Development of Dicationic Bisguanidine-Arylfuran Derivatives as Potent Agents against Gram-Negative Bacteria

Antibiotic resistance among bacteria is a growing global challenge. A major reason for this is the limited progress in developing new classes of antibiotics active against Gram-negative bacteria. Here, we investigate the antibacterial activity of a dicationic bisguanidine-arylfuran, originally developed as an antitrypanosomal agent, and new derivatives thereof. The compounds showed good activity (EC50 2–20 µM) against antibiotic-resistant isolates of the Gram-negative members of the ESKAPE group (Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter spp.) and Escherichia coli with different antibiotic susceptibility patterns, including ESBL isolates. Cytotoxicity was moderate, and several of the new derivatives were less cytotoxic than the lead molecule, offering better selectivity indices (40–80 for several ESKAPE isolates). The molecular mechanism for the antibacterial activity of these molecules is unknown, but sensitivity profiling against human ESKAPE isolates and E. coli collections with known susceptibility patterns against established antibiotics indicates that it is distinct from lactam and quinolone antibiotics.


Introduction
Bacterial resistance to antibiotics currently represents one of the biggest threats to global public health with more than 1 million people worldwide dying each year because of drug-resistant infections [1]. For this reason, the design and development of new classes of antibiotics are essential. Gram-negative bacteria have the highest resistance indices of all pathogenic bacteria, and development of new antibiotics to tackle them is urgently needed [2]. However, a recent report from the World Health Organization (WHO) [2] reveals a weak pipeline for antibiotics. The 60 products currently in development (50 antibiotics and 10 biologics) provide little advantage over existing treatments, and very few target Gram-negative bacteria.
The guanidine group, an organic base which is hydrophilic in nature, is commonly found in biologically active compounds, including antibiotics [3]. At physiological pH, the guanidine moiety is positively charged. The presence of this charge may lead to an electrostatic interaction between positively charged guanidine-containing compounds and, e.g., the negatively charged bacterial cell surface [3]. This immediate binding to the components of the cytoplasmic membrane or the cell wall causes the loss of biological functions of phospholipids, which can result in reduced membrane integrity. The resulting increase in membrane permeability leads to lysis and cell death [4]. Furthermore, the presence of a positive charge in guanidine derivatives may favor binding to intracellular targets, e.g., the minor groove of DNA [5].
2 functions of phospholipids, which can result in reduced membrane integrity. The resulting increase in membrane permeability leads to lysis and cell death [4]. Furthermore, the presence of a positive charge in guanidine derivatives may favor binding to intracellular targets, e.g., the minor groove of DNA [5].
The Pathogen Box (Medicines for Malaria Venture) is a further development of the Malaria Box collection [6]. This collection has been successfully used, e.g., to screen for small molecules active against pathogenic mycobacteria [7]. In the Pathogen Box, the dicationic 2,5-bis(2-chloro-4-guanidinophenyl)furan (1, MMV688179, Figure 1) caught our attention because it has two guanidinium groups in its structure [8]. The compound has an affinity for A/T-rich DNA [8]. Antifungal, antimycobacterial [8], and antiparasitic [9,10] activity has been reported for this compound, and it has been found to be active against the Gram-negative bacterium Burkholderia pseudomallei [11], the causative agent of the tropical disease melioidosis. However, its activity against E. coli and other pathogenic Gram-negative bacteria from the ESKAPE group, commonly causing serious infections around the world and exhibiting troublesome antibiotic resistance patterns, has not previously been examined. A variety of antimicrobial guanidine-containing compounds have been reported [3], targeting the bacterial envelope [12][13][14] or key bacterial proteins, such as DNA gyrase [15], lipid A and fatty acid biosynthesis enzymes [16], the bacterial cell division protein FtsZ4 [17], the NorA efflux pump [18], or targeting the ribosomal decoding rRNA site [19]. Because 1 has two guanidinium groups in its structure (Figure 1), we anticipate that bis-guanidine dicationic compounds bearing an arylfuran framework could be novel antibacterial agents. In this study, we describe the design, synthesis, and antibacterial evaluation of a series of dicationic derivatives with general structure A ( Figure 1). The compounds were tested against a panel of five laboratory strains (one Gram-positive and four Gram-negative) and ten different Gram-negative bacteria isolates of human origin, representing E. coli and the ESKAPE group of species. In addition, we determined the cytotoxicity against MCF-7 and HepG2 human cell lines. To broadly characterize the mechanism of action of the new substances, we performed high-resolution microbial phenomics profiling of selected compounds and the known antibiotic cefotaxime (CTX) against two E. coli libraries (ECOR and ESBL) with characterized susceptibility patterns against established antibiotics.

Chemistry
A set of compounds was prepared to address the key structure-activity relationships of the bis-arylfuran scaffold. Different substituents of the phenyl ring were investigated, as well as an isostere of the guanidine group. Asymmetric compounds were also explored ( Figure 2). The first of these was a series of 2,5-bis(4-guanidino-aryl)furan derivatives Because 1 has two guanidinium groups in its structure (Figure 1), we anticipate that bis-guanidine dicationic compounds bearing an arylfuran framework could be novel antibacterial agents. In this study, we describe the design, synthesis, and antibacterial evaluation of a series of dicationic derivatives with general structure A ( Figure 1). The compounds were tested against a panel of five laboratory strains (one Gram-positive and four Gram-negative) and ten different Gram-negative bacteria isolates of human origin, representing E. coli and the ESKAPE group of species. In addition, we determined the cytotoxicity against MCF-7 and HepG2 human cell lines. To broadly characterize the mechanism of action of the new substances, we performed high-resolution microbial phenomics profiling of selected compounds and the known antibiotic cefotaxime (CTX) against two E. coli libraries (ECOR and ESBL) with characterized susceptibility patterns against established antibiotics.

