Antibiotic resistance and inhibition mechanism of novel aminoglycoside phosphotransferase APH(5) from B. subtilis subsp. subtilis strain RK

Abstract

Bacterial resistance towards aminoglycoside antibiotics mainly occurs because of aminoglycoside phosphotransferases (APHs). It is thus necessary to provide a rationale for focusing inhibitor development against APHs. The nucleotide triphosphate (NTP) binding site of eukaryotic protein kinases (ePKs) is structurally conserved with APHs. However, ePK inhibitors cannot be used against APHs due to cross reactivity. Thus, understanding bacterial resistance at the atomic level could be useful to design new inhibitors against such resistant pathogens. Hence, we carried out in vitro studies of APH from newly deposited multidrug-resistant organism Bacillus subtilis subsp. subtilis strain RK. Enzymatic modification studies of different aminoglycoside antibiotics along with purification and characterization revealed a novel class of APH, i.e., APH(5), with molecular weight 27 kDa approximately. Biochemical analysis of virtually screened inhibitor ZINC71575479 by coupled spectrophotometric assay showed complete enzymatic inhibition of purified APH(5). In silico toxicity study comparison of ZINC71575479 with known inhibitor of APH, i.e., tyrphostin AG1478, predicted its acceptable values for 96 h fathead minnow LC₅₀, 48 h Tetrahymena pyriformis IGC₅₀, oral rat LD₅₀, and developmental toxicity using different QSAR methodologies. Thus, the present study gives novel insight into the aminoglycoside resistance and inhibition mechanism of APH(5) by applying experimental and computational techniques synergistically.

Electronic supplementary material

The online version of this article (10.1007/s42770-019-00132-z) contains supplementary material, which is available to authorized users.

Introduction

Now-a-days resistance to antibiotics is a serious global public health problem [1]. Antibiotic resistance is not just limited to a particular class of antibiotics, but associated with all classes of currently used antibiotics. This, multidrug resistance ability tends to emergence of resistant pathogens showing insensitivity towards available therapeutic drugs [1]. The mechanism by which bacteria develop resistance to different antibiotics is highly diverse [2]. The aminoglycosides correspond to complex family of broad spectrum antimicrobial agents consisting of an aminocyclitol nucleus (streptamine, 2-deoxystreptamine, or streptidine) linked to amino sugars through glycosidic bond [3, 4]. These antibiotics are primarily used to treat several infections caused by Gram-negative aerobic organisms, Staphylococci, and other Gram-positive organisms [5, 6].

Several reasons are known to develop resistance against aminoglycosides which includes the presence of aminoglycoside-modifying enzymes that inactivates aminoglycosides [2], reduced permeability towards aminoglycosides by modification of outer membrane [7], alteration of drug target site [8], or export of aminoglycoside outside the cell by active efflux pump [9]. Among them, the major mechanism of bacterial resistance towards aminoglycosides is the chemical modification of drug by bacterial enzymes [2]. There are three different types of aminoglycoside-modifying enzymes; each of these modifies aminoglycosides by transferring a functional group to the aminoglycoside antibiotics thereby inactivating its activity [2]. It includes aminoglycoside O-nucleotidyltransferases (ANTs), which transfer a nucleoside monophosphate (AMP) from donor substrate ATP to hydroxyl group of aminoglycoside molecule, aminoglycoside acetyltransferases (AACs) transfer acetyl group from acetyl-CoA to aminoglycoside, and the aminoglycoside phosphotransferases (APHs) catalyze the transfer of γ-phosphate group from ATP or GTP to antibiotic substrate [10]. Aminoglycoside phosphotransferases also known as aminoglycoside kinases (AKs) are one of the most common sources for aminoglycoside antibiotics resistance [11]. Various studies have reported about the molecular characterization of AK enzymes with aminoglycoside and nucleotide substrates [11–13]. These studies revealed structural and functional diversity of antibiotic binding sites, with higher degree of structural similarity for NTP binding site of AK enzymes [12]. Despite sequence variation, all aminoglycoside kinase enzymes adopt a common eukaryotic protein kinase (ePK)–like fold, which is the NTP binding site of AKs and ePKs [12, 13].

