Affinity of cefotiam for the alternative penicillin binding protein PBP3_SAL used by Salmonella inside host eukaryotic cells

Prestwick Chemical Library and selected compounds

The Prestwick Chemical Library (Prestwick Chemical Libraries, GreenPharma S.A.S., Orléans, France) was used to screen compounds that induce filamentation in the S. Typhimurium ΔPBP3 mutant having PBP3_SAL as the only PBP promoting cell division.10 The library has 1520 off-patent compounds approved by agencies such as the FDA and the EMA. The compounds were supplied at 10 mM in 100% DMSO.

Bacterial strains, eukaryotic cell lines and growth conditions

S. Typhimurium strains used were isogenic to the parental wild-type strain SV5015,13 a His⁺ derivative of the virulent strain SL1344 isolated from calf.14 The strains and plasmids used in this study are listed in Table 1. Bacteria were grown at 37°C in LB broth buffered with 80 mM MES [2-(N-morpholino) ethanesulfonic acid] and adjusted to pH 4.6. When necessary, ampicillin and kanamycin were added at 0.1 mg/mL and 0.03 mg/mL, respectively. The S. Typhimurium double mutants MD5063 [ΔmrdA (ΔPBP2) ΔftsI (ΔPBP3)] and MD2588 [ΔSTM1910 (ΔPBP2_SAL) ΔSTM1836 (ΔPBP3_SAL)] were constructed by bacteriophage P22-mediated transduction using donor strains MD2502 [ΔSTM1836::Km (ΔPBP3_SAL::Km)] and MD5049 [ΔmrdA::Km (ΔPBP2::Km)]. The P22 lysates obtained in MD2502 and MD5409 strains, were used to transduce recipient strains MD2576 [ΔSTM1910 (ΔPBP2_SAL)] and MD4356 [ΔftsI (ΔPBP3)], respectively. Subsequent removal of the kanamycin cassette from strains MD2580 [ΔSTM1910 (ΔPBP2_SAL) ΔSTM1836::Km (ΔPBP3_SAL::Km)] and MD5061 [ΔmrdA::Km (ΔPBP2::Km) ΔftsI (ΔPBP3)] (see Table 1), was performed as described.15

Open multimedia modal

Identification of drugs blocking cell division based on monitoring of OD₆₀₀ values

As a previous control, the growth of S. Typhimurium wild-type and ΔPBP3 isogenic strains was monitored in LB medium at pH 4.6 in the absence/presence of 0.001 mg/mL aztreonam using a Spark Automatic Microplate Reader (Tecan Trading AG, Switzerland). Differences in final OD₆₀₀ values were correlated with the presence of filamented bacteria in which cell division was blocked by the antibiotic. Response to compounds of the Prestwick Library was directed to the monitoring of growth curves and final OD₆₀₀ values. Drugs having a bacteriostatic effect were discarded from further analyses. Those compounds causing bacteriolytic effect at the initial dose used (1:100 dilution of the 10 mM stock) were rescreened at lower concentrations and reanalysed for increased values in the final OD₆₀₀ measurements. Growth curves were monitored in the automatic reader for a minimum period of 8 h. Control samples consisted of bacteria grown in LB medium pH 4.6 containing 1% DMSO.

Antibiotic susceptibility tests

MIC values were determined using MIC strips (MIC Test Strip, Liofilchem, Roseto degli Abruzzi, Italy) in LB plates at pH 4.6. Because no cefotiam MIC strips are commercially available, MIC values for this antibiotic were determined in 96-well microplates by serial dilutions of the antibiotic in liquid LB pH 4.6 medium and overnight incubation at 37°C. The starting inoculum was 6 × 10⁵ cfu per well.

Preparation of membrane extracts for BOCILLIN-FL (Boc-FL) labelling

S. Typhimurium ΔftsI (ΔPBP3) and ΔSTM1836 (ΔPBP3_SAL) isogenic strains were grown overnight at 37°C in LB pH 4.6. Cultures were diluted 1:100 in 200 mL fresh media and bacteria grown up to exponential phase (OD₆₀₀ ∼0.2–0.3). Bacteria were harvested by centrifugation (6000×g, 10 min, 4°C) and washed in 50 mM sodium phosphate buffer pH 4.6. After subsequent centrifugation (12 000×g, 15 min, 4°C), bacteria were suspended in 20 mL of 50 mM sodium phosphate buffer. Cells were disrupted by passing through a French press and lysates centrifuged at low speed (4000×g, 10 min, 4°C) to remove unbroken cells. The supernatant was further centrifuged (150 000×g, 35 min, 4°C) and pellets containing membrane material were suspended in 150 μL of 50 mM sodium phosphate buffer pH 4.6. The protein concentration was measured using Pierce 660 nm Protein Assay reagent (Thermo Scientific).