Chemistry
A set of compounds was prepared to address the key structure-activity relationships of the bis-arylfuran scaffold. Different substituents of the phenyl ring were investigated, as well as an isostere of the guanidine group. Asymmetric compounds were also explored ( Figure 2). The first of these was a series of 2,5-bis(4-guanidino-aryl)furan derivatives which were synthesized (Scheme S1) using the corresponding di-amino compounds 1b and 3b-9b as common precursors.
Initially, two different methodologies were tested to obtain 2,5-bis(4-nitroaryl)furans (1a, 3a-9a): (1) a Suzuki cross-coupling reaction using the corresponding furan-2,5-diboronic acid pinacol ester with a substituted aryl bromide, and (2) a direct palladium-catalyzed arylation using furan with a substituted aryl bromide. In both cases, monoaryl furan was the major product obtained. This led us to use the synthetic approach previously reported by Stephens et al. [8,20] with some modifications. The synthesis of the amino compounds was achieved in two steps (Scheme S1). Firstly, a Stille coupling using 2,5-bis(trin-butylstannyl)furan and a substituted 4-bromonitroarene was performed to form the corresponding 2,5-bis(4-nitrophenyl) furans (1a, 3a-9a) in good to excellent yields (45-80%). From these intermediates (1a, 3a-9a), the nitro compounds were then reduced using iron powder with ammonium chloride to obtain the desired diamino compounds (1b, 3b-9b) in excellent overall yields. The final transformation to obtain the target was carried out in two steps (Scheme S1). The diamines were first reacted with Boc-protected S-methylthiourea in the presence of mercuric chloride (1c, 3c-9c), followed by Boc-deprotection of the guanidine derivatives using 4 M HCl in dioxane. Ultimately, a good overall yield of the target compounds (1, 3-9) was obtained.
To investigate whether the activity is modulated by the cationic moiety when it is not directly attached to an aromatic system, we modified 3 by adding an additional carbon atom to extend the space between the aryl group and the guanidine moiety. This required the synthesis of the diamine 10b (Scheme S2), which was prepared in a two-step process involving palladium-catalyzed direct arylation using furan with 4-bromobenzonitrile, followed by reduction of the cyano group by treatment with lithium aluminum hydride. The bis-guanidine compound 10 was prepared (Scheme S2) using the same synthetic strategy used for 1.
Next, we exchanged the guanidinium residue of the bis-arylfuran scaffold for a bio-isostere. Here we chose to use the squaryldiamide moiety, as this group has been identified as a new potential bioisostere for unsubstituted guanidine functionality in peptidomimetics [21]. The 1,2-diaminocyclobutene-3,4-dione (squaryldiamide) derivatives were prepared using the diamine compounds through two synthetic steps (Scheme S3). Initially, two different methodologies were tested to obtain 2,5-bis(4-nitroaryl)furans (1a, 3a-9a): (1) a Suzuki cross-coupling reaction using the corresponding furan-2,5-diboronic acid pinacol ester with a substituted aryl bromide, and (2) a direct palladium-catalyzed arylation using furan with a substituted aryl bromide. In both cases, monoaryl furan was the major product obtained. This led us to use the synthetic approach previously reported by Stephens et al. [8,20] with some modifications. The synthesis of the amino compounds was achieved in two steps (Scheme S1). Firstly, a Stille coupling using 2,5-bis(trin-butylstannyl)furan and a substituted 4-bromonitroarene was performed to form the corresponding 2,5-bis(4-nitrophenyl) furans (1a, 3a-9a) in good to excellent yields (45-80%). From these intermediates (1a, 3a-9a), the nitro compounds were then reduced using iron powder with ammonium chloride to obtain the desired diamino compounds (1b, 3b-9b) in excellent overall yields. The final transformation to obtain the target was carried out in two steps (Scheme S1). The diamines were first reacted with Boc-protected S-methylthiourea in the presence of mercuric chloride (1c, 3c-9c), followed by Boc-deprotection of the guanidine derivatives using 4 M HCl in dioxane. Ultimately, a good overall yield of the target compounds (1, 3-9) was obtained.
To investigate whether the activity is modulated by the cationic moiety when it is not directly attached to an aromatic system, we modified 3 by adding an additional carbon atom to extend the space between the aryl group and the guanidine moiety. This required the synthesis of the diamine 10b (Scheme S2), which was prepared in a two-step process involving palladium-catalyzed direct arylation using furan with 4-bromobenzonitrile, followed by reduction of the cyano group by treatment with lithium aluminum hydride. The bis-guanidine compound 10 was prepared (Scheme S2) using the same synthetic strategy used for 1.
Next, we exchanged the guanidinium residue of the bis-arylfuran scaffold for a bioisostere. Here we chose to use the squaryldiamide moiety, as this group has been identified as a new potential bioisostere for unsubstituted guanidine functionality in peptidomimetics [21]. The 1,2-diaminocyclobutene-3,4-dione (squaryldiamide) derivatives were prepared using the diamine compounds through two synthetic steps (Scheme S3). The diethyl squarate was first treated with the corresponding diamine compounds (1b, 3b-5b) to displace one ethoxy residue and produce the corresponding intermediates 11a-14a. These intermediates were then treated with ammonia to displace the second ethoxy group and produce the squaryldiamide derivatives 11-14 in good overall yields.
To modulate the activity of the bis-arylfuran scaffold, we synthesized asymmetric furan derivatives, replacing one of the aromatic rings with a residue containing an amide attached to an aromatic heterocycle. A series of 5-arylfuran-2-yl-indoline guanidine derivatives were synthesized. Scheme S4 summarizes the method used for the preparation of asymmetric furan guanidine compounds 16-20. Like the guanidine compounds, the asymmetric derivatives required the corresponding amino or diamine compounds 16b-20b as a common precursor. The preparation of these intermediary amines was performed in three steps starting with an amide reaction between 2-furoyl chloride and 5-nitroindoline or indoline to form the corresponding amides 2 and 15 in moderate yields. Then, an arylation reaction [21] was performed to add a substituted aryl bromide group to the 5-furanoyl amide derivatives. The bis and mono nitro compounds 16a-20a were obtained in reasonable yield through this coupling reaction. Finally, the nitro compounds were reduced with iron to provide the desired amine compounds 16b-20b. The asymmetric guanidine compounds 16-20 were prepared from the amine and di-amine (Scheme S4) via the same general synthetic route used for 1. Finally, the NMR analyses together with the HRMS studies confirm that all the synthesized compounds have a high degree of purity.

