Functionalised Bicyclic Tetramates Derived from Cysteine as Antibacterial Agents

Abstract

Routes to bicyclic tetramates derived from cysteine permitting ready incorporation of functionality at two different points around the periphery of a heterocyclic skeleton are reported. This has enabled the identification of systems active against Gram-positive bacteria, some of which show gyrase and RNA polymerase inhibitory activity. In particular, tetramates substituted with glycosyl side chains, chosen to impart polarity and aqueous solubility, show high antibacterial activity coupled with modest gyrase/polymerase activity in two cases. An analysis of physicochemical properties indicates that the antibacterially active tetramates generally occupy physicochemical space with MW of 300–600, clogD_7.4 of −2.5 to 4 and rel. PSA of 11–22%. This work demonstrates that biologically active 3D libraries are readily available by manipulation of a tetramate skeleton.

Introduction

Tetramates are of interest^1–4 principally for their antibacterial activity,^{5, 6} and we have reported recently that systems derived from the amino acids, serine 1a,⁷ threonine 1b,⁸ and cysteine 1c⁹ provide useful templates for application in synthetic and medicinal chemistry, which in some cases exhibit potent antibacterial activity. Amongst various methodologies to these derivatives,¹⁰ our route makes use of the preferential formation of the malonamides of t-butyl oxazolidine/thiazolidine templates, formed as the cis-2,5-isomer 2a-c rather than the alternative trans-2,5- 3a-c, which mimises the steric impact of the bulky t-butyl group, to direct a very efficient chemoselective Dieckmann ring closure (Route a, Scheme 1) in preference to the alternative (Route b, Scheme 1).¹¹ Of interest, though, is that the analogous thiazolidine system derived from cysteine and an aromatic aldehyde behaves differently (Scheme 2), not least because there is a shift in preference away from the cis-thiazolidine 6 to the trans- isomer 7, the latter of which now preferentially cyclizes to give tetramate 8, equivalent to Route b in Scheme 1; this shift is, however, not complete and some of the cis-isomer 6, after epimerization at C5, closes under these conditions leading to the enantiomer of 8, and therefore giving some erosion of enantioselectivity.¹² The overall outcome is that acyl tetramates 8 (Scheme 2),¹³ known for their metal chelating¹⁴ and antibacterial activity,¹⁴ are accessed directly, rather than by acylation of an unfunctionalised tetramate nucleus.¹⁵ Since we had earlier explored in detail the structure activity relationship of tetramates substituted at C-7 but limited to t-butyl substitution at C-2, these thiazolidine compounds provided access for the first time to compounds with an aryl substituent at C-2 and this gave the opportunity to explore the scope of antibacterial bioactivity at this position. The results of that work are reported here.

Scheme 1.

Scheme 2.

Results and Discussion

Synthesis of the required tetramate proceeded from cysteine 1c to a mixture of the cis- and trans- diastereomeric malonamides 6a-g and 7a-g, with a preference for the latter trans- isomers in three cases, followed by Dieckmann cyclisation to 7-ethoxcarbonyltetramates 8a-g (Scheme 2); there is up to 20% attrition of e.e. during this process, arising from cyclisation of the minor but unseparated cis-isomer 6a-g.¹² These esters were expected to be suitable for conversion to the amide by direct aminolysis, a process which had been used successfully previously,¹⁶ and this was immediately successful for synthesis of systems 9a-v (see Table 1) despite the high level of functional group density around the ring system which might not have given a chemoselective process, although elevated temperature was needed for successful reaction.

Table 1:

Synthesis of tetramates 9a-v according to Scheme 2.

Compound	Ar	R	Yield (%)	Tautomeric ratioa
9a	Ph	1-Adamantyl	45	2.7:1
9b	p-BrC6H4		70	2.6:1
9c	m-BrC6H4		38	2:1
9d	p-FC6H4		56	2.7:1
9e	p-NO2C6H4		40	2.4:1
9f	2Cl-4-FC6H3		60	2.5:1
9g	2-C4H3O		92	3:1
9h	Ph	4-Cyclohexylphenyl	54	15:1
9i	p-FC6H4		45	13:1
9j	p-NO2C6H4		59	12.5:1
9k	2Cl-4-FC6H3		73	15:1
9l	Ph	4-Chloro-2-methylphenyl	53	only 1 form
9m	p-FC6H4		40	only 1 form
9n	p-NO2C6H4		37	only 1 form
9o	p-BrC6H4		40	only 1 form
9p	m-BrC6H4		38	only 1 form
9q	2Cl-4-FC6H3		35	only 1 form
9r	p-BrC6H4	4-Morpholinophenyl	69	only 1 form
9s	2-C4H3O		75	only 1 form
9t	p-BrC6H4	4-Aminotetrahydropyranyl	70	5.2:1
9u	Ph	Cyclohexyl	49	3.1:1
9v	2Cl-4-FC6H3		40	3.1:1

^aMajor form is tautomer pair AB -see Figure 4 and Table 2.

Chemoselective formation of the amide product, and not the alternative enamine arising by attack at the ketone, was evident by loss of the ester function, and also confirmed by HMBC correlation (Figure 1); thus, C-6 correlates with H-4 and H-5, while C-8 correlates with H-2 and H-5 as expected. For compounds 9s and 9t, C-9 correlates with the β-CH in the amine with respect to C-9, confirming the assignment of the carbonyl functional groups.

Figure 1.

HMBC correlation data of selected carboxamide tetramates.

