Emerging technologies in several fields like energy, pharmaceuticals, catalysis, and textiles have increased demand for highly efficient nanomaterial-based systems, which are cost effective and easy to use. Focusing on the energy front, it has been estimated that the use of nanomaterial-based systems will increase by 2050 due to the demand for more electric vehicles.
Recent advances have made nanoscale lithiated nickel manganese cobalt oxide (Li x Ni y Mn z Co 1- y-z O 2 , 0 < x,y,z < 1, abbreviated as NMC) a viable option as a highly efficient battery cathode material (1)(2)(3). NMC has a layered structure that is similar to nanoscale lithium cobalt oxide (LCO) (4) and has gained interest due to its superior ability to transport lithium ions to provide better conduction, and the reduced impact of mechanical stress during lithium intercalation and deintercalation (5)(6)(7). As NMC has huge commercial benefits, its large-scale production and use also increases the likelihood that it will enter the natural environment during manufacturing, usage, and waste disposal. The costs and energy requirements for recycling lithium ion-or NMC-based battery materials are high and large quantities of these materials go into landfills.
Nickel, manganese, cobalt, and lithium ions often leak from the buried batteries, which ultimately contaminate land, ground water, and other water bodies. As such, it is critical for us to understand the effects of NMC on the ecosystem and organisms in the environment. We and others have previously reported that NMC exhibits toxic effects to many species including bacteria and aquatic organisms (8)(9)(10)(11). Here, we investigate the roles of reactive oxygen species (ROS), DNA damage, and mutation frequency in the response of the bacteria Shewanella oneidensis to NMC.
NMC nanoparticles have sheet-like morphology (12) with an average size of 8422 nm measured along the basal plane (10). Full characterization details for NMC used in these studies is provided in the Supporting Information. An electric vehicle typically has 40-50 kg of NMC nanomaterial in the battery pack (7). Thus, improper materials disposal could result in the release of tens of kilograms of NMC. Indeed, leachate from landfills containing lithium ion batteries has been found to have toxic levels of various heavy metals linked to battery materials (59). The concentration of these materials in the environment is highly variable, but our previous work has shown that concentrations of > 25 mg/L are toxic to bacterial cells (12).
NMC nanomaterials have toxic effects on growth of the environmentally-relevant bacterial species Shewanella oneidensis MR-1. S. oneidensis is a Gram-negative bacterium that is ubiquitously present in the environment including soil, sediment and aquatic systems and possesses metal cycling and remediation properties. Reports suggest that in addition to being toxic, NMC decreases cellular respiration in bacteria, as measured by oxygen consumption and also leads to DNA damage within eight hours of exposure in S. oneidensis, as studied by the comet assay and high-resolution DNA adductomics (10,13). Interestingly, NMC nanoparticles cannot enter the bacteria as determined by high resolution scanning electron microscopy and transmission electron microscopy and are believed to exert toxic effects by their presence in the vicinity of the cells (10). NMC is transformed in liquid as shown by surface composition studies using XPS and metal dissolution by ICP-OES. NMC undergoes incongruent dissolution resulting in metal ion release and ROS generation (10). Relatedly, H 2 O 2 generation from freshly suspended lithium cobalt oxide nanomaterials (similar to NMC, but without Ni and Mg)
has been shown to led to ROS damage in bacterial cells (14). H 2 O 2 is cell permeable and is known to itself cause an increase in cellular production of other ROS, such as hydroxyl radicals, in bacteria.
We previously reported that chronic exposure to NMC leads to the development of resistance (or adaptation) in bacteria where these organisms can grow in the presence of NMC concentrations that were toxic in earlier exposures (12). The ion equivalents of NMC (the ions released from NMC during 72 hr of exposure) were not as toxic as the particles themselves, indicating an effect that is specific to nanoparticle exposure (12). We found that during the process of adaptation to NMC, a portion of the bacterial population become filamented and can increase to 10-30 m in length (wild-type 2-3 µm) with a minor population being elongated to 80-100 m (12). This extreme filamentation was not observed upon exposure to the ion equivalents of NMC, again indicating a nanoparticle-specific consequence.
Filamentation is a common bacterial response to stress from a variety of conditions such as DNA damage (15), starvation (16)(17)(18), exposure to antibiotics (19,20), changes in pH, low temperature, host immune response, or onset of the SOS response (21)(22)(23). The SOS response is an inducible global response triggered in bacteria upon DNA damage where cell division is arrested and the expression of several DNA repair proteins is induced to promote DNA integrity for improved survival at the cost of increased mutagenesis. While filamentation leads to division arrest, it also allows the cells to replicate and repair DNA damage to ensure that a repaired chromosome is passed to progeny. This process has largely been studied in Escherichia coli where division is halted until the DNA is repaired, which is sensed by SulA, a protein that remains bound to the essential cell division protein, FtsZ, until repair has been completed (15,24,25). Bacterial filamentation has also been correlated with the SOS response, reactive oxygen species (ROS), (26) and DNA damage.
