Antibacterial Barbituric Acid Analogues Inspired from Natural 3-Acyltetramic Acids; Synthesis, Tautomerism and Structure and Physicochemical Property-Antibacterial Activity Relationships

The synthesis, tautomerism and antibacterial activity of novel barbiturates is reported. In particular, 3-acyl and 3-carboxamidobarbiturates exhibited antibacterial activity, against susceptible and some resistant Gram-positive strains of particular interest is that these systems possess amenable molecular weight, rotatable bonds and number of proton-donors/acceptors for drug design as well as less lipophilic character, with physicochemical properties and ionic states that are similar to current antibiotic agents for oral and injectable use. Unfortunately, the reduction of plasma protein affinity by the barbituric core is not sufficient to achieve activity in vivo. Further optimization to reduce plasma protein affinity and/or elevate antibiotic potency is therefore required, but we believe that these systems offer unusual opportunities for antibiotic drug discovery.


Introduction
The use of natural products as leads for antibacterial drug discovery is enjoying a resurgence of interest, forced by the failure of existing drug discovery strategies, the particular requirements of antibacterial therapies, the emergence of virulent bacterial strains and the paucity of new development OPEN ACCESS candidates working their way through the drug development pipeline [1,2]. In this regard, the tetramic acid scaffold (especially with a 3-acyl side chain moiety) [3][4][5][6][7][8] is of particular interest, since naturally occurring 3-acyltetramic acids such as reutericyclin (bacterial membrane disruption) [5], streptolydigin (bacterial RNA polymerase (RNAP) inhibitor) [6], kibdelomycin (bacterial type II topoisomerase inhibitor) [7] and signermycin B (the dimerization domain of histidine kinase WalK inhibitor) [8] all exhibit antibacterial activity with novel modes of action. Aiming to develop both the biological activity and bioavailability of these systems, methodology for the modification of ring substituents in natural 3-acyltetramic acids (R1-R4 in 1a) [9][10][11][12][13][14] as well as the replacement of the 3-acyl side chain group by 3-carboxamide (1b) [13,[15][16][17][18] and 3-enamine functionalities (1c, Figure 1) [13,19] has been reported. These investigations revealed that 3-acyl 1a and 3-carboxamide 1b substitutions can impart good antibacterial activity, resulting from novel modes of action including bacterial membrane disruption, inhibition of bacterial RNAP or undecaprenyl pyrophosphate synthase (UPPS), while 3-enamine 1c exhibits much weaker antibacterial activity and without a clear mode of action. Furthermore, it has been found that the 5-membered tetramic acid core scaffold may also be replaced by the 6-membered piperidine-2,4-dione unit with 3-acyl (1d) [14] and 3-carboxamide (1e) [16,18] pendant functionality, and that these systems show similar antibacterial activity and mode of action compared to tetramates 1a,b. Although 3-acyl and 3-carboxamide tetramate analogues reported in our previous papers [13][14][15] showed good antibacterial activity, novel modes of action and acceptable toxicity, their high plasma protein binding (PPB) affinity interfered with further biological study in vivo for antibiotic drug discovery. In order to overcome this PPB affinity, we decided to conduct further structural optimization of the core scaffold, by moving from the tetramic acid to the 6-membered barbituric acid system. This scaffold possesses similar chemical structure compared to tetramic acids and piperidine-2,4-diones (especially around the C(3) position), but importantly has additional polar functional groups at the 5and 6-positions, and we expected that these groups might help to reduce PPB affinity. Further validation of this proposal comes from the fact that arylidene barbituric acids 1f are reported to have mild antibacterial activity [20][21][22][23]. In this paper, with the inspiration from related analogues 1a-f, novel barbituric acids 2-28 (with 3-acyl), 29-32 (with 3-carboxamide) and 33-37 (with 3-enamine) have been prepared and their tautomeric behavior, antibacterial activity and structure-activity relationships (SARs) have been studied. Furthermore, in order to understand trends of biological activity for further drug optimization, their physicochemical property-activity relationships have been investigated and compared with the tetramic acids reported in our previous papers [13][14][15]19] as well as clinical antibiotics. To the best of our knowledge, the antibacterial activity of 3-acylbarbituric acids with only limited functionality has been reported in the literature [20,21], and the 3-carboxamide and the 3-enamine analogues are as yet completely unreported.

