Molecular Diversity via Tetrasubstituted Alkenes Containing a Barbiturate Motif: Synthesis and Biological Activity

Abstract

The synthesis of a molecularly diverse library of tetrasubstituted alkenes containing a barbiturate motif is described. Base-induced condensation of N¹-substituted pyrimidine-2,4,6(1H,3H,5H)-triones with 5-(bis(methylthio)methylene)-2,2-dimethyl-1,3-dioxane-4,6-dione gave 3-substituted 5-(methylthio)-2H-pyrano[2,3-d]pyrimidine-2,4,7(1H,3H)-triones (‘pyranopyrimidinones’), regioselectively. A sequence of reactions involving ring-opening of the pyran moiety, displacement of the methylthio group with an amine, re-formation of the pyran ring, and after its final cleavage with an amine, gave tetrasubstituted alkenes (3-amino-3-(2,4,6-trioxotetrahydropyrimidin-5(2H)-ylidene)propanamides) with a diversity of substituents. Cleavage of the pyranopyrimidinones with an aniline was facilitated in 2,2,2-trifluoroethanol under microwave irradiation. Compounds were tested against Escherichia coli, Staphylococcus aureus, the yeast Schizosaccharomyces pombe, and the pathogenic fungus Candida albicans. No compounds exhibited activity against E. coli, whilst one compound was weakly active against S. aureus. Three compounds were strongly active against S. pombe, but none was active against C. albicans.

1. Introduction

Invasive fungal infections (IFIs) cause the deaths of about 1.5 million people every year [1] and are particularly prevalent in intensive care units, affecting elderly and/or immunocompromised patients. As with antibacterial agents [2], resistance to current antifungal drugs is a significant clinical problem [3]. The relatively few clinically useful antifungal drugs include inhibitors of ergosterol biosynthesis (e.g., azoles), the polyene macrolide antibiotics compromising ergosterol function (e.g., natamycin), and echinocandins (e.g., caspofungin) affecting cellular β-glucan synthesis. These compounds suffer from several liabilities, including fungal resistance (azoles, polyenes), lack of broad-spectrum activity (echinocandins), and toxicity (polyenes). There is a clinical need for improved antifungals, which either represent new structural classes or are novel members of existing classes [4].

The creation of molecular diversity [5,6] and privileged scaffolds [7] leading to new ‘drug-like molecules’ is an important direction of research with the promise that entirely new agents will arise. We have found a strategy for the construction of tetrasubstituted alkenes (derivatives 1 of 3-amino-3-(2,4,6-trioxotetrahydropyrimidin-5(2H)-ylidene)propanoic acid, Figure 1) containing a barbiturate motif. These compounds were obtained via intermediate 2H-pyrano[2,3-d]pyrimidine-2,4,7(1H,3H)-triones. The compounds described have been tested against selected organisms, revealing promising antifungal activity.

Our strategy for the construction of tetra-substituted alkene 1 permits the systematic incorporation of structural units from amine nucleophiles, giving a variety of products. This approach enables molecular diversity to be explored by judicious selection of starting materials leading to structural variations at up to five points of appendage diversity within two distinct scaffolds. A focussed library of compounds has been generated based on the structure of derivative 1 that varies the group at each of R¹, R², R³, R^4, and R⁵ with respect to the parameters of group size and polarity. Jeong and Moloney [8] and others [9,10] have recently described analogous compounds with an antibacterial activity that are dissimilar in structure from those described here and accessed synthetically in a different manner. Notably, novel barbituric acids exhibited antibacterial activity, especially against resistant Gram-positive strains, such as Staphylococcus aureus (methicillin-resistant, MRSA), Enterococcus faecalis (vancomycin-susceptible, VSE), E. faecium (vancomycin-resistant, VRE), and Streptococcus pneumonia (multidrug-resistant, MDRSP) [8]. A major difference from the previous studies is that our synthetic approach directly produces esters and amides corresponding to structure 1, avoiding the parent carboxylic acid, which is expected to be prone to decarboxylation.

