Routine environmental nanotoxicology observations are often made without considering the dynamic transformation of the nanomaterials over time, which may lead to observations that cannot be deconstructed adequately into each variable separately. This work investigates the dynamic transform of lithium cobalt oxide (LiCoO 2 ) nanosheets in solutions by both abiotic reactive oxygen species (ROS) generation and Co ion release, and delineates their respective biological impacts in a model bacterium, B. subtilis. The observed additional changes in oxidative stress genes and DNA damages in B. subtilis coincide with the burst of H 2 O 2 in fresh nanoparticle suspensions, which provides direct evidence to connect the abiotic ROS generated by the nanoparticles to the oxidative stress responses in organism. Therefore, the study illustrates a new approach to evaluate nanotoxicity and reveals the importance of evaluating abiotic ROS generation in complex metal oxide nanoparticles.
With the widespread use of nanoscale materials in a variety of fields, research into their biological and environmental impact becomes increasingly important. Several metal oxides, such as TiO2 and ZnO, due to their applications as photocatalysts 1,2 and in food industry and medical applications 3,4 , have been more broadly studied for their biological and environmental impacts. As energy demand grows and fossil fuel resources dwindle, lithium-ion batteries with promising high cell potential, high gravimetric and volumetric capacity, and good cycling performance have taken the center stage recently. 5,6 Consequently, a class of lithium intercalating complex metal oxide nanomaterials has emerged, and is produced in large quantities as battery cathodes. 7 To date, lithium cobalt oxides, LiCoO2, is one of the most ubiquitously used complex metal oxides -from electric vehicles to consumer microelectronic devices. A lack of economic incentives and infrastructure for recycling of these materials 8,9 especially calls for studies to examine the environmental and biological impact of these novel nanomaterials at the end of their life cycle and entrance into the environment.
Metal oxide nanoparticles can lead to cytotoxicity through dissolution of metal ions when placed in media, largely due to their high surface-to-volume ratios. For instance, lithium nickel manganese cobalt oxide has been shown to release toxic levels of nickel, manganese, and cobalt ions that impact bacterial respiration 10,11 and the lifecycle of Daphnia magna. 12 Yet, often, the dissolved ions cannot fully recapitulate the biological impacts induced by the nanoparticles. In eukaryotic cells where nanoparticles may be internalized, cytotoxicity has been linked to nanoparticleinduced intracellular reactive oxygen species (ROS) generation and cellular oxidative stress. 13 Metal oxide nanoparticles can induce ROS due to their reactive surfaces, semiconductor electronic properties, or through the release of redox active transition metal ions triggering biomolecule redox reactions. 14 Species, such as superoxide, hydroxyl radical, and hydrogen peroxide, once formed intracellularly at levels that overwhelm the antioxidant systems, often result in oxidative stress exhibited by DNA damage or lipid peroxidation. [14][15][16] The overproduction of ROS and imbalance of antioxidant defence system can lead to different diseases and cell death. 17 Therefore, examining the mechanisms leading to the production of intracellular ROS and the biological consequences has been proposed as a paradigm for NP toxicity. 1,18 Some metal oxides, especially semiconductors and photocatalysts, have been shown to spontaneously generate abiotic ROS. 3,[18][19][20] We have previously detected evidence for abiotic ROS produced from lithium nickel cobalt oxide suspensions as the material undergoes incongruent dissolution. 21 The production of ROS, especially superoxide, by TiO2 has been demonstrated both in response to UV radiation 19,22 and in the dark. 3 These findings raise the question of whether abiotic ROS can directly induce intracellular ROS and trigger oxidative stress responses in model organisms. However, to delineate the biological responses due to ROS vs. metal ions is a challenging question where few studies have attempted to answer, 1 and requires a highly systematic approach.