As anticipated, the stereochemical relationship of the starting bicyclic tetramate esters 8 was conserved in the amide systems, and a trans- relationship across the bicyclic ring between H-2 and H-5 was determined from NOE analysis (Figure 2); this outcome was further confirmed by single crystal X-ray diffraction studies of adamantyl derivative 9f (Figure 3).¹⁷

Figure 2.

NOE analysis of selected carboxamide tetramates.

Figure 3:

The structure of 9f from single crystal X-ray diffraction studies. Displacement ellipsoids drawn at 50% probability.

The products were generally found to exist as tautomeric forms, a phenomenon which has been observed previously, and which could be assigned from careful NMR examination.^{15, 18} In their NMR spectra, the internal tautomeric pairs (A and B, C and D) could not distinguished and were observed as an averaged hybrid, while the external tautomeric pairs (AB and CD) had distinct ¹³C chemical shift differences of C-6, C-8 and C-9 (Figure 4), so that the tautomeric forms could be readily distinguished (Table 2).

Figure 4.

NOESY correlation data for 9k suggesting tautomer A as the major tautomer. The distance between H-5 and enol O-H for each tautomer was calculated for the MM2-energy minimized molecular model using Chem3D v.15.

Table 2.

¹³C chemical shifts of tautomers of selected examples of carboxamide tetramates.

Compound Number	Solvent	13C chemical shifts (ppm)			Tautomeric form
		C-6	C-8	C-9
9a	CDCl3	187.8	172.2	165.8	major (AB)
		191.1	178.0	166.4	minor (CD)
9b	CDCl3	188.4	172.4	165.9	major (AB)
		191.1	178.2	166.4	minor (CD)
9f	CDCl3	188.7	172.0	166.0	major (AB)
		191.0	177.8	166.5	minor (CD)
	MeOD	189.5	175.2	166.9	only 1 form (AB)
9k	CD2Cl2	185.2	171.7	164.3	major (AB)
		191.5	178.2	165.4	minor (CD)
9q	CD2Cl2	184.7	171.7	164.3	only 1 form (AB)
9t	CD2Cl2	186.6	172.5	165.9	major (AB)
		191.6	178.8	166.5	minor (CD)

The ¹³C chemical shifts of C-6, C-8 and C-9 of the minor tautomers were invariably more downfield than their major tautomers, with the smallest Δδ¹³C observed for the exocyclic amide C-9. The tautomeric behaviour is also solvent dependent, as seen from the different chemical shift values of 9f in CDCl₃ and CD₃OD. In polar solvents (e.g. methanol-d₄, acetone-d₆, DMSO-d₆ and CD₃CN), only one set of resonances was observed for each carbon, and appears to be that of an averaged hybrid structure arising from facile equilibrium of the tautomeric forms. NOESY data obtained for 9k in dry CD₂Cl₂ revealed that the major tautomer for 9k was A (Figure 4), based on the correlation from H-5 to proximal enol O-H, which would not have been possible in other tautomeric forms B, C and D (Figure 4). This conclusion is supported by single crystal X-ray diffraction studies of tetramate 9f (Figure 3, vide supra), which was found to exist in the endoenolic form A. Since the chemical shift pattern observed for the major/minor tautomers is similar for all compounds, they are all assumed to adopt the major tautomeric form A, and this finding is similar to that found in related systems.¹⁵ Minimum energy conformations were generated for the four tautomeric forms A-D for compounds 9k using MM2 methodology and confirmed tautomer A to be the thermodynamically more stable (Table 3 and Table 1, SI).

Table 3.

MM2-minimized energy of tautomeric forms A-D for selected examples 9f, 9k and 9q.

Compound Number	Tautomeric ratio	Minimum Energy of tautomeric forms (kcal/mol)
		A	B	C	D
9f	2.5:1	18.93	39.86	26.83	39.32
9k	15:1	6.93	31.28	13.04	30.75
9q	only 1 form	0.80	25.54	7.05	26.23

Having demonstrated the generality and selectivity of this approach, of interest was its application to more elaborate side chain systems; to this end, the products 9w-g’ were accessed by direct aminolysis of esters 8a, b and f using the required amines (Scheme 2 and Table 4), mostly available from commercial sources. In order to incorporate a higher level of polarity, mono-Fmoc protected 10b, prepared from dapsone 10a, an inhibitor of dihydropteroate synthetase,¹⁹ was used for the aminolysis of tetramate ester 8a to give carboxamide tetramate 9a’, which on subsequent deprotection gave 9b’ in 40 % yield (Scheme 3). The design of carboxamide 9d’ was inspired by naturally occurring tetramates kibdelomycin²⁰ and amycolamicin,²¹ and was synthesised by aminolysis of tetramate 8a with glycosylated amine derivative 12b (Scheme 4). Although it has been previously reported that the desired glycosylated amine 12b could be directly synthesised from 4-aminophenol 11a and bromogalactose,^{22, 23} Fmoc-protected amine 11b was found to be more suitable for this purpose by Lewis acid-mediated (BF_3.OEt₂) glycosylation; the equatorial substitution of the aryl group was confirmed by the trans-diaxial ³J_H1-H2 coupling constant (8.0 Hz). Base-mediated deprotection of the amine gave the desired product 12b which was reacted with tetramate 8a to yield carboxamide 9d’ in 60 % yield, and methanolysis under basic conditions afforded 9e’.