We postulated that ROS may be a critical player in the filamentation of S. oneidensis upon NMC exposure as ROS may be generated by interaction of surface groups of NMC with molecular oxygen, which then participates in subsequent reactions to form additional ROS species (27). Thus, dissolution of metal-containing nanomaterials could be one cause of ROS formation, (28) which has been correlated with their toxicity (29)(30)(31)(32)(33)(34). Most aerobically respiring organisms have built-in mechanisms to maintain redox balance including detoxification enzymes such as catalase, peroxidase, glutathione reductase, superoxide dismutase, and thiol metabolites. However, these systems can be overwhelmed when the cell encounters high concentrations of exogenous ROS, resulting in its accumulation and oxidative stress.
Ultimately, widespread damage can result, such as protein and lipid damage, disruption of metal homeostasis, DNA strand breakage, and single nucleotide modifications (13,14,(35)(36)(37).
The present study examines the roles of ROS in the response of S. oneidensis to NMC with focus on filamentation, DNA damage, and mutation. the plate and inoculated into 5 mL of MM for primary cultures and grown at 30 ºC, shaking at 250 rpm, for 24 h. Overnight cultures were diluted to an optical density (OD 600 ) of 0.1 at 600 nm (GENESYS 20 spectrophotometer, ThermoFisher Scientific) for sub-culturing/passages.
The first sub-culture was performed with a 10% dilution (1:10 v/v) of the overnight diluted culture in fresh MM and is referred to as "passage A". After 72 h of growth, the culture was again diluted to 0.1 OD and sub-cultured into passage B. This was repeated until passage D.
All the samples were prepared in triplicate and bacterial growth assessed by measuring OD 600 .
NMC nanoparticles were synthesized as previously published (10,12). When needed, a fresh stock of dispersed NMC solution was prepared at the concentration of 2 mg/mL in minimal media with sonication for 10 min. The dispersed NMC was added to the cultures (25 mg/L) 10 h after inoculation for the first passage (passage A). For subsequent passages, NMC was added at the time of bacterial inoculation as in the previous study (12). Similarly, ion equivalents were prepared from fresh stock solutions of LiOH, NiCl 2 , MnSO 4 , and CoCl 2 and added to the cultures to achieve a final concentration as present following dissolution of NMC over 72 h in minimal media (12).
ROS was measured using a cell-permeant dye 2',7'-dichlorodihydrofluorescein diacetate (H 2 DCFDA) or DCFDA dye (ThermoFischer Scientific/Invitrogen, D399). NMC-exposed, Ion eqv.-exposed and unexposed bacterial samples were stained by adding 5 µM final concentration to 200 µL of culture in Eppendorf tubes, mixed by brief vortex (~2 sec) and incubated in the dark for 30 min at room temperature. For microscopy, 5-10 µL stained culture, unfixed, was spread on a glass slide, covered with coverslip, and imaged under fluorescence microscope (Olympus) with 100X magnification and 1.4 numerical aperture. Brightfield and fluorescent images were taken using the brightfield channel and FITC fluorescence channel, respectively, at fixed exposure time of 500 ms. For plate reader-based assay, 200 µL DCFDAstained cultures were transferred to a black 96-well flat bottom black (Greiner) and read in a plate reader (Tecan) at ex/em of 488/535 nm. The fluorescence readings were blank subtracted as well as corrected for the bacterial count obtained using colony forming units (CFU) for each sample. The data was analyzed and plotted using GraphPad Prism software. For flow cytometry, the samples were prepared in larger volume by aliquoting 1 mL from the cultures directly into the flow cytometry tubes and stained with DCFDA at the final concentration of 5 µM and incubated at room temperature for 30 min in the dark. An autofluorescence control was also taken for all the samples without addition of the DCFDA dye. Samples were analyzed by BD LSR II H4710 flow cytometer with a 488 nm excitation laser (20 mW blue laser), 525/50 BP emission filter with FITC (488 E) settings and 10,000 events were captured at medium flow rate for all the samples. A P1 gate was constructed by considering the spread of the population in the unstained samples (without the addition of DCFDA dye) and these samples were termed as autofluorescence controls. This P1 gate was placed in such a way to avoid maximum cells from all the autofluorescence samples. Thus, only the cells with fluorescence in DCFDA dye stained samples can be monitored in P1 gate. Data acquisition and data analysis were performed using BD FACSDiva TM software.