The question is further complicated by the dynamic nature of both NP transformation and their biological consequences. Although studies have demonstrated the kinetics of metal ion dissolution from metal oxides, 10,11,21,23 few have paid equal attention to monitor the time-dependent ROS generation from nanoparticles. 1 In addition, when organisms are exposed to nanomaterials for extended period of time (e.g. days), the studies often result in examining the impact from two coupled variables over time: material transformation and biological responses. Cui et al. demonstrated the dynamic oxidative stress responses from trout gill cells over 48 hrs in a series of elegant single-cell gene expression experiments where different genes were triggered at different time points upon exposure to LiCoO2. 24 These findings further highlighted the needs for investigating the dynamic process of LiCoO2 transformation and the cellular responses.
In this study, we monitor the dynamic transformation of LiCoO2 in terms of Co release and abiotic ROS generation, and examine the biological impact of the transformed LiCoO2 towards a model bacterium, B. subtilis. B. subtilis is a ubiquitous Gram-positive bacterium that plays major roles in the terrestrial carbon cycle to supply nutrients to plants, and has well-characterized genomes. In contrast to eukaryote cells where nanoparticles can be internalized, 25,26 bacteria do not usually take up the nanoparticles. 10,21,27,28 Therefore, using bacterial models eliminates the complication that material intracellular transformation may differ from that characterized in situ. By using species-specific probes and biochemical assays, we aim to detect and identify the spontaneously generated ROS in growth medium as well as intracellular ROS, and establish connections between them. In order to isolate the two time-dependent variables of material transformation and biological responses, we designed experiments to allow LiCoO2 NP suspensions to age in the absence of bacterium for 1 hour and 48 hours, and characterized ion release and abiotic ROS generation at two different time points. We then treated B. subtilis with the 1-hr and 48-hr aged LiCoO2 suspensions for a short duration of 30 minutes to minimize the variations in biological responses over time. We assessed oxidative stress markers, such as DNA damage and changes in gene expression in B. subtilis upon exposure to LiCoO2 suspensions. Results indicate that additional oxidative stress in bacterium that cannot be fully accounted for by the Co ion released were observed in freshly suspended LiCoO2 solutions, overlapping with the burst of H2O2 generation in solution, which suggests that abiotic ROS generation from LiCoO2 indeed led to additional cellular oxidative injury. This study highlights the importance to characterize the dynamic variables independently in order to understanding the multiple paths that lead to oxidative stress responses in organisms.
Characterization of as-synthesized lithium cobalt oxide nanoparticles.
LiCoO2 nanoparticles were characterize morphologically and structurally. Micrograph images from TEM show the typical morphology of these nano-structures (Fig 1 ) where a high magnification view of one of such structures shows the sheet-like morphology (Fig 1b). BET analysis yielded a surface value of 125 m 2 /g for the LiCoO2 nanoparticles. The collected powder XRD pattern, published previously, 29 can be indexed to the R3 ̅ m space group, as expected for this crystal structure. However, we note that the assynthesized LiCoO2 without the high temperature annealing step has lower crystallinity, which more-closely represents spent cathode materials after numerous electrochemical cycles, the condition at which environmental exposure occurs. Dynamic Light Scattering (DLS) measurements of 5 mg/L suspensions of particles in ultrapure water yielded diffusion coefficient and -potential values of 0.8 ± 0.1 µm 2 /s and -2.0 ± 1.0 mV, respectively. We report diffusion coefficient (in µm 2 /s) of these particles as an indicator for size because it is a direct measurement from the DLS, while hydrodynamic diameter is calculated from diffusion coefficient assuming the particles to be spherical in shape, a poor assumption to make in this case.
Previously, both experimental 25,30 and computational studies 31 have demonstrated that LiCoO2 nanoparticles can undergo partial dissolution in aqueous media. The extent and the kinetics of dissolution are dependent on nanomaterial surface properties, solution pHs and constituents. Prior studies [10][11][12]25,26 examining the biological impact of this class of complex metal oxides with various model organisms have repeatedly indicated that the amount of Li + released, although significantly higher than those of the transition metal ions (e.g. Co 2+ , Ni 2+ and Mn 2+ ), has minimal impact to the Please do not adjust margins Please do not adjust margins organisms tested, B. subtilis included. 11 Therefore, we focus on the specific impact of cobalt ion released from the material in this study. ICP-MS analysis was used to quantify the time-dependent cobalt ion dissolution in the LiCoO2 suspensions in a B. subtilis growth minimal medium. Fig 2 shows the amount of Co 2+ released from 5.0 and 50.0 mg/L LiCoO2 after suspension for 1-hr (blue) and 48-hr (red) periods. As expected, higher amounts of Co 2+ ions were released in solution at the higher LiCoO2 concentration, and as the nanoparticles were suspended for a longer period of time compared to a freshly suspended (1-hr) LiCoO2 suspension.