Members of Bacillus genus ubiquitously occur in an environment. Thus, occurrence of antibiotic resistance in these bacteria could lead to severe clinical manifestation as they are known as causative agent for several pathological complications [14, 15]. It is thus necessary to understand exact antibiotic resistance mechanism of AKs in detail at atomic level in these organisms. Thus, in the present study use of various experimental techniques helped to understand the enzymatic cause of resistance from newly isolated resistant organism Bacillus subtilis subsp. subtilis strain RK. The NTP binding site of ePKs is structurally conserved with AKs, which is the most extensively studied drug target site [13]. On this basis in our earlier study, we have investigated computationally a potent virtually screened inhibitor ZINC71575479 by targeting the NTP-binding site of one of the known APH and tested its binding affinity towards different APH from diverse MDR strains in comparison with known inhibitor tyrphostin AG1478 [12, 16]. This lead-like molecule (ZINC71575479) when tested experimentally showed enzymatic inhibition of purified novel APH(5) enzyme isolated from Bacillus subtilis subsp. subtilis strain RK, thus validating the in silico results. We believe that these results could open new avenues to investigate the enzymatic cause of resistance and design potent inhibitors against enzymes, which impart antibiotic resistance.

Materials and methods

Screening and identification of aminoglycoside-resistant bacteria

The aminoglycoside-resistant bacteria were isolated from soil by performing serial dilution and agar spread plate techniques [17]. 0.1 ml of serial dilutions ranging from 10⁻¹ to 10⁻⁵ was spread uniformly on Mueller–Hinton (MH) agar plates aseptically containing 50 μg/ml of streptomycin concentration. These plates were incubated at 37 °C for 24 h and observed for the appearance of streptomycin-resistant organism after incubation. The streptomycin-resistant species was then subcultured to get a pure form for its 16S rDNA gene sequence identification [18]. The obtained 16S rDNA gene sequence was then further analyzed for homology and phylogeny. NCBI’s BLASTn program was used with nr database of GenBank to find the homologous 16S rDNA gene sequences to target sequence [19]. Based on the maximum identity score, 20 sequences were selected and aligned using ClustalW [20]. The multiple sequence alignment file generated using ClustalW was then used to create phylogram using MEGA 4.0 (Molecular Evolutionary Genetic Analysis) to study the evolutionary relationship of the isolated streptomycin-resistant organism [21].

Determination of minimum inhibitory concentration and protein overexpression profiling study

The minimum inhibitory concentration (MIC) for streptomycin-resistant strain was determined by agar dilution method using MH agar. This is the most widely used medium for MIC testing and meets the requirement of NCCLS (National Committee for Clinical Laboratory Standards) [22, 23]. The MH agar plates were prepared by diluting the stock of streptomycin (1 mg/ml) into MH agar medium to meet desired concentration ranging from 50 to 500 μg/ml. Then, 0.1 ml of inocula of B. subtilis subsp. subtilis strain RK was allowed to spread on each plate, and plates were incubated at 37 °C for 24 h for determination of MIC. Similar procedure was followed for determination of MIC for B. subtilis subsp. subtilis strain RK against gentamicin, kanamycin, and amikacin to detect multidrug resistance ability.

Cell free lysate of streptomycin-resistant B. subtilis subsp. subtilis strain RK was prepared for protein overexpression profiling as earlier study [23]. Similar procedure was repeated to obtain cell free lysates at different concentrations of streptomycin (100 μg/ml, 200 μg/ml, 300 μg/ml, and 400 μg/ml) to check the consistency of overexpressed protein [23]. The sodium dodecyl sulfate polyacrylamide gel electrophoresis technique (SDS-PAGE) [24] was used for analysis of whole cell protein profile of streptomycin-resistant B. subtilis subsp. subtilis strain RK, which eventually helped in locating the overexpressed protein.