Hydroxyphenyl fluorescein, also known as 2-[6-(4′-hydroxy)phenoxy-3H-xanthen-3-on-9yl]benzoic acid or HPF (Sigma-Aldrich, H4290) was used for the estimation of hydroxyl radicals in the samples. Unexposed bacterial cultures, as well as those exposed to NMC or ion eqv., were aliquoted in fresh Eppendorf tubes, 200 µL of all samples in triplicate and stained with HPF at the final concentration of 5 µM for 15 min in the dark at room temperature. The samples were transferred to a black 96-well flat bottom black (Greiner) and read in a plate reader (Tecan) at ex/em of 490/520 nm. The fluorescence readings were blank subtracted from the respective MM blank, NMC in MM blank or ion eqv. in MM blank, as well as corrected for the bacterial count obtained using colony forming units (CFU) for each sample. The data were analyzed and plotted using GraphPad Prism software. For flow cytometry analysis, 1 mL of all the samples was aliquoted directly into the flow cytometry tubes and stained with HPF at the final concentration of 5 µM and incubated at room temperature for 15 min in the dark.
An autofluorescence control was taken for all the samples without addition of HPF dye. This was used the set a gate outside the autofluorescence signals in FSC/SSC to get the fluorescence- positive samples in the gated population. The samples were analyzed by BD LSR II H4710 flow cytometer with a 488 nm excitation laser (20 mW blue laser), 525/50 BP emission filter with FITC settings and 10,000 events were captured at medium flow rate for all the samples. P1 gate was constructed as described in previous paragraph. Data acquisition and data analysis were performed using BD FACSDiva TM software.
Amplex Red assay kit (Invitrogen, A22188) was used to estimate the H 2 O 2 concentration of the NMC/ion eqv.-exposed and unexposed cultures. All reagents were prepared as per manufacturer's protocol with the stock concentrations of 10 mM Amplex red reagent and 10 U/ml HRP solution. Standards for H 2 O 2 were also prepared from 0.1 µM to 10 µM in order to generate a standard curve to calculate the concentrations of H 2 O 2 in the samples. A H 2 O 2 scavenger, dimethyl thiourea-containing controls were also prepared to ensure the presence of H 2 O 2 in the samples. From the NMC/ion eqv.-exposed and unexposed cultures, 50 µL were transferred to individual wells in 96-well black plate as well as the H 2 O 2 standard solutions were also added to different wells. To all the samples, 50 µL of Amplex Red/HRP working solution was added to attain a final working concentration of 100 µM Amplex Red reagent and 0.2 U/ml HRP, followed by incubation at room temperature for 30 min in the dark. The fluorescence readings were taken using a plate reader (Tecan) at ex/em of 530/590 nm every 15 min over 1 hr. The readings thus obtained were blank subtracted and the limit of detection and limit of quantitation were calculated from the triplicate readings. Following the blank subtraction, the H 2 O 2 standard curve was plotted using GraphPad Prism and the straight-line equation was obtained using the linear regression settings. released per cell has also been calculated using the CFU count for all the culture samples.
Abiotic NMC-only, as well as ion-only controls in minimal media were also used to estimate the H 2 O 2 release from NMC/ion eqv. in minimal media.
Bacterial lipid membrane was visualized using commercially available fluorescent dye FM TM Pyridinium Dibromide), (Invitrogen, T13320). From the NMC/ion eqv.-exposed and unexposed cultures, 200 µL of culture was aliquoted and 2 µL of FM4-64 dye was added to samples to attain a concentration of 1 µg/mL, vortexed briefly. These samples were incubated at room temperature for 15 min in the dark. From the stained samples, 10 µL was spread on a glass slide, covered with a coverslip and fluorescence imaged using the TRITC channel with ex./em. of 544/570 nm on an inverted microscope (Olympus) with 100 X magnification and 1.4 numerical aperture.
A fluorescent D-amino acid (FDAA) was used to image the peptidoglycan layer in bacterial cells and vesicles. RADA (Orange-red TAMRA-based FDAA, R&D systems-biotechne, cat.
No. 6649) was used at the final concentration of 1 µM and added to 200 µL of culture and kept at room temperature for 15 min in the dark. Cells were imaged as described previously during FM4-64 staining. Ten µL of the stained cultures was spread on a glass slide, covered with a coverslip and fluorescence imaged using the TRITC channel on an inverted microscope (Olympus) with 100 X magnification and 1.4 numerical aperture.
