S. aureus strains (Newman or SA113) were grown in MHB (Hardy Diagnostics) at 37°C under aerobic (shaking) conditions unless otherwise noted. S. aureus ΔkatA and ΔmprF, were created previously (Peschel et al., 2001; Park et al., 2008) with ΔkatA from the Newman parental strain and ΔmprF from SA113. ZnO NPs were purchased from Sigma-Aldrich (catalog no. 544906) and the specific lots used in this study were MKCG5504 and MKBD9523. Confirmation of the predominant surface morphology of ZnO crystals was done by scanning electron microscopy (SEM) utilizing a JEOL FE-SEM instrument (JEOL, Peabody, MA, USA) at an operating voltage of 15 kV and a probe current of 9.6 A. Length calculations were determined by employing the ImageJ software (version 1.51, NIH, USA) on the SEM images. All other antimicrobials (H₂O₂, daptomycin, ZnCl₂, and MgCl₂) were also purchased from Sigma-Aldrich.

Cultures were grown to an optical density (OD) of 0.4 at 600 nm then diluted 1:200 to approximately 1×10⁶ cfu/ml for all assays. For ZnO NP assays, bacteria were added to microcentrifuge tubes with the indicated concentrations of ZnO NPs in a final volume of 400 μL. The microcentrifuge tubes were continuously inverted for 16–20 h at 37° C to maximize interactions between ZnO NPs and S. aureus. Microcentrifuge tubes containing the same concentrations of ZnO NPs in 400 μL MHB without S. aureus were co-incubated at the same time. After incubation, the microcentrifuge tubes were centrifuged at 100 rpm for 2 min to pellet the majority of the ZnO NPs but not S. aureus. 200 μL of supernatant was then transferred into a 96-well plate and the OD₆₀₀ values were measured using a Fluostar Omega plate reader (BMG Labtech). The background turbidity from residual ZnO NPs was removed by subtracting the OD₆₀₀ readings for the ZnO control tubes from the tubes containing S. aureus and the same concentration of ZnO NPs. For MIC assays using H₂O₂, daptomycin, ZnCl₂, or MgCl₂, the assays were performed in 96-well plates in a final volume of 200 μL and incubated under static conditions for 16–20 h. For daptomycin assays, MHB was supplemented with 50 μg/mL CaCl₂. To control for background turbidity caused by the salts (MgCl₂ and ZnCl₂), the OD₆₀₀ of wells with the same concentration of salt without S. aureus were subtracted from the OD₆₀₀ readings from the wells containing the same concentration of salt with S. aureus.

Conditioned media was made by incubating 20 mg/mL of ZnO NPs in sterile MHB in clear conical tubes at room temperature. At indicated times, the media was mixed to resuspend the ZnO NPs and a 1 mL aliquot of the media was removed and centrifuged at 16,000 rcf for 5 min to pellet the ZnO NPs. 100 μL of the supernatant from this aliquot was then added to an individual well of a 96-well plate. Log phase S. aureus was then diluted 1:100 in unconditioned MHB and 100 μL was added to the same wells for a final concentration of 50% conditioned media and approximately 1×10⁶ cfu/ml S. aureus in a total volume of 200 μL. As a positive control, S. aureus was also grown in unconditioned MHB and as a negative control, 200 μL of conditioned media (no S. aureus) was placed in another well. The plate was incubated under static conditions at 37° C and OD₆₀₀ values were measured after 16–20 h. To generate a dose curve of conditioned media, log phase S. aureus was diluted 1:100 and 100 μL was added to each well. Then 100 μL of media containing different ratios of conditioned: unconditioned media was added for a total volume of 200 μL containing approximately 1×10⁶ cfu/ml S. aureus in a final concentration of 50, 25, and 12.5% conditioned media. Plates were then incubated under static conditions at 37°C for 16–20 h before reading the OD₆₀₀ values.

Chelated media was made by adding 30 mg/mL of Chelex beads (Sigma Aldrich) to the media (either unconditioned MHB or conditioned MHB) and inverting continuously for 1 hour at room temperature. The mixture was then sterile filtered to remove the beads and the process was repeated. Cultures were grown to log phase (OD₆₀₀ of 0.4) and separated into four aliquots. Each of these aliquoted cultures were pelleted and washed in PBS two times before being resuspended in the equivalent volume of either MHB, conditioned media, chelated media, or chelated conditioned media. These aliquots were then diluted 1:50 in their respective media (MHB, chelated MHB, conditioned MHB, or chelated conditioned MHB). 50 μL of the 1:50 cultures was then added to a 96-well plate containing 150 μL of the matching media type to produce a 1:200 final dilution (approximately 1×10⁶ cfu/ml) of S. aureus in 100% of each media type. To produce a mix of 25% regular MHB and 75% of each media type (chelated MHB, conditioned MHB, or chelated conditioned MHB), 50 μL of the S. aureus culture diluted 1:50 in MHB was added to a 96-well plate containing 150 μL of either chelated media, conditioned media, or chelated conditioned media. The 96-well plate was incubated under static conditions at 37° C and OD₆₀₀ values were measured after 16–20 h.