Evaluation of Cytotoxicity and Antimicrobial Activity against Gram-negative and Gram-positive Non-pathogenic Bacterial Strains
The cytotoxicity of all compounds was evaluated using the human MCF-7 and HepG2 cell lines (Table 1; dose-response curves are shown in Figure S1). In general, all derivatives were moderately cytotoxic with an effective concentration of 50% (EC 50 ) greater than 25 µM in both cell lines. Most of the new derivatives (3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16, 17, 18) were less cytotoxic than the lead (1).  The diethyl squarate was first treated with the corresponding diamine compounds (1b, 3b-5b) to displace one ethoxy residue and produce the corresponding intermediates 11a-14a. These intermediates were then treated with ammonia to displace the second ethoxy group and produce the squaryldiamide derivatives 11-14 in good overall yields.
To modulate the activity of the bis-arylfuran scaffold, we synthesized asymmetric furan derivatives, replacing one of the aromatic rings with a residue containing an amide attached to an aromatic heterocycle. A series of 5-arylfuran-2-yl-indoline guanidine derivatives were synthesized. Scheme S4 summarizes the method used for the preparation of asymmetric furan guanidine compounds 16-20. Like the guanidine compounds, the asymmetric derivatives required the corresponding amino or diamine compounds 16b-20b as a common precursor. The preparation of these intermediary amines was performed in three steps starting with an amide reaction between 2-furoyl chloride and 5-nitroindoline or indoline to form the corresponding amides 2 and 15 in moderate yields. Then, an arylation reaction [21] was performed to add a substituted aryl bromide group to the 5furanoyl amide derivatives. The bis and mono nitro compounds 16a-20a were obtained in reasonable yield through this coupling reaction. Finally, the nitro compounds were reduced with iron to provide the desired amine compounds 16b-20b. The asymmetric guanidine compounds 16-20 were prepared from the amine and di-amine (Scheme S4) via the same general synthetic route used for 1. Finally, the NMR analyses together with the HRMS studies confirm that all the synthesized compounds have a high degree of purity.

Evaluation of Cytotoxicity and Antimicrobial Activity against Gram-negative and Gram-positive Non-pathogenic Bacterial Strains
The cytotoxicity of all compounds was evaluated using the human MCF-7 and HepG2 cell lines (Table 1; dose-response curves are shown in Figure S1). In general, all derivatives were moderately cytotoxic with an effective concentration of 50% (EC50) greater than 25 µM in both cell lines. Most of the new derivatives (3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16, 17, 18) were less cytotoxic than the lead (1). The diethyl squarate was first treated with the corresponding diamine compounds (1b, 3b-5b) to displace one ethoxy residue and produce the corresponding intermediates 11a-14a. These intermediates were then treated with ammonia to displace the second ethoxy group and produce the squaryldiamide derivatives 11-14 in good overall yields.
To modulate the activity of the bis-arylfuran scaffold, we synthesized asymmetric furan derivatives, replacing one of the aromatic rings with a residue containing an amide attached to an aromatic heterocycle. A series of 5-arylfuran-2-yl-indoline guanidine derivatives were synthesized. Scheme S4 summarizes the method used for the preparation of asymmetric furan guanidine compounds 16-20. Like the guanidine compounds, the asymmetric derivatives required the corresponding amino or diamine compounds 16b-20b as a common precursor. The preparation of these intermediary amines was performed in three steps starting with an amide reaction between 2-furoyl chloride and 5-nitroindoline or indoline to form the corresponding amides 2 and 15 in moderate yields. Then, an arylation reaction [21] was performed to add a substituted aryl bromide group to the 5furanoyl amide derivatives. The bis and mono nitro compounds 16a-20a were obtained in reasonable yield through this coupling reaction. Finally, the nitro compounds were reduced with iron to provide the desired amine compounds 16b-20b. The asymmetric guanidine compounds 16-20 were prepared from the amine and di-amine (Scheme S4) via the same general synthetic route used for 1. Finally, the NMR analyses together with the HRMS studies confirm that all the synthesized compounds have a high degree of purity.

Evaluation of Cytotoxicity and Antimicrobial Activity against Gram-negative and Gram-positive Non-pathogenic Bacterial Strains
The cytotoxicity of all compounds was evaluated using the human MCF-7 and HepG2 cell lines (Table 1; dose-response curves are shown in Figure S1). In general, all derivatives were moderately cytotoxic with an effective concentration of 50% (EC50) greater than 25 µM in both cell lines. Most of the new derivatives (3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16, 17, 18) were less cytotoxic than the lead (1). The diethyl squarate was first treated with the corresponding diamine compounds (1b, 3b-5b) to displace one ethoxy residue and produce the corresponding intermediates 11a-14a. These intermediates were then treated with ammonia to displace the second ethoxy group and produce the squaryldiamide derivatives 11-14 in good overall yields.
To modulate the activity of the bis-arylfuran scaffold, we synthesized asymmetric furan derivatives, replacing one of the aromatic rings with a residue containing an amide attached to an aromatic heterocycle. A series of 5-arylfuran-2-yl-indoline guanidine derivatives were synthesized. Scheme S4 summarizes the method used for the preparation of asymmetric furan guanidine compounds 16-20. Like the guanidine compounds, the asymmetric derivatives required the corresponding amino or diamine compounds 16b-20b as a common precursor. The preparation of these intermediary amines was performed in three steps starting with an amide reaction between 2-furoyl chloride and 5-nitroindoline or indoline to form the corresponding amides 2 and 15 in moderate yields. Then, an arylation reaction [21] was performed to add a substituted aryl bromide group to the 5furanoyl amide derivatives. The bis and mono nitro compounds 16a-20a were obtained in reasonable yield through this coupling reaction. Finally, the nitro compounds were reduced with iron to provide the desired amine compounds 16b-20b. The asymmetric guanidine compounds 16-20 were prepared from the amine and di-amine (Scheme S4) via the same general synthetic route used for 1. Finally, the NMR analyses together with the HRMS studies confirm that all the synthesized compounds have a high degree of purity.

Evaluation of Cytotoxicity and Antimicrobial Activity against Gram-negative and Gram-positive Non-pathogenic Bacterial Strains
The cytotoxicity of all compounds was evaluated using the human MCF-7 and HepG2 cell lines (Table 1; dose-response curves are shown in Figure S1). In general, all derivatives were moderately cytotoxic with an effective concentration of 50% (EC50) greater than 25 µM in both cell lines. Most of the new derivatives (3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16, 17, 18) were less cytotoxic than the lead (1). The diethyl squarate was first treated with the corresponding diamine compounds (1b, 3b-5b) to displace one ethoxy residue and produce the corresponding intermediates 11a-14a. These intermediates were then treated with ammonia to displace the second ethoxy group and produce the squaryldiamide derivatives 11-14 in good overall yields.
To modulate the activity of the bis-arylfuran scaffold, we synthesized asymmetric furan derivatives, replacing one of the aromatic rings with a residue containing an amide attached to an aromatic heterocycle. A series of 5-arylfuran-2-yl-indoline guanidine derivatives were synthesized. Scheme S4 summarizes the method used for the preparation of asymmetric furan guanidine compounds 16-20. Like the guanidine compounds, the asymmetric derivatives required the corresponding amino or diamine compounds 16b-20b as a common precursor. The preparation of these intermediary amines was performed in three steps starting with an amide reaction between 2-furoyl chloride and 5-nitroindoline or indoline to form the corresponding amides 2 and 15 in moderate yields. Then, an arylation reaction [21] was performed to add a substituted aryl bromide group to the 5furanoyl amide derivatives. The bis and mono nitro compounds 16a-20a were obtained in reasonable yield through this coupling reaction. Finally, the nitro compounds were reduced with iron to provide the desired amine compounds 16b-20b. The asymmetric guanidine compounds 16-20 were prepared from the amine and di-amine (Scheme S4) via the same general synthetic route used for 1. Finally, the NMR analyses together with the HRMS studies confirm that all the synthesized compounds have a high degree of purity.