Live cell imaging was performed on 1.5% agarose pads prepared with minimal media using a glass bottom µ-dish, 35 mm, #1.5H (170 µm +/-5 µm) D 263 M Schott glass, sterilized (Ibidi, Cat. # 81158). Bacterial cultures at 72 h were used for imaging, 100 µL culture was evenly spread on the sterile glass bottom dish, by tilting or spotting. A 1.5% low melting temperature agarose solution was made in minimal media and poured over the 100 µL culture in the glass bottom dish. This was solidified at room temperature for 30 min. Cells were imaged under the microscope with 100 X magnification, equipped with an environmental chamber to maintain a temperature of 30 ºC. Images were captured at regular intervals of 5 min to generate a timelapse combined image.
For all experiments, 20 µL of the cultures were mixed with 180 µL MM. This was considered as -1 dilution. From this sample, 100 µL was taken into a fresh tube and 900 µL MM added and considered as the -2 dilution and likewise serially diluted to a -6 dilution. From all the dilutions, 10 µL was spotted on LB agar plates, dried, and incubated at 30 ºC for 12-16 h. Separated colonies were counted (not merged or joined), ranging from 1-30. Colonies were also matched with the immediate next dilution for consistency of the counting. Colony counts were back calculated to determine the number of the cells per sample.
Single cell gel electrophoresis analysis, a.k.a., comet assay, was conducted on S. oneidensis cells from multiple passages upon re-exposure to NMC, following published protocols (13,14).
Briefly, bacterial cells were grown and exposed to NMC or NMC + thiourea for 72 h at each passage. Forty microliters of a bacterial suspension and low-melting agarose (LMA) mixture (1:10 ratio) were placed in a well of a Comet assay slide (Travigen ) and spread evenly. Upon solidifying, an LMA layer containing 0.5% lysozyme was placed on top and solidified. The slide was incubated at 30 o C for 30 min, and immersed in a lysing solution containing 2.5 M NaCl, 100.0 mM EDTA, 10.0 mM Tris-HCl, 1% sodium N-lauryl sarcosine, 0.6% Triton X-100 at pH 10.0 for 1 h, followed by an enzyme digestion solution containing 2.5 M NaCl, 10.0 mM EDTA, 10.0 mM Tris-HCl, and 0.5 mg mL -1 proteinase K at pH 7.4 at 37 o C for 2 h.
NMC/ion eqv.-exposed and unexposed cultures were set up in triplicate and 20 mL of the cultures at 72 h were used for harvesting the cells by centrifugation at 4000 x g for 10 min at room temperature. The bacterial pellet was resuspended in 200 µL of the supernatant. From the resuspended mixture, 20 µL were kept for CFU plating and the remainder (180 µL) was plated on antibiotic-containing plates with 200 µg/ml nalidixic acid, 25 µg/ml rifampicin, and 100 µg/ml erythromycin. Once plated, samples were incubated at 30 ºC for 48-72 h. Colonies observed on the antibiotic-containing plates were considered as the resister mutants. To calculate the resister generation frequency, the number of colonies found on the antibiotic plates was divided by the total number of cells plated (from CFU plating) on that particular plate. Higher values represent higher resister generation frequency or mutation rate (38,39).
Colonies obtained from antibiotic-containing plates were picked and cultured in LB media.
Genomic DNA was isolated from liquid inoculated cultures using a DNA purification kit following the manufacturer's protocol (Promega, Wizard® Genomic DNA Purification Kit, A1120), the final elution was performed in ultra-pure, autoclaved milliQ water. Genomic DNA was used in a polymerase chain reaction (PCR) with specific primers for the mutation hotspot region in the antibiotic resistance-determining region of the genes. For nalidixic acid, the hotspot region named QRDR (quinolone resistance-determining region) in the gyrA gene (40)(41)(42) was PCR amplified, and the hotspot region RRDR for rifampicin resistance-determining region in the rpoB (39,43,44) gene was PCR amplified using high fidelity DNA polymerase, Phusion polymerase (New England Biolabs, E0553S). The PCR-amplified product was purified using a PCR purification kit (GeneJET PCR purification kit, Thermo Scientific, K0701) and the sequenced was amplified using the specific primers (Table S1) by Sanger sequencing (ACGT Inc., DNA sequencing services). PCR amplifications and sequencing were performed for the colonies obtained from the resister plates from all samples, as well as the starter culture from the glycerol stock to confirm the sequence of the parent strain or culture used (never exposed or never sub-cultured). Mutations were identified by aligning the sequences using an online multiple sequence alignment tool, Clustal Omega from EMBL-EBI(45) and mismatches were examined.