In addition to ion release, previous investigations of complex metal oxides have revealed that, depending on the composition of the material, reactive oxygen species can be generated spontaneously upon material dissolution. 21 ROS are known to cause oxidative stress resulting in cellular damage in bacteria. 32 Therefore we aimed to first detect and identify if any ROS were generated simultaneously by LiCoO2 nanoparticle suspensions in bacterial growth media.
Fig 3 shows the results from a selection of ROS probes used to detect and identify the presence of specific species in LiCoO2 suspensions over time. Amplex Red TM assay, 33,34 a horseradish peroxidase (HRP)based assay that has been previously applied to detect H2O2 generated from TiO2 nanoparticle suspensions 35 was used to detect and quantify H2O2 in LiCoO2 NP suspensions. Fig 3a shows results from the quantification of H2O2 using Amplex Red, indicating the formation of abiotic ROS in bacterial growth medium with LiCoO2 suspension, and the identity of the ROS is likely to be H2O2. Statistical analysis shows that the amounts of H2O2 generated from the freshly (1-hr) suspended LiCoO2 NPs at both 5 and 50 mg/L level are significantly different from that of the control (blank medium), and the amounts of H2O2 detected in a 48-hr old suspension are significantly lower than those from freshly suspended LiCoO2 at both concentrations (P < 0.0001, two-way ANOVA). The results suggest that ROS is indeed generated at an early stages of LiCoO2 NP dissolution in aqueous media, yet the abiotic ROS signal decreases over time as the LiCoO2 is left in solution for longer period of time.
Although the use of fluorescent probes is a common and effective approach to detecting ROS, 36 one of many possible errors that can arise when using such a probe molecule is false-positive detection via the probe reacting at a potentially catalytic surface, such as LiCoO2 NPs. 37 To ensure that positive results of the Amplex Red assay were due to the transformation of Amplex Red to resorufin upon reaction with free H2O2 and not on the nanoparticle surfaces, we performed diffusion assays with LiCoO2 suspension. We assembled an apparatus (Fig S1 ) to spatially separate the LiCoO2 nanoparticles from the Amplex Red-HRP solution using a finely porous hydrophobic filter membrane. Fluorescence of the resulting solution above the filter membrane (i.e. no direct particle exposure) is used to detect H2O2 with 1 μM H2O2 below the filter membrane as a standard. Under this experimental design, an increase in fluorescence would be from the reaction of Amplex Red with free H2O2 and not with the nanoparticle surfaces, assuming i) H2O2 readily diffuses through the membrane, ii) LiCoO2 cannot diffuse through the membrane (25 nm pores are small relative to particle diameter), and iii) HRP will not appreciably diffuse through the membrane within the time scale of the experiment. Although Amplex Red may diffuse through the membrane, conversion to the fluorescent product requires the HRP catalyst. Although we expect cobalt ions may also diffuse through the membrane, Co 2+ alone did not induce fluorescence intensity change with Amplex Red in control experiment. Fig 3b shows representative fluorescence spectra of each sample tested with the normalized fluorescence intensity normalized to the background in the inset. The presence of LiCoO2 NPs results in an increase in fluorescence compared to the AR-HRP control solution. Addition of 1 μM H2O2 spike produces a similar effect, indicating that H2O2 diffuses through the membrane over the 1 hr period and reacts with AR-HRP. The presence of LiCoO2 particles alone (no AR-HRP) shows no fluorescence, with small background intensity coming from scattering of the excitation. These results show that positive Please do not adjust margins Please do not adjust margins detection of H2O2 in our Amplex Red experiment is indeed due to the presence of abiotic H2O2 and not direct interaction of Amplex Red with the LiCoO2 NPs.