Evaluation of Cytotoxicity and Antimicrobial Activity against Gram-negative and Gram-positive Non-pathogenic Bacterial Strains
The cytotoxicity of all compounds was evaluated using the human MCF-7 and HepG2 cell lines (Table 1; dose-response curves are shown in Figure S1). In general, all derivatives were moderately cytotoxic with an effective concentration of 50% (EC50) greater than 25 µM in both cell lines. Most of the new derivatives (3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14, 16, 17, 18) were less cytotoxic than the lead (1).  The diethyl squarate was first treated with the corresponding diamine compounds (1b, 3b-5b) to displace one ethoxy residue and produce the corresponding intermediates 11a-14a. These intermediates were then treated with ammonia to displace the second ethoxy group and produce the squaryldiamide derivatives 11-14 in good overall yields.
To modulate the activity of the bis-arylfuran scaffold, we synthesized asymmetric furan derivatives, replacing one of the aromatic rings with a residue containing an amide attached to an aromatic heterocycle. A series of 5-arylfuran-2-yl-indoline guanidine derivatives were synthesized. Scheme S4 summarizes the method used for the preparation of asymmetric furan guanidine compounds 16-20. Like the guanidine compounds, the asymmetric derivatives required the corresponding amino or diamine compounds 16b-20b as a common precursor. The preparation of these intermediary amines was performed in three steps starting with an amide reaction between 2-furoyl chloride and 5-nitroindoline or indoline to form the corresponding amides 2 and 15 in moderate yields. Then, an arylation reaction [21] was performed to add a substituted aryl bromide group to the 5furanoyl amide derivatives. The bis and mono nitro compounds 16a-20a were obtained in reasonable yield through this coupling reaction. Finally, the nitro compounds were reduced with iron to provide the desired amine compounds 16b-20b. The asymmetric guanidine compounds 16-20 were prepared from the amine and di-amine (Scheme S4) via the same general synthetic route used for 1. Finally, the NMR analyses together with the HRMS studies confirm that all the synthesized compounds have a high degree of purity.
Lead compound 1 bearing an electron-withdrawing Cl group on the phenyl ring and its isosteres 5, 7, and 8 bearing a CF3, CN, and COOCH3 group on the phenyl ring, had low antibacterial activity against all tested strains (Tables S1 and S2; Figure S2A,D,G,J). Compounds bearing an electron donating group on the phenyl ring such as CH3 (4) or OCH3 (6) had moderate antibacterial activity. Both compounds had an EC50 of around 28.5 µM against Gram-negative P. carotovorum and around 14.4 µM against P. putida (Table  S1). Additionally, 4 had an EC50 of 8.2 µM against Gram-negative P. caledonica and was the most potent compound against the Gram-positive strain B. subtilis with an EC90 of 8.5 µM (Table S1), which is 18 times more potent than ampicillin ( Figure S4). When the phenyl ring in 4 was replaced with a pyridine ring (e.g., in 9) the activity almost disappeared (Tables S1 and S2; Figure S2B,E,H,K). Interestingly, 3 does not have any substituent on the aromatic ring and had good activity against the Gram-negative strains P. putida and P. caledonica (Table S1). Compound 10 is characterized by a methylene between the guanidino group and the aromatic system with respect to 3 (Table S1) and had a good EC90 value of 5 µM against P. caledonica. The squaryldiamide-based compounds (11)(12)(13)(14) were the least active against all bacteria tested, indicating that the guanidino group is essential for antibacterial activity. Compounds 17 and 19 were the most potent against the Gram-negative E. coli (EC90 values of 4.3 and 2.5 µM, respectively; Table S1), being 20-and 35-fold more potent than ampicillin ( Figure S4). Compound 18 was the most active of the asymmetric series of compounds against Gram-negative P. putida and P. caledonica (Table S1) and Gram-positive B. subtilis (Table S2).

Antibacterial Activity against ESKAPE and E. coli Isolates
To evaluate the antimicrobial activity of the dicationic derivatives against clinically relevant bacteria, compounds 1, 3, 4, 6, 8, 10, 16 and 17, which showed good activity against the non-pathogenic bacteria, were selected to be tested against 10 different Gramnegative isolates from the ESKAPE group and E. coli. The results are summarized in Tables 2 and 3 and Figure 3 (dose-response curves are shown in Figures S5-S12). Cefotaxime (CTX) was used as positive control ( Figure S13). All compounds, except 10, had moderateto-good antibacterial activity against these isolates. 20 37.8 28.5 Half maximal effective concentration (EC 50 ) in µM for each compound. Values >1000 and >100 µM represent the maximum compound concentration tested in the cytotoxicity assays, without observing 50% inhibition. MCF-7: Michigan Cancer Foundation-7 cell line; HepG2: Hepatoma G2 cell line. Cytotoxicity dose-response curves for all compounds are shown in Figure S1.
Lead compound 1 bearing an electron-withdrawing Cl group on the phenyl ring and its isosteres 5, 7, and 8 bearing a CF 3 , CN, and COOCH 3 group on the phenyl ring, had low antibacterial activity against all tested strains (Tables S1 and S2; Figure S2A,D,G,J). Compounds bearing an electron donating group on the phenyl ring such as CH 3 (4) or OCH 3 (6) had moderate antibacterial activity. Both compounds had an EC 50 of around 28.5 µM against Gram-negative P. carotovorum and around 14.4 µM against P. putida (Table  S1). Additionally, 4 had an EC 50 of 8.2 µM against Gram-negative P. caledonica and was the most potent compound against the Gram-positive strain B. subtilis with an EC 90 of 8.5 µM (Table S1), which is 18 times more potent than ampicillin ( Figure S4). When the phenyl ring in 4 was replaced with a pyridine ring (e.g., in 9) the activity almost disappeared (Tables S1 and S2; Figure S2B,E,H,K). Interestingly, 3 does not have any substituent on the aromatic ring and had good activity against the Gram-negative strains P. putida and P. caledonica (Table S1). Compound 10 is characterized by a methylene between the guanidino group and the aromatic system with respect to 3 (Table S1) and had a good EC 90 value of 5 µM against P. caledonica. The squaryldiamide-based compounds (11)(12)(13)(14) were the least active against all bacteria tested, indicating that the guanidino group is essential for antibacterial activity. Compounds 17 and 19 were the most potent against the Gram-negative E. coli (EC 90 values of 4.3 and 2.5 µM, respectively; Table S1), being 20and 35-fold more potent than ampicillin ( Figure S4). Compound 18 was the most active of the asymmetric series of compounds against Gram-negative P. putida and P. caledonica (Table S1) and Gram-positive B. subtilis (Table S2).

