Bacterial cells from an overnight culture were centrifuged (4,000 rpm, 10 min) and washed with saline solution. Afterward, the optical density (at 610 nm) was adjusted to 0.2 ± 0.02 in saline solution. Then, bacterial cells were placed in contact with the biocidal solutions for 30 min at 25°C and 160 rpm. Four different concentrations of each biocide were used, namely: the concentration determined as the MBC, one concentration above the MBC, and two concentrations below the MBC. Therefore, the selected concentrations were 30, 40, 100, and 160 mg/L for BAC and 5, 8, 10, and 35 mg/L for DBNPA. Controls were performed with both live cells only (negative control) and dead cells only (positive control). To obtain dead cells, bacterial cells were autoclaved at 121°C for 20 min. The autoclaving process led to microbial inactivation percentages of: ~100, 99, and 87% in terms of culturability, ATP and metabolic activity, respectively. Henceforth, the live cells suspension and the dead cells suspension will be called “C⁻” and “C⁺,” respectively.

After biocide exposure, the cells were neutralized as described in section “Minimum Bactericidal Concentration Determination,” centrifuged (4,000 rpm, 5 min), washed twice (to remove biocide) and then resuspended in saline for further analysis.

To fully understand the mechanisms of antimicrobial action of the selected biocides, the following parameters were evaluated after biocide exposure, namely: (i) culturability; (ii) membrane integrity; (iii) metabolic activity; (iv) cellular energy; and (v) structural and morphological changes.

Bacterial cells were serially diluted in saline solution and plated on plate count agar (PCA) using the drop plate method (Reed and Reed, 1948). Then, the plates were incubated at 30°C for 24 h. Afterward, colonies were enumerated and the results expressed as logarithm of the colony forming units per milliliter (CFUs/mL). Three independent experiments were performed for each condition.

The LIVE/DEAD® Baclight™ kit (Invitrogen) was used to assess membrane integrity of bacterial cells after exposure to biocides. In brief, an aliquot of 1 mL was filtered through a 0.2-μm Nucleopore® (Whatman) black polycarbonate membrane and stained with 250 μl of SYTO9™ and 50 μl of propidium iodide (PI) in the dark (Barros et al., 2021). After reacting for 7 min, the stains were filtered and the samples mounted on a slide with immersion oil. Then, samples were observed using a 100x objective oil immersion fluorescence objective of a LEICA DMLB2 microscope incorporated with a mercury lamp HBO/100W/3 and a CCD camera. The optical filters used were a combination of a 480–500 nm excitation filter and a 485 nm emission filter (Chroma 61000-V2 DAPI/FITC/TRITC; Ferreira et al., 2011). A IM50 software (LEICA) was used to record images. At least 15 images were captured for each sample and the experiments were repeated in three different occasions.

Stock solutions of resazurin (400 μM) were prepared in sterile ultrapure water and stored at −20°C. In brief, 190 μl of samples and 10 μl of resazurin were added in the dark to each well. Then, the fluorescence was recorded at λexcitation = 570 nm and λemission = 590 nm every 30 min for 24 h using a FLUOstar® Omega microtiter plater reader (BMG LABTECH, Germany). Blank assay was performed using saline solution. The results were presented in terms of relative fluorescence units (RFU). Three independent experiments with triplicates were performed for each case.

Adenosine triphosphate was measured using BacTiter-Glo™ Microbial Cell Viability Assay (Promega, Madison, Wisconsin, United States) according to the manufacturer’s instructions. Briefly, 100 μl of samples was mixed with 100 μl of BacTiter-Glo™ reagent in white 96-well plates and allowed to interact for 5 min at room temperature. Then, luminescence was recorded using the lens from a FLUOstar® Omega microtiter plater reader (BMG LABTECH, Germany) and the output values were recorded in relative light units (RLU). Background luminescence was removed by recording luminescence of control wells without cells. Three independent experiments with triplicates were performed for each case.