The singlet oxygen sensor green (SOSG) fluorescence dye was employed to discern the production of singlet oxygen in the LiCoO2 dissolution and is highly specific to singlet oxygen detection. This dye manifests in weak blue color initially and emits green fluorescence with the presences of singlet oxygen (excitation/emission: 504/525 nm). 19 Lastly, the generation of superoxide was monitored by observing the reduction in absorbance at 259 nm of a nitro blue tetrazolium solution (NBT). Superoxide, if present, can react with NBT to form a precipitate, formazan, resulting in a decrease in the absorbance at 259 nm. 35 Fig 3d shows that the NBT absorbance signal does not decrease in the presence of LiCoO2 suspension regardless of concentration and time-point, which suggests no superoxide formation by the LiCoO2 nanoparticles in solution.
LiCoO2 suspension in bacterial growth medium. More significantly, we were able to identify the species generated as hydrogen peroxide and quantify its concentrations in LiCoO2 suspensions. The H2O2 diffusion experiment also critically illustrated that the observed fluorescence signal using a dye was not the results of optical interference by the nanomaterials, or by nanoparticle surface catalyzed chemical reaction of the dye molecules. The chemical nature of LiCoO2 determines that cobalt is in the Co 3+ state. In fact, we have detected a Co(III)-EDTA complex spectroscopically when LiCoO2 NP was dissolved in the presence of EDTA in solution (Fig S2). Upon dissolution, Co 3+ is likely to reduce to the more soluble form of Co 2+ . We have also observed the redox activity of LiCoO2 upon suspension in aqueous media is able to oxidize the electron transporters, nicotinamide adenine dinucleotide (NADH) (unpublished). Therefore, we hypothesize that the Co(III) reduction is then likely to trigger water oxidation and generate H2O2.
Interestingly, by examining the solution constituents of the nanoparticle suspension at two different time points (1-hr vs. 48-hr), we were also able to probe the dynamic process of both ion release and ROS formation from complex metal oxides in aqueous medium. Although studies have indicated the importance to monitor ion release from metal or metal oxide nanoparticles through dissolution over time because of ion-induced toxicity to organisms, 10,11 there have been few studies examining abiotic ROS formation over time. We note that parallel attention is needed to monitor the generation of abiotic ROS in these nanoparticle suspensions over time as well as ion release, in order to develop a holistic view of the chemistry of material transformation.
On the other hand, the LiCoO2 particles after suspension in media do not undergo noticeable changes morphologically compared to those freshly synthesized (Fig S3). The layered structure of the nanosheets remains visible. -potential (ZP) analysis of medium-exposed LiCoO2 also yielded comparable values as shown in Table 1. The exposed particles have more negative ZP values than that of the pristine particles, which is expected due to the surface-adsorbed phosphate species to LiCoO2. 30 The diffusion coefficients of the mediumexposed particles after 1 hr and 48 hr exposure are also comparable with that of the pristine particles, suggesting minimal variations in particle size. This experimental evidence suggests that any transformations due to exposure to the medium is likely surface limited.