Synthesis
The starting barbituric templates 40a-c were prepared by known methods, while templates 40d,e were commercially available (Scheme 1). Thus, urea 39a-c was condensed with malonic acid in the presence of acetic acid and acetic anhydride to provide barbituric acids 40a-c, respectively [24]. N-Disubstituted 40b,c, along with 3-acetyl 2b,c [20,25] as minor products, respectively, could be purified by flash column chromatography, while N-monosubstituted 40a was best obtained by precipitation in ethyl acetate solution. For this reaction, urea 39c was efficiently obtained from amine 38 and ethyl isocyanate [20,26], while ureas 39a,b were commercially available. With templates 40a-e in hand, the synthesis of 3-acyl, 3-carboxamide and 3-enamine tetramic acids using recently reported approaches were successfully applied to the synthesis of the corresponding barbituric acid analogues [13][14][15]19,27]. 3-Acyl analogues 3-28 (with the exception of 5a) were prepared via direct 3-acylation of templates 40a-e with the required carboxylic acid promoted by 1.1 equivalent of DCC and 1.2 equivalent of DMAP, while stepwise 3-acylation via O-acylation using the acid chloride in the presence of triethylamine followed by acyl migration promoted by DMAP (1.2 equivalent) gave analogue 5a. Although other synthetic methods for 3-acylbarbituric acids have been reported [20][21][28][29][30], this direct acylation approach provides efficient access to systems with a wide variety of substituents at the acyl group. Furthermore, it is also applicable to N-unsubstituted, -mono and di-substituted barbituric acids. 3-Enamines 33-37 were prepared by reaction of the corresponding 3-acyl analogue with an amine in refluxing toluene [25,[31][32][33]. In the case of compound 36, methanol instead of toluene was required as solvent. Alternatively, 3-alkoxycarbonyl barbiturates 42a-d needed as starting materials for 3-carboxamides 29-32 were conveniently prepared from the corresponding barbituric acids by treatment with butyl chloroformate in the presence of 1.2 equivalents of DMAP. Conventional direct amine exchange of the 3-alkoxycarbonlys in toluene allowed preparation of 3-carboxamides 29-32. To the best of our knowledge, this is the first example of the preparation of 3-carboxamidobarbituric acids.

Tautomerism
Similar to tetramic acid derivatives [14,15,19,27], 3-acyl, 3-carboxamide, 3-alkoxycarbonyl and 3-enamine barbituric acids can exist as endoand exo-enol and keto tautomers in solution ( Figure 2). In the case of barbiturates derived from symmetrical barbituric acids 40b,d,e, one set of peaks in their NMR spectra was observed, while in the case of asymmetric barbituric acids 40a,c, split signals (rather than two sets from two tautomeric isomers) were observed. The enol tautomer was assigned from the chemical shift of the C(3)-carbon (80-90 ppm for sp 2 carbon) as well as the absence of the H(3)-proton signal. The observation of one set of signals for endoand exo-enol tautomers supports the fact that equilibration between endoand exo-enol tautomers is fast on the NMR time scale, resulting in coalescence of the peaks of the two enol tautomers.
In order to identify the favored enol-form, the ground state energies of simplified analogues, 3-acyl 2a, 3-carboxamide 43a, 3-alkoxycarbonyl 43b and 3-enamine 43c in endoand exo-enol tautomers were calculated ( Figure 2). The exo-enol form of 3-acyl 2a and endo-enol form of 3-alkoxycarbonyl 43b were found to be more stable than the alternative enol form, and these results are similar to the favoured tautomer of tetramic acids [15,27]. In contrast, 3-carboxamidobarbiturate 43a favours the exo-enol tautomer, while the corresponding 3-carboxamidotetramate favours the endo-enol tautomer [15]. In the case of 3-enamine 43c, the ground state energy of the endo-enol tautomer is much less stable than the exo-enol tautomer (compare the geometry of 3-enamine 43c between endoand exo-enol tautomers in Supplementary Figure 1g,h in the Supporting Information). From this computational result, 3-acyl, 3-carboxamide and 3-enamine barbituric acids all preferentially exist as exo-enol tautomers, and 3-alkoxycarbonyl as endo-enol tautomers. In addition, HMBC NMR spectra of representative analogues were acquired and the correlations for the main ring were established (Supplementary Figure 2 in Supporting Information). In this assignment, the free carbonyl (around 160 ppm) and hydrogen-bonded carbonyl (165-170 ppm) on C(2) and (C4) were readily identified.