Table 4:

Synthesis of tetramates 9w-g’

Compound	Ar	R	Yield (%)
9w	Ph		30
9x			74
9y			75
9z			72
9a’			30
9b’			40
9c’	p-BrC6H4		40
9d’	Ph		60
9e’	Ph		54
9f’	2Cl-4-FC6H3		30
9g’	2Cl-4-FC6H3		36

Scheme 3.

Scheme 4.

The carboxamidotetramates 9a-g’, when purified by silica gel column chromatography, gave broad signal resonances in their ¹H NMR spectra, consistent with metal chelation, as has been observed previously.⁷ As a result, carboxamides were routinely purified on silica column with 1 % Et₃N in the eluent. Residual Et₃N was removed by washing with 5 % citric acid, giving the purified product with sharp and well-resolved ¹H spectra; an example is given for compound 9d (Figure 5). Alternatively, it was also possible to purify these carboxamides on silica column without any Et₃N in the eluent, provided that they were subsequently washed with 2 M HCl to give the metal-free form.

Figure 5.

¹H NMR spectra of tetramate 9d (a) post-column purification, before acid-wash and (b) post-column purification, after acid-wash.

Since peak broadening was a recurrent observation after column purification, the metal-chelated tetramate and acid-washed tetramate for compound 9a were analysed by Inductively Coupled Plasma Mass Spectrometry. The metals Na, Mg, Fe and Zn but especially Ca were found in high abundance in the metal-chelated tetramate, which were substantially removed by washing with 2M HCl (Table 5 and Table 2, SI). Clearly these tetramates are easily capable of forming metal salts with diverse cations; the metal chelating behaviour of 3-acyltetramates has been reviewed recently,^{2, 6, 13, 14, 24} and chelation both of natural products^25–29 especially with antibacterial activity³⁰ and of simpler systems^31–36 has also been reported while in our own work, calcium salt formation (but not other metals) during the isolation of tetramates had been observed earlier.³⁷

Table 5.

Metal content observed in metal-chelated and acid-washed forms of carboxamide 9a.

	Metal content (ppm)
	Mg	Ca	Fe	Zn	Na
Metal-chelated	7026.1	20483.7	38.2	71.2	3385.6
Acid-washed	15.9	24.8	34.8	9.9	18.8

In an attempt to make use of the C-2 bromoaryl substituent for Suzuki-Miyaura (SM) cross-coupling reactions for further derivatisation, reaction of both thiazolidines 8b and 9b with 2-furanylboronic acid was examined, but only unreacted starting material was recovered; this was in contrast to oxazolidine 13a (Scheme 5) which gave the expected coupled product 13b under the same conditions. This failure of SM cross-coupling on thiazolidine-derived tetramic acids was attributed to catalyst deactivation by the cyclic thioether function, and alternative reaction conditions were clearly necessary. While successful SM cross-coupling on thiophene systems had been documented using PPh₃ and AsPh₃,³⁸ no reports on systems containing non-aromatic thioethers such as those of the thiazolidine-derived tetramates could be found in the literature. However, a more recent report on Suzuki cross-coupling on biotin-derived systems³⁹ together with a detailed review by Buchwald et al.⁴⁰ suggested XPhos as an alternative, and this ligand proved to be successful, with the desired product 14a being formed along with the return of some unreacted starting material (Scheme 5 and Table 6). Although it has been reported that the active form of the catalyst for SM cross-coupling involving dialkylbiaryl phosphine ligands is the monoligated L₁Pd(0) species rather than the more highly coordinated L₂Pd(0),^{40, 41} a ratio of L:Pd=1.5:1 was not sufficient to drive the reaction to completion even after a reaction time of 48 h. Increasing the L:Pd ratio to 3:1 led to completion of the reaction and this approach was successfully applied to the synthesis of a range of carboxamide tetramates 14a-l with elaborated pendant groups at C-2 derived from 9b and 9s (Table 6). The boron reagents used for the above transformations were commercially available acids or their pinacol esters, except for 14c where the boron reagent was the glycol ester derived from the corresponding boronic acid, prepared according to literature procedure.⁴² Carboxamide tetramate 14k was synthesised as a mixture of diastereomers from the corresponding commercially available racemic 1-(tetrahydropyran-2-yl)-1H-pyrazole-5-boronic acid pinacol ester. It would appear the additional nucleophilicity of XPhos gives a higher electron density at Pd(0) in the monoligated L₁Pd(0) species, and increases the rate of oxidative addition leading to overall successful coupling.⁴⁰

Scheme 5.

Table 6:

Suzuki-Miyaura cross-coupling of carboxamide tetramates 9b and 9t.

Carboxamide tetramate (R=)	Compound Number	R1	% Yield	Compound Number	R1	% Yield
	14a		72	14g		70
	14b		74	14h		78
	14c		80	14i		53
	14d		45	14j		70
	14e		78	14k		60
	14f		74
	14l		30