Identification of aminoglycoside-modifying enzyme and HPLC analysis

The aminoglycoside-modifying enzyme (AME) identification was carried out initially by microbiological assay [23]. The Proteus vulgaris was used as a sensitive organism for AME detection through microbiological assay by confirming its sensitivity towards aminoglycosides. The reaction mixture for identification of phosphorylating, adenylating, or acetylating type of AME was prepared according to earlier study [25]. In the present study, we have used K-Na-phosphate buffer instead of PMK buffer [25]. The reaction mixture was incubated for 1 h at 37 °C, and then, enzymatic reaction was stopped by boiling for 3 min. The reaction mixture containing modified antibiotics was then used for microbiological assay on MH agar plates with their respective standard antibiotics of same concentration. For detection of AME, plates were incubated at 37 °C for 24 h to see difference in zone of inhibition of modified antibiotic in comparison with standard antibiotics. The Shimadzu prominence HPLC system equipped with degasser DGU-20A 5R, low-pressure quaternary pump LC 20 AD, and photo diode array detector SPD-M20 A was used for HPLC analysis of streptomycin, gentamicin, kanamycin, amikacin, and their modified forms. The separation of antibiotics was achieved by using reverse phase C-18 column (Enable 250 × 4.6 mm, 3 μm). All the HPLC-grade solvents used for the analysis were purchased from Sigma-Aldrich. Solvents were filtered through 0.2-μ sterile CN membrane filters (C-152 M Axiva) using filtration assembly (Riviera glasses Pvt. Ltd Mumbai, India) and degassed by ultrasonic cleaner (Revotek). The mobile phase consisted of acidified water (A) and organic mobile phase acetonitrile (B) with 80:20% respectively by considering flow rate of 1 ml/min. All the antibiotics and their modified products were identified by difference in their retention time, comparing the UV–Visible spectra and spiking with standards at 254 nm. All the data acquisition and post run analysis was performed by Lab solution software [26, 27].

Purification and characterization of aminoglycoside phosphotransferase

The purification of aminoglycoside phosphotransferase (APH) was performed by ion exchange chromatography technique using DEAE cellulose resin. The crude cell free lysate with APH activity for purification was obtained as earlier procedure [23]. The elution of column was carried out by linear gradient of NaCl with concentration ranging from 0.1 to 0.6 M prepared in K-Na-phosphate buffer with flow rate of 1 ml/min. Total, 70 fractions were collected and protein concentration of eluted fractions was determined by noting the absorbance at 280 nm using double beam UV spectrophotometer. Thus, fractions showing highest protein peak were then further confirmed for APH activity by qualitative and quantitative methods like microbiological assay and coupled spectrophotometric assay as described earlier [23, 28]. The molecular weight of the purified enzyme from active fraction was estimated by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) [23, 24]. Phosphorylase b (molecular weight, 97,400 Da), bovine serum albumin (molecular weight, 66,000 Da), pyruvate kinase from rabbit muscle (molecular weight, 58,000 Da), ovalbumin (molecular weight, 43,000 Da), carbonic anhydrase (molecular weight, 29,000 Da), and Trypsin soyabean inhibitor (molecular weight, 20,100 Da) were used as standard proteins to determine the molecular weight of purified APH.

Elucidation of reaction mechanism of purified APH by MS/MS analysis

The antibiotic modification mechanism of purified APH from B. subtilis subsp. subtilis strain RK was studied by using mass spectrometry. The instrument used in this study was UHPLC–QTOFMS, an Agilent Series 1290 infinity rapid resolution LC system interfaced with electrospray ionization (ESI) to an Agilent 6540 UHD Accurate Mass Q-TOF LC/MS [26]. Chromatographic separation was achieved using different columns initially; however, to optimize separation conditions and to acquire maximum metabolites, ZORBAX RRHD Eclipse Plus reversed phase C18 analytical column (100 mm × 2.1 mm, 1.8 μm particle size) was kept at 40 °C temperature. The injected sample volume was 2 μl in positive ionization mode. The mobile phases used were eluent A 0.1% formic acid in water and eluent B 0.1% formic acid in methanol. A flow rate of 0.4 ml/min was used for elution, but during column equilibration, it increased to 0.5 ml/min. Data was collected using Mass Hunter Workstation software (Agilent, version B.05.00). To keep up mass accuracy during run time, a reference mass solution with purine and hexakis (1H, 1H, 3Htetrafluoro propoxi) phosphazine was used, with m/z 121.0508 and m/z 922.0097 reference ions respectively in positive ionization mode and m/z 112.3985 and m/z 1034.9881 in negative ionization mode.

Inhibition study of purified APH by virtually screened inhibitor

The most potent virtually screened inhibitor with ZINC ID 71575479 investigated from our earlier study [16] was purchased from Sigma-Aldrich to study in vitro enzymatic inhibition of purified APH. This inhibitor was selected based on our earlier results obtained through virtual screening, molecular docking, molecular dynamics simulation, and binding free energy calculation study performed on one of the known APH [16]. In vitro enzymatic inhibition of purified APH by virtually screened inhibitor was studied using coupled spectrophotometric assay [28, 29]. The components of the reaction mixture for coupled spectrophotmetric assay were same as described earlier [28]. For inhibition study of APH, an extra cuvette was used comprising all components like earlier study [28, 29] along with virtually screened inhibitor. The monitoring of the reaction was carried out by detection of oxidation of NADH at 340 nm spectrophotometrically. The inhibition result of virtually screened inhibitor ZINC71575479 [16] was compared with tyrphostin AG1478, a known inhibitor of APH as a control [12].