Prolonged exposure to NMC leads to extensive filamentation in S. oneidensis S. oneidensis was cultured in minimal media and exposed to 25 mg/L of NMC for four passages where cells were sub-cultured every 72 h (passage A-D; Fig. 1a). This NMC concentration was chosen based on previous work where the response of S. oneidensis was assessed at various concentrations of NMC and the first exposure to 25 mg/L NMC resulted in significant lethality, which diminished as the organism started to adapt (12). Cells from all the passages were imaged with a brightfield microscope to measure length. As seen before, the average cell length remained constant through passages A and B (Fig. 1b,c,f,g) and began to increase during the third re-exposure to NMC (Fig. 1d,h). In passage D, NMC-exposed cells exceeding 40 µm in length were observed although the population is not homogenous at any point in the experiment (Fig. 1e,i). We did not observe any further increase in cell length or a larger proportion of cells with greater lengths in further passages (data not shown). Hence, we focused our studies on passage D.
Given that increased cell length or filamentation is often a stress response, we hypothesized that the observed changes in S. oneidensis resulted from exposure to exogenous ROS (i.e., from NMC dissolution) and/or organismal generation of ROS. To test this supposition, we treated samples with a cell permeable ROS scavenger, thiourea (TU), at a nonlethal concentration to examine its impact on bacterial filamentation (0.1 mM; Fig. S1a andS1b). Thiourea indeed decreased the extent of filamentation in NMC-exposed cells and the length of the bacterial cells did not increase significantly compared to the passaged control (Fig. 1d and1e). Because we had to balance the effects of thiourea on the cells with its utility as a scavenger, we could not examine higher concentrations to determine if this would completely prevent cell elongation. However, the observed decrease in filamentation in the presence of thiourea indicated a role of ROS during NMC exposure.
We also sought to evaluate if the bacterial cells were actively producing ROS using a cell permeable ROS-sensitive dye, 2',7'-dichlorodihydrofluorescein diacetate (H 2 DCFDA or DCFDA). Cells from all passages at 72 hr post inoculation were incubated with DCFDA and imaged. NMC-exposed bacteria in all passages exhibited higher fluorescence as compared to unexposed cells in that passage (Fig. 2a and2b). To investigate ROS levels in the population as compared to individual cells, the fluorescence intensities of DCFDA-stained cells were measured using flow cytometry, as well as a fluorescent plate reader method. Flow cytometry confirmed the increase in ROS upon NMC exposure in all passages. A shift in the population with higher DCFDA fluorescent signal was observed in NMC-exposed cultures (red population in the P1 gate; Fig. 2c and2d; Fig. S2). Importantly, addition of ROS scavengers like TU to the growing cultures decreased ROS levels in the population, confirming the presence of ROS in the NMC-exposed samples (e.g., shift of red population to left by 10.7% in NMC-exposed populations; compare Fig. 2d and2g). Again, because we needed to use a sub-lethal concentration of TU, ROS levels were not expected to drop completely (Fig. S1). We next investigated ROS production when S. oneidensis was exposed to the metal ions equivalent to what is dissolved from NMC during the course of a passage. (12) Ion exposure also resulted in an increase in ROS-specific fluorescence but not as marked as that upon NMC exposure (median fluorescence values 15% higher in NMC-exposed population than ion-exposed samples), indicating an NMC-specific effect (Fig. 2e; Fig. S3). As before, addition of TU decreased the level of ROS (red population shifts left by 8.2%; Fig. 2e and2h).
Similar results were obtained using a fluorescence-based plate reader method where NMC-exposed cells showed a higher DCFDA-specific fluorescence intensity per cell (per cell data has been calculated from C.F.U.) as compared to the unexposed cells in all the passages (Fig. 2i-2l). Ion-exposed cells also exhibited higher DCFDA-specific fluorescence intensity per cell than unexposed cells, but this fluorescence was lower than the NMC-exposed cells (Fig. 2i-2l). Assays were performed with continuously passaged cells (A to D) meaning that experiments were performed across many days making it difficult to quantitatively compare across data sets as some variability is unavoidable. Negative controls such as minimal media blank, minimal media with NMC, or minimal media with ions samples showed negligible fluorescence and were used for background subtraction from their respective test samples.
DCFDA is most informative about the overall level of ROS in the cells, including hydroxyl radicals, hydrogen peroxide, and superoxide radicals. Superoxide radicals in the presence of superoxide dismutase are converted to hydrogen peroxide, which ultimately gives rise to hydroxyl radicals in the presence of free ferrous ion, through the Fenton reaction (46).
