Non-Beta Lactam Inhibitors of the Serine Beta-Lactamase blaCTX-M15 in Drug Resistant Salmonella typhi

Abstract

Antibiotic resistance by bacterial pathogens against widely used β-lactam drugs is a major concern to public health worldwide resulting in high health care cost. The present study aimed to extend previous research by investigating the potential activity of reported compounds against S. typhi β-lactamase protein. 74 compounds from computational screening reported in our previous study against β-lactamase CMY-10 were subjected to docking studies against blaCTX-M15. Site-Identification by Ligand Competitive Saturation (SILCS)-Monte Carlo (SILCS-MC) was applied to the top two ligands selected from molecular docking studies to predict and refine their binding conformations against blaCTX-M15. The SILCS-MC method predicted affinities of −8.6 and −10.7 kcal/mol for Top1 and Top2, respectively, indicating low micromolar binding to the blaCTX-M15 active site. MD simulations initiated from SILCS-MC docked orientations were carried out to better characterize the dynamics and stability of the complexes. Important interactions anchoring the ligand within the active site include pi-pi stacked, amide-pi, and pi-alkyl interactions. Simulations of Top2-blaCTX-M15 complex exhibited stability associated with a wide range of hydrogen bond and aromatic interactions between the protein and ligand. Experimental β-lactamase (BL) activity assays showed that Top1 has 0.1 u/mg BL activity and Top2 has BL activity of 0.038 u/mg with minimum inhibitory concentration of 1 mg/ml. The inhibitors proposed in this study are non-β-lactam-based β-lactamase inhibitors which exhibit the potential to be used in combination with β-lactam antibiotics against multidrug resistant clinical isolates. Thus, Top1 and 2 represent lead compounds that increase the efficacy of β-lactam antibiotics with low dose concentration.

Graphical Abstract

1. Introduction

In the late 1970s to early 1980s Salmonella typhi isolates that were resistant to antibiotics including ampicillin, trimethoprim sulfamethoxazole, and chloramphenicol were observed for the first time¹. Currently, there are about 21 million cases and 222,000 typhoid deaths reported annually around the world. Among these statistics, Asia contributes about 90% of typhoid fever cases with a high ratio of morbidity and mortality. Notable is the haplotype of S. typhi known as H58 that is globally spreading². Most of the prevalence of H58 has been reported across Africa, Oceania and Southeast Asia. Local outbreaks of this strain have been linked to other sub-lineages as well^3,4. Resistance problems cultivated among such strains have been enhanced by certain clinical measures and the horizontal transfer of antimicrobial genes (AMR), facilitated by plasmid and transposable elements.

Similar to other Gram-negative bacteria, one of the most concerning mechanisms of resistance adopted by S. typhi is the emergence of Extended Spectrum β-lactamases (ESBLs) that have originated due to the excessive use of third generation Cephalosporins. In a recent study, blaCTX-M15 producing S. typhi isolates resistant to Cephalosporin have been reported which are labelled as XDR (extensively drug resistant). These XDR isolates are comprised of composite transposon and IncY plasmid having qnrS and blaCTX-M15 genes.

In 2019, the outbreak of an XDR strain of S. typhi in Pakistan was highlighted by an official US report⁵ to which 339 cases were assigned between November 2016 and September 2017. Reportedly, S. typhi is on the watchlist because of its resistant behavior against sporadic third-generation Cephalosporin and fluoroquinolone. Local government and authorities are exerting efforts to monitor sanitation processes, diagnosis, and proper medication against typhoid fever⁶. Currently, the treatment of choice for deadly superbugs is a third-generation Cephalosporin such as ceftriaxone/cefotaxime as prescribed in a WHO report published on 27 December 2018⁷.

The most important class of drugs used for the treatment of bacterial infections are β-lactam antibiotics^8,9. Continuous use of such inhibitors is jeopardized by antimicrobial resistance mechanism mediated by β-lactamases. Four classes of β-lactamases are involved in the mechanism of resistance which include Ambler classes A, B, C, and D. Although these classes have diverse primary structures and catalytic profiles but they are conserved and have overlapping substrate profiles¹⁰. Among the SBLs, ESBLs are of clinical concern as they have the ability to hydrolyze extended spectrum Cephalosporins and the monobactam aztreonam (AZT)¹¹. Currently, a combination of β-lactam antibiotic and β-lactamase inhibitor is used to treat infections caused by β-lactamase-producing bacteria. β-lactamase inhibitors improve the activity of antibiotics by binding and inactivating β-lactamase enzyme. β-lactamase inhibitors stabilize and extend the antibiotic’s spectrum of activity by causing irreversible inhibition of different classes of β-lactamase enzymes. Published studies show that β-lactam-based inhibitors such as clavulanic acid are widely given in combination with penicillin to target antimicrobial resistance. Literature evidence suggests that the activity spectrum of β-lactam based inhibitors is limited to non-carbapenemase enzymes of class A and C SBLs unlike the inhibitors having non-β-lactam structural backbones which are effective against KPC carbapenemase in addition to class A and C β-lactamases. These non-β-lactam β-lactamase inhibitors include boron-based inhibitors which have demonstrated positive activity against serine proteases. A study by Dobozy¹² and coworkers showed the inhibition of penicillinase enzyme from Bacillus cereus with boric acid. Later on reversible inhibitors such as 3-aminophenylboronic acid and phenylboronic acid were developed which demonstrated activities in millimolar range against β-lactamase enzyme of B. cereus¹³. Following this, various experimental¹⁴ and in silico studies¹⁵ were conducted to identify novel potent scaffolds using fragment-based approaches against AmpC β-lactamases. Virtual screenings and covalent docking studies were also conducted to identify noncyclic boronic acid inhibitors against clinically relevant β-lactamases¹⁵. Even though boronic acids have proved to be highly potent against β-lactamases but their major limitation is the potential off-target effects due to inhibition of mammalian serine proteases¹⁶. Another class of compounds 1,6-Diazabicyclo[3,2,1]octanes were also introduced in 1990’s as β-lactamase inhibitors containing bicyclic carbamoyl moiety which demonstrated potential inhibition capacity against class A and C enzymes¹⁷. Subsequently, avibactam was approved by FDA which was given in combination with ceftazidime for the treatment of infections caused by susceptible Gram-negative pathogens. Even though avibactam has proved itself as an effective inhibitor against class A and C, and some of class D enzymes, its efficacy against pathogens containing class D enzymes OXA-23 and OXA-24/40 is very low. This demands extensive modifications and development of avibactam analogs with improved activity¹⁶. Currently, most of the β-lactamase inhibitors that have been introduced as potent are highly susceptible to β-lactamase-catalyzed hydrolysis^18–19,20. However, designing a single inhibitor that has a wide range of activity spectrum against variety of β-lactamase enzymes is still cumbersome.