Several naturally occurring tetramic acids bearing glycosyl residues are known, including virgineone,⁴³ kibdelomycin,⁴⁴ amycolamicin,⁴⁵ aurantoside^{46, 47} and streptolydigin.⁴⁸ While the exact role of the sugar moiety of these tetramates is not known, it may give improved bioavailability with increased solubility and transportation, or help modulate toxicity and aid in bacterial target interaction.⁴⁹ Thus, investigating the effect of glycosylated tetramates on antibacterial activity was of interest, and the systems developed thus far offered the feasibility of C-2 substitution of the tetramate ring. In order to demonstrate an alternative process using late-stage ring formation, bicyclic tetramates with C-2 pendant glycones were derived from β-D-galactose pentaacetate (Scheme 6). Acetobromo-α-D-galactose 15b was prepared from β-D-galactose pentaacetate 15a according to the literature procedure;⁵⁰ axial bromination at the anomeric carbon was confirmed from the ³J_(H1-H2)-coupling constant of H-1 (d, J=4.1 Hz). Solid-liquid phase transfer-catalysed aryl glycosylation of 4-hydroxybenzaldehyde with acetobromo-α-D-galactose 15b gave the desired aldehyde 15c, according to a modified literature procedure; BnNBu₃Cl proved to be a better phase transfer catalyst than Bu₄NBr while the presence of H₂O led to a remarkable reduction in yield.⁵¹ The β-anomer was obtained exclusively, where the equatorial substitution of the phenoxide at the anomeric carbon was confirmed by ³J-coupling constant of the anomeric C-H (d, J=8.0 Hz). The glycosylated aldehyde 15c was then condensed with ʟ-cysteine methyl ester hydrochloride to yield the expected thiazolidine in 85 % yield with a cis:trans diastereomeric ratio of 1.6:1, and this material was N-acylated to give the corresponding malonamide in 52 % yield, which was in turn cyclised to afford the glycosylated tetramate 16 but only if the crude product was isolated without acidic work-up. Base-catalysed methanolysis gave tetraol 17 in quantitative yield (Scheme 6); neither hydrolysis nor transesterification of the ethyl ester at C-7 was observed under these reaction conditions. Aminolysis of tetramate ester 16 afforded carboxamide tetramates 18 with the acyl groups of the sugar moiety intact; methanolysis of the former resulted in the very polar carboxamide tetramate 19.

Scheme 6.

A tetramate with a disaccharide pendant group was derived from commercially available lactose octaacetate (Scheme 7); this followed a similar approach to the monosaccharide derivative above. Commercially available lactose octaacetate 20a (as a 2.9:1 mixture of α- and β- anomers at the glucopyranosyl end) was subjected to BiBr₃-catalysed bromination and afforded 20b as the α-anomer in 92 % yield, confirmed by ³J-coupling constant of the anomeric C-H. Aryl glycosylation gave aldehyde 20c, which was condensed with ʟ-cysteine methyl ester hydrochloride, acylated and cyclised to afford tetramate 21.

Scheme 7.

Due to the acid-labile nature of phenyl glycosides prepared above and the resulting difficulty associated with the acidic work-up procedure required to remove chelating metal ions, an alternative route to the incorporation of glycones via a benzyl alcohol linking group was explored (Scheme 8). The controlled reduction of terephthalaldehyde with NaBH₄ between 0–2°C afforded 4-(hydroxymethyl)benzaldehyde in 70 % yield according to a modified literature procedure.⁵² Lewis acid-catalysed glycosylation with β-D-galactosepentaacetate 15a gave aldehyde 22 as the β-anomer in 71% yield along with the byproduct 4-(acetoxymethyl)benzaldehyde in 28 % yield. Following the usual synthetic route to access the bicyclic tetramate core, condensation of aldehyde 22 with cysteine ethyl ester yielded the expected thiazolidine, which was acylated and cyclized under basic conditions to afford benzyl glycosylated tetramate 23, which proved to be stable to acidic work-up.

Scheme 8.

In order to permit late stage C-2 substituent manipulation, tetramates 26a and 26b were derived from tetramates 25a,b by silyl deprotection, which were obtained from the silyl ether of 4-hydroxybenzaldehyde 24a or 4-(hydroxymethyl)benzaldehyde 24b respectively (Scheme 9). Conversion of 25a to the corresponding carboxamides 27a-c by aminolysis with the required amine proceeded using the approach described above, in modest to good yield. The deprotection of silyl ether 27a to obtain tetramate 28a, however, required some optimisation. While TBAF was a successful desilylating agent, chromatographic purification was difficult, possibly due to the basicity of fluoride ion in an aprotic solvent⁵³ leading to deprotonation of the product tetramic acid. However, alternative conditions using KF as a fluoride source were more successful. Two other carboxamide tetramates 28b and 28c were also synthesised using this approach.

Scheme 9.

The presence of a phenolic group at the C-2 position provided a platform for further functionalization of carboxamide tetramates via Williamson ether synthesis. In the presence of K₂CO₃, nucleophilic substitution of ethyl bromoacetate and 1-bromo-2-butyne by the phenoxide derived from 28a gave carboxamide tetramates 29a and 30 respectively (Scheme 9). HMBC correlation data confirmed the substitution at the phenolic but not at the enolic alcohol group. Ester hydrolysis of 29a afforded the carboxylic acid bearing tetramate 29b with improved hydrophilicity.

Since the pyroglutamate skeleton is found in antibacterial natural products (e.g. oxazolomycin^{54, 55} and pramanicin⁵⁶), the synthesis of a pyroglutamate-tetramate hybrid system was also examined. This was achieved by condensing pyroglutaminol 31,⁵⁷ prepared by the literature procedure,^{58, 59} with terephthalaldehyde by reflux in toluene using a Dean-Stark apparatus to give a mixture of products 32 and 33 in 1:2.4 ratio (Scheme 10); the lactam-functionalized aldehyde 33 was isolated by column chromatography as a single diastereomer, whose stereochemical assignment was in agreement with previous reports on N,O-acetal formation of 31 with benzaldehyde.^{60, 61} Aldehyde 33 was in turn converted to the desired bicyclic tetramate ester 34a in the usual way, which was converted to carboxamide 34b with 1-adamantylamine. The stereochemical assignment of the tetramate system was confirmed by NOE experiments (Scheme 10). The relative stereochemistry of both heterocyclic systems was found to be trans-, in keeping with structures reported in the literature.⁶²

Scheme 10.