Hydroxyl radicals are extremely reactive due to their high one-electron reduction potential (~+1.8 V) (47). We employed a hydroxyl radical-specific dye, hydroxyphenyl fluorescein (HPF), to probe this reactive species. We found that hydroxyl radical levels were significantly higher in NMC-exposed bacteria as compared to unexposed cultures in all the passages (Fig. 3a-3d). Ion-exposed cells also exhibited higher HPF-specific fluorescence intensity per cell than unexposed cells, but this fluorescence was lower than the NMC-exposed cells (Fig. 3a-3d). Flow cytometry also confirmed that the hydroxyl radical levels were higher during NMCexposure, which was observed by a shift of 33% to a more fluorescent population (red colored population shift) as compared to the WT cells (P1 gate, Fig. 3e and 3f; Fig. S4). The P1 gate was assigned on the basis of auto-fluorescent samples (unstained) and not the WT controls, thus, the WT sample showed minimal ROS in the cultures, which might be due to the regular respiratory-related process in the cells. The ion-exposed cells exhibited lower hydroxyl radical levels than upon NMC exposure (Fig. 3g). The median fluorescence of NMC-exposed samples in the P1 gated population was 32.0% higher than that of the ion-exposed samples in the P1 gate (compare Fig. 3f and3g), confirming higher hydroxyl radical content in the presence of NMC. Cells cultured in the presence of a hydroxyl scavenger, TU, along with NMC or ions showed a decrease in the hydroxyl radical level compared to growth without the scavenger (Fig. 3i andj). This decrease was observed by a shift of the red population towards the left by 27.3%
in the case of NMC-exposed populations and 2.30% in the case of ion-exposed samples.
Complete reduction is not expected due to the use of a sub-lethal concentration of TU (see above). The ion equivalent control also showed HPF fluorescence, which is higher than the unexposed cultures but lower than NMC-exposed samples, confirming a combinatorial effect of released ions and a nanoparticle-specific effect during NMC exposure (Fig. 3g). The observed presence of hydroxyl radicals likely contributes to lethality/toxicity in these cultures as it is one of the most toxic ROS and can lead to cell death and/or DNA mutagenesis (48).
Given that hydroxyl radicals are formed upon NMC exposure, we sought to determine if this can be correlated to the levels of hydrogen peroxide in the cultures. An Amplex-Red assay was used to quantify the hydrogen peroxide concentration and compare among differing conditions.
A standard curve was generated using H 2 O 2 solutions, enabling absolute quantification of H 2 O 2 in the bacterial cultures (Fig. S5a). Amplex Red is cell impermeable and therefore only reports on H 2 O 2 that has diffused out of the cells. Cells exposed to NMC showed the highest levels of H 2 O 2 as compared to unexposed and ion exposed samples (Fig. 4a). Cell cultures from all four passages showed higher H 2 O 2 in the NMC-exposed samples as compared to the unexposed cultures (Fig. 4b and S6a-S6c). A gradual increase in the concentration of H 2 O 2 was observed from passages A to D, where passage D exhibited the highest H 2 O 2 concentration from NMCexposed bacteria (Fig. 4b), confirming that NMC exposure leads to an increase in H 2 O 2 over time. Addition of the H 2 O 2 scavenger, dimethyl thiourea (DMTU), in the cell cultures decreased the H 2 O 2 concentrations, confirming the presence and detection of H 2 O 2 . Cultures grown in the presence of metal ions alone also resulted in more H 2 O 2 than unexposed cells but less than NMC-exposed cultures, except in passages B and C (Fig. 4a andS6a-S6c).
With the confirmation of ROS formation in bacterial cultures upon NMC exposure, we sought to determine if it was generated by the nanomaterial, the bacteria, or both. Given that several nanomaterials are known to generate ROS, it is possible that the observed increase in (Fig. S5b). At the later timepoints in the exposure experiments, passages C and D, the amount of H 2 O 2 increases, which is indicative of bacterial production of ROS (Fig. 4b and4c).
generation by unexposed bacteria is minimal (Fig. 4a).
When bacteria encounter environmental toxins, an SOS response is activated resulting in the generation of high levels of ROS within the cellular milieu (26). We anticipate that NMC exposure activates the SOS response as this material is initially lethal to most cells (12). We postulate that the elevated levels of ROS in later passages is the result of SOS activation as the cells started to adapt (H 2 O 2 released in passage C and D highlighted by yellow arrows; NMConly indicated with brown arrow, Fig. 4c). Together, these data confirm that H 2 O 2 is produced both by NMC transformation in the media and by prolonged exposure of S. oneidensis to this toxic material.
We used the same assay to evaluate abiotic H 2 O 2 production from the metal ion solutions, which we found to be low relative to NMC solutions (Fig. S6d), which indicates that the vast majority of the H 2 O 2 released in cultures exposed to metal ions is contributed by the bacterial cells. Most of the passage data suggest that the levels of H 2 O 2 are higher in NMCexposed cultures than in ion-exposed ones. The exception to this trend is passage B, in which the amount of H 2 O 2 from NMC-and ion-exposed cultures is not significantly different. Indeed, the Amplex red assay confirmed that NMC exposure induces the formation of more H 2 O 2 from bacteria as compared to its ion equivalents, indicating a nanomaterial-specific effect.