In silico toxicity study of ZINC71575479 and tyrphostin AG1478

Toxicity Estimation Software Tool (T.E.S.T) developed at the US EPA (Environmental Protection Act) [30] was implemented to study in silico toxicity predictions for virtually screened and experimentally known inhibitor of APH, i.e., ZINC71575479 and tyrphostin AG1478 respectively. T.E.S.T allows estimating toxicity values using different advanced QSAR (quantitative structure–activity relationship) methodologies like hierarchical clustering, food and drug administration, nearest neighbor, single model, and consensus method [30]. For in silico toxicity study, the prediction values included parameters like, LC₅₀ (fathead minnow, 96 h), IGC₅₀ (T. pyriformis, 48 h), LD₅₀ (oral, rat), and developmental toxicant. These experimental data sets from T.E.S.T [30] were chosen as toxicity endpoints for both the inhibitors (ZINC71575479 and tyrphostin AG1478) to carry out predictions.

Statistical analysis

Statistical analysis was performed using Sigmaplot version 11.0 for Windows (Systat Software, Inc., San Jose, CA, USA).

Results

Isolation and identification of streptomycin resistant microorganism

MH agar plates showed uniculture colonies of streptomycin-resistant bacteria after incubation at 37 °C for 24 h. These uniculture colonies of bacteria resistant to streptomycin antibiotic were isolated from soil sample near hospital area by spreading it on MH agar plates. 16S rDNA sequencing (1316 bp) and phylogenetic analysis study of the newly isolated streptomycin-resistant bacteria strain showed its close relationship and homology to Bacillus subtilis subsp. subtilis with branch length and taxon separation value as 0.000000 (Fig. 1). Hence, identified strain was named as Bacillus subtilis subsp. subtilis strain RK and 16S rDNA sequence was deposited in GenBank with the accession number KJ849237.1.

Fig. 1.

Phylogenetic tree of Bacillus subtilis subsp. subtilis strain RK denoted by its GenBank accession number KJ849237.1 along with other closely related Bacillus subtilis subsp. subtilis strains inferred by using neighbor-joining method

Minimum inhibitory concentration and protein overexpression profiling study

The resistant strain Bacillus subtilis subsp. subtilis strain RK showed high MIC value, i.e., 500 μg/ml for streptomycin, whereas comparatively low MIC value, i.e., 100 μg/ml noticed for gentamicin, kanamycin, and amikacin (Fig. S1). The cell free lysate obtained from streptomycin-resistant strain Bacillus subtilis subsp. subtilis strain RK in presence of different concentrations of streptomycin, i.e., 100 to 400 μg/ml, showed a typical overexpression protein profiling pattern, when compared in absence of streptomycin. The SDS-PAGE analysis of cell free lysate showed similar type of protein overexpressed at different concentrations of streptomycin, thus suggesting the role of protein in antibiotic resistance mechanism of Bacillus subtilis subsp. subtilis strain RK (Laemmli 1970) (Fig. 2).

Fig. 2.

Whole cell free lysate protein overexpression profiling of resistant Bacillus subtilis subsp. subtilis strain RK at different concentrations of streptomycin (100–400 μg/ml). Arrow indicates overexpressed protein in presence of streptomycin at different concentrations