Antibacterial Activity against ESKAPE and E. coli Isolates
To evaluate the antimicrobial activity of the dicationic derivatives against clinically relevant bacteria, compounds 1, 3, 4, 6, 8, 10, 16 and 17, which showed good activity against the non-pathogenic bacteria, were selected to be tested against 10 different Gram-negative isolates from the ESKAPE group and E. coli. The results are summarized in Tables 2 and 3 Antibiotics 2022, 11, 1115 7 of 26 and Figure 3 (dose-response curves are shown in Figures S5-S12). Cefotaxime (CTX) was used as positive control ( Figure S13). All compounds, except 10, had moderate-to-good antibacterial activity against these isolates.    Tables 2 and S2. For each species, two isolates with different susceptibility profiles to commonly used antibiotics were chosen. For CTX, the difference in sensitivity in each pair is obvious, ranging from 18× for the A. baumannii isolates to 1600× for K. pneumoniae (Table S3). Likewise, resistance to meropenem is drastically different in each pair of isolates; from 8× for A. baumannii to 256× for K. pneumoniae, except for Enterobacter where the two isolates are equally resistant (Table S3). The same is seen for ciprofloxacin resistance for E. coli (8000×), K. pneumoniae (1000×), and P. aeruginosa (32×) ( Table S3). By contrast, for sensitivity to all the bisguanidinearylfuran compounds, the difference in each pair was less than two-fold for all species, except for Enterobacter (7× difference). This clearly demonstrates that the antibacterial efficacy of the new diaryl compounds is not diminished by the mechanisms that cause resistance against established antibiotics.
When comparing the activity of 3 with 10, the addition of a carbon between the aromatic system and the guanidino moiety did not have an impact on the activity against the ESKAPE isolates. Finally, the asymmetric 16 and 17 had moderate activity against all the strains evaluated. Compound 3 displayed the highest selectivity index, between 13 and 83 across all ESKAPE isolates (Table 3). This represents an improvement over 1 by a factor between 1.7 and 9, achieved in all cases through weaker cytotoxicity. Therefore, it is important to analyze the differences in cytotoxicity when evaluating the practical antibiotic potential of the new compounds.

Sensitivity Profiling against Two Collections of E. coli Strains with Defined Antibiotic Susceptibility Patterns
As an approach to finding the mechanism of action of the new substances, we performed high-resolution microbial phenomics profiling of selected compounds 1, 4, 6, 10, and 16 and the known antibiotic CTX against two E. coli libraries with known patterns of antibiotic susceptibility. The E. coli reference collection (ECOR) contains 72 strains isolated from a wide variety of environments and geographical locations [22] including representatives of the seven E. coli phylogroups [23]. Eighteen of the strains are resistant to one known antibiotic (12 antibiotics tested) and 14 are resistant to two antibiotics, with resistance to sulfisoxazole, tetracycline and streptomycin being the most common [24]. A set of 96 extended spectrum beta-lactamase (ESBL) strains were isolated at Sahlgrenska University Hospital in Gothenburg, Sweden between 2011 and 2012, some of which have been previously published [25,26]. These are likely mostly closely related since they were isolated from a small patient population and are in large part uncharacterized except for their identification as ESBL E. coli strains. The results of these experiments are summarized in Figures 4 and S14. In Figure 4, we see the growth yield of the strains in these collections upon exposure to diaryl compounds relative to our reference strain (E. coli ATCC #25922) and normalized for growth without any added compound. Primarily, there was no widespread resistance to the diaryl compounds in either strain set. As expected, most ESBL strains are highly resistant to the lactam CTX, but are no more resistant to any diaryl compound than the control strain ( Figure 4A). Importantly, the profiles across the strains of CTX resistance do not covary with the resistance profiles for any of the diaryls ( Figure 4A,B). Strains strongly resistant to CTX showed normal sensitivity to the diaryl compounds (e.g., GU1114, GU1078), whereas strains that were particularly sensitive to several of the diaryls display normal CTX sensitivity (GU1117, GU1068, GU2320).