Cefotiam competition in Boc-FL binding assays

Cefotiam hydrochloride (ref. 66309-69-1, Molekula GmbH, Munich, Germany) was added to 0.02 mg membrane extracts prepared in 50 mM sodium phosphate buffer pH 4.6 to reach final concentrations of the antibiotic in the 0.00005–0.005 mg/mL range. Sample volume was 10 µL in all cases. Binding conditions were pH 4.6 for 10 min at 30°C, as described.12 Subsequently, 20 μM Boc-FL (Molecular Probes) was added and the sample incubated for 20 min at 30°C. Samples were finally processed by adding Laemmli buffer and boiled for 5 min. Proteins were resolved by SDS-PAGE in 8% (w/v) acrylamide gels. The gel was washed with 30% (v/v) methanol/10% (v/v) acetic acid, and fluorescence was detected on a Typhoon 8410 variable-mode imager (General Electric) with an excitation wavelength of 588 nm and a 520BP40 emission filter. Quantification of gel bands was performed using ImageJ software (National Institutes of Health, Bethesda, MD, USA). Bands with no signal changes in the presence of cefotiam were used as loading controls.

Microscopy analyses

Overnight bacterial cultures were diluted at OD₆₀₀ ∼0.01 in LB pH 4.6 and grown in 96-well plates for 8 h in the presence of aztreonam (ref. 78110-38-0, Molekula GmbH, Munich, Germany) or cefotiam hydrochloride at varied concentrations. A volume of 120 μL of each culture was harvested (4300×g, 5 min, room temperature), washed in PBS pH 7.4 and fixed with 3% (w/v) paraformaldehyde. Images were acquired on an inverted Leica DMI 6000B microscope with an automated CTR/7000 HS controller (Leica Microsystems) and an Orca-R2 CCD camera (Hamamatsu Photonics).

Bacterial infection of fibroblasts

NRK-49F fibroblasts (ATCC CRL-1570) were propagated in DMEM containing 10% (v/v) FBS at 37°C in a 5% CO₂ atmosphere as previously described.16 Fibroblasts were infected for 20 min at a multiplicity of infection of 10:1 with wild-type or mutant strains, all of them previously grown overnight in LB pH 4.6 without shaking. After this time, fibroblasts were washed three times with PBS pH 7.4, and fresh tissue culture medium containing 0.1 mg/mL gentamicin to kill remaining extracellular bacteria was added. At 2 hours post-infection (hpi), tissue culture medium was further replaced with fresh medium containing 0.01 mg/mL gentamicin and 0.001 mg/mL cefotiam hydrochloride. Cefuroxime at 0.02 mg/mL was also used as an alternative cephalosporin. Unlike cefotiam, cefuroxime has higher affinity for binding to PBP3 compared with PBP3_SAL.10 At 2 and 24 hpi, fibroblasts were lysed in a 0.1% Triton X-100 solution and the number of viable intracellular bacteria was calculated by plating of 10-fold serial dilutions using PBS pH 7.4. Volumes plated were 100 μL onto LB pH 4.6 plates to calculate number of viable intracellular bacteria or 5 µL for the drop assay.

Statistical analyses

Data were analysed with GraphPad Prism, version 8.0, software (GraphPad Inc.). A t-test was used for data analysis. Significance was established at P values ≤0.05.