Antibacterial Activity
Minimum inhibition concentration (MIC) values for the in vitro in vitro antibacterial activity of 73 barbiturates was determined (shown in Table 1) against Gram-positive bacteria such as Staphylococcus aureus (methicillin sensitive S1, vancomycin susceptible S26, non-resistant S4 and methicillin-resistant in vivo, MRSA, S2), Enterococcus faecalis (vancomycin susceptible, VSE, E1), E. faecium (vancomycin resistant, VRE, E2) and S. pneumonia (erythromycin susceptible P1 and multi drug resistant, MDRSP, P9) as well as Gram-negative bacteria such as Pseudomonas aeruginosa, Escherichia coli (efflux-positive Ec50 and -negative Ec49) and Haemophilus influenzae (efflux-positive H3 and -negative H4). In general, the activity trend for barbiturates is similar to that for tetramates [13][14][15]19,34]; firstly, none of the analogues was active against both P. aeruginosa and efflux-positive and -negative E. coli, (MIC ≥ 32 µg/mL) while the activity against the other strains depended on their ring substituents. Secondly, templates 40a-e, 3-alkoxycarbonyls 42a-d, and O-acyl derivative 41 did not exhibit antibacterial activity against any strains, while the activity of 3-acyls 2-28 ( Figure 3), 3-carboxamides 29-32 and 3-enamines 33-37 ( Figure 4) depended both upon the identity of the bacterial strains as well as their chemical substituents, with 3-acyls and 3-carboxamides tending to be more effective than 3-enamines. Third, N-disubstituted barbituric acids (especially 3-acyls) exhibited excellent activity whereas N-monosubstituted and N-unsubstituted analogues were inactive (see below in detail). These two results reveal that the functional group located on the C(3) position, as seen for tetramates (e.g., 3-acyl and 3-carboxamide), as well as the N-substitution in the barbituric acid templates, are critical factors for the observation of antibacterial activity. Lastly, of particular importance is that, depending on the substituents, the analogues exhibited excellent antibacterial selectivity against resistant and susceptible strains (S1, S26, S2, E1, E2, P1 and P9). By comparison, the activity of ciprofloxacin against MRSA S2 and VRE E2 and amoxicillin against MDRSP P9 dropped more than 50-fold compared to that of the non-resistant strain [15]. In conclusion, 3-acyls 7a,b and 3-carboxamides 32 possessing adamantyl groups exhibited excellent antibacterial activity against Gram-positive strains and Gram-negative H. influenzae (MIC; up to 0.25 µg/mL).  Despite replacement of the tetramic acid with the more hydrophilic barbituric acid core, PPB affinity of barbiturates was only slightly reduced when compared with that of tetramates [13][14][15]19]; overall, MICs of barbiturates against S. pneumonia P9 in the presence of 2.5% horse blood were only slightly shifted from the values without blood (Table 1). For example, (±)-5a,b, 6a,b, 7a,b, 8a,b and 18b,g against S. aureus S26 in the presence of 10% human serum exhibited weak activity (MIC = 64 µg/mL) which were approximately 4-fold worse compared to those without serum. Moreover, it appears that the 3-acyl group might be better than the 3-carboxamide group for PPB binding in this family ((±)-5b; 8 to 64 µg/mL versus 31a; 8 to >64 µg/mL, and 6a; 16 to 64 µg/mL and 7a; 1 to 64 µg/mL versus 32; 1 to >64 µg/mL by 10% human serum).
From the previous finding that 3-acyltetramic acids possessing substituted phenyl groups with a C3-C4 chain length bridge exhibited good antibacterial activity [14], the SAR of 17-19 were studied in detail. It was found that lipophilicity is a crucial factor for cell permeability in the whole-cell antibacterial assay (see below for details). The more lipophilic analogues 17c (compared with 17a,b), 18b (compared with 17a and 18a) and 19a (compared with 18g and 19b) all exhibited better activity. Of particular interest is that di-substituted phenyl 18e (R1 and R4) and 18f,g (R1 and R2) exhibited better activity than di-substituted phenyl 18c (R2 and R3) and 18d (R3 and R4), even though they all possess similar steric effects and lipophilicity. This result indicates that the activity is sensitive to the nature of phenyl substitution (especially at R1 position). In addition, compounds 21-24 possessing polar atoms on the 3-acyl group and/or shorter bridge than 17-19 exhibited poor activity, with the exception of compound 20.