For the ultrastructure analysis, cells were fixed in a solution of 2% glutaraldehyde (#16316; Electron Microscopy sciences) with 2.5% formaldehyde (#15713; Electron Microscopy sciences) in 0.1 M sodium cacodylate buffer (pH 7.4) for 24 h, at room temperature, and post fixed in 2% osmium tetroxide (#19190; Electron Microscopy Sciences) diluted in 0.1 M sodium cacodylate buffer. After centrifugation, the pellet was resuspended in Histogel™ (Thermo, HG-4000-012) and then stained with aqueous 1% uranyl acetated solution overnight, dehydrated and embedded in Embed-812 resin (#14120; Electron Microscopy sciences). Ultra-thin sections (50 nm thickness) were cut on a RMC Ultramicrotome (PowerTome, United States) using Diatome diamond knifes, mounted on mesh copper grids (Electron Microscopy Sciences), and stained with uranyl acetate substitute (#11000; Electron Microscopy Sciences) and lead citrate (#11300; Electron Microscopy Sciences) for 5 min each. Samples were viewed on a JEOL JEM 1400 transmission electron microscope (JEOL, Tokyo, Japan) and images were digitally recorded using a CCD digital camera Orius 1100W (Tokyo, Japan). The transmission electronic microscopy (TEM) was performed at the HEMS core facility at i3S, University of Porto, Portugal with the assistance of Ana Rita Malheiro and Rui Fernandes.

The culturability of cells previously exposed to BAC and DBNPA was evaluated and the results are presented in Figure 2.

The results illustrated in Figure 2 show that total inactivation (no CFU counts) of bacterial cells was obtained for: (i) 100 and 160 mg/L of BAC; (ii) 10 and 35 mg/L of DBNPA; and (iii) autoclaved cells (C⁺). These results are not surprising as these concentrations correspond to the MBC’s and concentrations above the MBC (Figure 1) or, in the case of C⁺, to the positive control (dead cells).

Figure 2A also shows that the lowest BAC concentrations tested, 30 and 40 mg/L, affected the culturability of cells as they resulted in a 0.3 and 1.3 log₁₀-reduction, respectively. Similarly, the lowest DBNPA concentrations (5 and 8 mg/L) also reduced the culturability in 0.2 and 1.3 log₁₀, respectively.

The membrane integrity was evaluated using two different dyes: SYTO 9 and PI. PI is only able to penetrate cells with damaged membranes; it binds to nucleic acids and emits a red fluorescent light. SYTO9, which also binds to nucleic acids, is used as a viability marker as it enters all cells, regardless their membrane integrity, staining cells green (Ding et al., 2017).

Figure 3A shows an increase of PI-stained cells with increasing BAC concentrations. The positive control presented 79 and 21% of SYTO9 and PI-stained cells, respectively. On the other hand, total loss of membrane integrity (100% PI uptake) was observed for the concentrations equal to or above 30 mg/L, as well as for the negative control (p < 0.0001). Fernandes et al. (2020) also studied the membrane integrity of P. fluorescens after being exposed to BAC (10 and 100 mg/L) and found similar values of PI uptake percentage (84–100%).

On the other hand, no statistical differences (p > 0.05) were observed in the PI and the SYTO9 uptake between C⁻ and DBNPA-exposed cells (Figure 3B). These PI uptake values (25, 15, 20, and 15%, respectively for 5, 8, 10, and 35 mg/L) are also similar to the ones observed in the negative control (C⁻). Since the C⁻ corresponds to live cells in the growth phase, these results suggest that the DBNPA did not have a significant impact on the membrane integrity of P. fluorescens. Additionally, and as expected, the autoclaved cells (C⁺) showed total loss of membrane integrity (100% PI uptake).

It is important to note that the PI uptake percentages observed for C⁻ (18–21%) are consistent with data reported in the literature (Fernandes et al., 2020; Barros et al., 2021). The question why live cells can be stained with PI remains unanswered (Shi et al., 2007). Some authors (Shi et al., 2007) showed that active growing cells can have, for short periods of time, their membrane’s integrity affected. Other studies (Kirchhoff and Cypionka, 2017) attributed the PI uptake by intact live cells to the existence of a high membrane potential.

In short, results from Live/Dead assay clearly demonstrate that the two tested biocides have a different impact at the cells membrane level: BAC disrupts cells membranes, while DBNPA does not affect the cells membrane integrity (independently of inactivating the cells, or not). The interpretation of these results will be further developed in sub-section “The Issue of Viable but Non-culturable Cells,” when addressing the issue of viable but not culturable cells.

