A preliminary investigation into bacterial viability using scanning electron microscopy–energy-dispersive X-ray analysis: The case of antibiotics

“The metabolic stages of bacterial development and viability under different stress conditions induced by disinfection, chemical treatments, temperature, or atmospheric changes have been thoroughly investigated. Here, we aim to evaluate early metabolic modifications in bacteria following induced stress, resulting in alterations to bacterial metabolism. A protocol was optimized for bacterial preparation using energy-dispersive X-ray (EDX) microanalysis coupled with scanning electron microscopy (SEM), followed by optimizing EDX data acquisition and analysis. We investigated different preparation methods aiming to detect modifications in the bacterial chemical composition at different states. We first investigated Escherichia coli, acquiring data from fresh bacteria, after heat shock, and after contact with 70% ethanol, in order to prove the feasibility of this new strategy. We then applied the new method to different bacterial species following 1 h of incubation with increasing doses of antibiotics used as a stress-inducing agent. Among the different materials tested aiming to avoiding interaction with bacterial metabolites, phosphorous-doped silicon wafers were selected for the slide preparation. The 15 kV acceleration voltage ensured all the chemical elements of interest were excited. A thick layer of bacterial culture was deposited on the silicon wafer providing information from multiple cells and intra-cellular composition. The EDX spectra of fresh, heat-killed, and alcohol-killed E. coli revealed important modifications in magnesium, potassium, and sodium. Those same alterations were detected when applying this strategy to bacteria exposed to antibiotics. Tests based on SEM–EDX acquisition systems would provide early predictions of the bacterial viability state in different conditions, yielding earlier results than culture.”

EDX measurement conditions: Selection of settings

Bacterial deposits were observed using the Tabletop SEM TM4000 Plus (Hitachi High-Tech, Japan) combined with the AZtecOne EDX system (Oxford Instruments, United Kingdom). The preparations were loaded into the TM4000 Plus with a 10 mm distance between the sample surface and the detector. Glucose was used as a sample material for the Monte Carlo simulation, since it has an elemental composition similar to bacterial cells. The acceleration voltage was set at 15 kV, with the highest beam current mode (LensMode 4) under vacuum conditions (<30 Pa). Each bacterial deposit was observed at ×300 magnification. Map-sum EDX spectra of each image area were taken using a mapping mode at 256 times image resolution, three times frame count, and 200 μs pixel dwell time. EDX spectra from Si substrate were also measured for spectral analysis.

Sample preparation and EDX measurements: Selected conditions

Si was selected as a material for the substrate, given its negligible concentration in typical bacterial cells, avoiding interactions and overlapping elements between the bacteria and the slide. This overlap removal will ensure the absence of interaction between EDX signals from the substrate and the sample, thus eliminating biased results.

The acceleration voltage of the electron beam is one of the most important parameters in EDX measurement. The electron beam is scattered in the sample while generating X-rays, which are detected as an EDX signal. The Monte Carlo simulation (Supplementary Figure S3) revealed that when the acceleration voltage was 5 kV, the penetration depth of the electron beam was <500 nm, which mainly provides signal from the bacterial surface. With an acceleration voltage of 15 kV, the penetration depth was about 3 μm. With the larger penetration volume, the EDX signal is generated from more bacterial cells and thus provides averaged information from multiple cells, including the intra-cellular atomic composition. The acceleration voltage was set at 15 kV with the highest beam current mode. This setting ensured excitation of all the chemical elements of interest and obtaining as many X-ray signals as possible.

EDX spectra of E. coliK. pneumoniae, and E. cloacae incubated without and with increasing imipenem concentrations were acquired (Figure 3). In some cases, aluminum (Al) peaks (1.49 keV) were detected, owing to the interaction between scattered electrons and the sample support made of Al. Signal fluctuation around 1.74 keV resulted from the subtraction of Si peaks. Obvious decreases in the Mg and K peaks, and a rise in the Na peak were also detected for the susceptible strains (E. coli Q5586, K. pneumoniae Q5580, and E. cloacae P9549), with increasing imipenem concentrations (MIC: 0.25, 1 and 0.25 mg/L respectively; Figures 3A,D,G). We correlated these modifications to the metabolic response of the bacteria to the induced chemical stress. However, these modifications were not observed in the resistant strains (E. coli P1872, K. pneumoniae Q2247, and E. cloacae P9548; MIC: >32, 4 and 2 mg/L respectively; Figures 3B,E,H).These results were confirmed by culture and SEM imaging. However, the K and Na concentrations express a high variability among the spectra from the same bacterial deposit, due to their presence in the culture medium. Therefore, the measurement of Mg was selected as the follow-up chemical element for our metabolic profiling. Regarding the susceptible strains Q5586, Q5580, and P9549, the Mg/C ratio showed no significant change from the control (p > 0.05) below the MIC. Exceeding this limit, the Mg/C ratio tends to decrease monotonically with increasing imipenem concentrations, which was more pronounced in the case of K. pneumoniae and E. cloacae (p < 0.0001; Figures 3C,F,I). The Mg/C ratio of the resistant strains showed no significant changes compared to the imipenem-free controls below the MIC, where K. pneumoniae Q2247 and E. cloacae P9548 showed a decreasing tendency which is less prominent than the susceptible isolates (p < 0.01 and p > 0.05, respectively; Figures 3C,F,I). These results indicate that the antibiotic-concentration dependency of the Mg/C ratio corresponds well to conventional MICs. When evaluating the correlation of the K/C and Na/C ratios with Mg/C for K. pneumoniae Q5580, the plots were distributed along a line with a positive and negative slope, respectively (Figure 4). These correlations imply the leakage of cytoplasmic cations caused by bacterial lysis. Similar correlations were also observed in the case of E. coli Q5586 and E. cloacae P9549 (not shown).

Figure 3

www.frontiersin.orgFIGURE 3. EDX spectra and Mg/C ratio of Gram-negative bacilli incubated with imipenem. EDX spectra of (A, D, G) susceptible (Q5586, Q5580 and P9549) and (B, E, H) resistant (P1872, Q2247, and P9548) strains of E. coliKlebsiella pneumoniae and Enterobacter cloacae, respectively. Each curve represents the average of three spectra taken from the same bacterial deposit. Spectra are offset by 5 (C, F, I) Mg/C ratio of tested strains calculated from the respective EDX spectra. Error bars: standard deviations. Blue and red triangles: MIC of tested strains. *p < 0.01; **p < 0.001; **p < 0.0001; (ns): not significant.

Leave a Comment