2. Results

2.1. Synthesis of Pyranopyrimidinones 2a–e

Regioselective condensation between readily available barbiturate derivatives (substituted pyrimidine-triones, 3) and compound 4 [11,12] gave pyranopyrimidinones 2 (Scheme 1). When R¹ = alkyl (Me 2c, Et 2b) or substituted alkyl (benzyl 2a, 4-fluorobenzyl 2d) and R² = H, one regioisomer was formed almost exclusively, as validated by crystal structure analysis for compound 2c (Appendix A, Figure A1). Reactions of compound 4 with amines displacing one or both methylthio groups proceed by Michael addition, followed by the elimination of methanethiol [11,13,14].

2.2. Synthesis of Alkenes 5a–5l

Heating 2e with benzylamines in ethanol gave alkenes 5a–d (Scheme 2). Crystal structure analysis of 5c confirmed its structure (Appendix A, Figure A2), which shows there is an intramolecular hydrogen bond between the amino NH and neighboring pyrimidine trione carbonyl group. The amide proton of 5c forms an intermolecular hydrogen bond to the amide carbonyl of an adjacent molecule (See Appendix A for a fuller description of this structure).

With the less nucleophilic anilines, it was preferable to proceed via oxidation of 2e by m-chloroperbenzoic acid to 6, which on heating with an aniline derivative in ethanol gave alkenes 5f–l (Scheme 3).

2.3. Synthesis of Compounds 8a–f, 9a–f and 10a–q

Given the failure to effect stepwise substitution of the methylthio group and cleavage of the pyrone moiety when 2d or 2e were treated with amines (cf. Scheme 2), a different approach was adopted. Thus, heating compound 2a or 2b with sodium methoxide in methanol afforded methyl esters (7a,b). Compound 7 was reacted with an amine in ethanol to give alkenes 8a–f. Heating 8 with NaOH in methanol gave pyranopyrimidinones 9a–f. A solution of 9 and the corresponding amine in 2,2,2-trifluoroethanol (TFE) was subjected to microwave irradiation to give the target alkenes 10a–q (Scheme 4). The structures and purities of the newly synthesized compounds were confirmed by ¹H/¹³C-NMR spectroscopy and liquid chromatography-mass spectrometry, with further validation by crystal structure analysis in some cases (Appendix A, Figure A3, Figure A4 and Figure A5).

The structures of 7b, 9c, and 10b were determined by single-crystal X-ray crystallography (Figure A3, Figure A4 and Figure A5; see Appendix A for a fuller description of the structures) and were representative of the compound groups 7, 9, and 10, respectively. As observed with 5c, an amino (NHR³) to carbonyl intramolecular hydrogen bond was apparent in 9c and 10b, forming a 6-membered ring. The configuration of the C=C bond in crystalline 7b was the Z isomer, arising from nucleophilic attack by methoxide on the pyrone ring carbonyl of 2b. Although compound 10b exhibited Z configuration in its crystal structure, the ¹H-NMR in d₆-dimethylsulfoxide at ambient temperature showed that this compound and its congeners are mixtures of E and Z isomers in solution, the resonances from which coalesced at 90 °C.

2.4. Biological Assays

Disc bioassays were performed on compounds 5a–l, 8a–f, 9a–f, and 10a–q against the Gram-negative bacterium Escherichia coli (DH5α), the Gram-positive bacterium Staphylococcus aureus (RN4220), the unicellular eukaryotic yeast Schizosaccharomyces pombe (SAK950) and the human pathogenic fungus Candida albicans ATCC 90028. The most interesting results against S pombe are summarised in

Table 1. Selected bioassay results against Schizosaccharomyces pombe.

Compound	Average Zone of Inhibition (Radius, mm)
5b	1.5
5c	2
8b	1
8c	1.5
8e	0.5
9b	10
9c	5
9e	2.5
9f	3
10a	0.5
10b	5
Nystatin	8

. Scanned images of bioassay plates are shown in Appendix B (Figure A6).