Efficiency of Compounds, Relationship between Structure and Function
For decades, no new classes of antibiotics effective against Gram-negative bacteria have reached the market. The current rise in antibiotic resistance among such bacteria therefore threatens to deplete the remaining clinical treatment options. Especially troubling is the fact that acquired co-resistance to several antibiotics in one bacterial population (multidrug resistance) is common [28]. Infections by some Gram-negative species are particularly problematic to treat with antibiotics. It is therefore promising that the human isolates of E. coli and ESKAPE species tested are sensitive to the new molecules presented here. The sensitivity patterns we observe across bacterial strains, which do not correlate with their sensitivity to established antibiotics, also indicate that the mechanism of action of diaryl compounds may be different from those.
It is noteworthy that the new compounds show good activity against A. baumannii. This opportunistic pathogen is notoriously difficult to treat with antibiotics, as it displays high level intrinsic resistance to many antibiotics [29], which has been attributed at least in part to abundant membrane-bound export pumps [30]. It will be important to identify which properties allow these molecules to escape such barriers to cellular uptake. Future work could examine, for instance, what distinguishes 1 and 3, which both show a particularly high efficiency against A. baumannii (Table 2), from the other related compounds in the design series.
K. pneumoniae is strongly implicated in nosocomial infections and accumulates multiple plasmid-borne antibiotic-resistance genes [31], making this bacterial species a major medical problem. It is therefore encouraging that several of the compounds presented here (e. g., 1, 3, 4, 6, 8, 16) are effective against multidrug resistant K. pneumoniae (Tables 2 and S3), with selectivity indices between 11 and 24 (3, 6, 8, 16; Table 3). P. aeruginosa, in addition to widespread antibiotic resistance, has a high propensity to form biofilms, adding further difficulty to the clinical treatment of infections [32]. Here, we have only examined planktonic P. aeruginosa, however, we see that 1 and 3 were more effective than CTX against both isolates of P. aeruginosa (Table 2).
Compound 3 was more effective than the lead compound against some of the bacterial isolates (Table S1) and is less cytotoxic than 1 (Table 1). Even in cases where 3 did not show more antibacterial activity than 1 (K. pneumoniae, P. aeruginosa, Enterobacter), this combination results in better selectivity indices. Therefore, 3, which lacks substituents in the aromatic systems, represents a favorable compromise in situations where the chlorine substituents in 1 (Figure 2), which may confer toxicity, have been eliminated (Table 3). Compound 8 also has better selectivity indices than 1, against K. pneumoniae, A. baumannii, and Enterobacter (Table 3). This molecule has methoxycarbonyl groups replacing the chlorine substituents in the lead compound ( Figure 2). The improved selectivity indices of 8 over 1 is mainly due to reduced cytotoxicity (Tables 1-3). By contrast, 4 and 6, which carry electrondonating methyl or methoxy groups on the phenyl ring (Figure 2), displayed greater activity than 1 against several laboratory strains (Table S1). However, their selectivity indices were less favorable because of higher cytotoxicity (Tables 1 and 3).
Interestingly, antibacterial compounds carrying an aminoguanidine group have been reported to potentiate norfloxacin in Staphylococcus aureus, suggesting a possibility to develop the guanidine-containing molecules reported here as co-drugs [18].

Mechanism of Action
The mechanism for the antibacterial effect of this group of compounds is not clear. The lead compound and derivatives were originally designed to target A/T-rich DNA sequences [8]. However, 1 is comparably effective against the eukaryotic parasites Plasmodium falciparum (EC 50 for erythrocyte stage = 590 nM) [10], which has a genome with an exceptionally low G/C content of 19.8% [33], and against Trypanosoma cruzi (290 nM) [34], which has a G/C content of 51.0% [35]. Additionally, there is no obvious correlation between the genomic G/C content of the bacterial species examined here and their sensitivity to the compounds used in this work (Table S4). And while these compounds were thought to selectively bind in the DNA minor groove [8] there is no structural difference between prokaryotic and eukaryotic DNA that could account for an antibacterial effect. A DNA-binding compound would also be suspected to be genotoxic. However, we are not aware of any reports on mutagenic or carcinogenic activity of dicationic bisguanidine-arylfurans. Together, this argues against DNA-binding as the main mechanism for the observed antibacterial activity.
The mode of action of a compound can be probed indirectly by comparing with the sensitivity of bacteria to established antibiotics with known action mechanisms. It is evident from analyzing the E. coli strains (Figure 4) that the resistance profiles for the diaryl compounds across the strain collections are distinct from that of CTX, the lactam compound used here as a reference. This conclusion is also supported by the resistance pattern of the ESKAPE isolates (Tables 2 and S3), where the aryl compounds were equally effective against isolates that are resistant or sensitive to established antibiotics. CTX is a lactam in the cephalosporin subgroup. Meropenem is also a lactam compound, though belonging to the carbapenem subgroup, while ciprofloxacin is a quinolone and acts by interfering with bacterial DNA replication. This argues that whatever mechanisms underlie the resistance of these isolates of ESKAPE species and E. coli to established antibiotics, they are not effective against the diaryl compounds in this study. It is therefore plausible that the mode of action of the diaryl compounds is distinct from commonly used antibiotics, at least from lactam compounds and quinolones.
Given all the above, it also has to be considered that a small molecule may interact with multiple cellular targets at any given time. Which of these will result in the biologically relevant effect may in addition vary with extrinsic and intrinsic factors acting on the cell, and on uptake and intracellular localization.

General Experimental Information for Synthesis and Compound Characterization
General reagent and solvents for the synthesis of compounds were purchased from commercial sources and used as supplied, unless otherwise stated.
Purification by flash column chromatography was performed on a Selekt (Biotage, Uppsala, Sweden) automated instrument with Sfär KP-amino D or Sfär silica D cartridges (Biotage, Uppsala, Sweden), mobile phase consist of pentane (solvent A) and ethyl acetate (solvent B). The final compounds were purified by reverse phase flash column chromatography performed on an Isolera (Biotage, Uppsala, Sweden) automated instrument with Sfär C18 D cartridges (Biotage, Uppsala, Sweden), mobile phase consist of water (solvent A) and acetonitrile (solvent B). The standard gradient consisted of x% solvent B for 1 columns volume, x% to y% B for 10 column volumes, and then y% B for 2 column volumes. x and y are defined in the characterization section of the compound the interest.

General Procedure A
In a sealed 20 mL microwave vial, 2,5-Bis-(trimethylstannyl)furan (0.5 mmol), aryl bromide (1.0 mmol) and tetrakis(triphenyphosphine)-palladium(0) (0.025 mmol) in anhy-drous dimethylformamide (10 mL) was evacuated and backfilled with N 2 (×3) and heated for 14 h at 100 • C. Upon cooling, the mixture was filtered through Celite, the Celite rinsed with chloroform, and the residue was reduced under vacuum. Then 100 mL of chloroform, and 50 mL of 10% aqueous potassium fluoride was added and the mixture was stirred at room temperature for 0.5 h. The organic layer was separated and dried over sodium sulfate, filtered, and concentrated under reduced pressure. The crude product was purified by flash chromatography (25 g Sfär silica D cartridge, 15-75% B, R f = 12 column volumes) to give the desired compounds.

General Procedure B
To a solution of 2,5-Bis(4-nitrophenyl)furan derivatives (1a, 3a-9a) (0.3 mmol) in THF (3 mL) and EtOH (3 mL), ammonium chloride (3 mL, 0.3 M) and iron (1.75 mmol) were added After stirring at 60 • C for 4 h, the reaction was allowed to cool to room temperature and the heterogeneous mixture filtered through Celite and the Celite was rinsed with ethyl acetate. The solution was concentrated to half-volume, then diluted with ethyl acetate (20 mL) and washed with sodium hydroxide solution (1 M, 20 mL). The organic layer was separated, the aqueous phase was extracted with ethyl acetate (2×), the combined organic phases dried over sodium sulfate, filtered and the solvent evaporated. The crude product was purified by flash chromatography (11 g Sfär KP-amino D cartridge, 15-90% B, R f = 10 column volumes) to give the desired compounds.