Previous studies have shown that transition metal oxides do not enter bacterial cells, in contrast to their interactions with eukaryote cells. 10,25 Instead, the ions released from these materials often can recapitulate the impact on bacterial respiration 10,11,38 and viability. 39 Therefore, we studied the effect of LiCoO2 nanoparticles and the corresponding amount of ions released over time by monitoring the viability of B. subtilis. A growth-based viability (GBV) analysis was performed to quantify the relative amount of viable bacterial cells under different exposure conditions to LiCoO2 nanoparticles by the periodic measurement of OD600. 40 In this assay, the viability of bacterial cultures exposed to nanoparticles is assessed by comparing the delay in the culture regrown in fresh nutrient-rich media against a preconstructed calibration curve relating the delay to the number of viable cells. The assay is especially effective in evaluating nanotoxicity to bacterial species because it eliminates the concerns for nanomaterial aggregation in nutrient-rich media, and reduces optical interferences by nanomaterials in normal growth assays. We hypothesized that cobalt dissolution contributes significantly to the
Cell viability and toxicity studies, although important indicators of the nanomaterial's impact, are often end-point measurements that do not provide detailed mechanistic insights of the more subtle biological changes. Hence, we examine the biological impact to LiCoO2 by characterizing Co ion internalization and intracellular ROS generation guided by the observations of ion release (Fig 2 ) and the abiotic ROS generation (Fig 3). We first test the hypothesis that as the amount of Co 2+ released in solution increases over time, a higher influx of Co 2+ is internalized in bacterial cells. Therefore, the LiCoO2 dissolution will likely result in increased metal ion influx in bacterial cells. Newport Green™ DCF is a cell permeable fluorescent dye used for the detection of divalent metal ion internalization in cells 41 and has been previously successfully employed in bacterial species to monitor the internalization of divalent transition metal ions. 39,42 In the intracellular environment, superoxide is often formed on the oxygen reduction pathway in a single-electron transfer redox reaction when molecular O2 adventitiously oxidizes redox enzymes. 45,46 Intracellular superoxide, once formed, can be further reduced to H2O2 in another one-electron redox process by superoxide dismutase, SOD. 45 Therefore, the generation and consumption of superoxide are likely to be a dynamic process in B. subtilis, which may explain the concentration-and suspension-ageindependent behavior observed.
Because H2O2 can both permeate through the cell wall from LiCoO2containing media and be generated from intracellular superoxide reduction, we investigate the fate of intracellular H2O2 in B. subtilis using a luminescence dye, ROS-Glo TM . Unfortunately, due to an optical interference of the dye in the presence of LiCoO2 NPs, indicated by a color change upon mixing, we were unable to conduct the parallel experiment using LiCoO2 NP-treated bacterial cells. Instead, we conducted experiments by dosing the bacterial culture with 30 µM H2O2 to observe the luminescence signals from cells upon treatment. The results from cells spiked with H2O2 distinctively show that H2O2 is not accumulated in bacterial cells, but is likely further converted (Fig S6). The intracellular concentration of H2O2 has been previously described as the difference between influx+intracellular formation and efflux+scavenging. 47 The diffusion rate of H2O2 across bacterial membrane can be matched by the rate of Alkyl hydroperoxide reductase (Ahp) or catalase turnover at the micromolar concentration levels of H2O2, which lowers intracellular H2O2 concentrations than that of the external environment. 48 Our results from ROS-Glo TM indicate a dynamic and active conversion of H2O2 in B. subtilis.
Intracellular H2O2 can either be converted to H2O and O2 by catalase, or react with labile Fe 2+ in the intracellular environment through the Fenton reaction to produce •OH. 34,48 Therefore, we further investigate the downstream biological impact of intracellular ROS. Although H2O2 does not normally damage DNA directly, 45 the highly transient and electrophilic •OH is known to attack the electron-dense DNA molecules that lead to DNA damage. 49 Hence, although it is experimentally challenging to directly detect •OH intracellularly, by monitoring the extent of bacterial DNA damage, we can probe the impact of this intracellular ROS. Single-cell gel electrophoresis (i.e, comet assay) at neutral pH environment is a sensitive method to detect and compare the extent of double-strand DNA breakage caused by exposure to LiCoO2 NPs. The fragmented bacterial DNA exhibits a tail-like morphology upon nucleic acid staining, and the tail length is indicative of the extent of damage. 50,51 We have previously used this method to assess the genotoxicity of nanomaterials to bacteria successfully. Transition metals are well-known to induce oxidative stress through generating intracellular ROS. 34 Such properties of metal or metal oxide nanoparticles have also been explored as antibiotic agents towards bacterial species. [52][53][54] Due to their multivalent nature, many transition metal ions can trigger redox chemistry in cellular environment by disrupting enzymatic functions and damaging biomolecules, such as proteins and DNA. In vitro studies have shown that Co 2+ have been linked to single-strand breaks in salmon sperm DNA through Fenton-like reactions, and to create putative intrastrand cross-links of DNA. 55 Although Co(II) did not generate