Physicochemical Property-Antibacterial Activity Relationships
Examination of physicochemical property-activity relationships [35], especially in order to understand any trends for bacterial cell permeability, including transportation by efflux pump and PPB affinity, for antibiotic discovery was made. Figure 5 presents a plot of ClogD7.4 against molecular surface area (MSA) of 3-acyl (50 analogues), 3-carboxamide (7 analogues) and 3-enamine (6 analogues) barbituric acids 2-37 along with the corresponding tetramic acids (326 analogues) from our previous reports [13][14][15]19], classified into active (MIC ≤ 4 µg/mL), mild (4 < MIC ≤ 32 µg/mL) and inactive (MIC > 32 µg/mL) analogues. In addition, Supplementary  in the Supporting Information presents the plots of ClogP, polar surface area (PSA) and relative-PSA (rel-PSA = PSA/MSA), respectively (see also Supplementary Table 1 in Supporting Information for physicochemical properties in detail). As shown in Figure 5A and Supplementary Figures 3A-5A, analogues with a wide range of physicochemical properties permeate Gram-positive bacteria whereas the cell permeability of Gram-negative bacteria is limited to a much narrower range of physicochemical properties (with an especially higher threshhold for lipophilicity). This is clearly indicated by the limited activity against efflux-positive H. influenzae H3 shown in Figure 5B and the inactivity against P. aeruginosa and Escherichia coli. Since analogues with a wider range of physicochemical properties exhibit better antibacterial activity against efflux-negative H. influenzae H3 than the efflux-positive strain ( Figure 5B,C), and the fact that tetramic acids exhibited antibacterial activity against TolC negative E. coli and Klebsiella pneumonia in our previous study [13], the main obstacle to Gram-negative bacteria cell permeability appears to transportation by efflux-pumps.  [13][14][15]19] against (A) MRSA; (B) H. influenzae 3 and (C) efflux-negative H. influenzae 4. Active, mild and inactive mean that the values are MIC ≤ 4 µg/mL, 4 µg/mL < MIC ≤ 32 µg/mL and MIC > 32 µg/mL, respectively.
In our previous analysis with tetramates [13,14,19] presented as blue-filled circles in Figure 5B and C and Supplementary Figure 3B,C, tetramates possessing less lipophilic (ClogD7.4 < 3.0 and ClogP < 4.0) and smaller (MSA < 620 Å 3 ) characteristics tended to be less easily transported by efflux pumps in H. influenzae. Although the active barbiturates are in this zone of lipophilicity and MSA, they were slightly more easily transported than tetramates. This may be due to the fact that barbiturates possess higher PSA (78 < PSA < 100 Å 2 ) than tetramates (65< PSA < 90 Å 2 ), are in the same range of MSA (420 < MSA < 650 Å 3 ) and this results in higher rel-PSA (15 < rel-PSA < 20% versus 10 < rel-PSA < 15%). This phenomena agrees with the observation that compounds possessing higher topological PSA are more easily transported by multidrug resistance-associated protein 1 (MRP1/ABBC1) [37].