The current study is therefore focused on the candidate gene blaCTX-M15, a serine beta-lactamase (SBL) that is extensively involved in the resistance mechanism of beta-lactam drugs. The resistance mechanism of the MBLs is due to hydrolysis of all classes of beta-lactam antibiotics²¹. Therefore, there is a need for novel non-beta-lactam inhibitors against the MBLs²². Herein, in-silico structural docking combined with experimental analysis was used to characterize new active compounds, yielding two compounds that are active against the MBL blaCTX-M15. Such compounds have the potential to be used in combination with lactam antibiotics against frontline resistant bacteria.

2. Methodology

2.1. Computational Screening

2.1.1. Plasmid Retrieval and Candidate Gene Exploitation

The plasmid of S. typhi isolate 22420_1_10_Pak60006_2016’s with the accession codes LT882486 and LT906492 respectively, was retrieved from the gene bank of NCBI database and translated via Prodigal software https://github.com/hyattpd/Prodigal²³. The aim was to identify the desired targets that are responsible for increased resistance against beta-lactam drugs which were then characterized against antibiotic resistance by applying comprehensive antibiotic resistance drug database (CARD) server https://card.mcmaster.ca/ontology/36007²⁴.

2.1.2. Molecular Docking

The proposed sequence was checked for availability of a crystal structure in the RCSB protein database (PDB)²⁵. blaCTX-M15 with PDB ID 5T66 having high resolution was obtained in monomeric form and investigated for further characterization. The structure was prepared, and the active site residues reported in the literature were selected for molecular docking studies¹¹.

Structure based docking was performed with seventy-four selected compounds against the protein target using Genetic Optimization for Ligand Docking (GOLD) 5.2²⁶. The relative docking affinity of the molecules was calculated using the GoldScore fitness function. GoldScore fitness function is the original function provided with GOLD that predicts ligand binding position while taking into account the van der Waals energy, H-bond energy, ligand torsion strain, and metal interactions. The GoldScore function²⁷, as described by Jones and coworkers (1997), was designed such that the higher values indicate better the performance of the ligand. Results were analyzed using UCSF Chimera²⁸ and Discovery Studio (DS) https://www.3ds.com/products-services/biovia/.

2.1.3. SILCS Simulations

SILCS simulations^29,30 were performed to explore the distributions of chemical functional group affinity pattern on the surface and active site of blaCTX-M15 using PDB ID 5T66. SILCS uses a combination of Grand Canonical Monte Carlo/Molecular Dynamics (GCMC/MD) to map the functional group requirements of the target protein in terms of free energies. The protein was solvated by water that contained eight solutes of different chemical classes with the system equilibrated using energy minimization and MD simulation. The protein was described using the CHARMM36m protein force field³¹ and CHARMM General Force Field (CGenFF)³² was used for the solute molecules (benzene, propane, methanol, formamide, imidazole, dimethylether, acetate, and methylammonium) along with the CHARMM TIP3P water model³³. The individual equilibrated systems were then subjected to GCMC/MD simulations³⁴ in which water and chemical solutes compete for binding sites on the target protein resulting in the generation of functional group affinity patterns referred to as FragMaps. The full SILCS simulations involves 10 × 100 cycle GCMC/MD simulations where each cycles involves 200,000 GCMC steps and 1 ns of MD from which a total of 1 microsecond of MD sampling over the 10 systems is obtained from which the grid free energy (GFE) FragMaps are obtained. Full details of the SILCS methodology are described in reference³⁴.

