a–fvia Pd–C catalysed hydrogenation was very diastereoselective, it suffered from four major drawbacks. Firstly, all the β-oxoesters 5b–f and oximes 6a–f – with the exception of 5a – had to be synthesised. Secondly, a sizeable amount of Pd–C was required to convert the oximes 6a–f to allo-phenylserines 7a–f, and thirdly, the yields were modest. Finally, chloro- and bromo-substituted allo-phenylserine derivatives further reacted by hydrogenolysis under the reduction conditions to give dehalogenated side-products. An alternative, based upon a report by Chaumont et al.27 of the erythro-diasteroselective synthesis of Boc-protected allo-phenylserine derivatives via a decarboxylative aldol reaction between the hemi-malonate derivative of diethyl (Boc-amino)malonate 27c and various aldehydes (Scheme 5),27 was therefore examined; starting 27c could be readily synthesised by reacting inexpensive diethyl aminomalonate hydrochloride 27a with Boc2O. Boc-protected allo-phenylserine 9a synthesised this way was fully identical to that both prepared by Chaumont et al.27 and the material prepared earlier (Scheme 1). This approach proved to be applicable to a wide variety of substituted allo-phenylserine derivatives 9a–m (Scheme 5 and Table S11, ESI†).
Boc-protected-allo-phenylserine derivatives 9a–l were deprotected using TFA and were stirred at room temperature for 2 h to furnish allo-phenylserines 28a–k as the trifluoroacetate ammonium salts in quantitative yields (Scheme 5 and Table S12 ESI†). The exception was 9l, which gave a complex mixture of side-products, and was thus not used further, and 9m was discontinued due to its low yield. The allo-phenylserine derivatives 28a–k were then reacted with pivaldehyde using the established protocol outlined above, to obtain oxazolidines 10a,g–p in good yields as mixtures of diastereomers (Scheme 5 and Table S13, ESI†). The spectroscopic consistencies of oxazolidines 10a,g–p with those of 10a–f derived from β-oxoesters 5a–f prepared above confirmed their cis-relationship (see Tables S5 and S13, ESI†). The major and minor diastereomers of oxazolidines 10g–p were readily assigned by comparison to the independently prepared and characterised materials described above. These compounds were then N-acylated with ethyl malonyl chloride using the established protocol to obtain malonamides 11a,g–p in good yields as predominantly one diastereomer (Scheme 5 and Table S14, ESI†), again consistent with that of malonamides 11a–f. Comparison to earlier compounds and NOESY studies confirmed their relative stereochemistry. These malonamides 11a,g–p were then reacted with DBU in THF overnight (Scheme 5) to form tetramates 14a,g–m in good yields (Table S15, ESI†). However, surprisingly, tetramate 14n was formed as a very minor product, and was not used further, and tetramate 14o did not form at all, presumably due to hindrance at the ortho-position. Additionally, the Dieckmann cyclisation of the 3-pyridyl substituent 11p was not successful.