Please do not adjust margins significant amount of •OH when reacting with H2O2, a Co(I) mediated Fenton-like reaction has been proposed. 56 In addition, Co 2+ can preferentially bind to specific sequences in DNA, e.g. the 5' G of GG sequences or the middle G of GGG. 34 We have also observed doublestrand breakage induced by Ni 2+ and Co 2+ , 11 as well as a variety of oxidative stress-related putative DNA adducts in two bacterial species upon exposure to nanoscale nickel manganese cobalt oxides in previous studies. 42 Taken together, transition metal oxides generating multivalent metal ions that can enter bacterial cells and lead to bacterial DNA damage through Fenton-like reactions is likely a common toxicity mechanism for such materials. alone. H2O2, an uncharged species that can penetrate membranes 47 and enter bacterial cells, has been known to induce cellular stress whenever it is present in their extracellular habitat. With as little as 1 μM of intracellular H2O2, crippling levels of DNA damage has been reported in E. coli. 48 In addition to examine the resulting DNA damage in bacterial cells, gene expression changes related to bacterial oxidative stress, metal regulation and DNA repair mechanisms were also evaluated to help understand the biological response to both LiCoO2 NP suspensions and their corresponding amount of Co 2+ released. The specific functions of each gene examined are included in Table S1 in the ESI. B. subtilis was exposed to fresh (1 hr) and aged (48 hr) suspensions of LiCoO2 NPs and Co 2+ at 5 and 50 mg/L concentration levels to complement the DNA damage analysis. The heat map in Table 2 shows that, among the three categories of genes examined, the ones related to oxidative stress pathways are most severely altered. None of the genes related to metal regulation or DNA repair is significantly changed compared to control.
In a previous global analysis of oxidative stress genes 57 , it has been demonstrated that protection against H2O2 in B. subtilis was largely mediated by the induction of proteins controlled by the PerR regulon, representing the primary stress response. Among members of the PerR regulon, only katA, mrgA, and zosA can be strongly induced by the H2O2 treatment. 58,59 Remarkably, katA and mrgA, encoding the vegetative catalase, KatA, and the metalloregulation DNA-binding stress protein, MrgA, respectively, are the most significantly altered genes in all testing conditions in our study. A previous study examining changes in the CAT gene in a benthicdwelling organism, C. riparius, upon exposure up to 100 mg/L LiCoO2 has also reported significant down regulation of the CAT genes. 26 In addition, tpx, another gene that is significantly changed in Fig 8a and8b, encodes proteins that has been suggested to be a thiol peroxidase, has been linked to peroxide detoxification. 57 . We also note that, interestingly, although we have observed an increase in intracellular O2 -signals using a fluorescence probe (Fig 6a), no significant fold changes in the oxidative stress gene sodA, related to the superoxide anion, was observed. This difference could be due to the dynamic conversions of ROS in the intracellular environment. We note that the formation of intracellular superoxide from molecular oxygen could be catalyzed by Co 2+ , 56 and unlike H2O2, superoxide anion is not membrane permeant. The gene expression profile especially highlights the differences in cellular response to the stimulus of H2O2 (katA, mrgA and tpx), likely in response to the abiotic generation of H2O2 in the growth medium.
Oxidative stress in bacteria has been a well-reviewed area. 32,45,48,66 It was recently proposed that induction of the OxyR regulon (PerR for B. subtilis) is one of the most reliable markers for oxidative stress in bacteria, as it is the bacterial detection system for intracellular H2O2. 66 We note that the connection between the generation of abiotic ROS in growth media and the intracellular ROS detected in model organism needs to be carefully established. Previous study has demonstrated that even when NPs do not generate abiotic ROS spontaneously, they may still induce intracellular ROS in mammalian cells and induce mitochondrial apoptosis. 1 Here, we demonstrate that the abiotic ROS (H2O2 in this case) generated in the process of NP transformation in solution can also trigger intracellular ROS and lead to additional oxidative stress responses in cells. Since both the abiotic ROS generation and intracellular ROS responses are dynamic, their correlation and interdependence can only be probed properly when both processes are monitored over time.