Of interest is that direct Buchwald-Hartwig amination of bromo-substituted tetramates 8b and 9b using any of XPhos, RuPhos and XantPhos was not successful. Instead, p-bromobenzaldehyde 35a was functionalized via Buchwald amination/amidation to give aldehydes 35b,c and these were then used to synthesise the tetramate core de novo using a similar route to that described above, giving esters 36a,b which were converted to carboxamides 37a,b as before (Scheme 11).

Scheme 11.

In one case, deprotection of the N,S-acetal was examined; in comparison to oxazolidine-based N,O-acetal systems, thiazolidine-based N,S-acetals have been shown to have higher stability to acids,⁶³ and their deprotection often requires harsh conditions such as heating under reflux with 5M HCl for 72 h.⁶⁴ However, Onoda et al. have shown that Boc-protected thiazolidines can be deprotected at the N,S-acetal in good yield with 15 eq of TFA in water-saturated CH₂Cl₂ under ambient conditions within 2 h.⁶⁵ Deprotection of carboxamide tetramate 9a was investigated with TFA in CH₂Cl₂, but even after a reaction time of 24 h, only starting material remained, and while the addition of water accelerated the reaction, complete deprotection was not observed in 24 h. However, deprotection of thiazolidine via the Corey-Reichard protocol⁶⁶ was successful giving carboxamidotetramate 38 in 70 % yield (Scheme 12). This observation highlights the need for stronger conditions for the deprotection of the more stable thiazolidine-derived carboxamide tetramates.

Scheme 12.

Antibacterial properties

Mode of action of tetramates

Naturally-occurring tetramates, such as kibdelomycin and amycolamycin, are inhibitors of bacterial topoisomerase IV (topo IV) and DNA gyrase,^{20, 44, 67} whereas the tetramates streptolydigin and tirandamycin are inhibitors of RNA polymerase (RNAP).^{68, 69} Some tetramates have been shown to provide dual inhibition of both RNAP and undecaprenyl pyrophosphate synthase (UPPS).⁷ All four of theses enzymes are essential for viability. The primary function of topo IV is the decatenation of daughter chromosomes that are multiply-linked, during the final stages of DNA replication,⁷⁰ and the role of DNA gyrase is catalysis of negative supercoiling of DNA.⁷¹ Bacterial RNAP is an attractive target in antibacterial chemotherapy as the bacterial RNAP subunit sequences are highly conserved while being different from eukaryotic RNAP subunit sequences. This permits broad-spectrum activity and therapeutic selectivity.⁷² An examination of the mode of action of the tetramates prepared in the current work was therefore of interest.

Inhibition of possible bacterial target enzymes by some bicyclic tetramates was studied for Staphylococcus aureus topo IV and RNAP, Escherichia coli RNAP and Mycobacterium tuberculosis gyrase. A selection of compounds (full data is given in Tables 3–5, SI) initially was tested at a fixed concentration of 100 μM for the inhibition of topo IV and gyrase, and the percentage of DNA decatenated or supercoiled in the presence of each test compound was calculated. A lower percentage of decatenation or supercoiling indicated a higher level of inhibition of the target enzymes. The data were compared to those for the known topo IV and gyrase inhibitor, ciproflaxacin. Concentrations resulting in half-maximal inhibition (IC₅₀s) for some of the more potent inhibitors (vide infra) were determined and are presented in Figure 6(a). Tetramates 9a, 9h and 9m were tested for their inhibition of S. aureus and E. coli RNAP,^73–75 and the data obtained are presented in Figure 6(b).

Figure 6.

(a) IC₅₀ values of tetramates against M. tuberculosis gyrase and S. aureus topoisomerase IV. Compounds 9a and 9m were weak inhibitors of M. tuberculosis gyrase and 9d’ was a weak inhibitor of S. aureus topoisomerase IV. (b) IC₅₀ values of tetramates against S. aureus RNAP and E. coli RNAP.

Comparison of mimimal inhibitory concentrations (MICs) for antibacterial activity against methicillin-resistant S. aureus (Table 10, SI; see also Figure 7; vide infra) with IC₅₀s for inhibition of S. aureus RNAP and topo IV (Figure 6) indicates that compounds exhibiting high antibacterial activities also exhibit high S. aureus RNAP and topo IV inhibitory activities, consistent with the possibility that antibacterial activity may be attributable to RNAP and topo IV inhibitory activity. Compounds 9a, 9h and 9m inhibited both RNAP and S. aureus topo IV, consistent with a possible dual mode of action for these compounds. Some compounds that showed high efficacy in enzyme inhibition assays did not show high efficacy in whole-cell assays (e.g 9d’, 19). This potentially may be due to low cell permeability or elimination via efflux mechanisms, leading to reduced bioavailability in the bacterial cell. In general, E. coli RNAP was less sensitive to tetramates than S. aureus RNAP, and the M. tuberculosis gyrase was less sensitive to tetramates than S. aureus topo IV.

Figure 7.

Bioactivity of carboxamide tetramates with/without HSA.