Overall, we conclude that NMC increased H 2 O 2 levels in cell cultures by both the direct release of abiotically-produced H 2 O 2 and by inducing stress, causing H 2 O 2 generation by the bacteria, perhaps through the SOS response. It is well-known that SOS activation can alter or halt normal bacterial processes, such as cell division, to repair DNA damage (15). Indeed, effects on cell division can ultimately lead to cellular filamentation, which we have observed both in this study and in earlier work (12,49) (Fig. 1e and1i).
To investigate the spatial distribution of ROS in stressed bacteria, we treated NMCexposed cells with DCFDA followed by microscopy visualization, from all the four passages.
Most cells in passage D showed increased fluorescence signal compared to NMC unexposed ones with no fluorescence signals (calculated as fluorescence per unit length, also compare Fig. 2a-2b), with a small population of highly elongated cells (>25 µm) exhibiting lower or no fluorescence (Fig. S7a). These data are indicative of heterogeneity in the ROS levels within the cells, which correlates with the spread in the DCFDA/HPF-stained population observed with flow cytometry (Fig. 2d). We also observed that a proportion of the cells contained vesicles ranging from 0.1-1 µm in diameter. These vesicles contained DCFDA fluorescence indicating that they harbor ROS (Fig. 5a-5d; Fig. S7b-S7m). Staining of the cells with Hoechst dye showed the presence of DNA in cells of all sizes but absence of DNA in ~70% of the vesicles (Fig. 5e andg; Fig. S8). To determine if the vesicles are membrane bound, we stained them with FM4-64 and confirmed the presence of lipid membrane at the periphery of the vesicles (Fig. 5h).
To further evaluate the make-up of the vesicle architecture, we next investigated whether they are encompassed by the crucial cell wall structure, peptidoglycan. Peptidoglycan can be labelled using fluorescent D-amino acids (FDAAs), which become incorporated into the stem peptide of the peptidoglycan chains (50,51). Using the orange-red TAMRA-based FDAA, RADA, to label the peptidoglycan layer, we found that the vesicles exhibited fluorescence confirming the presence of peptidoglycan (Fig. 5f andS8). Formation of vesicles that contain both peptidoglycan and lipid membrane often results from pinching off of a portion of the cells, typically at the division site or towards the poles (52). As these vesicles contained ROS, we hypothesize that the adapted cells may be using them to remove excess ROS, which would otherwise be highly toxic. Consistent with this hypothesis, we also found that some vesicles eventually burst while still attached to the cells, which was visualized with live-cell time-lapsed imaging (Fig. 6a-6o). Once a vesicle bursts, the intensity of the DCFDA stain of this cell drops dramatically, which is not observed in cells with no vesicles (Fig. 6a-6g).
Previous work has indicated that abiotic ROS generated from another transition metal oxide, lithium cobalt oxide, can induce bacterial DNA damage in B. subtilis (14). As such, we examined the extent of DNA damage from multiple passages of NMC-exposed S. oneidensis using the comet assay to assess double-strand DNA breakage. The distribution of DNA tail lengths of single cells from bacterial culture in different passages upon exposure to NMC compared to those from control (WT) is illustrated as a violin plot (Fig. 7a-d). In all four passages, NMC exposure induced significantly longer bacterial DNA tails than those from the control conditions, suggesting more severe DNA damage upon NMC exposure across all Page 20 of 39 Environmental Science: Nano passages. The addition of thiourea (TU) to NMC-containing cultures reduced the extent of DNA damage, indicated by the shorter overall DNA tail lengths. The difference is most stark in passage D (Fig. 7d). The impact of TU as an ROS quencher on mitigating bacterial DNA damage (Fig. 7 and S9) echoes the results obtained in the evaluation of bacterial filamentation (Fig. 1) and in reducing intracellular ROS signals in flow cytometry (Fig. 2 and3). Therefore, NMC exposure across multiple passages induced increased intracellular ROS, as measured from the levels of H 2 O 2 and hydroxyl radicals. Since DNA is a well-known biomolecular target of hydroxyl radicals (53), DNA damage in all passages is observed, and the damage can be mitigated in the presence of a ROS scavenger.