Altogether, our detailed analysis reveals that LiCoO2 NPs in aqueous growth medium can spontaneously release cobalt ions and generate H2O2 when freshly suspended in solution. An initial burst of H2O2 in solution is followed by a subsequent decrease over time, while cobalt ion concentration increases over the time period monitored. Solution ROS generation from nanomaterials has previously been demonstrated largely in semiconductors and photosensitive materials. 1,3,19 LiCoO2, as a model complex metal oxide, has not been previously identified as an ROS generator. The half reaction of LiCoO2 reduction to generate Co 2+ in water has a standard reduction potential of E 0 = +2.14 V vs. SHE, which is likely to drive the oxidation half reaction of H2O to form H2O2 (E 0 = -1.76 V vs. SHE) thermodynamically. Therefore, it is critical to identify and quantify abiotic ROS formation, as well as to develop principles to predict biological impact upon NP exposure. We note that the fate of solution ROS, depending on its identity, can also be further influenced by media constituents, for instance pyruvate has been shown to sequester H2O2 to remediate cytotoxicity in mammalian cells, 67 and radical scavenger, Trolox, or serum proteins can mitigate lipid peroxidation by ROS generated from TiO2 nanoparticles. 68 Overall, our study has revealed the intriguing dynamic oxidative stress responses in B. subtilis in response to the aqueous transformation of LiCoO2. Although NP toxicity through dissolution and associated release of toxic ions into solution has been a wellestablished mechanism to organisms, 18,69,70 the spontaneous formation of ROS in aqueous media by NPs coupled to ion release has been less investigated, 1,3 especially in the absence of light or other energy sources. By using a bacterial model, and allowing the NP suspensions to transform independently from biological exposure, we successfully demonstrate the biological impacts of both ion release and the spontaneous formation of abiotic ROS by LiCoO2. By acknowledging the dynamic nature of nanomaterial transformation which may trigger different subtle oxidative stress responses in organisms, we have designed the experiments that allow us to evaluate these variables independently. Although our attempt at tackling the dynamic process of nanomaterial transformation is coarse in this work (two time points), the information revealed provides us with insights into the subtle molecular-level responses in model organisms that have previously been obscured. Our approach is a crucial step towards a new way to evaluate nanotoxicity of complex metal oxides that will allow us to predict the environmental and biological impacts of these nanomaterials more accurately. Please do not adjust margins Please do not adjust margins
Nanoparticle synthesis and transformation in bacterial growth medium Synthesis of lithium cobalt oxide nanomaterials. We synthesized sheet-like nanoparticles of LixCoO2 as described in detail in previous publications. 29,30 Only a brief description is provided here. First Co(OH)2 nanoparticles were prepared through a precipitation between lithium hydroxide, LiOH, and cobalt nitrate hexahydrate, Co(NO3)2•6H2O. The Co(OH)2 precursor nanoparticles were then transformed into lithium cobalt oxide, LixCoO2 by addition to a molten salt flux of LiNO3:LiOH. The reaction in the molten salt flux was allowed to continue for 30 minutes, then quenched with water. The isolated precipitate was dried at 30 °C in a vacuum oven overnight, and stored in a glovebox under argon atmosphere when not in use. To quantify the cobalt ions released from the LiCoO2 after aging, 5 and 50 mg/L diluted LiCoO2 solutions prepared from the 1000 mg/L stock were centrifuged at 4000 ×g for 10 min. A fraction of the supernatant was removed and ultra-centrifuged at 200,000 ×g for 30 min. The supernatant was then removed to measure for cobalt ion content using the Inductively Coupled Plasma-Mass Spectrometry (ICP-MS), as done previously. 39 Abiotic ROS generation from lithium cobalt oxide nanoparticles Solution hydrogen peroxide detection. To detect the formation of hydrogen peroxide, Amplex Red ® reagent (Thermo Fisher) was used according to the manufactural procedure. Amplex Red reagent and 0.2 U/mL Horseradish Peroxidase (HRP) were mixed with either standard hydrogen peroxide solutions, or LiCoO2 suspensions at various concentrations to incubate in the dark at room temperature for 30 minutes. 71 Fluorescence intensity was measured (535 nm / 590 nm) from a series of hydrogen peroxide standards to construct a calibration curve for the quantification of solution hydrogen peroxide generated in LiCoO2 suspensions.