The selected top two ligands from molecular docking studies were docked using SILCS-Monte Carlo (SILCS-MC) to predict and refine their binding conformations and binding affinities against blaCTX-M15. The SILCS-MC method subjects ligands to sampling of translational, rotational, and intramolecular dihedral degrees of freedom limited to rotatable bonds in the field of the SILCS grid free energy (GFE) FragMaps that defines the calculated binding affinity based on the ligand grid free energy (LGFE) with the intramolecular energy determined using the CGenFF energy function²⁹. Conformational sampling was performed using full docking in a 5 Å radius sphere centered on the center of mass of the ligand C6S from PDB ID 5T66 (−40.603, 3.073, 19.592). The conformation and predicted binding affinity based on the most favorable LGFE score based on the generic atom-classification scheme. SILCS-MC is also described in detail in Goel et al.³⁴

2.1.4. Molecular Dynamic Simulations

The refined binding conformations of the selected top two ligands obtained from SILCS simulations were subjected to 100 ns long production MD simulations. The MD simulations were performed using AMBER16³⁵. The inhibitor parametrization was done through the general AMBER force field (GAFF) and blaCTX-M15 protein parameters were from the ff14SB force field³⁶. The simulation systems involved box borders a minimum of 12 Å from blaCTX-M15 non-hydrogen atoms in a TIP3P water box. The addition of Na+ ions to the system was used to neutralize it. For 500, 1000, 1000, and 300 steps, the system hydrogen atoms, solvation box, carbon alpha atoms, and all non-heavy atoms were minimized, respectively. Following that the system was heated to 300 K (NVT)³⁷ for 20 ps using Langevin dynamics to achieve system temperature³⁸. The simulation used a time step of 2 fs with 5 kcal/mol-A² harmonic restraints on carbon alpha atoms. The system was relaxed for 100 ps of NVT MD followed by an additional 50 ps of equilibration in the NPT ensemble using Berendsen barostat. Finally, a production run of 100 ns was performed. CPPTRAJ was used to examine the generated trajectories for structural parameters³⁹. The hydrogen bonds produced along the trajectories were depicted in VMD and defined based on a distance of 3.5 Å between the donor (D) and acceptor (A) non-hydrogen atoms.

2.2. Experimental Testing

2.2.1. Antibacterial Disc Diffusion Assay

The compounds antibacterial efficacy was determined using agar disc diffusion technique⁴⁰. A testing culturing media, Mueller Hington Agar (MHA, Oxoid, England) was used. 1 mg of each inhibitor was dissolved in 1 ml of DMSO to make compound stock solutions (dimethyl sulfoxide, Duksan pure chemicals, Korea). As a positive control, cefixime was utilized, and its stock solution was made by dissolving 1 mg of cefixime powder in 1 ml of DMSO. Inoculating fresh test strains, inoculum in sterilized 1 ml of normal saline solution yielded bacterial test strains. The inoculum’s turbidity was then adjusted to 0.5 McFarland (1% BaCl₂ and 1% H₂SO₄). Salmonella typhi (ATCC 14028TM^®) and clinical Salmonella typhi bacterial strains were employed as test strains. Using sterilized cotton swabs, a homogeneous bacterial lawn was created. Filter paper discs (6 mm in diameter, Whatman International Ltd, England) were put on the plates at a 90° angle. On the discs, test compound, a positive control (5 μl, final concentration 5μg/5 μl or 1 μg/l), and a negative control were administered. The plates were then incubated for 24 hours at 37°C. The potency of antibacterial substances was determined by measuring the inhibitory zone in millimeters (mm). The antimicrobial assay was carried out in triplicate, with three values statistically examined to determine the mean and standard deviation (SD). Table 1 shows the minimal inhibitory zone (mm).

Table 1.

Antibacterial activities of compounds Top1 and Top2 against ATCC and clinical bacterial isolates.

Compounds	CHEMBRIDGE ID	BACTERIAL ISOLATES (ATCC and Clinical) zone of inhibition (mm)
		S-typhi ATCC	S-1 (Clinical)	S-2 (Clinical)	S-3 (Clinical)	S-4 (Clinical)	S-5 (Clinical)	S-6 (Clinical)
		M ± SD	M ± SD	M ± SD	M ± SD	M ± SD	M ± SD	M ± SD
Top1	5747390	16.6± 0.4	8.6 ± 0.4	7.6 ± 0.4	8.6 ± 0.4	9.3 ± 0.4	7.6 ± 0.4	7.3 ± 0.4
Top2	32025297	14.6± 0.4	9.3 ± 0.4	8.3 ± 0.4	9.3 ± 0.4	8.2 ± 0.4	7.3 ± 0.4	8.2 ± 0.4
positive Control	Cephalosporin	19.6± 0.4	12.6 ± 0.4	11.6 ± 0.4	11.6 ± 0.4	10.6 ± 0.4	10.6 ± 0.4	12.6 ± 0.4
Negative control	DMSO	0.0± 0.0	0.0± 0.0	0.0± 0.0	0.0± 0.0	0.0± 0.0	0.0± 0.0	0.0± 0.0

2.2.2. Beta-lactamase inhibition assay with IC₅₀ and K_i determination

This inhibition assay is based on the hydrolysis of the chromogenic cephalosporin substrate Nitrocefin, which produces a colored product visible at OD=490nm. This is inversely correlated with beta lactamase activity. By using the abcam^® beta-lactamase kit’s methodology, a standard curve for beta lactamase with linear regression was obtained. The abcam^® beta-lactamase kit’s procedure was followed to prepare the positive control solution (nitrocefin) and reaction mix (ab197003). We added 0.5 g in 100 μl to get a molar stock solution which was then converted into μM concentration. The IC50 was calculated by competitive inhibition experiments with addition of beta-lactamase substrate nitrocefin and competitive inhibitors in separate experiment. Before receiving various dosages of clavulanic acid, Top1, Top2, and blaCTX-M15 were incubated at 1 nM purity. The rate of nitrocefin hydrolysis was assessed by observing the change in absorbance brought on by the breakdown of the β-lactam ring at 490 nm. Plotting the percentage of residual enzyme activity on nitrocefin versus the log of inhibitor concentration allowed us to determine the IC₅₀ values. Using the Cheng-Prusoff adjustment, the inhibition constant, Ki, was determined from the IC₅₀ value as follows⁴¹.

$$Ki=\text{ IC}50/(1+\left(\left[\text{S}\right]/\text{Km}\right)$$ (1)

The nitrocefin concentration and Michaelis-Menten constant are represented here by S and Km, respectively. Nitrocefin’s parameters for kinetics K_m and k_cat were obtained from Michaelis-Menten analysis.