Identification of aminoglycoside-modifying enzyme and HPLC analysis

The microbiological assay was used to examine the aminoglycoside modifying ability of cell free lysate of resistant Bacillus subtilis subsp. subtilis strain RK which showed a phosphorylation type of modification (Fig. S2). Cell free lysate of resistant Bacillus subtilis subsp. subtilis strain RK did not show any evidence of aminoglycoside, acetylating or adenylating activity. The zone of inhibition for standard streptomycin, kanamycin, amikacin, and gentamicin was found higher in comparison with their respective modified forms from reaction mixture, thus indicating phosphorylation type of modification of antibiotics by aminoglycoside phosphotransferase from cell free lysate (crude) (Fig. S2a-d). HPLC technique implemented for detection of phosphorylated products of various aminoglycosides as substrate illustrated the biochemical characteristics of aminoglycoside phosphotransferase enzyme as earlier studies [26, 27]. The results of HPLC analysis of standard antibiotics streptomycin, gentamicin, kanamycin, and amikacin with their respective modified products from reaction mixture are shown in Fig. 3. The phosphorylation of aminoglycoside antibiotics by aminoglycoside phosphotransferase resulted into shift in the retention time of the modified products in comparison with their respective standards. The retention time of standard streptomycin and its phosphorylated modified product was observed as 8.106 and 5.061 respectively (Fig. 3a, b). Likewise, retention time of standard gentamicin and its modified product noticed as 8.115 and 4.900 respectively (Fig. 3c, d). Standard kanamycin and its modified product showed retention time 8.132 and 4.895 respectively (Fig. 3e, f), whereas for standard amikacin and its modified product was found as 8.126 and 4.902 respectively (Fig. 3g, h). Thus, this shift in retention time clearly demonstrates the phosphorylation type of antibiotics modification from reaction mixture as shown in HPLC chromatogram (Fig. 3).

Fig. 3.

HPLC chromatograph showing modification of aminoglycosides, with shift in their standard retention time. a Peak showing standard streptomycin with 8.106 as RT extracted at 254 nm. b Peak indicated by arrow of modified product from reaction mixture with RT 5.061. c and d showing peak of gentamicin with RT of 8.115 and modified product with RT of 4.900 respectively. e Peak of kanamycin with RT of 8.132 extracted at 254 nm. f Peak of modified product of kanamycin with RT of 4.895. g HPLC chromatograph of amikacin with RT of 8.126. h HPLC chromatograph showing modified product of amikacin with RT of 4.902

Purification and characterization of aminoglycoside phosphotransferase

Aminoglycoside phosphotransferase extracted from Bacillus subtilis subsp. subtilis strain RK was purified by ion exchange column chromatography technique using DEAE cellulose. The elution pattern of enzyme with NaCl gradient (0.1 to 0.6 M) is shown in Fig. 4. The elution profile showed three major peaks at fraction number 11, 37, and 45 out of 70 fractions when measured spectrophotometrically at 280 nm (Fig. 4). Out of these three fractions, fraction number 11 showed highest concentration of protein, i.e., 0.28 mg/ml. Coupled spectrophotometric assay with streptomycin as a substrate confirmed aminoglycoside phosphotransferase activity for the 11th fraction with gradual timewise oxidation of NADH monitored at 340 nm, whereas no oxidation of NADH was observed for 37th and 45th fractions showing absence of aminolgycoside phosphotransferase activity in those fractions (Fig. 5). Microbiological assay of 11th fraction confirmed the aminoglycoside phosphotransferase activity by comparing the critical zone of standard antibiotics streptomycin, kanamycin, gentamicin, and amikacin with the zone of phosphorylated products of respective antibiotics from reaction mixtures (Fig. S3a-d). Column chromatography purification of APH using DEAE-cellulose resin revealed homogenous separation of APH from 11th fraction with an apparent molecular mass of approximately 27 kDa on sodium dodecyl sulfate-polyacrylamide gel electrophoresis [24] as shown in Fig. 6. Single intact band of APH on SDS-PAGE from 11th fraction confirmed purity of APH (Fig. 6).

Fig. 4.

DEAE-cellulose column chromatography purification elution profile of aminoglycoside phosphotransferase from Bacillus subtilis subsp. subtilis strain RK showing three major protein peaks at fraction numbers 11th, 37th, and 45th out of 70 fractions

Fig. 5.

Enzyme activity of fractions 11th, 37th, and 45th by coupled spectrophotometric assay. Aminoglycoside phosphotransferase activity from 11th fraction was confirmed by gradual oxidation of NADH at 340 nm after specific interval of time

Fig. 6.