Metabolic activity of cells was determined using the resazurin dye assay. Resazurin is commonly used to assess bacterial viability. Basically, resazurin is a weakly-fluorescent dye that is reduced to a pink highly-fluorescent dye (resorufin). This reduction only occurs in the presence of metabolically active cells (Bueno et al., 2002; Csepregi et al., 2018). Therefore, the intensity of fluorescence can be correlated to the presence of metabolically active cells (viable cells).

The results obtained for resazurin assays (in RFU units) are presented in Figure 4.

Figure 4A, shows that BAC has a significant effect (p < 0.0001) on the reduction of cells’ metabolic activity as compared with the negative control (C⁻). More specifically, a 60–80% decrease in fluorescence intensity can be observed between BAC-exposed cells (for all tested concentrations) and the live ones from C⁻. Although the impact on cells’ metabolic activity is noticeable for concentrations below the MBC (100 mg/L), 160 mg/L was the only concentration presenting similar values to the autoclaved cells (C⁺; p > 0.05). These results suggest that the MBC concentration is not enough to decrease the cells metabolic activity to the levels observed in autoclaved cells.

Similarly, cells exposed to increasing DBNPA concentrations (Figure 4B) significantly decreased (p < 0.0001) the fluorescence intensity when comparing to C⁻. These results indicate that DBNPA affected the metabolic activity of cells (reductions between 58 and 78%), although without causing membrane damage. Also, the differences between C⁺ and the DBNPA concentrations tested were statistically significant, except for 35 mg/L (p > 0.05). Again, only the concentration above the MBC seems to be able to decrease the cells’ metabolic activity to levels comparable with the ones observed for the autoclaved (dead) cells.

Adenosine triphosphate bioluminescence assay is a powerful tool to assess cell viability (Lomakina et al., 2015). ATP is a key carrier of energy in cells, it is essential for cell respiration and its synthesis is immediately stopped after cell death (Lomakina et al., 2015). The ATP kit has a reagent that lyses cells. The principle of the method is that when cells are lysed ATP is released and the oxidation of luciferin (an organic substrate) by luciferase (an enzyme) uses ATP from viable cells to produce light. So, a higher luminescence (RLU) signal can be correlated to the presence of higher levels of ATP and, therefore, a higher number of viable cells (Lomakina et al., 2015; Kumar and Ghosh, 2019). As described in the methodology section, cells are lysed upon contact with one of the reagents present in the ATP test kit, enabling the detection of ATP. It is important to highlight that the ATP measured in this work (results shown in Figure 5) is the intracellular ATP (ATP present inside living cells), since upon biocide exposure, the samples were centrifuged and the supernatant discarded.

The luminescence results presented in Figure 5A show that increasing BAC significantly decreased the measured luminescence when compared to the positive control (p < 0.0001), suggesting a decrease on the ATP levels. Although from Figure 1A it was observed that 40 mg/L of BAC was far below the MBC (which was found to be 100 mg/L), the differences between 40, 100, 160 mg/L and C⁻ were not statistically significant (p > 0.05). These results seem to indicate that BAC concentrations equal to or above 40 mg/L decrease the ATP to levels similar to the ones from autoclaved dead cells (C⁺). In the work of Aragones et al. (2012), when P. aeruginosa was tested with 100 mg/L of BAC for 5 min of contact time, a 0.3 log₁₀-reduction of luminescence was observed. Although this value is low compared to the 1.9 log₁₀-reduction obtained in the present study, the contact time used was also lower and the microorganism (a Gram-negative) was not from the same species.

The comparison between Figure 5A (BAC) and Figure 5B (DBNPA) seems to indicate a smoother decreasing tendency between DBNPA concentration and luminescence decrease (Figure 5B) as compared to BAC, which can be related to the lytic action of BAC that leads to the release of intracellular contents (including ATP). Although luminescence decreases with increasing DBNPA concentrations (Figure 5B), there is a statistically difference (p < 0.0001) among all the tested concentrations and both controls (C⁺ and C⁻). The only exception is observed between the 35 mg/L and C⁺, which did not present statistically significant differences (p > 0.05). These results suggest that a 35 mg/L concentration of DBNPA is needed to decrease ATP to levels similar to the ones observed in cells submitted to an autoclaving process (C⁺).