General Procedure E
To a solution of 2,5-Bis(4-aminophenyl)furan derivatives (1b, 3b-5b) (0.2 mmol) in 2.5 mL of ethanol was added 3,4-diethoxy-3-cyclobutene-1,2-dione (0.40 mmol) and zinc trifluoromethanesulfonate (0.08 mmol) in ethanol (2.5 mL). The solution was stirred at room temperature for 3 h. Then, the reaction mixture was evaporated under reduced pressure and the residue was subsequently dissolved in ethyl acetate and washed several times with aqueous ammonium chloride (1 M), and then with water. The organic layer was separated, dried over sodium sulfate, filtered, and triturated several times with pentane and diethyl ether. The resultant solid was dried under vacuum to give the desired compounds.

General Procedure G
To a solution of 2-furoyl chloride (2.8 mmol) in dichloromethane (15 mL) was added 5-nitroindoline or indoline (2.8 mmol) and triethylamine (1.6 mL), the mixture was stirred at room temperature for 3 h. The reaction mixture was monitored by GCMS, until the starting material was consumed. Then, the mixture was concentrated under reduced pressure and the crude product was purified by flash chromatography (25 g Sfär silica D cartridge, 15-50% B, R f = 8 column volumes) to give the desired compounds.

General Procedure H
In a sealed 20 mL microwave vial, the aryl bromide (1.1 mmol), furan-2-yl(indolin-1-yl)methanone (2 or 15) (0.9 mmol), potassium acetate (2.75 mmol), and palladium(II) acetate (0.02 mmol) were dissolved in dimethylacetamide (5 mL) and the resulting reaction mixture was evacuated and backfilled with nitrogen several times. The reaction was stirred at 150 • C for 20 h. Upon cooling to room temperature, the residue filtered through Celite and the Celite rinsed with ethyl acetate. The organic phase was washed with water and brine, dried over sodium sulfate, concentrated under reduced pressure. The crude product was purified by flash chromatography (25 g Sfär silica D cartridge, 25-100% B, R f = 12 column volumes) to give the desired compounds.

General Procedure J
To a solution of (5-aminoindolin-1-yl)(5-(4-aminophenyl)furan-2-yl)methanone derivatives (16b-18b) or (5-aminoindolin-1-yl)(5-(phenyl)furan-2-yl)methanone (19b) or (indolin-1-yl)(5-(4-aminophenyl)furan-2-yl)methanone (20b) (0.10 mmol) in dimethylformamide (5 mL) at 0 • C was added mercury (II) chloride (0.21 mmol), 1,3-bis(tert-butoxycarbonyl)-2-methyl-2-thiopseudourea (0.19 mmol) and triethylamine (0.48 mmol). The resulting mixture was stirred at 0 • C for 1 h and 18 h at room temperature. The reaction mixture was diluted with ethyl acetate, filtered through Celite and the Celite rinsed with ethyl acetate. microplate-reader (BMG Labtech, Offenburg, Germany) with excitation filter set to 544 nm and emission filter to 590 nm. Cells exposed to only the equivalent concentration of DMSO were used as negative control. Bleed-through of fluorescence from resorufin between wells in the microtiter plate fluorescence reader, was measured and found to be <1% between adjacent wells. The 384-well plates were used to avoid this fluorescence bleed-through, achieved by skipping a well in-between bacterial cultures and compound dilutions. To check for quenching of fluorescence by any of the investigated compounds, grown bacterial cultures were mixed after 1 h incubation with resazurin and the compound of interest at the highest concentration to be assayed, and the measured fluorescence compared with samples without compound added. All tests of compound activity were performed in three independent replicates for more accurate determination of the half (EC 50 ) and 90% maximal effective concentration (EC 90 ).
The antibacterial activities of eight compounds (1, 3, 4, 6, 8, 10, 16, and 17) were tested against 10 different isolates of Gram-negative bacteria from human and other sources, two E. coli (CCUG #67180 and CCUG #17620/ATCC #25922, control strain), two K. pneumoniae (CCUG #58547 and CCUG #225T), two P. aeruginosa (CCUG #17619 and CCUG #59347), two A. baumannii (CCUG #57035 and CCUG #57250), E. cloacae (CCUG #6323T), and E. hormaechei (CCUG #58962) in a 96-well plate format. The compounds to test were 3-fold diluted in six steps in cation-adjusted Mueller-Hinton (ca-MH) broth to final concentrations spanning from 30 to 0.12 µM, i.e., a 1667-416,667 times dilution from the 50 mM stocks. Bacterial mass from the ten isolates grown overnight on horse blood or Mueller-Hinton agar plates was suspended in ca-MH and adjusted to a final inoculum cell density of~5 × 10 5 CFU/mL. For all assays, Biolog redox dye A diluted 100× from the stock solution was used for measurements of all ten isolates (up to a total volume of 120 µL per well). All plates included one well per isolate with only inoculum and dye but no test compound (i.e., positive control) and one well per isolate with inoculum, dye and DMSO diluted 1667×, corresponding to the DMSO concentration in the wells with the highest concentration of test compound. All measurements were performed using the Omnilog microplate reader (Biolog, Hayward, CA, USA) where the 96-well plates were read at 15 min intervals for a total of 24 h at 37 • C. Each compound was run in triplicate on three independent assay plates for more accurate determination of the half (EC 50 ) and 90% maximal effective concentration (EC 90 ).
Non-linear regression dose-response inhibition following a log (agonist) vs. response-Find ECanything was performed using GraphPad Prism version 9.2.0 for Windows, Graph-Pad Software, San Diego, CA, USA, www.graphpad.com (accessed on 20 July 2022).