2.2.3. Molecular Identification Assay

Six clinical Salmonella typhi bacterial strains were collected from Pakistan Institute of Medical Sciences (PIMS) in Islamabad. The MDR clinical bacterial isolates were tested using a molecular assay to see if they generate β-lactamase blaCTX-M15.

According to Sambrook and Russell’s standard protocol, plasmid DNA was extracted from six clinical isolates⁴². The plasmids were subsequently separated using a FIGE Mapper Electrophoresis System in 1.0 percent agarose (Bio-Rad, Hercules, CA). A Gel Extraction Kit was used to purify them (Whatman, INC, Model V16–2). For PCR amplification, purified plasmids were used as a source of template DNA.

To identify a gene of interest, primers from the literature were used against the required plasmid DNA template⁴³. As forward and reverse primers, in 4 mM MgCl, 250 pM of each primer (3´-CGCAAATACTTTATCGTGCTGAT-5´) and C-XhoI (5´-GATTCGGTTCGCTTTCACTTT-3´) were employed in this study. Following the amplification, the products were investigated in Seakem LE agarose at a concentration of 2%. (BMA, Rockland, ME). They were also used to perform PCR amplifications using a thermal cycler.

2.2.4. Cytotoxic Activity

The toxicity of both compounds Top1 and Top2 was investigated against brine shrimp using the protocol of reference⁴⁴. Sea water was used in bi-compartment perforated tank with Artemia salina eggs (1.5 gm/L) at one end. The other end of the tank was covered with aluminum foil and the next portion was exposed to light under 30 °C for 48 hours for hatching. About ten shrimps were collected and put in a sterile sea water test-tube. Different concentration of both compounds with a dilution i.e., 100 ug/ml, 75 ug/ml, 50 ug/ml, 25 ug/ml was added along positive control (Doxorubicin) with incubation time of 24 hrs at 37 °C. Dead nauplii were counted and compound LC₅₀ (lethal dose 50) was calculated that shows ≤ 50% of mortality. The percent mortality was estimated using formula,

$$\%\mathit{\text{\hspace{0.17em}mortality}}=\mathit{\text{Dead\hspace{0.17em}shrimp}}/\mathit{\text{Total\hspace{0.17em}shrimp}}\times 100$$ (2)

2.2.5. Scanning Electron Microscopy (SEM).

Using Murtey and Ramasamy (2016) protocol the sample was prepared. Centrifugation was performed to acquire solid pellet of three clinical bacterial strains⁴⁵. After washing the pallet twice, it was dehydrated using gradient solvent; ethanol for 20 minutes. This investigation was set to check both the compounds (Top1 and Top2) potency along with cefixime at 1:1 μg/μl and control group using blank sample.

3. Results and Discussion

Seventy-four putative CMY-10 inhibitors exhibiting enhanced BL-activity identified in our previous study were selected for screening against blaCTX-M15⁴⁶. Computational methods were initially applied followed by experimental identification and characterization of a subset of those compounds as effective inhibitors. These compounds have the potential to be developed into inhibitors or used in combination therapy as they have the potential to treat resistant strains of S. typhi. The complete flow of work is shown in (Figure 1).

Figure 1.

Workflow of computational and experimental work.

3.1. Computational Screening

3.1.2. Plasmid Retrieval

Plasmid retrieved from the literature having a sequence size of 84492 bp was translated using prodigal. The protein sequences were checked against virulence factor databases to identify the candidate gene responsible for multi-drug resistance. It was inferred that the gene blaCTX-M15 is responsible as that it may potentially hydrolyze β-lactam drugs. Literature evidence indicates novelty of the target among Salmonella typhi infections especially in Pakistan and other developing countries⁹. Further investigation shows antibiotic resistance ontology. We used the CARD reference data to do a BLAST analysis and identify 6 antibiotic resistance genes⁴⁷. These include strict, loose and perfect antibiotic resistance genes i.e., APH families, sul2 (strict), QnrS1, TEM1 and blaCTX-M15 (Figure 2). The TEM1 gene confers resistance to monobactam and cephalosporin antibiotics, as well as penam. In contrast, the CTX-M-15 gene provides resistance against cephalosporin and penam antibiotics. Additionally, the QnrS1 gene imparts resistance to sulfonamide antibiotics, while the APH(3”)-Ib and APH(6)-Id genes perform aminoglycoside antibiotic inactivation (Table S1).

Figure 2.

Identification of the drug target region on plasmid including descriptions of the molecular basis for resistance genes with comprehensive antibiotic resistance targets. RGI (resistance genes identifier) circular plot depicting the resistance genes with exact matches to the CARD reference sequences; herein, “perfect” denotes a high degree of sequence similarity between a gene contained in the input data and a reference resistance gene kept in the database. “Loose” hits show discovery of novel resistance genes and the Strict algorithm is used for high-confidence detection of resistance genes.