The concentration of Zn²⁺ in conditioned media was measured by ICP-MS at the University of Texas at Dallas Chemistry Core Facility. MHB was conditioned with 20 mg/mL ZnO NPs for either one-day or at least one-month. An aliquot of the conditioned media was then chelated as described above. 100 μL of each sample was diluted with 4,900 μL of blank for analysis and a 50x correction factor applied to the data. Two independent samples were measured for each condition and reported as mean ± standard deviation.

Empty dialysis buttons (Hampton Research), elastic bands, and 3.5 MWCO semipermeable dialysis tubing (ThermoFisher) were sterilized by UV radiation (254 nm) for at least 15 min on all sides prior to assay set up. Either 0 mg or 30 mg of ZnO NPs was loaded into sterilized dialysis buttons and sterile MHB was added until the button was filled. The dialysis button was then sealed with dialysis tubing soaked in sterile water and secured with elastic bands. Excess tubing was cut to size with disinfected scissors. Log phase S. aureus Newman (approximately 2×10⁶ cfu/ml) was grown in the presence of an empty dialysis button, 30 mg free ZnO (no button), or 30 mg of ZnO inside a dialysis button in a total volume of 1.5 mL MHB. To ensure proper sterilization of the dialysis buttons and aseptic technique, an empty dialysis button was incubated in MHB in the absence of bacteria and served as the negative control. To control for background turbidity of free ZnO NPs, 30 mg of ZnO NPs (no button) was incubated in the absence of S. aureus in 1.5 mL of MHB. The assays were conducted in individual wells of a 24-well tissue culture plate that was incubated at 37°C shaking at 200 rpm for 16–20 h. After 16–20 h, OD₆₀₀ values were measured. To minimize background from residual ZnO particles, the OD₆₀₀ of free ZnO NPs incubated in the absence of S. aureus was subtracted from the OD₆₀₀ of free ZnO NPs incubated with S. aureus.

GraphPad Prism (San Diego, CA, USA) was used for all statistical analysis. Statistical significance set as p<0.05 unless otherwise indicated.

Several previous studies have demonstrated the antimicrobial activity of ZnO NPs against S. aureus using a variety of ZnO NP sources and assay conditions (Raghupathi et al., 2011; Kadiyala et al., 2018; Dadi et al., 2019; de Lucas-Gil et al., 2019; Ali et al., 2021). The ZnO NPs used in this study were purchased from Sigma-Aldrich and no treatment was done on them prior to experimentation. The certificates of analysis from the two lots used have nearly identical parameters including a complexometric titration of 79.7 percent zinc (lot MKBD9523) or 79.8 percent zinc (lot MKCG5504) and an average particle size of 67 nm and specific surface area of 16 m²/g (both lots). We used SEM to measure size distribution and shape and found a wide range of sizes with an average of 52 nm (lot MKCG5504) and 48 nm (lot MKBD9523), but with a standard deviation of 39 and 46 nm, respectively, and a granular/mixed shape (Supplementary Figure S1). Despite the size range, our measurements align with the reported values and particle size is less than 100 nm. One challenge in assay setup is that ZnO NPs have low solubility and do not stay suspended in solution. If added to the well of a plate, particularly under static conditions, they will accumulate at the bottom, potentially creating unequal interactions between the ZnO NPs and bacteria in the well. In order to address this, we utilized a ZnO NP assay with continuous inversion that we had previously established (Johnson et al., 2022) as we found better mixing with the ZnO NPs than orbital shaking in a well plate. As seen before, this resulted in inhibition of bacterial growth at an MIC of 1.25 mg/mL (15.4 mM) for ZnO NPs obtained from Sigma-Aldrich (Figure 1B).