As shown in Table 1 and Figure A6, compounds 9b, 9c, and 10b were strongly active against S pombe, with 9b being comparable to nystatin. Significant inhibition was also observed for compounds 5b, 5c, 8b, 8c, 8e, 9e, 9f, and 10a. However, compounds 8a, 8d, 8f, 9a, 9d, 5a, 5d–l, and 10c–q were all inactive. None of the compounds was active against C. albicans. No compounds exhibited activity against E. coli, and only one compound (10a) was active, albeit weakly, against S. aureus (data not shown).

3. Discussion

An efficient route (Scheme 4) is described to afford a diverse library of alkenes 10 via esters 8 and pyranopyrimidinones 9. Compounds derived from 8 were readily obtained from 4. The route enables systematic variation of up to five substituents (R¹, R², R³, R⁴, and R⁵, compared to structures in 10). Experimental details for the synthesis of all compounds described is given in Appendix C.

For the reactions of 3 with 4, where one possible regioisomer was formed preferentially, it is proposed that the carbanion derived from deprotonation of compound 3 at C-3 displaces one methylthio group to give an intermediate (Scheme 5), which can cyclise with loss of acetone, followed by CO₂, to give 2a–d or the regioisomer where R¹ and R² are interchanged. The preference for the regioisomer shown may be due to the reacting oxygen atom near to R² (= H) being less sterically shielded than the corresponding oxygen near to R¹ (= alkyl).

For efficient reactions of pyranopyrimidinones 9a–f with amines to afford alkenes 10a–q (Scheme 4), it was expedient to use TFE as a solvent with microwave irradiation. Previous studies have highlighted the advantages of TFE for reactions where its exceptional solvating and hydrogen bond donor properties are beneficial, as well as its suitability for microwave irradiation [15,16]. No reaction was observed when dimethylformamide was used as the solvent in attempts to convert compound 9 or its derivatives into 10.

Biological evaluation of the compounds (see Table 1 and Figure A6, Appendix B) identified 9b, 9c, and 10b had significant hits against S. pombe, with 9b similar to the potency of nystatin. The S. pombe strain, developed for screening purposes, is a mutant strain with seven drug exporter genes deleted to make it hypersensitive to challenge with toxins [17,18]. It is interesting that the substituents R¹, R^2, and R³ were the same for 8b, 9b, and 10a, and for 8c, 9c, and 10b. The difference between the compounds 8b/8c and 10a/10b is a methyl carboxylate in compounds 8b/8c versus a morpholino-amide in 10a/10b. Compounds 8b/8c, 9b/9c, and 10a/10b differ solely by a methyl group in one versus a methoxy in the other. No compound where R¹ = Et showed comparable activity to the best benzyl-substituted compounds. Comparing results for compounds 8, 9, and 10 indicates that substituents R¹ and R³ were the primary drivers for potency. For compounds 10, only when R⁴ contained a morpholino moiety was any activity observed. Further studies will be aimed at identifying the biological target for these compounds and extend structure-activity relationships by further exploring the parameters of group size, polarity, and hydrogen bonding capability. The ultimate aim is to identify lead compounds against pathogenic fungi.

4. Materials and Methods

4.1. General

All the materials used were purchased from Sigma-Aldrich (Merck Life Science UK Limited, Gillingham, Dorset, UK), Acros Organics (Fisher Scientific, Loughborough, Leicestershire, UK), or Fluorochem (Hadfield, Derbyshire, UK) and used without further purification. All reactions were monitored by thin-layer chromatography on 0.25 mm silica gel plates (60GF-254) and visualized with UV light. ¹H and ¹³C-NMR spectra were recorded on a Bruker Avance300 spectrometer (Bruker UK Limited, Banner Lane Coventry CV4 9GH, UK) operating at 300.13 and 75.48 MHz, respectively, or a Bruker Avance400 spectrometer (Bruker UK Limited, Banner Lane Coventry CV4 9GH, UK) operating at 399.78 and 100.54 MHz, respectively, using TMS as an internal standard in DMSO-d₆ or CDCl₃ solutions. Chemical shifts were reported in delta (δ) units, parts per million (ppm) downfield from tetramethylsilane. LC-MS analyses were performed on a Shimadzu 2010EV (Shimadzu UK Limited, Buckinghamshire MK12 5RD, UK) instrument operating in negative (-ve) electrospray ionization (ESI) mode with UV detection at 254 nm.