Physicochemical Property-Plasma Protein Binding Affinity Relationships
In order to investigate physicochemical property-PPB affinity relationships, plots of MIC difference against MSA, PSA, rel-PSA, ClogP and ClogD7.4 of barbiturates (32 analogues) used in this study along with tetramates (208 analogues) from our previous studies [13][14][15]17] were made ( Figure 6). In this analysis, the MIC difference is defined as the value from MIC against S. pneumonia P9 in the presence of 2.5% blood divided by MIC without blood (inactive analogues against any one of these panels were not considered). Since PPB affinity is affected by multiple interactions in many proteins such as human serum albumin, lipoprotein, glycoprotein and globulins in blood, there is no linear correlation between the MIC difference and the physicochemical properties in Figure 6. However, this analysis, which uses a larger number of analogues than in our previous analysis [13,14], shows clearer trends (especially against MSA and PSA). As with the ClogP-PPB affinity relationship in the literature [38,39], PPB affinity is more closely related to lipophilicity as represented by rel-PSA, ClogP and ClogD7.4 ( Figure  6C-E) than MSA and PSA ( Figure 6A,B). It appears that less lipophilic analogues (rel-PSA > 16%, ClogP < 1.5 and ClogD7.4 < −0.69) exhibit higher probability of having lower PPB affinity (MIC difference ≤ 2) while those analogues with high PPB affinity (MIC difference > 30) are positioned in the area of highly lipophilic regions (rel-PSA < 13%, ClogP > 2.9 and ClogD7.4 > 2.9, Figure 6C-E). Although the trends of PPB affinity with steric effect (MSA) and PSA are weaker than lipophilicity, the analogues in the range of MSA between 600 and 900 Å 3 and PSA being less than 100 Å 2 exhibit a higher probability of possessing high PPB affinity ( Figure 6A,B). However, the barbiturate library (red-filled circles) tended to exhibit lower PPB affinity compared to the tetramate library because of their less lipophilic (rel-PSA > 14%, ClogP < 2.0 and ClogD7.4 < 1.0) and smaller MSA (MSA < 650 Å 3 ) character on average.  [13][14][15]17]. The MIC difference is defined as MIC with 2.5% blood/ MIC without blood against S. pneumonia 9.
In fact, potency enhancements of 3-acyltetramic acids [13,14] arising from improvement of lipophilicity and molecular size (ClogD7.4 >2.0 and MSA >600 Å 3 ) are unlikely to provide both lower PPB affinity as well as efflux pump transport. In contrast, we believe that active barbiturates and 3-carboxamide tetramic acids [13,15] exhibit amenable PPB affinity for in vivo activity because of their similar physicochemical properties with clinical antibiotics used for oral or injectable administration as well as being an anionic microspecies under weakly basic conditions (see below). Indeed, the ability to control PPB affinity by adjustment of physicochemical properties proved to be limited; therefore, although appropriate physicochemical properties might be necessary for overcoming PPB affinity, there appear to be other factors involved. One possible hypothesis is that the main core (tetramic and barbituric acids) with its inherent acidic character might be responsible for binding to serum albumin, the major protein in blood, at the sites for aromatic carboxylic acids such as salicylates and ibuprofen. In order to understand whether this is the case, computational and NMR study of serum albumin affinity with our analogues is under investigation.