Production of H₂O₂ by ZnO NPs has historically been the predominant theory proposed for the antibacterial properties of ZnO (Xu et al., 2013; Sirelkhatim et al., 2015; Siddiqi et al., 2018; Tiwari et al., 2018; Lallo Da Silva et al., 2019; Gudkov et al., 2021). The NP surface acts as a catalyst surface to initiate a cascade of oxidation reactions that results in the production of H₂O₂. This cascade begins as dissolved oxygen molecules are transformed into superoxide radical anions (O₂⁻). The O₂⁻ molecules then react with H⁺ to form HO₂ which can pick up electrons and become hydroxyl peroxide anions (HO₂⁻). HO₂⁻ is inherently unstable and will react with H⁺ to form H₂O₂ (Lallo Da Silva et al., 2019; Li et al., 2020). In addition to being more stable than the intermediates, H₂O₂ is also neutrally charged and can diffuse through the bacterial membrane to cause intercellular damage (Lallo Da Silva et al., 2019; Li et al., 2020). To evaluate the role of H₂O₂ in ZnO NP toxicity, we used a strain of S. aureus where the catalase gene (katA) was removed via allelic replacement to generate ΔkatA (Park et al., 2008). Catalase is responsible for breaking down H₂O₂ into H₂O and O₂ and loss of the gene results in increased susceptibility to H₂O₂ (Park et al., 2008). We hypothesized that if H₂O₂ is responsible for ZnO-mediated growth inhibition, then ΔkatA should be more susceptible to ZnO NPs than its parental strain (Newman). We confirmed that ΔkatA had increased susceptibility to H₂O₂ (Figure 1A), but we saw no difference in ZnO NP susceptibility as compared to the parental strain (Figure 1B) indicating that H₂O₂ is not responsible for ZnO-mediated toxicity.

Physical contact that disrupts cellular integrity is another commonly proposed mechanism of ZnO toxicity. Previous studies have seen evidence of physical interactions between ZnO NPs and the bacterial cell surface and reported damage to the bacterial cell membrane after exposure to ZnO NPs (Feris et al., 2010; Sirelkhatim et al., 2015; Joe et al., 2017; Siddiqi et al., 2018; Akbar et al., 2019; Lallo Da Silva et al., 2019; Gudkov et al., 2021). Additionally, several studies have specifically implicated electrostatic interactions in mediating this contact (Feris et al., 2010; Arakha et al., 2015). To investigate the role of electrostatic interactions in S. aureus, we used a mprF deficient strain of S. aureus (ΔmprF) (Peschel et al., 2001). MprF is responsible for synthesizing and translocating lysyl-phosphatidylglycerol, a positively charged phospholipid, to the cell membrane. Loss of mprF results in a cell envelope that is more negatively charged compared to the parental strain resulting in increased susceptibility to cationic antimicrobial peptides such as defensins and the antibiotic daptomycin (Peschel et al., 2001; Ernst et al., 2009). We hypothesized that if electrostatic interactions on the bacterial envelope mediate the antibacterial activity of ZnO NPs, ΔmprF would exhibit increased susceptibility to ZnO NPs compared to the SA113 parental strain. We confirmed that ΔmprF showed increased susceptibility to daptomycin relative to the parental SA113 strain (Figure 2A) but found that there was no difference in susceptibility with ZnO NPs (Figure 2B) under our assay conditions.

Although electrostatic interactions between bacterial and ZnO NP surfaces did not mediate toxicity, it is possible that physical contact and disruption of the membrane occur independent of electrostatic interactions. To evaluate the necessity of physical contact, we conditioned the media by incubating ZnO NPs in MHB in clear conical tubes for up to 2 months, allowing any soluble factors produced by ZnO NPs to accumulate in the media. ZnO NPs were then removed via centrifugation at the indicated times and the ZnO-free, conditioned supernatant was evaluated for the ability to inhibit bacterial growth (Figure 3A). No statistically significant difference in S. aureus growth was seen between unconditioned MHB and conditioned MHB after 1 day of conditioning (Figure 3A, upper left panel), but by 2 weeks the ZnO NP-conditioned media exhibited partial growth inhibition (Figure 3A, upper right panel). At one-month, only small amounts of growth were seen in conditioned media although it was not significantly different than the negative control lacking S. aureus (Figure 3A, lower left panel). By 2 months, no growth was seen when S. aureus was added to conditioned media (Figure 3A, lower right panel). This suggests that ZnO NPs release a soluble species in a time-dependent manner that contributes to bacterial growth inhibition.