Design of a screening method to identify drugs selectively inhibiting PBP3_SAL versus PBP3

Unlike in E. coli, it is possible to generate S. Typhimurium mutants defective in PBP3 due to the presence of an alternative PBP, named PBP3_SAL, which accomplishes cell division in acid pH.10 To select drugs targeting PBP3_SAL with higher affinity than PBP3, we designed a screening method to search for compounds inducing filamentation, a manifestation of cell division blockage. The candidate compound was expected to cause filamentation in an S. Typhimurium ΔPBP3 mutant expressing only PBP3_SAL for division but not in wild-type bacteria that express both PBP3 and PBP3_SAL when growing in acidified (pH 4.6) LB medium.12 To detect differences in filamentation using high-throughput screening, we monitored the growth curve of wild-type and ΔPBP3 bacteria in the presence of aztreonam, a β-lactam that binds PBP3 with high affinity and blocks cell division at very low concentrations.17 The growth curves in the presence of a low dosage of aztreonam (0.001 mg/mL, equivalent to 2.3 µM) reached higher final optical density values (Figure 1a). This change was consistent with the presence of long filaments (Figure 1a). A dose–response assay for aztreonam confirmed the higher affinity for PBP3 compared with other PBPs. Thus, aztreonam caused S. Typhimurium filamentation starting at 0.0000078 mg/mL (0.018 µM) up to 0.05 mg/mL (115 µM), the highest concentration used (Figure S1, available as Supplementary data at JAC Online). No massive cell lysis was observed even at 115 µM of this antibiotic.

Open multimedia modal

Based on these preliminary controls, we screened the Prestwick Chemical Library consisting of 1520 off-patent compounds. This library contains diverse anti-infective (antiviral, antiprotozoal, antifungal and antibacterial) compounds as well as drugs used in clinics to treat a number of pathologies, all supplied at 10 mM in DMSO. S. Typhimurium wild type and its isogenic ΔPBP3 mutant were exposed to these compounds at 1:100 dilution of the stock (100 µM). A few compounds increased final optical density values compared with control cultures not exposed to drug. These compounds were further discarded because no differential effect on wild-type and ΔPBP3 strains was observed. Compounds of this first group included β-lactam antibiotics like aztreonam, which validated our previous control assays (see Figures 1a and S1), cloxacillin, amoxicillin, piperacillin and cefazolin. A second group of β-lactams included in the Prestwick Library, comprising cefixime, cefotetan, ceforanide and cefotiam, caused lysis of wild-type and ΔPBP3 bacteria at 100 µM. Aiming to find different responses in these two strains, these β-lactams were subsequently tested at more diluted concentrations starting from 100 µM up to 3.125 µM. We also included an isogenic ΔPBP3_SAL to discern whether the differences, if found, were related to the presence/absence of PBP3_SAL.

Within this second group of β-lactams, cefotiam hydrochloride (MW = 598.56) was the only antibiotic causing a differential effect. In LB pH 4.6 medium, a condition in which S. Typhimurium wild type expresses PBP3 and PBP3_SAL,12 cefotiam hydrochloride at 0.00375 mg/mL (6.25 µM) increased final optical density values in wild-type and ΔPBP3 strains, but not in the ΔPBP3_SAL mutant (Figure 1b). We next acquired cefotiam hydrochloride powder at ≥98% purity from commercial sources. The response of bacteria exposed to cefotiam was then examined by microscopy in wild-type, ΔPBP3 and ΔPBP3_SAL strains. Only those strains expressing PBP3_SAL (wild-type and ΔPBP3) filamented in LB pH 4.6 medium containing 0.00032 mg/mL cefotiam hydrochloride whereas such an effect was not seen in the ΔPBP3_SAL strain expressing only PBP3 (Figure 1c). The data obtained in liquid culture were confirmed in LB pH 4.6 agar plates containing distinct concentrations of cefotiam hydrochloride (Figure 1d). In the assays involving solid media we also tested isogenic single and double mutants lacking distinct PBPs involved in morphogenesis: PBP2, PBP3, PBP2_SAL and PBP3_SAL. Those strains having PBP3_SAL as the single enzyme for division (ΔPBP3 genetic background) exhibited higher susceptibility to cefotiam (Figure 1d). Taken together, these observations suggested that the second-generation cephalosporin cefotiam could bind more efficiently to PBP3_SAL than to PBP3.