Antibacterial activity: hole-plate method

Whole-cell antibacterial assays were performed using the hole-plate method with Gram-positive S. aureus DS267 or Gram-negative E. coli X580, using Cephalosporin C as a positive control. The samples were prepared as 4 mg/mL solutions of 70% DMSO in MeOH, with serial dilution to the desired concentrations where necessary. The relative potency was estimated by reference to positive controls prepared with Cephalosporin C.⁷⁶ Many of the carboxamide analogues 9a-v showed antibacterial activity against S. aureus, but none showed activity against E. coli (Table 6–9, ESI). The presence of an adamantyl carboxamide side chain on the tetramate core resulted in a significant enhancement of antibacterial activity and such tetramates 9a-f were active at a concentration of 1 μg/mL. Tetramates 9h-9k with a 4-cyclohexylphenyl group or 9l-9q with a 4-chloro-2-methylphenyl group demonstrated good antibacterial activity, although much less potent compared to their adamantyl analogues, with the 4-cyclohexylphenyl group being the more active. The decrease in antibacterial activity observed for 9r, 9s and 9w further confirms the requirement for a bulky pendant group at C-7 of the bicyclic tetramate core.

Since some tetramates have been shown to lose antibacterial activity in the presence of human serum albumin (HSA),⁷ further assays were run in the presence of HSA. These data were compared with the bioactivity of each sample in the absence of HSA and selected examples are given in Figure 7 along with full data in Table 6–9 (ESI). While 9a-9f showed good antibacterial activity at 1 μg/mL in the absence of HSA, there was a complete loss of activity in the presence of 4 mg/mL of HSA (where the tetramate:HSA ratio is approximately 1:30, data not shown). These data suggest that carboxamides are possible ligands for HSA. Nevertheless, of interest was the return of bioactivity for these tetramates when the concentration of the test compounds was increased to 5 μg/mL (in which tetratmate:HSA ratio is approximately 1:6). Similarly, tetramates 9h-9q exhibited no antibacterial activity at their original concentrations when tested in the presence of HSA, but their activity was regained by increasing the compound concentrations. Interestingly, tetramates 9g,9r,9s and 9w which were less potent compared to the other tetramates discussed above retained their potency in the presence of HSA.

The assay samples were prepared in 70% DMSO in MeOH for reasons of solubility, followed by mixing with albumin solution, and resulted in a final solvent composition of 35:15:50 of DMSO:MeOH:H₂O in the well (35% DMSO). It is known that higher concentrations of DMSO can lead to protein denaturation,^{77, 78} and this might have interfered with the above analysis. Carboxamides 9a-9f showed good antibacterial activity at 1 μg/mL, and these samples could be dissolved in a minimum amount of DMSO at such low concentrations. Thus, the antibacterial activity of carboxamides 9a-9f in the presence of HSA was tested where the final solvent composition was 2% DMSO in H₂O. The zones of inhibition measured were in agreement with the data obtained in the presence of 35% DMSO, confirming that the antibacterial activity observed for the tetramates in the presence of HSA was not a result of the compound unbound to denatured HSA.

Broth dilution method

Broth assays were performed to determine MICs of compounds (Table 10, ESI). The carboxamido tetramates 9a-g’ showed either weak (MIC= 250 μg/mL) or no activity against Gram-(−) strains, in agreement with the data obtained from the hole-plate method, while some exhibited promising antibacterial activity against Gram-(+) strains. Among the C-7 functionalized tetramates, the most potent carboxamide tetramates 9b-f, h, l-m, q, u-v, z showed an MIC of 0.49 μg/mL against at least one organism, significantly better than the tetramate ester analogues 8a-g,¹² highlighting the importance of a carboxamide side chain group for whole-cell antibacterial activity. However, the presence of more polar carboxamides at C-7 led to a loss of activity (see 9d’ and 9e’). Since an MIC value of 16–32 μg/mL against a test organism is considered a suitable minimum for optimisation,⁷⁹ several tetramates therefore offer suitable starting points for core scaffold elaboration. The selectivity of these tetramates for prokaryotic cells over mammalian cells was analysed via a primary cytotoxicity screening against HaCaT cell line (see ESI). Compounds that are worth progressing from this stage should possess ≥10-fold higher antibacterial activity over cytotoxicity,⁷⁹ and on this basis, carboxamido tetramates 9e, 9f, 9g, 9h, 9l, 9m and 9p are of interest.

The tetramates with C-2 functionalization obtained from SM cross-coupling were assayed for antibacterial activity against Gram-(−) bacterial strains E. coli, K. pneumoniae and P. aeruginosa and Gram-(+) strains MRSA, Enterococcus and Streptococcus pneumonia but no activity was observed. There was, however, potent antibacterial activity against Gram-(+) strains MRSA, Enterococcus and Streptococcus pneumoniae. In particular, compounds 14d, 14h and 14k displayed improved antibacterial activity against MRSA while 14a, 14d and 14h showed improved activity against Enterococcus sp. compared to 9a. The glycosylated tetramates were either inactive or only weakly active (MIC= 250 μg/mL) against Gram-(−) strains tested, but displayed potent antibacterial activity against Gram-(+) strains MRSA, Enterococcus and Streptococcus pneumoniae. Contrary to the trend in antibacterial activity observed for the tetramate esters 8a-g,¹² tetramate ester 16 proved to be more active than its adamantyl analogue 18. Increasing polarity at C-2 as in 17 conferred no bioactivity. Introduction of a disaccharide analogue also led to the complete loss of antibacterial activity of tetramate ester 21. Further, it was interesting to note that the aglycone analogue 26a was slightly more potent than the glycosylated tetramate 16. In the case of glycosyl tetramate 23, in which an acylated galactose is linked to the tetramate via a benzyl ether, improved antibacterial activity was observed compared to 16. Both 23 and its aglycone analogue 26b displayed similar level of antibacterial activity. Thus, while the presence of a glycone does not lead to a significant compromise in bioactivity, it seems that a significant increase in polarity at C-2 leads to a detrimental reduction in the antibacterial activity (e.g. 17). The exact role of the glycosidic residue is not known, but it may assist in improving polarity, bioavailability, modulating toxicity and bacterial target interaction.⁴⁹ Conversion of the tetramate ester 16 to its carboxamide analogues 18 led to a loss in activity. However, the absence of the glycone moiety in these carboxamide tetramates restored antibacterial activity as observed for 28a, indicating that the presence of a glycone is not favourable for antibacterial activity of carboxamide tetramates. This outcome for amides is in contrast to that observed for tetramate esters.