Physicochemical Property-Activity Relationships of Small Molecule Antibacterial Agents
Although physicochemical properties of antibacterial agents have been discussed in detail in the previous literature [1,40], an examination of the desirable properties for small molecule antibacterial agents, especially to understand cell permeability and PPB affinity, was made using cheminformatic analysis. In order to achieve a more reliable comparison with our library (Mw < 650 Da), antibiotic families with a large molecular weight (Mw > 600 Da), such as glycopeptides which act at the peptidoglycan layer and do not therefore require the penetration of a lipid membrane, and macrolides, have been excluded. In this study, 8 bins for each antibiotic family along with a separate bin for topical antibiotic agents but without consideration of their family, were created. The plots of lipophilicity descriptors (ClogD7.4, and ClogP) and polar surface descriptors (PSA and rel-PSA) against steric effect descriptor (MSA) were made (Figure 7, see also Supplementary Table 2-10 in Supporting Information for the physicochemical properties in detail). In this analysis, most antibiotics, with the exception of topical agents, tended to have a higher limit for lipophilicity (ClogD7.4 < 2.0 and ClogP < 3.0) and a lower limit for polar surface area (PSA > 60 Å 2 and rel-PSA > 13%). Furthermore, it is noteworthy that the bigger antibiotics (MSA > 650 Å 3 ) tend to have a stricter limit for lipophilicity (ClogD7.4 < −4.0 and ClogP < 0) and polar surface area (PSA > 120 Å 2 and rel-PSA > 20%). That these margins might result from PPB affinity is supported by the fact that topical agents, for which PPB affinity is less important, are free from these boundaries, although of course other factors such as intrinsic antibacterial activity and ADMET are clearly important (for example, antibiotics with higher antibacterial activity compensate for higher PPB affinity for oral or injectable use). However, the lipophilicity (ClogD7.4 and ClogP) and the PSA are in inverse proportion to the MSA while the rel-PSA is not affected by the MSA, generally displaying values between 12%-38%. The analogues as shown in Figure 5  From a consideration of their physicochemical properties, on the other hand, the 8 bins for oral or injectable antibiotics can be classified into 3 sub-types. The first one is aminoglycosides populating the most hydrophilic (lowest ClogP and ClogD7.4 and highest PSA and rel-PSA) and the biggest (MSA > 600 Å 3 ) regions. Due to this hydrophilicity (ClogD7.4 < −12), their oral administration has been generally limited as a result of problems with absorption. The second class includes Gram-negative active β-lactams and tetracyclines, and this class of antibiotics possesses lipophilicity and MSA between aminoglycosides and the third class of antibiotics. The third class includes antibiotics of natural origin such as Gram-positive only active penicillins and amphenicols as well as synthetic origin antibiotics such as fluoroquinolones, sulfa drugs and oxazolidinones, possessing higher lipophilic character and smaller MSA than both of the other classes. Of particular interest is that the third class displays a narrow zone of physicochemical properties (−3.0 < ClogD7.4 < 2.0; 0 < ClogP < 3.0; 60 < PSA < 120 Å 2 , 270 < MSA < 650 Å 3 ) with the exception of rel-PSA, while the two other classes populate a wider range of the physicochemical space, with the preference for lower lipophilicity and bigger MSA than the third class. The physicochemical space populated by the third class is amenable to general drug design, and appears to be the most suitable for antibiotic development. As shown in Figure 5 and Supplementary Figures 3-5, our active barbiturates in this study and 3-carboxamide tetramic acids from the previous study [13,15] also satisfy these properties, in contrast to active 3-acyltetramic acids which generally possess higher ClogP and ClogD7.4 values [13,14]. From a consideration of MSA of the third class of antibacterial agents (MSA < 650) as well as from the fact that smaller molecules (MSA < 620 Å 3 ) tend to be less easily transported by the efflux pump in H. influenzae as described above ( Figure 5B,C), and that analogues with 600 < MSA < 900 Å 3 exhibit a higher probability of higher PPB affinity ( Figure 6A), both lipophilicity and molecular size might be crucial factors for antibiotic discovery in which smaller analogues (MSA < 600 Å 3 ) are likely to have a benefit for both PPB affinity and cell permeability.

Ionic State of Small Molecule Antibacterial Agents
Since for bacterial cell permeability (including the transportation by efflux pump), ionic state (related to pKa) might be expected to be a major factor, the major microspecies at pH 7.4 were calculated (data not shown, see Supplementary Figures 6-14 for the structures of antibacterial agents). Of particular interest is that Gram-negative active agents favoured anionic(s) and zwitterionic(s) microspecies, while aminoglycosides exist as cationic forms, generally as a result of the presence of the amine functionality. Fluoroquinolones, tetracyclines and Gram-negative active β-lactams exist as various forms of zwitterionic and anionic microspecies. This might result from the fact that their key skeletons include an acidic carboxylic acid (fluoroquinolones and β-lactams) or two acidic enols and a basic amine group (tetracyclines), which make them easily able to form anionic or zwitterionic species under weakly basic conditions (pH = 7.4). In addition, sulfa drugs mainly exist as an anionic form on the nitrogen in the sulfone amide group, in some cases with a minor amount of the uncharged form. In this family, sulfaguanidine (used in the treatment of gastrointestinal infections) exceptionally exists as a neutral microspecies under weakly basic conditions but is cationic in acidic conditions (pH = 1.0). In addition, amphenicols also exist as a mixture of a neutral and an anionic (on the nitrogen of amide functionality) microspecies. On the other hand, Gram-positive only active antibiotics generally exist as an anionic (rather than a zwitterionic, Gram-positive only β-lactams), neutral (oxazolidinones) or a mixture of a neutral and cationic (lincosamides and 2,4-diaminopyrimidines) microspecies, and tend to be less charged state than Gram-negative active agents. In general, Gram-negative antibacterial agents tend to be more charged than the Gram-positive only agents to penetrate outer membrane via porins as well as prevent efflux.