Crystal structure data were collected at 150 K on a Rigaku Oxford Diffraction Xcalibur Atlas Gemini Ultra diffractometer equipped with a sealed tube X-ray source (λ_CuKα = 1.54184 Å) and an Oxford CryostreamPlus open-flow N₂ cooling device. Intensities were corrected for absorption using a multifaceted crystal model created by indexing the faces of the crystal for which data were collected [19]. Cell refinement, data collection, and data reduction were undertaken via the software CrysAlisPro (Rigaku Oxford Diffraction, Tokyo, Japan).

All structures were solved using XT [20] and refined by XL [21] using the Olex2 interface [22]. All non-hydrogen atoms were refined as anisotropic, and hydrogen atoms were positioned with idealized geometry, with the exception of those bound to heteroatoms, the positions of which were located using peaks in the Fourier difference map. The displacement parameters of the hydrogen atoms were constrained using a riding model with U_H set to be an appropriate multiple of the U_eq value of the parent atom.

4.2. Synthesis

See Appendix A.

4.3. Biological Assays

The bacterial strains (Escherichia coli (DH5α), the Gram-positive bacterium Staphylococcus aureus (RN4220)) were grown in Luria Bertani (LB) medium (10 mL) overnight and then diluted 1:50 in fresh LB medium and grown for 6 h. S pombe was grown in YE6S medium overnight and then diluted 1:10 in fresh YE6S medium and grown for 6 h. C albicans was grown in Sabourand medium overnight and then diluted 1:10 in fresh Sabourand medium and grown for 6 h, then diluted 1:10 in fresh medium and the diluted culture was used in the assay.

For all organisms, 400 µL of culture was spotted onto a solid medium in a standard square plate and spread evenly over the surface by shaking with sterile glass beads. For the bacterial strains, Oxoid Nutrient Agar plates were used, whilst for S. pombe YE6S agar plates were used, and for C. albicans, Sabourand agar plates were used. Filter paper discs to which 5 µL of each test compound (10 mg/mL in DMSO, i.e., ~0.02 M for each compound) were placed on the plates, which were incubated overnight at 30 °C.

Control discs were used where 5 µL of the positive controls rifampicin (1 mg/mL in DMSO) and nystatin (50 mg/mL in DMSO, 0.054 M) were added.

After incubating overnight, the plates were inspected for a halo (zone of inhibition) surrounding the discs, and the plates were scanned. The S. pombe bioassay plates were incubated for a further 24 h and then reassessed and scanned. Owing to the slower growth of S. pombe (~3–4 h doubling time) compared to the bacterial strains (20–30 min doubling time) and the more vigorous C. albicans, a longer incubation period was required to produce a confluent lawn of S. pombe cells.

5. Conclusions

Routes are described herein affording a diverse library of tetrasubstituted alkenes 3-amino-3-(2,4,6-trioxotetrahydropyrimidin-5(2H)-ylidene)propenamides via pyranopyrimidinones (Scheme 1, Scheme 2, Scheme 3 and Scheme 4). Biological assays against selected bacterial and fungal organisms revealed significant activity for compounds 9b, 9c, and 10b against the unicellular eukaryotic yeast Schizosaccharomyces pombe, with compound 9b matching the potency of nystatin. This study has revealed novel structural motifs targeting the fungal organism Schizosaccharomyces pombe and provides the groundwork for identifying a lead against pathogenic fungi.