SDS-PAGE analysis of aminoglycoside phosphotransferase (active form) from 11th fraction. Molecular weight of the purified enzyme was determined by comparing with standard molecular weight markers (from bottom to top) of 20.1, 29, 43, 58, 66, and 97.4 kDa (lane A). The purified enzyme from 11th fraction showed molecular weight of 27 kDa approximately with single intact band (lane B)

Elucidation of aminoglycoside phosphorylation mechanism of purified APH from Bacillus subtilis subsp. subtilis strain RK

HRMS and MS/MS analysis of the phosphorylated reaction mixture of purified APH obtained different metabolites, which were characterized against Agilent METLINE Personal compound database. Both raw HPLC-QTOFMS and MS/MS data helped for data mining based on molecular formulae to obtain the fragmentation pattern of these metabolites [26] (Fig. 7a, b). Targeted MS/MS of standard streptomycin antibiotic from the reaction mixture showed m/z value as 582.27293. This value helped to identify the structure of streptomycin with formulae C₂₁H₃₉N₇O₁₂ when compared with Agilent METLINE in-house database (Fig. 7a). Similarly, targeted MS/MS analysis also revealed the structure of phosphorylated product streptomycin phosphate with m/z value of 663.2471 on comparison with Agilent METLINE Personal compound database (Fig. 7b). Thus, the obtained structures of metabolites from reaction mixture by targeted MS/MS analysis helped to illustrate the resistance mechanism of APH as occurred in Bacillus subtilis subsp. subtilis strain RK (Fig. S4). Modified product from reaction mixture was obtained as streptomycin phosphate with phosphate group at 5th position of streptamine ring. Thus, on this basis, enzyme was nomenclatured as APH(5) as per aminoglycoside-modifying enzyme nomenclature system (Fig. 7b and Fig. S4).

Fig. 7.

MS/MS fragmentation pattern for standard streptomycin and its modified products. a Full-scan mass spectra of standard streptomycin with m/z value of 582.2729. b MS/MS fragmentation pattern for modified streptomycin phosphate with m/z value of 663.2471 all obtained in positive ion mode

In vitro inhibition study of APH by virtually screened inhibitor

Virtually screened lead-like molecule ZINC71575479 obtained from our earlier study [16] showed complete enzymatic inhibition of APH(5) in presence of ATP and streptomycin when applied experimentally. The coupled spectrophotometric assay showed linear reaction in presence of ZINC71575479 throughout the reaction time with no oxidation of NADH indicating complete inhibition of APH(5) (Fig. 8). The reaction usually goes linear in absence of APH which is indicated by no oxidation of NADH at 340 nm (Fig. 8). Due to action of APH on ATP, phosphate group gets transferred to antibiotic streptomycin releasing ADP, which then accepts phosphate group from phosphoenol pyruvate released by action of pyruvate kinase, thus accelerating reaction in forward direction. Hence, considerable oxidation of NADH in coupled spectrophotometric assay occurs mainly with active form of APH (Fig. 8), whereas inhibitory action of ZINC71575479 on purified APH allows linear reaction without oxidation of NADH when measured spectrophotometrically at 340 nm. The result of ZINC71575479 was found similar to the inhibition results obtained for known inhibitor of APH, i.e., tyrphostin AG1478, with linear reaction without oxidation of NADH (Fig. 8), thus confirming the inhibition ability of ZINC71575479.

Fig. 8.

Coupled spectrophotometric assay showing in vitro inhibition of APH(5) from Bacillus subtilis subsp. subtilis strain RK by virtually screened inhibitor ZINC71575479 and its comparison with known inhibitor tyrphostin AG1478

In silico toxicity study of ZINC71575479 and tyrphostin AG1478

For in silico toxicity analysis of ZINC71575479 and tyrphostin AG1478, different experimental data sets of T.E.S.T like 96 h fathead minnow LC₅₀, 48 h Tetrahymena pyriformis IGC₅₀, oral rat LD₅₀, and developmental toxicity are used to predict toxicity values using different QSAR methodologies [30] (Table 1). Absolute values obtained for different toxicity endpoints from the consensus method represent the average of the predicted toxicities from all different QSAR methods as per T.E.S.T software [30]. ZINC71575479 showed absolute values of toxicity for fathead minnow LC₅₀ (96 h) mg/l, T. pyriformis (48 h) mg/l, and oral rat LD₅₀ mg/kg as 123.88, 134.35, and 669.79 respectively, whereas developmental toxicity result was non-toxicant (Table 1 and Fig. 9). Interestingly, tyrphostin AG1478 showed absolute values of toxicity for fathead minnow LC₅₀ (96 h) mg/l, T. pyriformis (48 h) mg/l, and oral rat LD₅₀ mg/kg as 0.73, 2.89, and 672.12 respectively with developmental toxicity result as toxicant. Comparison of test compounds with similar chemicals from the test set helped to increase the confidence of the predicted values. The mean absolute error (MAE) obtained for similar chemicals with known experimental values was in close proximity with the MAE for the entire test set, thus suggesting appropriateness of the predicted values for the test chemicals such as ZINC71575479 and tyrphostin AG1478 (Fig. 9).