Cytotoxicity Assays
The cytotoxicity levels of all compounds were evaluated against human Michigan Cancer Foundation-7 (MCF-7) and Hepatoma G2 cell line (HepG2) cell lines. MCF-7 is an extensively characterized breast cancer cell line isolated in 1970 [41], while HepG2 cells were derived from a hepatoma in 1975 [42]. Both cell lines grow robustly during in vitro culture and have been widely used for cytotoxicity testing. Cells of hepatic origin, such as HepG2, are of particular relevance for toxicity studies as many drugs accumulate in the liver during metabolic conversion [41,42]. Cells were grown in Dulbecco's Modified Eagle Medium supplemented with 10% fetal calf serum and kept in exponential growth, as previously reported [43]. Before the assay, cells were reseeded into 96-well microtiter plates at a density allowing continued exponential growth and let to settle for 24 h. The compounds were added from a stock solution in DMSO, for a final concentration of 0.3% v/v of the solvent in the culture medium. After 24 h of incubation in presence of the compound, cell viability was assayed using PrestoBlue Cell Viability Reagent (resazurin-based solution, ThermoFisher, Waltham, MA, USA) according to the manufacturer's instructions. A POLARstar Omega microplate-reader (BMG Labtech, Offenburg, Germany) was used to measure resorufin fluorescence at 544 nm excitation/590 nm emissions. Each assay contained a DMSO control at the equivalent starting concentration, positive control (uninhibited cell growth) and negative control (cell medium only). Survival was expressed as percentage of the solventonly control. EC 50 values for each compound were calculated from three independent replicate experiments using 2-fold dilution intervals. Non-linear regression dose-response inhibition following a log(agonist) vs. response-Find ECanything was performed using GraphPad Prism version 9.2.0 for Windows, GraphPad Software, San Diego, California USA, www.graphpad.com. The Selectivity Index (SI) for each compound and bacterial strain was calculated as the ratio between the mean of the EC 50 values for the two human cell lines and the EC 50 for the bacterial strain in question ((EC 50 MCF-7 + EC 50 HepG2)/2)/EC 50 bacterial strain). The higher the value, the more selective is the compound against the different bacterial strain.

High-Resolution Microbial Phenomics (Scan-o-Matic) Assays
E. coli colonies were deposited as initially isogenic populations at initial population sizes of~100,000 cells, with 1536 colonies deposited in systematic colony arrays on each plate on top of a solid matrix composed of LB medium supplemented with a sublethal concentration (60 µM) of five different compounds, 1, 4, 6, 10, and 16, and a known antimicrobial, cefotaxime (CTX; 2 µg/mL), using automated pinning by robot. The compound concentrations were empirically chosen to strongly but not completely inhibit colony growth on agar of the more sensitive strains in the collections, in order to enable quantitation of the difference in growth yield between more and less resistant strains [44,45]. Of these colonies, 384 were identical controls used to correct for spatial bias between and within plates. For CTX and 4, 10, and 16, each lineage was cultivated as six biological replicates on different plates. For 1, nine biological replicates were done. Population expansion for each colony was followed by measuring cell numbers at 10 min intervals using the Scan-o-Matic framework, version 2.0 [44] with an E. coli calibration curve [45]. From each colony growth curve, the total cell yield after 8 h was extracted (growth yield). Experiments included automated transmissive scanning and signal calibration in 10 min intervals, as described [44]. The absolute population yields were log(2) transformed and normalized to the corresponding measures of adjacent controls (fourth position E. coli CCUG #17620/ATCC #25922 strain) on each plate, while data for missing or mis-quantified colonies were discarded. The relative growth yield of each strain (total n = 164) of the screened ECOR (n = 72) and ESBL (n = 92) libraries on the tested compounds was normalized to their corresponding growth without compound and the resulting ratios were clustered and visualized using heatmaps constructed using the R package ComplexHeatmap v. 2.8.0 [27].

Conclusions
There is a paucity of antibiotics effective against Gram-negative bacteria among which multiple resistance has spread. We have identified a group of small molecules that have demonstrated promising activity against antibiotics-resistant Gram-negative pathogens of the ESKAPE group and E. coli, which pose a major clinical concern. As with all new antibacterial agents, resistance to these molecules is likely to eventually occur among bacterial populations. Even if bacterial gene products dedicated to inactivate these new compounds do not exist, resistance can develop through e.g., increased efflux, blocked uptake, and increased target expression. Further, the efficacy against persister cells or biofilms has not been tested.
The molecular target of these diaryl compounds remains to be established, however they represent a promising resource for further development of antibiotics, or as potentiators of other antibiotics. Our modifications of the lead molecule for reduced cytotoxicity and greater activity against a broader set of bacterial species represent important steps in this direction.

Supplementary Materials:
The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/antibiotics11081115/s1, Scheme S1: Synthetic pathway to 1, 3-9a; Scheme S2: Synthetic pathway to 10a; Scheme S3: Synthetic pathway to 11-14a; Scheme S4: Synthetic pathways to 16-20a; Table S1: Effective concentrations of each compound against the Gram-negative and Gram-positive set of laboratory strains; Table S2: Selectivity indexes of each compound against the Gram-negative and Gram-positive set of laboratory strains; Table S3: MIC values of established antibiotics for the clinical isolates of ESKAPE species and E. coli tested in this paper; Table S4: Genomic G/C content (%) of bacterial species; Figure S1: Cytotoxicity dose-response curves of all compounds against MCF-7 (A, B and C) and HepG2 (D, E and F) cell lines; Figure S2: Antibacterial activity dose-response curves of all compounds against Gram-negative E. coli (A, B and C), P. putida (D, E and F), P. carotovorum (G, H and I) and P. caledonica (J, K and L); Figure S3: Antibacterial activity dose-response curves of all compounds against Gram-positive B. subtilis (A, B and C); Figure S4: Measurements of the antibacterial activity of ampicillin, as a positive control antibiotic tested against Gram-negative Escherichia coli (A) and Gram-positive Bacillus subtilis (B); Figure S5: Antibacterial activity dose-response curves of 1 against the 10 Gram-negative bacteria; Figure S6: Antibacterial activity dose-response curves of 3 against the 10 Gram-negative bacteria; Figure S7: Antibacterial activity dose-response curves of 4 against the 10 Gram-negative bacteria; Figure S8: Antibacterial activity dose-response curves of 6 against the 10 Gram-negative bacteria; Figure S9: Antibacterial activity dose-response curves of 8 against the 10 Gram-negative bacteria; Figure S10: Antibacterial activity dose-response curves of 10 against the 10 Gram-negative bacteria; Figure S11: Antibacterial activity dose-response curves of 16 against the 10 Gram-negative bacteria; Figure S12: Antibacterial activity dose-response curves of 17 against the 10 Gram-negative bacteria; Figure S13: Antibacterial activity dose-response curves of the known antibiotic cefotaxime (CTX) as a control against the 10 Gram-negative bacteria; Figure S14