3.1.3. Molecular Docking

Gold docking was performed on the 74 compounds targeting the active site of blaCTX-M15 (Table S2). Two of those compounds, having Chembridge ID: 5747390 (Top1) and 32025297 (Top2), exhibited high Goldfitness scores of 75.8 and 57.5, respectively, against the target protein. The Goldfitness score for the rest of compounds are presented in the supplementary material (Table S2). Both the compounds have multiple hydrogen bond interactions with the active site residues of the protein including Ser73, Asn107, Asn135, Ser240, Lys76, Gly242, and Ser133. Additionally, alkyl-pi, amide-pi stacked, and pi-pi stacked interactions occur between the aromatic rings of the ligands and protein residue Tyr108 and Gly241. These represent additional interactions playing a role in anchoring the ligand within the binding pocket (Figure 3). The druglikeness of the compounds was computed using SwissADME⁴⁸. According to this analysis compound Top1 is indicated by Egan rule as drug like (WLogP ≤ 3.92, XLOGP3 ≥,3.95, hydrogen bonds donor 2, water Solubility = moderatly soluble) with PAINS alert of 0⁴⁹. Compound Top1 was met all the rules of druglikeness (Lipinski, Ghose, Veber, Egan and Muegge) with molecular weight 388.43 g/mol, Num. heavy atoms 29, Num. rotatable bonds 7, Num. H-bond acceptors 5, Num. H-bond donors 2, TPSA 102.55 Å2, WLOGP 2.13, XLOGP3, 2.20, Water Solubility = soluble with 0 PAINS alert⁵⁰.

Figure 3.

Depiction of top docked complexes highlighting the interacting residues of the active site. Ligand is represented in yellow and blue and protein is in cyan ribbon representation (a) Top1 compound (Chembridge ID: 5747390) (b) Top2 compound (Chembridge ID 32025297).

3.1.4. SILCS Simulations

After GOLD molecular docking the binding modes of the compounds were futher verified using the SILCS method³⁴, which characterizes the functional group requirements of the active site in terms of GRE FragMaps. The results show apolar and negatively charged Fragmaps in the active site of blaCTX-M15 receptor (Figure 4). Overlay of the crystal binding mode of cyclic boronate 1 from the experimental blaCTX-M15–1C complex (PDB ID 5T66, ligand C6S) shows the ligand to align with the different FragMap types indicating that the SILCS method properly characterizes the functional group affinity pattern of the binding site (Figure 4A). For example, the cyclohexane carbaldehyde group positions itself adjacent to the apolar Fragmaps whereas the fused bicyclic boronate ring as well as the acid groups overlap with the negative Fragmaps. Accordingly, the SILCS-MC method was used to redock the two compounds into the binding site. This docking yielded LGFE scores, the SILCS metric indicative of binding affinity, of −8.6 and −10.7 kcal/mol for Top1 and Top2, respectively. While the LGFE scores do not directly correspond to experimental binding affinities, these values are consistent with low to sub micromolar binding affinities. Notably, the orientation of Top1 in the binding site is flipped relative to that predicted by GOLD docking (Figure 4B) while the binding orientations of Top2 are similar for the two docking methods (Figure 4C).

Figure 4.

(a) X-ray crystal orientation of inhibitor C6S from PDB ID 5t66 and SILCS-MC docked orientations of (b) Top 2 and (c) Top 1 in the binding pocket of blaCTX M15. The protein is shown as a white solvent accessible surface along with the SILCS FragMaps for negative (orange), apolar (green), hydrogen bond donor (blue) and hydrogen bond acceptor (red) functional groups. Negative and apolar FragMaps are shown at a GFE cutoff of −0.9 kcal/mol and hydrogen bond donor and acceptor maps at a cutoff of −0.6 kcal/mol. 2D images of the ligands are shown as insets in the respective panels.

SILCS offers the capability to both qualitatively and quantitatively understand the contributions of individual atoms in ligand binding³⁴. Therefore, analysis of the docked orientations was undertaken to indicate the regions of the ligands that contribute to binding based on the overlap of the ligand functional groups with the corresponding FragMaps. The Top1 docked complex shows the 4-methyl-2,5-dihydropyrimidin-2-amine group to be in a region the includes apolar, hydrogen bond acceptor and hydrogen bond donor groups consistent with the aromatic nature of the heterocycle that includes the exocyclic hydrogen bond donor NH moiety. Similarly, hydrogen bond donor and acceptor maps encompass the N-(4-methyl-4H-pyrazol-3-yl)acetamide moiety with apolar FragMaps around the terminal phenyl group. In the Top2 docked complex the 1,2,4-triazole group is aligned with hydrogen bond acceptor FragMaps with the two phenyl rings attached to the triazole are encompassed by apolar FragMaps to different degrees. In addition, hydrogen bond acceptors FragMaps correspond with the 3 methoxy groups on the terminal phenyl group. On the other end of the molecule hydrogen bond acceptor and donor maps coincide with the amide group and the negative FragMaps are overlaid on the acid group. The lack of apolar FragMaps on the phenyl ring that contains the acid group indicates a larger role of the acid over the ring in driving ligand binding. Overall, the analysis of the two compounds and the SILCS FragMaps indicate moieties on the molecules that make the most substantial contributions to binding.