Table 1.

Predicted values for different experimental data sets of T.E.S.T obtained by implementing multiple QSAR methodologies for ZINC71575479 and experimentally known inhibitor of APH tyrphostin AG1478

Method	Fathead minnow LC50 (96 h) predicted value -Log10(mol/l)		T. pyriformis IGC50 (48 h) predicted value -Log10(mol/l)		Oral rat LD50 predicted value -Log10(mol/kg)		Developmental toxicity predicted value
	ZINC71575479	Tyrphostin AG1478	ZINC71575479	Tyrphostin AG1478	ZINC71575479	Tyrphostin AG1478	ZINC 71575479	Tyrphostin AG1478
Consensus	3.29	5.63	3.25	5.04	2.55	2.67	0.50	0.68
Hierarchical Clustering	4.24	5.72	2.12	5.55	2.86	2.77	0.51	0.64
Single model	2.03	5.53	N/A	N/A	N/A	N/A	0.48	0.71
Group distribution	3.32	5.50	4.10	4.75	N/A	N/A	N/A	N/A
Nearest neighbor	3.55	5.78	3.53	4.81	2.24	2.58	N/A	N/A

N/A not applicable

Fig. 9.

Predictions for similar chemicals from the test sets for different experimental data sets with mean absolute error (MAE). a Similar chemicals to ZINC71575479 from the test sets for fathead minnow LC₅₀ (96 h). b ZINC71575479 similar chemicals for T. pyriformis IGC₅₀ (48 h). c Similar chemicals to ZINC71575479 for oral rat LD₅₀. d Similar chemicals to tyrphostin AG1478 from the test sets for fathead minnow LC₅₀ (96 h). e Tyrphostin AG1478 similar chemicals for T. pyriformis IGC₅₀ (48 h). f Similar chemicals to Tyrphostin AG1478 for oral rat LD₅₀ (*MAE for entire set, a 0.55, b 0.33, c 0.43, d 0.55, e 0.33, and f 0.43)

Discussion

A prevalent means to acquire bacterial resistance towards aminoglycoside antibiotics is through the expression of bacterial aminoglycoside kinases (AKs), which modify these antibiotics by transferring phosphate group to free hydroxyl groups of these drugs [31]. Organisms from Bacillus genus are known to cause various nosocomial infections along with dreadful clinical manifestations like bacteremia, septicemia, endocarditis, central nervous system infections, ocular infections, and local infections [32]. Severity of resistance can directly be correlated to the MIC value obtained for various antibiotics. Earlier report illustrates MIC value for different organisms from Bacillus genus as determined by agar dilution method [33]. Bacillus anthracis showed MIC value of 0.25–0.5 μg/ml for gentamicin, whereas MIC values from 0.25 to 1.0 μg/ml for Bacillus cereus as per earlier studies [33]. The newly isolated resistant organism Bacillus subtilis subsp. subtilis strain RK from our study showed higher MIC value for potent aminoglycoside antibiotic streptomycin, whereas low MIC values for gentamicin, kanamycin, and amikacin suggesting the existence of higher MICs in Bacillus genus for streptomycin (Fig. S1). Such kind of higher MIC values for aminoglycosides reported earlier among clinical isolates of Acinetobacter calcoaceticus subsp. anitratus [34]. Overexpression phenomenon of protein has been the key point in antibiotic resistance mechanisms acquired through multidrug efflux system or by aminoglycoside-modifying enzymes, as studied earlier in Stenotrophomonas maltophilia [35] and Enterococcus faecium [36]. Similar protein overexpression phenomenon was also encountered as a resistance factor in Bacillus subtilis subsp. subtilis strain RK against aminoglycoside group of antibiotics (Fig. 2), which was later identified as aminoglycoside phosphotransferase (APH) enzyme by microbiological assay and coupled spectrophotometric assay. Monitoring of the antibiotic modification (phosphorylation) by APH was demonstrated using HPLC on the basis of difference in retention time observed between standard antibiotics and their modified form. Similar approach was also used to study the resistance mechanism from Enterococcus acquired by aminolgycoside 2″-phosphotransferase type IIIa [37]. The use of HPLC analysis to detect aminoglycoside modification by APHs has also been reported earlier for one of the well-known aminoglycoside phosphotransferase APH(6)-Id, a streptomycin-inactivating enzyme, where streptomycin and its phosphorylated form distinctly showed different retention time [27].