Compound | MIC(S. aureus) (μg mL−1) | Cell lines EC50 (μg mL−1) | Therapeutic ratioa | |||
---|---|---|---|---|---|---|
HeLa | heK-293 | CaCo | MDCK | |||
a Calculated from MIC against S. aureus using the HeLa cell line (EC50 HeLa cell line/MIC(S. aureus)). | ||||||
14a | 15.6 | 15.6 | 31.3 | 62.5 | 31.3 | 1 |
14b | 7.8 | 15.6 | 15.6 | 31.3 | 31.3 | 2 |
14c | 15.6 | 7.8 | 3.9 | 31.3 | 15.6 | 0.5 |
14d | 15.6 | 15.6 | 31.3 | 62.5 | 31.3 | 1 |
14e | 15.6 | 15.6 | 31.3 | 62.5 | 31.3 | 1 |
14f | 15.6 | 15.6 | 15.6 | 31.3 | 31.3 | 1 |
23a | 0.25 | 15.6 | 15.6 | 31.3 | 31.3 | 62 |
23b | 0.25 | 15.6 | 15.6 | 31.3 | 31.3 | 62 |
23c | 3.9 | 15.6 | 31.3 | 31.3 | 62.5 | 4 |
23d | 0.25 | 7.8 | 15.6 | 7.8 | 31.3 | 31 |
23e | 0.25 | 7.8 | 15.6 | 15.6 | 31.3 | 31 |
23f | 0.25 | 7.8 | 7.8 | 15.6 | 15.6 | 31 |
However, C7-carboxamides 23a–f displayed potent antibacterial activities against MRSA, similar to previously reported C7-carboxamides,12,21 and of interest was that a C5-ethyl ester group was well tolerated, particularly when compared with threo-derived C7-carboxamide 24a,12 although none exhibited antibacterial activity against E. coli. The antibacterial activity against MRSA for C7-carboxamides 23a,b,d–f was extremely potent – MIC values of 0.25 μg mL−1 – with the exception of tetramate 23c, which had a MIC value of 3.9 μg mL−1 (Table 1), suggesting that hydrophobic groups at the para-position worsen bioactivity. The physicochemical property space of the most potent compounds (MIC < 4 μg mL−1) was confined to a small area of chemical structure space (523 < Mw < 579 g mol−1; 3.9 < clogP < 5.9; 105 < PSA < 114 Å2; 777 < MSA < 898 Å2; HBD = 2; HBA = 5, see Fig. 6 and Table S16, ESI†), with many compliant with the Ro5 – even though the initial design of these compounds did not explicitly seek to obey Lipinski's parameters. The high clogP values of many of these may also explain, at least in part, the lack of activity observed against Gram (−) E. coli bacteria. These bicyclic tetramates were significantly more lipophilic and less polar than other antibiotic classes (Fig. S5, ESI†) and the structure–activity profile indicates distinct regions of the bicyclic nucleus responsible for antibacterial activity and/or metal chelation (Fig. 7).
Fig. 7 Overview of the chemical components necessary for potent antibacterial activity against MRSA, derived from an in-depth SAR analysis of the N,O-bicyclic tetramates. |
This compound series showed some cytotoxicity against HeLa, heK-293, CaCo and MDCK human cell lines, and the therapeutic ratio (calculated from MIC against S. aureus using the HeLa cell line (EC50 HeLa cell line/MIC(S. aureus)) (Table 1) shows that C5-ethyl ester tetramates 14a–f possessed low selectivity, although this was mainly due to the lack of potency observed against MRSA. However, carboxamides 23a,b,d–f showed high selectivity for bacteria, with a therapeutic ratio of either 31 or 62. Panduwawala et al.4 and Jeong et al.21 had earlier demonstrated that carboxamidotetramic acids typically lost their antibacterial activity in the presence of human serum albumin (HSA). When bicyclic tetramates 23a–f were tested against MRSA in the presence of 1/3 the serum albumin concentration as that found in human blood, for carboxamides 23a,b, 23d–f the activity dropped 16-fold, and in the case of 23e, 33-fold. However, compound 23c only showed a 4-fold decrease in antibacterial activity and noteworthy is that it is the most lipophilic in the carboxamide series 23a–f. It was clear from these results that MIC shifted to higher values in the presence of HSA, consistent with the known behaviour of plasma protein binding small molecules; it has earlier been shown that PPB strongly correlates with increased hydrophobicity.29 However, good bioactivity (MIC < 4 μg mL−1) is still observed in the presence of 1/3 the HSA concentration as that found in blood, and this result is important, indicating that the aromatic substitution placed at the C4-position of the bicyclic tetramate could not only be tolerated, but could also be crucial to adjusting the physicochemical properties of these bicyclic tetramates to lend itself towards potential in vivo studies and therapeutic applications.
Footnote |
† Electronic supplementary information (ESI) available: Experimental details; 1H and 13C NMR spectra; X-ray crystallographic data (CIF); bioassay and cheminformatic data. 2235613–2235618. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3ob00376k |
This journal is © The Royal Society of Chemistry 2023 |