Quantitative contributions of individual atoms or moieties in the compounds to the overall predicted LGFE scores are based on the atomic GFE contributions. Presented in Figure 5 are GFE contributions of the different moieties in the two compounds, where the atomic GFE contributions have been summed over the rings and their substituents or the linker regions between the rings. With Top1 the GFE contributions are more balanced with the disubstituted benzoic acid ring contributing −2.90 kcal/mol with the trimethoxyphenyl group on the other end of the molecule contributing −3.48 kcal/mol. The phenyl ring makes a small contribution of −0.34 kcal/mol consistent with the lack of apolar FragMaps around the moiety (Figure 4C) suggesting that removing this group would have a small impact on the binding affinity while minimizing the molecular weight which may be beneficial for bioavailability. In Top2 the terminal pyrimidine ring and its -NCH3 substituent make the largest contribution to the overall affinity, −4.04 kcal/mol, with the phenyl ring only contributing −0.69 kcal/mol, a value less than the linker region of the molecule which contributes −1.06 kcal/mol. Analysis of the FragMaps overlaid on Top1 in Figure 4B show the presence of negative maps around the ring indicating that the addition of an acid group to the ring would improve ligand affinity. These results indicate how FragMaps may be used in future studies to facilitate modifications to the ligands to further improve their binding affinities as well as maximize bioavailability.

Figure 5.

GFE contributions of the rings with their substituents and the linker regions of (a) Top1 and (b) Top2. Atomic GFE contributions are summed over all the classified atoms in the rings and their substituents or over all the classified atoms in the linker regions between the rings. The sum of the GFE contributions yields the LGFE scores for each ligand.

3.1.5. Molecular Dynamics simulations

To better characterize the dynamics and stability of the complexes MD simulations for Top1 and Top2 were carried out, initiated from the SILCS-MC docked orientations. Initially 200 ns simulations were run for both the complexes, but they were extended to 400 ns for Top 1 for ensuring the stability and anchoring of ligand in the binding pocket. In the complex with Top1 it was observed that the RMSD of ligand continuously fluctuated throughout the MD simulation (Figure 6a). This indicates that the initial ligand-receptor interactions were changed throughout the simulation and ligand exhibited multiple binding orientations while re-positioning itself 5 times during the 200 ns simulation as shown in Figure 7. This is consistent with the differences in the GOLD and SILCS docked orientations and the less favorable SILCS LGFE score vs. Top2 discussed above, although the overall orientation of the ligand obtained by SILCS was retained throughout the MD simulation. Nonbonded interactions involved pi stacking of the aromatic rings which appear to maintain the ligand within the cavity of receptor’s active site. Phenyl triazole and dimethoxybenzoate occupy the hydrogen bond acceptor regions depicted by the Fragmaps suggesting the crucial role of acidic functional groups in driving ligand binding. Tyr79 and 103 are playing a significant role by forming pi-pi stacked interactions with the aromatic ring of the ligand. The functional group S-methyl methylcarbamothioate initially engaged in hydrogen bond interactions with Asn78 constantly flips at 180° back and forth to form electrostatic interactions with multiple residues of the binding pocket such as Ser44, 211, and Arg248 causing increased entropy penalty. This highlights the region of the ligand that is more flexible and contributes to a larger magnitude of entropy loss. The information can be further exploited for lead optimization. For instance, restricting the movement of S-methyl methylcarbamothioate region by adding unsaturated groups and ring systems can generate structurally rigid lead compound with improved affinity.

Figure 6.

Root-mean-squared difference (RMSD) of the protein (blue) and ligand (black) non-hydrogen atoms from the (a) Top1 and (b) Top2-complex MD simulations. RMSD values were calculated following alignment of the protein non-hydrogen atoms with the 5T66 crystallographic coordinates.

Figure 7.

Top1 ligand orientations during the MD simulation. (a) Overall binding orientation of the ligand in the pocket at 5 ns (yellow), 50 ns (green), 100 ns (magenta), 150 ns (pink) and 200 ns (orange), (b) zoomed view of ligand conformation with selected functional groups labeled from selected time frames from the MD simulation. Details of the Top1-protein interactions at (c) 50 ns, and (d) 200 ns.

The MD simulation of the Top2-protein complex showed the interactions to be much more stable as compared to Top1 (Figure 6b, 7). Analysis of the simulation shows that receptor residues mainly interacting with the ligand include Pro242, Gly213, Ser246, Gly212, Ser211, Asn144, Thr145, Tyr214, Lys243 and Ile147. The role of pi-pi stacked and pi-alkyl interactions contribute to anchoring the ligand within the binding pocket. In addition, hydrogen bonding interactions are also present between receptor and ligand contributing to further stability of the complex. Consistent with the SILCS FragMap analysis above and the more favorable LGFE scores, the wide range of interactions between the ligand and the protein contribute to the overall stability of the ligand in the binding site (Figure 8).

Figure 8.

Top2 ligand orientations during the MD simulation. (a) Overall binding orientation of the ligand in the pocket at 150 ns (yellow) (b) 200 ns (cyan) (c) Details of the Top2-protein interactions at 200 ns. (d) zoomed view of ligand conformation at 200 ns.

3.2. Experimental Analysis

Based on the computational analysis indicating Top1 and Top2 to binding stably to the protein active site experiments were undertaken on these compounds to evaluate their biological activity.