Different aminoglycoside phosphotransferases (APHs) have been reported earlier on the basis of their antibiotic modification ability which includes APH(2″), APH(3″), APH(6), APH(9), APH(4), APH(7″), and APH(3′) [38]. Our findings obtained through targeted MS/MS analysis of APH reaction metabolites revealed the resistance mechanism from Bacillus subtilis subsp. subtilis strain RK at atomic level catalyzed by a novel class of aminoglycoside phosphotransferase, i.e., APH(5). Thus, APH(5) reported in our study phosphorylates streptomycin by transferring a phosphate group from ATP to the 5-OH of streptamine ring of streptomycin, yielding an inactivated streptomycin phosphate and ADP [26]. Though earlier study has shown that various known ePK inhibitors can inhibit AK enzymes, the cross-reactivity between them represents an obstacle for their use [31]. To overcome this obstacle in our earlier study, we had implemented virtual screening technique to design a potent lead-like molecule against APH [16], which is applied experimentally in the present study against pure APH(5) from Bacillus subtilis subsp. subtilis strain RK. Earlier, structure-based ligand designing contributed successfully to develop specific marketed drugs like human immunodeficiency virus (HIV) protease inhibitor viracept and anti-influenza drug relenza [39]. Similarly, in silico screening approach has also been used earlier to identify the potential inhibitors of the Aminoglycoside 6′-N-acetyltransferase type Ib (AAC(6)′-Ib) by exploring ChemBridge chemical library, which showed in vitro inhibition by screened compounds in competitive manner to kanamycin A and noncompetitive manner to acetyl CoA [40]. Virtual screening has also been implemented earlier to study inhibition of beta-lactamase enzyme [41]. However, this approach was restricted to in silico level without in vitro testing of most stable screened compound [41]. In the present study, application of virtually screened inhibitor ZINC71575479 showed complete enzymatic inhibition of APH when tested in vitro (Fig. 8), which therefore demonstrates that computer docking is a good initial step to identify compounds that inhibit the phosphotransferase activity of APH. This has inhibited the ability of bacteria to resist the toxic effect of aminoglycosides, which in turn lowered the MIC of potent aminoglycosides in presence of ZINC71575479. Any lead molecule to act as a drug should obey certain pharmacological criteria; most important is its toxicity towards humans and animals. Different QSAR methodologies from T.E.S.T software helped to determine the LC₅₀, IGC₅₀, oral rat LD₅₀, and developmental toxicity values [30]. The predicted values of toxicity fall in the range of developmental non-toxicant for virtually screened inhibitor ZINC71575479 and developmental toxicant for known inhibitor of APH, i.e., tyrphostin AG1478 (Table 1). This strongly suggests the potency of virtually screened inhibitor ZINC71575479 as a drug-like molecule.

Thus, investigation of cause of resistance in B. subtilis subsp. subtilis strain RK identified the role of aminoglycoside phosphotransferase also known as aminoglycoside kinase with overexpression phenomenon. Thus, occurrence of resistance in such ubiquitous organism towards aminoglycoside with higher MIC value indicates the severity of resistance. The LCMS analysis clearly illustrates the phosphate group transfer reaction at the 5th position of streptamine nucleus of streptomycin which is a unique type of reaction observed with consideration to earlier known APHs. This signifies the role of novel aminoglycoside phosphotransferase, i.e., APH(5), involved in the resistance mechanism of newly isolated B. subtilis subsp. subtilis strain RK. In the present in vitro study, application of virtually screened inhibitor ZINC71575479 showed complete enzymatic inhibition of APH, validating the applicability of the in silico techniques to find out potent lead-like molecules against diverse resistance factors. The predicted values of toxicity for ZINC71575479 obtained from different experimental data sets were more or less similar to tyrphostin AG1478 a known inhibitor of APH. This strongly suggests the drug-like potency of virtually screened inhibitor ZINC71575479. Thus, in vitro and in silico approaches used in this study are useful to understand the antibiotic resistance and inhibition mechanism of antibiotic-resistant microorganisms.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

This article does not contain any studies with human participants or animals performed by any of the authors.