Etherification of 28a to 29a resulted in a 64-fold reduction in antibacterial activity. Ethyl ester hydrolysis of 29a to give carboxylic acid 29b resulted in a further 8-fold reduction in potency (MIC of 29b against MRSA was 250 μg/mL). However, etherification of 28a with 1-bromo-2-butyne afforded 30 without much compromise in antibacterial activity. It was also interesting to note that replacement of the adamantyl side chain with either phenyl (28b) or tetrahydropyranyl (28c) groups led to a reduction in potency, with no antibacterial activity being observed for the latter. An evaluation of adamantyl carboxamide tetramates with aliphatic heterocycles 34b, 37a and 37b appended at C-2 showed good antibacterial activity when compared to the corresponding tetramate esters. This observation is in agreement with the general trend observed for tetramate esters and their carboxamide analogues, highlighting the need for a bulky, hydrophobic C-7 pendant group. In a separate comparison of C-7 adamantyl carboxamides bearing morpholine, pyroglutamate and β-lactam moieties at C-2, morpholine substituted tetramate 37a was the most active while pyroglutamate substituted tetramate 34b was the least potent.

In the absence of a C-2 substituent, antibacterial activity of tetramate 38 against Gram-positive S. aureus is significantly reduced, and to obtain a similar zone of inhibition to that of 9a, a 1000-fold increase in concentration of 38 was necessary. However, upon deprotection, activity against Gram-negative E. coli could be observed; thus, at a concentration of 4 mg/mL, 38 showed activity, while 9a was completely inactive. It is noteworthy that deprotected tetramate 38 does not exhibit significant binding to albumin as its antibacterial activity is retained when treated with HSA. This improvement in activity observed for E. coli could be due at least in part to its increased polarity, which in turn leads to increased cell membrane permeability.

O’Shea and Moser have analysed the major antibacterial classes of compounds and have shown that mean values for physicochemical properties define distinct classes;⁸⁰ antibacterial efficacy does not, however, necessarily correspond to oral bioavailability codified in the Lipinski Rule of 5,⁸¹ primarily as a result of bacterial cell wall penetration.^{82, 83} A correlation of calculated physicochemical properties with the antibacterial activity of some known tetramates was undertaken (using Marvin (16.4.18.0), 2016, ChemAxon (http://www.chemaxon.com)) (Table 11, SI), and their physicochemical properties show general resemblance to fluoroquinolones, with relative PSA values of 13–25% and a MW range of 300–600Da. A similar comparison of physicochemical properties (MW, clogD_7.4 and rel. PSA) and antibacterial activity (where ‘activity’ was defined as MIC of ≤ 8 μg/mL⁸⁰ against at least one bacterial strain) of the tetramate compound library (Figure 8 and Table 11, SI) showed compounds that are active largely occupy physicochemical space with MW of 300–600, clogD_7.4 of −2.5 to 4 and rel. PSA of 11–22; in this, they also generally resemble fluoroquinolones.⁸⁰ The ‘active’ tetramate esters 16 and 23 with glycone pendants appear as outliers due to their higher MW in the region of 650–670. Compounds that are more hydrophilic (that is, with lower clogD_7.4 and higher relative PSA) did not exhibit antibacterial activity against Gram-positive strains, even though the mean MW of known Gram-positive antibacterial agents is around 800.⁸⁰ However, not all polar tetramates that fit in to the optimal chemical space of antibacterials with Gram-negative activity showed the expected activity, although the antibacterial activity of N,S-acetal deprotected tetramate 38 against E. coli suggests suitable polarity improves Gram-negative activity of tetramates, since uptake/penetration through the cell membrane might be better.⁷⁰ Tetramates with antibacterial activity possessed desired physicochemical properties required by Lipinsky’s Ro5, and therefore have potential to be developed as oral drug candidates, but these highly hydropbobic compounds are likely to be penetrative of the blood brain barrier on the basis of calculation of log BB values using a published algorithm.⁸⁴ Furthermore, the Fsp³ value (Fsp³ = sp³ hybridized carbon count/ total carbon count⁸⁵) of the tetramate scaffold 8 is 0.43; Lovering and co-workers have demonstrated that complexity as measured by Fsp³ increased for molecules in clinical progression and that drugs have an average Fsp³ of 0.47.⁸⁵ This places the tetramate scaffold in an ideal region of chemical space for development. Moreover, increased complexity appears to reduce toxicity of drug candidates,⁸⁶ and this was consistent with our observations: for the antibacterially active tetramates for which cytotoxicity against HaCaT mammalian cell line was determined, the mean Fsp³ calculated for tetramates with low toxicity (Fsp³ = 0.41), defined as those that possess ≥10-fold antibacterial activity (against at least one bacterial strain) over cytotoxicity, was higher than those with higher toxicity levels (Fsp³ = 0.38).⁸⁶ No reaction of tetramate 9f with benzylamine and cysteine under ambient conditions was observed, consistent with the lack of toxicity observed in these systems.