3.2.1. Antibacterial and Minimum Inhibitory Activity

Antibacterial assay investigation of both compound Top1 and Top2 showed active response against Gram-negative Salmonella typhi (ATCC^® 14028^™) and the clinical strains (Table 1). The mean zone of inhibition of 16.6 mm was recorded for Top1 and 14.6 mm for Top2 against Salmonella typhi (ATCC^® 14028^™). Whereas mean zone of inhibition 8.1 mm for Top1 and 8.4 against Top2 in response to clinical strains of Salmonella typhi. Furthermore, synergistic effects of both inhibitors were checked along the antibiotic Cephalosporin against all the 6 clinical strains of Salmonella typhi. Mean zone of inhibition was measured showing less resistance to the mixture of inhibitor and Cephalosporin (see Table 2). As evident, there is a significant increase in the zone of inhibition in the presence of Top1 or Top2 with Cephalosporin as compared to Cephalosporin alone. This indicates the possibility of a combination therapy where binding of proposed inhibitor to the β-lactamase and antibiotic to peptidases further can block the cell wall synthesis. Both the compounds showed zone of inhibition to some extent, as several studies have inferred that compounds (Top1) with pyrimidin-4-yl]-1H-pyrazol-1-yl and compound (Top2) having such moieties have been found to exhibit antibacterial properties, which is consistent with previous literature^51–53.

Table 2.

Synergistic assay of both compounds against clinical strains of Salmonella typhi

SYNERGISTIC ASSAY		BACTERIAL ISOLATES (Clinical) Salmonella typhi strains
		S-1	S-2	S-3	S-4	S-5	S-6
		Zone of inhibition (mm)
Top1+ Cephalosporin	Chembridge-ID-5747390	18.6	20.3	18.3	20.3	18.3	19.3
Top2 Cephalosporin	Chembridge-ID-32025297`	19.6	18.3	19.3	19.3	18.3	20.3
Positive Control	Cephalosporin	8.6	9.4	10.6	10.6	10.3	9.4
Negative control	DMSO	0	0	0	0	0	0

Quantitative Determination of blaCTX-M15 inhibition

In addition, both Top1 and Top2 demonstrated an efficient β-lactamase inhibition with a threshold of different concentration shown in Table 3. This result shows that both compounds may be resistant to BL hydrolysis thereby offering new compounds for consideration for combination therapy in addition to previously published non-β-lactam BLIs⁵⁴. The inactivation kinetic parameters were obtained using the reporter substrate nitrocefin. The kcat value for blaCTX-M15 was determined to be 25.9 min⁻¹ based on initial rates observed at saturating substrate concentrations analyzed using GraphPad prism software⁵⁵. In addition, the Top1 and Top2 half maximum inhibitory concentration (IC₅₀) values were determined with Clavulanic acid for blaCTX-M15. The results showed Top1 to have a lower IC50 value of 200 μM and inhibition constant, K_I, value = 41.15 μM, whereas Top2 had a similar K_I of 45.41μM. These results show the Top1 values to be comparable to that of clavulanic acid and have a high affinity as shown in Table3. Graphs of the percent hydrolysis as a function of concentration for the three inhibitors are shown in Figure 9. Notably, the effectiveness of the compounds as inhibitors was consistent with the suppression of in vitro growth.

Table 3.

Beta-lactamase inhibitory activity of compound Top1 and Top2 at different concentrations with IC50 and Ki value.

concentration ug/ml		BL inhibitory activity
	Log10	Top1	Top2	abcam Clavulanic acid
0.03	−1.5	0.921	0.875	0.9345
0.09	−1.0	0.829	0.758	0.8498
1	0	0.745	0.685	0.753
5	0.6	0.549	0.567	0.623
10	1	0.43	0.457	0.53
50	1.6	0.198	0.198	0.239
100	2	0.076	0.098	0.09
150	2.1	0.045	0.0547	0.02
200	2.3	0.0184	0.0432	0.015
IC 50 , μM		200	220.8	189.3
K i , μM		41.15	45.41	38.94

Figure 9.

Graphical representation of the hydrolysis of 100 mM nitrocefin used to measure the activity of blaCTX-M15 followed by 5-minute pre-incubation with Top1, Top2 and clavulanic acid inhibitor doses using non-linear regression curve.

3.2.2. Cytotoxic Assay

Compounds were subjected to experiments for cytotoxic effect at different concentration using a brine shrimp model (Table 4). Concentration dependent brine shrimp lethality assay of both compounds showed % of inhibition against the log concentration (logC)⁵⁶. Increase in concentration of both compounds show a small decrease in the survival of nauplii. Comparing with positive control shows both compounds to only be less toxic with LD₅₀ values of 333.3 μg/ml and 281 μg/ml versus 187 μg/ml for the Doxorubicin control⁵⁷. Whereas negative control (DMSO) shows LD₅₀ value of 1000 μg/ml highlighting no toxic response.

Table 4.