As NMC exhibited DNA damaging effects in the bacterial cells due to the production of ROS, we sought to determine if this ROS also promoted random point mutations in the genome. Rather than performing whole genome sequencing, we opted for a more straightforward selection-based strategy. NMC-exposed and unexposed cells were plated on high (lethal) concentrations of antibiotics as a selection criterion and the resister generation frequency was calculated (Fig. 8a andb). While NMC exposure could cause fitness-conferring mutations that are specific to this material within the genome, ROS-mediated mutations are random and would thus result in potentially advantageous changes in antibiotic-resistancedetermining regions of the genome. As a result, cells with mutations that aid in growth of the bacteria under higher antibiotic concentrations can easily be identified and characterized by sequencing the specific region of the genome associated with antibiotic resistance. For this purpose, we used three antibiotics that function through different mechanisms of action: nalidixic acid (DNA gyrase inhibition), rifampicin (RNA polymerase inhibition) and erythromycin (protein synthesis inhibition). We found that a larger number of colonies were able to survive at higher concentrations of all three antibiotics following NMC exposure as compared to the unexposed cultures (Fig. 8c; S10a and b). These data likely indicate a higher rate of random mutation among NMC-treated cells.
To determine the specific mutations within these strains, single colonies from the antibiotic-containing plate were selected, cultured in liquid media without antibiotics, the genomic DNA isolated, and the resistance-determining region for nalidixic acid was PCR amplified using high fidelity DNA polymerase. The PCR amplicons were sequenced using specific primers and the sequences compared. Colonies that were grown on nalidixic acidcontaining plates showed the presence of point mutations (Fig. 8d and Table S1 and 2), which have been reported in other bacterial species in a clinical setting that gained resistance against nalidixic acid (40)(41)(42). These 'C' to 'T' point mutations also translated to an amino acid change from serine to leucine for most of the colonies. These mutations are usually random in nature as the causative agent is ROS. Thus, some colonies exhibited different a point mutation of 'C' to 'G' resulting in change from serine to tryptophan. These amino acid alterations would affect the protein functionality, ultimately making the mutated bacterial cells resistant to the antibiotic.
Unpassaged cells (WT-stock) did not have these mutations (Fig. 8d), confirming that more mutations either developed during NMC treatment or on the antibiotic-containing plate. Clearly, the presence of NMC increases the potential of cells to inflict genomic mutations, which could be due to the higher levels of hydroxyl radicals (Fig. 3e andf). Higher resister generation frequency in NMC-exposed cells was also seen when cells were plated on rifampicin and erythromycin (Fig. S10).
These results are consistent with our DNA damage assay and indicate that S. oneidensis undergoes more rapid mutation, and thus has a higher mutation frequency upon NMC exposure, perhaps due to greater ROS concentrations. This increase in mutation frequency also implies that there may be random genome-wide mutations that could affect processes such as metabolism, as is indicated by our previously-reported data showing changes in the respiratory abilities of bacteria upon NMC exposure (10).
Nanomaterial-induced toxicity and resistance have largely been studied in the context of antibacterial agents for medicinal purposes. Relatively little is known about how prolonged exposure to engineered nanomaterials that are made for other purposes affects microbes.
However, acute exposure to many metal nanomaterials is toxic to various organisms, including bacteria. From our studies, it is evident that NMC is toxic to a ubiquitously present and environmentally relevant bacterial species, S. oneidensis. Initial exposure causes widespread cell death while prolonged contact results in bacterial filamentation, ROS generation likely due to triggering of the SOS response, and DNA damage. We hypothesize that higher intracellular ROS levels led to this DNA damage and may also cause transformation of other biomolecules.
The exact mechanism of ROS production by NMC is not yet fully understood. However, the similar nanomaterial, LCO, also produces H 2 O 2 upon dissolution that damages bacterial cells (14). It has been reported that NMC does not need to come into physical contact with the bacteria to cause damage and that Ni and Co ions released during dissolution (13), along with abiotically-produced ROS, can enter bacterial cells and trigger intracellular ROS responses.
Our data further support this model.
Our data also show that chronic exposure to NMC resulted in higher rates of DNA mutation. Indeed, we are the first to report a higher frequency of antibiotic resistance evolution in bacteria following nanoparticle exposure. There are reports of the emergence of antibiotic resistance in filamented (54)(55)(56) or multinucleated E. coli upon antibiotic treatment,(57) the former phenotype being seen in our studies. During filamentation, random mutations can be generated, and only those bacteria that gain a beneficial mutation can survive and become resistant towards specific antibiotics or stressors.
Our results indicate that S. oneidensis exposure to NMC is initially lethal to most of the bacterial population. However, extended exposure leads to adaptation and cell populations with phenotypic variations, likely due at least in part to an increased rate of random mutation. We also found that bacteria under NMC stress can generate membrane-bound vesicles, perhaps as a mechanism to expel ROS (58). These vesicles may also be released into the surroundings and act as signals to alert cells in the vicinity of toxic stress exposure. Overall, this study indicates the changes that bacteria undergo during nanomaterial exposure, variations in phenotype and acquisition of non-specific mutations, underscoring the importance of evaluating the environmental impact of technologically relevant engineered nanomaterials such as NMC.
Published on 14 M