Boc-FL bindings assays reveal higher affinity of cefotiam for PBP3_SAL

To further demonstrate the distinct behaviour of cefotiam for binding to PBP3 and PBP3_SAL, a competition binding assay with the fluorescent derivative Boc-FL antibiotic was performed in membrane extracts obtained from ΔPBP3 and ΔPBP3_SAL strains (Figure 2a). The assay revealed IC₅₀ values for cefotiam of 0.0004 mg/mL in the case of PBP3_SAL and 0.00375 mg/mL for PBP3 (Figure 2b). This result contrasted with our previous Boc-FL competition assays with other cephalosporins like cefuroxime, which exhibited higher affinity for PBP3.10 Cefotiam also showed avidity for binding to the high molecular weight bifunctional PBP1A and PBP1B, although with relatively high IC₅₀ values of 0.004 and 0.0028 mg/mL for the ΔPBP3 and ΔPBP3_SAL strains, respectively (Figure 2b). MIC values obtained for single and double mutants with deficiencies in PBP2, PBP3, PBP2_SAL or PBP3_SAL corroborated the higher susceptibility to cefotiam of those strains expressing PBP3_SAL as the only PBP involved in cell division (Table 2). Conversely, the determination in S. Typhimurium by the Etest assay of MIC values for other cephalosporins known to have high affinity for PBP3, like aztreonam, ceftriaxone, cefuroxime, cefotaxime and cefepime, revealed an opposite phenotype compared with the response to cefotiam. Thus, the susceptibility to these other β-lactams increased notoriously in strains lacking PBP3_SAL, i.e. having PBP3 as the only PBP for division (Table 2). The lack of PBP3_SAL associated with increases in MIC values in the order of more than 8-fold (cefuroxime, cefepime), 6-fold (aztreonam) or 2.65-fold (ceftriaxone, cefotaxime) (Table 2). The MIC to mecillinam (amdinocillin), a β-lactam that binds specifically to PBP2 in E. coli,18 showed less than 2-fold difference in strains expressing either PBP3 or PBP3_SAL (Table 2). The MIC to mecillinam was also very similar in strains expressing either PBP2 or PBP2_SAL (Table 2). Unlike PBP3 and PBP3_SAL, the pair PBP2/PBP2_SAL might therefore bind mecillinam with similar affinity. Altogether, these data indicated that, unlike the other cephalosporins tested, cefotiam shows higher affinity for PBP3_SAL compared with PBP3 and with no binding to other PBPs at concentrations in the 0.0001–0.001 mg/mL (0.16–1.6 µM) range.

Open multimedia modal

Open multimedia modal

Unlike cefuroxime, cefotiam is highly bactericidal against intracellular S. Typhimurium in a PBP3_SAL-dependent manner

PBP3_SAL is not produced by S. Typhimurium when growing extracellularly in neutral pH but its expression is up-regulated in acidic compartments of eukaryotic cells.10^,12 Because cefotiam showed increased affinity for PBP3_SAL, we reasoned that it could inhibit bacterial growth by blocking the activity of this PBP. To differentiate such predicted selectivity, we used the isogenic series of single and double S. Typhimurium mutants lacking PBP2, PBP3, PBP2_SAL or PBP3_SAL to infect the rat fibroblast NRK-49F cell line, in which we have extensively characterized mechanisms used by the pathogen to survive and persist intracellularly.16^,19^,20 Non-phagocytic cells consistent with fibroblasts are also targeted in vivo by S. Typhimurium in the intestinal lamina propria.16 The addition of 0.001 mg/mL (1.6 µM) cefotiam to the tissue culture medium at 2 hpi led to a drastic drop in the viability at 24 hpi of those strains dividing only with PBP3_SAL (ΔPBP3 genetic background). This was demonstrated by drop assays performed on agar plates containing the antibiotic (Figure 3a) as well as by plating and cfu counting corresponding to viable intracellular bacteria at 2 hpi and 24 hpi (Figure 3b).

Open multimedia modal

To further demonstrate the selective action of cefotiam against intracellular bacteria due to inhibition of PBP3_SAL, we also tested cefuroxime (Figure 3c), a cephalosporin that has higher affinity for PBP3 than PBP3_SAL.10 The bactericidal effect of cefuroxime in intracellular bacteria was the reverse to that of cefotiam, i.e. more pronounced for the ΔPBP3_SAL mutant expressing PBP3 (Figure 3d,e). Overall, these data demonstrated that cefotiam is an antibiotic that reaches the intracellular compartment colonized by S. Typhimurium inside host cells and that it inhibits selectively PBP3_SAL at the concentrations used.

Acknowledgements

We thank fruitful comments from members of García-del Portillo’s laboratory.