In this work, TEM was used to accurately visualize biocide impact on bacterial cell envelope. For this study, we examined both controls (C⁺ and C⁻) and the cells previously exposed to BAC and DBNPA at the MBCs (100 and 10 mg/L, respectively; Figure 6).

Figures 6A,B revealed that untreated bacteria (C⁻) have a dense cytosol and a well-defined and intact bacterial cell envelope.

From Figures 6C,D is possible to notice the appearance of many vacuoles (islands of cytoplasm) that are not present in the C⁻. In a recent study, Ziklo et al. (2021) observed the same islands of cytoplasm on P. aeruginosa cells after dodecyl-dimethyl-ammonium chloride (a QAC compound) treatment. Also, BAC exposure led to a loss in intracellular content, with genetic material spread across the cytoplasm. Although there is no clear evidence of membrane disruption, the leaflets of bacterial cell envelope cannot be distinguished. The same trend has already been observed for colistin (Nguyen et al., 2021) and Polymyxin B (Voget et al., 2015), two antimicrobials which also target the outer membrane.

The exposure of cells to DBNPA (Figures 6E,F) appeared to have no effect on the bacterial cell envelope, but resulted in the loss of cellular components, which can be observed by the appearance of translucent zones.

Autoclaved cells (C⁺) Figures 6G,H-exhibited a high decrease in intracellular electron density. Besides, a lot of vesicles can be observed. Although the different layers of membrane structure cannot be distinguished, a massive cell wall damage was observed.

In short, all treatments induced structural and morphological changes to P. fluorescens and TEM images are in agreement with the results from the preceding analysis.

Over the past two decades, many researchers have been discussing that cells can reach the viable but non-culturable (VBNC) state as a mechanism of defense and self-preservation (Keep et al., 2006; Alam et al., 2007; Li et al., 2014; Emerson et al., 2017; Fleischmann et al., 2021; Kirschner et al., 2021). Cells enter the VBNC state when submitted to harsh stress conditions (Highmore et al., 2018; exposure to biocides/antibiotics, temperature shifts, lack of nutrients). The more usual characterization of cells that enter the VBNC state is: (i) they retain their membrane integrity; (ii) they are still metabolically active (reduced levels of metabolic activity); (iii) they are able to produce proteins; but (iv) they are not able to grow on solid media (Oliver, 2010; Ayrapetyan et al., 2014, 2018; Fleischmann et al., 2021; Kirschner et al., 2021). Also, VBNC cells are more chemically and physiologically resistant than culturable cells (Signoretto et al., 2000; Li et al., 2014), making them a major threat in many areas, including public health. If somehow the environmental conditions are restored, these cells can become culturable again. Hence, the detection of VBNC cells is important to prevent resuscitation, possible infections and adjust the biocide/antimicrobial dosing.

For illustration purposes, it could be considered that viability (for VBNC characterization) would be determined through PI/SYTO9 uptake results. Taking into consideration the mechanisms of action of the two biocides, this approach would clearly “penalize”/overestimate DBNPA in terms of VBNC calculations (Fleischmann et al., 2021), since this biocide does not interfere with the cell’s membrane, during the contact time considered. On the other hand, VBNC cells resulting from BAC exposure could be underestimated. This issue is particularly relevant when assessing VBNC’s in real-life systems, where biocides with different mechanisms of action are combinedly applied. Thus, the use of the most appropriate viability marker should be further discussed when determining VBNCs. Nonetheless, Kirschner et al. (2021) also pointed out that care must be taken when quantifying VBNC cells, especially when such quantification is based on a single viability parameter analysis. Alternatively, a double staining procedure with 4,6-di-amidino-2-phenylindole (DAPI) and tetrazolium salt 5-cyano-2,3-ditolyltetrazolium chloride (CTC) could be used to discriminate VBNC cells (Besnard et al., 2000; Fleischmann et al., 2021). Unlike PI/SYTO9, CTC does not assess membrane integrity, rather it is reduced through the electron transport chain by actively respiring bacteria (Wideman et al., 2021).