We next wanted to investigate whether physical contact was necessary in a real-time assay as it is possible that physical interactions are more critical for toxicity when time is limited (i.e., there is not a month to generate sufficient soluble species for growth inhibition). To do this, we sequestered ZnO NPs inside a dialysis button covered with dialysis tubing which created a semi-permeable membrane that allowed for diffusion of small soluble species across the membrane but prevented any physical contact between the ZnO NPs and bacteria. To ensure that ZnO NPs were unable to diffuse through the dialysis tubing, sequestered ZnO NPs were first incubated in MHB in the absence of bacteria. We found no change in turbidity relative to MHB alone indicating the ZnO NPs remained inside the dialysis membrane (data not shown). We next added S. aureus to the assay and found that the sequestered ZnO NPs effectively inhibited bacterial growth relative to the S. aureus grown in the absence of any ZnO NPs (Figure 3B). This confirmed that ZnO NP could inhibit bacterial growth in the absence of physical interaction with the cell. To compare the effectiveness of growth inhibition in the absence and presence of physical contact, S. aureus was also incubated with free (non-sequestered) ZnO NPs under the same conditions. We found that while there was no statistical difference in S. aureus growth between sequestered and free ZnO NPs, only the growth inhibition with the free ZnO NPs showed no statistical difference from our negative control (MHB containing an empty dialysis button but no S. aureus). Therefore, while physical contact with ZnO NPs is not required for growth inhibition in our assay, it may still contribute to the effectiveness of the particles.

Although we saw no difference in growth between the parental S. aureus and ΔkatA in the presence of ZnO-NPs (Figure 1B), we wanted to further confirm that H₂O₂ is not the soluble species by incubating ZnO NP-conditioned media with the catalase-deficient S. aureus stain. Unsurprisingly, we saw no difference in growth between the parental S. aureus and ΔkatA in the presence of conditioned media (data not shown). Previous studies have shown dissolution of Zn²⁺ ions from ZnO NP surfaces that have been linked to microbial toxicity (Pasquet et al., 2014; Sirelkhatim et al., 2015; Siddiqi et al., 2018; de Lucas-Gil et al., 2019; Lallo Da Silva et al., 2019; Song et al., 2020; Gudkov et al., 2021). To assess whether accumulation of Zn²⁺ ions in the conditioned media is the soluble factor responsible for growth inhibition, we used Chelex beads to strip the conditioned media of all divalent cations. S. aureus grew well in unconditioned MHB (Figure 4, black bar) but the removal of all divalent cations in chelated unconditioned MHB completely inhibited growth (Figure 4, left dark gray bar). This was expected since the presence of some divalent ions is necessary for bacterial growth. As additional negative controls, we incubated S. aureus in ZnO NP-conditioned MHB (Figure 4, middle dark gray bar) and in chelated conditioned MHB (Figure 4, right dark gray bar) and no growth was seen in either condition. We next supplemented our experiments with a small amount (25% of total volume) of regular (non-chelated, unconditioned) MHB to restore enough divalent cations to support the growth of S. aureus in the media. When we supplemented the chelated MHB with regular MHB (Figure 4, left light gray bar), it resulted in growth that was no different than our positive control (Figure 4, black bar). However, supplementing with regular MHB was not sufficient for growth when conditioned MHB was included in place of chelated MHB (Figure 4, middle light gray bar) suggesting the presence of a toxic, soluble species only present in conditioned media. Chelating conditioned MHB completely removed this toxic substance resulting once more in full growth of S. aureus (Figure 4, right light gray bar) that was no different from our positive control (black bar). Removal of the toxic soluble species through chelation indicates that the toxicity is mediated by divalent cations.

To confirm that Zn²⁺ concentration is increased when MHB is conditioned with ZnO NPs, we measured its concentration using ICP-MS. We found that Zn²⁺ concentration increased 100x after one-day of conditioning (31 μM to 3,249 μM) and 300x after at least 1 month of conditioning (31 uM to 9,588 μM) and that this was reversed after chelation (Table 1). We next evaluated susceptibility of S. aureus Newman to Zn²⁺ and found that the MIC of ZnCl₂ was 0.625 mg/mL (4,590 μM) (Figure 5A). It is therefore not surprising that using conditioned media containing Zn²⁺ at a similar concentration would inhibit growth. To determine whether this was the result of an overabundance of divalent cations or if it was specific to Zn²⁺, we conducted an MIC with MgCl₂ and saw no inhibition of growth up to 10 mg/mL (Figure 5B). Together this suggests that it is the accumulation of Zn²⁺ ions released from ZnO NP surfaces that are mediating the antibacterial mechanism of ZnO NPs in MHB.

Now that we had identified the soluble species as Zn²⁺, we wanted to again look at the potential impact of electrostatic changes to the bacterial membrane using the ΔmprF mutant. However, we saw no difference with either our conditioned media (Figure 6A) or Zn²⁺ generated from ZnCl₂ (Figure 6B). Therefore, while electrostatic interactions may still play a role, they are distinct from what is observed with cationic molecules such as antimicrobial peptides or daptomycin and changes in cell membrane surface charge are not sufficient to effect changes in bacterial susceptibility to Zn²⁺.