Figure 8.

Correlation of physicochemical properties (MW, clogD_7.4, and rel. PSA) with antibacterial activity.

Experimental

General procedure: Esterification of L-serine and DL-cysteine.¹²

To a suspension of the amino acid (1.0 eq) in MeOH (2 mL/mmol) at 0 °C , SOCl₂ (1.2 eq) was added drop-wise under continuous stirring and warmed to rt, then heated at reflux for 1–3 h. The reaction mixture was concentrated in vacuo to obtain the respective amino ester.

General procedure: Synthesis of N-acylated thiazolidines 6,7.¹²

Step 1: To ʟ-cysteine methyl ester hydrochloride (1.0 eq) in petrol (25 mL/1 g), Et₃N (1.2 eq) and aldehyde (1.2 eq) were added. The mixture was heated at reflux, with continuous removal of water using a Dean-Stark apparatus, for 19 h. It was then filtered and washed with Et₂O. The combined filtrates were concentrated in vacuo and residue was purified by silica gel flash column chromatography (eluent: EtOAc/petrol) to give the required thiazolidine.

Step 2: A solution of ethyl hydrogen malonate (1.2 eq) in CH₂Cl₂ (2.5 mL/mmol) was added to a stirred solution of thiazolidine (1.0 eq), DCC (1.2 eq) and DMAP (0.1 eq) in CH₂Cl₂ (5 mL/mmol) at 0°C. The mixture was stirred at 0 °C for 15 min and then at rt for 15 h. The reaction mixture was filtered to remove dicyclohexylurea and the residue was washed with CH₂Cl₂. The combined filtrates were concentrated in vacuo and purified by silica gel flash column chromatography (eluent: EtOAc/petrol) to give N-acylated thiazolidines 6,7.

General procedure: Synthesis of bicyclic tetramates 8.¹²

KO^tBu (1.2 eq) was added to a solution of the N-acylated thiazolidine in THF and heated at reflux for 3 h. It was then cooled to rt, concentrated in vacuo and partitioned between Et₂O and water. The aqueous layer was extracted and acidified with 2M HCl (to pH 1) and extracted with EtOAc. The combined EtOAc extracts were washed with brine, dried over Na₂SO₄, filtered and concentrated in vacuo. The residue was purified by flash column chromatography (with 1% Et₃N) to give the desired product. The product was then dissolved in CH₂Cl₂ and washed with 5% citric acid. The organic fractions were dried over Na₂SO₄, filtered and concentrated in vacuo to yield the desired bicyclic tetramates 8a-g.

General procedure: Synthesis of carboxamide tetramates 9a-g’

To tetramic acid (1.0 eq) dissolved in THF/toluene or DMSO/toluene (1:9, 10 mL/mmol), amine (1.5 eq) was added. The solution was heated at reflux for 16 h, cooled to rt and concentrated in vacuo. The residue was purified by silica gel flash column chromatography (eluent: EtOAc/MeOH/1 % Et₃N). The product was dissolved in CH₂Cl₂ and washed with 5% citric acid. The organic fractions were dried over Na₂SO₄, filtered and concentrated in vacuo to yield the bicyclic carboxamide tetramate.

Conclusion

Thiazolidine derived tetramate esters were readily elaborated to carboxamido tetramates by direct uncatalyzed ester to amide exchange, and conditions were identified which permit skeletal elaboration by Suzuki-Miyaura coupling despite the presence of a cyclic thioether function. While tetramate esters were inactive against both Gram-positive and Gram-negative bacterial strains, the presence of a carboxamide pendant group at C-7 of the tetramate core conferred antibacterial activity against Gram-positive strains; however, the introduction of more polar substituents led to reduced or no activity as observed for 17 and 29b. The antibacterial activity spectrum observed for various C-2 functionalized tetramates demonstrated the ability to accommodate a range of substituents while retaining antibacterial activity, providing scope for physicochemical property optimization, and an analysis of physicochemical properties indicates that bioactive tetramates generally occupy physicochemical space with MW of 300–600, clogD_7.4 of −2.5 to 4 and rel. PSA of 11–22%. With regards to cytotoxicity, compounds with low toxicity had a higher Fsp3 value compared to those with high toxicity. Evaluation of antibacterial activity of tetramates in the presence of HSA suggests possible binding of tetramates to albumin may occur. Although the major tautomeric pair of these species has been identified in solution, it should not be assumed that this is also the binding tautomer at its enzyme target site, since a number of tautomeric forms may be in free equilibrium.

More generally, the identification of novel, non-planar, synthetically accessible heterocyclic systems has become of interest in drug discovery.^{85, 87, 88} The work described here shows that highly functionalised heterocyclic skeletons with several points of diversity, modelled on bioactive natural products, are available in a short synthetic sequence, and that further functional group elaboration may give structures which exhibit similar bioactivity profiles to that of the parent natural product, in this case antibacterial activity. Importantly, these compounds provide well-defined 3D templates which minimise aromatic ring count,⁸⁸ but maintain acceptable starting values of MW, PSA, numbers of rotatable bonds, H-bond acceptors and H-bond donors, while leaving ample scope for lead optimisation in the drug discovery process. They therefore offer suitable skeletons for application in fragment-based drug design^{89, 90} which allow escape “from flatland”.⁸⁵