General
All starting materials and 3-acyl 2a were commercially available. Melting points were checked by STUART SCIENTIFIC SMP1. The 1 H-, 13 C-and HMBC-NMR spectra were obtained using a Bruker AVB500 (500 MHz for 1 H-NMR and 126 MHz for 13 C-NMR) or DPX400 (400 MHz 1 H-NMR and 101 MHz for 13 C-NMR). Mass spectra (MS) and high resolution MS were obtained with Micro Mass LCT and GCT spectrometers under the conditions of electrospray ionization (ESI) and chemical ionization (CI) respectively, and values were reported as a ratio of mass to charge in Daltons. The energies at ground state were computed with Density Functional B3LYP (6-31G*) in Spartan 02 in each enol form and the physicochemical properties were calculated with MarvinSketch version 5.3.8. in which VG method for ClogP and ClogD7.4, van der Waals method for MSA were selected and sulfur atoms were excluded in the calculation of PSA. In vitro antibacterial activity was performed using standard methodology as described in our previous paper [13].

Synthesis of Urea 39c
To the solution of 2-methoxy ethylamine 38 (3.2 g, 42.2 mmol) in dichloromethane (50 mL) was slowly added ethyl isocyanate (3.0 g, 42.2 mmol) at 0 °C under nitrogen atmosphere and the mixture was stirred for 30 min. Concentration in vacuo followed by precipitation in ethyl acetate and petrol solution gave urea 30c (5.9 g, 40.2 mmol, 95% yield) as a solid (M.P.; 40.0 °C).

Synthesis of Barbituric Acid 40a
To the solution of N-allylurea 39a (2.0 g, 20.0 mmol) in acetic acid (60 mL) was added malonic acid (2.1 g, 20.0 mmol) at room temperature. The mixture was slowly added acetic anhydride (40 mL) at 60 °C for 30 min and the mixture was stirred at 90 °C for 5 h. Concentration in vacuo followed by precipitation in ethyl acetate gave barbituric acid 40a (960 mg, 5.71 mmol, 30%) as a solid (M.P.; 158 °C).

Synthesis of 3-alkoxycarbonyl Barbituric Acid Templates 41a-d
General procedure; To the solution of barbituric acid (1 eq) and DMAP (2.2 eq) in dichloromethane was slowly added butyl chloroformate (1.2 eq) at 0 °C under nitrogen atmosphere, and the mixture was stirred overnight at room temperature under nitrogen atmosphere. Then the mixture was washed with 2 M HCl. The organic layer was dried over MgSO4 and evaporated in vacuo to give 3-alkoxycarbonyl barbituric acid templates 41a-d contained about 10%-20% impurity. The impure 3-alkoxycarbonyl barbituric acid template was used for next step without further purification.

Synthesis of Compound 41
To the mixture of barbituric acid 40d (1.0 g, 6.40 mmol) and triethylamine (0.78 g, 7.70 mmol) in dichloromethane (50 mL) was slowly added 3,5,5-trimethylhexanoyl chloride (1.20 g, 6.72 mmol) at 0 °C under nitrogen atmosphere and the mixture was stirred for 3 h at room temperature. After completion of the reaction, the mixture was washed with 2 M HCl. The organic layer was dried over MgSO4 and evaporated in vacuo to give crude product. Further purification was carried out by recrystallization in the mixture of ethyl acetates and petrol giving pure compound 41 (1.56 g, 5.37

Synthesis of Compound (±)-5a
To the solution of compound 41 (500 mg, 1.69 mmol) in dichloromethane (30 mL) was added DMAP (250 mg, 2.02 mmol) at room temperature and the mixture was stirred overnight. After completion of the reaction, the mixture was washed with 2 M HCl and the organic layer was evaporated in vacuo. Short activity in vivo, although the physicochemical properties and ionic state are similar to antibiotic agents for oral and injectable use. Further optimization to reduce PPB affinity or elevate antibiotic potency is therefore required. On the other hand, physicochemical property-activity relationship analysis of clinical antibiotics shows that oral and injectable antibiotics may have a higher margin for lipophilicity depending on molecular size (MSA) in order to achieve low PPB affinity, while topical antibiotics are free from such constraints. In addition, Gram-negative active antibiotics tended to exist as more ionized microspecies than Gram-positive antibiotics to penetrate outer membrane. We believe that these systems offer unusual opportunities for antibiotic drug discovery, and these results suggest that a strategy based upon the use of natural products is viable [41,42], and highlights the importance of continuing development of methodologies to access tetramate-like systems [43]. The work contained here-in, and another recently reported study [44], indicate that barbiturates offer a core template of promise and worthy of further investigation.