Cytotoxic activity in a brine shrimp model of compounds Top1 and Top2 at different concentration with LD₅₀ value.

compounds	Concentration (μg/ml)	logC	No of nauplii	dead	live	% mortality	LD50 μg/ml
Top1	25	1.3	10	1	9	10	333.3
	50	1.6	10	1	9	10
	75	1.8	10	1	9	10
	100	2	10	2	8	20
Top2	25	1.3	10	1	9	10	281
	50	1.6	10	1	9	10
	75	1.8	10	2	8	20
	100	2	10	2	8	20
Positive control (Doxorubicin)	25	1.3	10	1	9	10	187
	50	1.6	10	2	8	20
	75	1.8	10	2	8	20
	100	2	10	3	7	30
							1000
Negative control (DMSO)	25	1.397940009	10	1	9	10
	50	1.698970004	10	1	9	10
	75	1.875061263	10	1	8	20
	100	2	10	1	9	10

Morphological analysis was performed using SEM experiments to analyze the changes in bacterial cell wall of untreated and control treated (Cefixime) samples in comparison with the cells exposed to the β-lactam with β-lactamase inhibitor combinations of Top1+Cefixime and Top2+Cefixime (Figure 10). Both compound Top1 and Top2 were applied during minimum inhibitory concentration assays in combination with Cefixime. Results inferred showed some apparent morphological variations in cells of the pathogens treated with non β-lactam/β-lactamase inhibitor combinations. A clear formation of pores was observed that depicted the destruction of cell wall. On the other hand, relatively smooth surface was found in the untreated and treated control (cefixime) samples in all the three strains as compared to treated sample with the β-lactam/β-lactamase inhibitor combinations. Hence, these findings further suggest that both inhibitor Top1 and Top2 have the potential for therapeutic efficacy in combination therapy against MDR clinical isolates.

Figure 10.

Scanning electron microscopy of Salmonella typhi clinical strains in triplicate against untreated, treated (control) and combinational therapy for compound Top1 and Top2. Red circle highlighting the pores that shows the changes during combination therapy in cell wall against compound Top1 and Top2. During this investigation close morphological changes have been observed upon different scale 2μm and 10μm respictvely.

Various research studies have highlighted the positive role of using β-lactamase inhibitors along with the antibiotics in treating multidrug resistant bacterial infections⁵⁸. The morphological changes observed in the SEM analysis suggest that the selected lead compounds have negatively affected cell wall synthesis in MDR clinical isolates. These images have been clearly adjusted and visualized upon different scale via SEM analysis (2μm and 10μm) depends upon the outcomes we have obtained. Experimental analysis including β-lactamase inhibition assay and antibacterial assay were also in agreement with the computational results suggesting that these compounds can significantly contribute to overcome antibiotic resistance through binding to blaCTX-M15. Both compounds investigated in the current study lack β-lactamase core structure and are thus capable of escaping different resistance mechanisms used by pathogens against β-lactam drugs.

Conclusions

Antimicrobial drug resistance is a worldwide health crisis that jeopardises achievements in many fields of medicine, including surgery, cancer treatment, organ transplantation, and preterm baby survival. The current study has explored a potential therapeutic target candidate, blaCTX-M15, a β-lactamase to identify non β-lactam based inhibitors that can inhibit Salmonella typhi serovar, high prevalence of which has been reported in Pakistan. Computational analysis predicts that the lead compounds Top1 and Top2 are high affinity binders of the target protein. SILCS simulations have revealed that the binding of compounds to the active site of blaCTX-M15 is driven by a collection of different types of interactions as indicated by overlap of the ligands with SILCS Fragmaps. This analysis highlights key regions of the ligands that are important for binding that can be used for further lead optimization. For example, in future studies modification of Top1 by introducing suitable ring structure at S-methyl methylcarbamothioate region will help improve binding affinity and reduce entropic penalty. Supporting the SILCS results are MD simulations showing the ligands to remain bound to the protein on the time scale of 200 ns. Results of experimental analysis validated computational findings showing the selected lead compounds to exhibit 0.1 and 0.038 BL inhibition activity with minimum inhibitory concentration of 1 mg/ml. Moreover, minimum inhibitory concentration of compounds given in combination with the lactam antibiotic resulted in increased inhibiton. This inhibitory effect of compounds during combination therapy has also been witnessed in SEM analysis in which damaged cell wall of clinical isolates of Salmonella typhi strains can be observed.

Clinically available inhibitors such as clavulanic acid, tazobactam, sulbactam, all share a common β-lactam ring and thus are at increased risk of getting hydrolyzed during the inhibition process⁵⁵. The two lead compounds proposed in the current study are non-β-lactam based β-lactamase inhibitors which can potentially work in a synergetic fashion when given in combination with a β-lactam antibiotic. Class A extended spectra β-lactamases contain a catalytic serine at the active site which makes a nucleophilic attack on the substrate wherein Glu166 is believed to act as a base. Docking, SILCS modeling, and MD simulations of the lead compounds show that both ligands occupy the binding cavity, thereby trapping the functionally important residues of the catalytic site. These observations combined with the experimental activity indicate the potential of Top1 and Top2 as biologically active lead compounds with the potential to be developed into therapeutically useful β-lactamase inhibitors.

Supplementary Material

SI

Table S1. AMR gene families with antibiotic resistance mechanism.

TableS2. Goldfitness score of all the screened compounds.

Abbreviations

SILCS-MC

Site identification by ligand competitive saturation-Monte Carlo

LGFE

ligand grid free energy

MD

Molecular Dynamics

SEM

Scanning electron microscopy

Data and Software Availability

Docking files along with SILCS and AMBER input files for simulations are available in supporting information. The plasmid was retrieved and screened for potential therapeutic target candidate, blaCTX-M15 https://www.ncbi.nlm.nih.gov/nuccore/LT906492 from the NCBI database. Tools (Prodigal) used to translate genomes are freely available at https://github.com/hyattpd/Prodigal. GOLD software and SILCS technology used in this study are proprietary. The SILCS software suite is available at no charge to academic users from SilcsBio LLC. Molecular Dynamics Simulations were performed using AMBER 16 suite. The AmberTools suite is free of charge and its components are mostly released under the GNU General Public License (GPL).