Solution singlet oxygen formation. The singlet oxygen sensor green (SOSG) (Thermo Fisher) can be used to detect the presence of singlet oxygen in solution by reacting specifically to singlet oxygen. 19 LiCoO2 suspension were mixed with the SOSG stock solution in a 0.1M HEPES buffer (pH 7.2). The generation of singlet oxygen was evaluated by measuring fluorescence at ex = 480 nm and em = 528 nm on a 96well fluorescence plate reader. Fluorescence intensity was background corrected and normalized to those from control wells.
Solution superoxide detection. The formation of superoxide in LiCoO2 suspension can be monitored by measuring the reduction of nitro blue tetrazolium (NBT). 34,35 An NBT stock solution was mixed with LiCoO2 suspensions and incubated at room temperature in the dark for 1 hour. The supernatant of the mixture was obtained through centrifugation (12,000 ×g, 5 min), and analyzed on a spectrophotometer at 259 nm. The formation of superoxide in solution is often indicated by the reduction in the peak intensity at 259 nm, resulted from the depletion of NBT upon reacting with superoxide.
Bacillus subtilis SB491 was purchased from the Bacillus Genetic Stock Center (Columbus, OH). Bacterial colonies were grown in solid lysogeny broth (LB) agar plates, and inoculated in LB growth medium overnight at 37 °C.
Lithium cobalt oxide nanoparticle toxicity to B. subtilis. Growthbased viability (GBV) was performed to assess the bacterial viability in the presence of aged LiCoO2. 40 B. subtilis was inoculated and grew in liquid LB at 37 °C overnight and harvested at mid-log phase. The cell pellets were then washed with Dulbecco phosphate-buffered saline (D-PBS) and resuspended in minimal medium with dextrose. The bacterial culture was then diluted to an OD600 nm of 0.05 with minimal medium. Bacterial culture at OD 0.05 was exposed to desired LiCoO2 suspension conditions for 30 minutes while agitated, and reinoculated in fresh LB media for 16 hours. The growth curves were compared and analyzed to assess the impact of LiCoO2 or Co 2+ to bacterial viability.
General intracellular ROS formation. To detect the formation of intracellular reactive oxygen species in general, the non-specific ROS fluorescence probe, 2',7'-dichlorodihydrofluorescein diacetate, Please do not adjust margins Please do not adjust margins DCFH2-DA, was used according to manufactural procedure. Bacterial cells were harvested at mid-log phase and adjusted to OD600 of 0.2. DCFH2-DA stock solution was added to the cell culture to a final dye concentration of 20 µM, followed by an incubation at 37 o C for 1 hour in the dark. After incubation, cells pellets were resuspended in fresh minimal medium to eliminate the unreacted dye. LiCoO2 suspensions were added to the dye-loaded cells and incubated with cells for 30 minutes. Lastly, the LiCoO2-treated bacterial culture were washed and resuspended in fresh minimal medium for OD600 and fluorescence measurements (485 nm / 525 nm).
Dihydroethidium assay for intracellular superoxide. To detect the formation of intracellular superoxide formation, Dihydroethidium (DHE) (Sigma Aldrich) was used. 3,44 Bacterial cell cultures suspended in minimal medium with dextrose at OD600 0.6 were incubated with 5 µM DHE solution in the dark at 37 o C for 30 minutes. LiCoO2 suspensions at various concentrations were added to the dye-loaded cell cultures and incubated with cells for 30 minutes. Bacterial cell pellets were then centrifuged and resuspended in fresh minimal medium for OD600 and fluorescence measurements (500 nm / 580 nm).
Co 2+ internalization in bacterial cells. To monitor the internalization of Co 2+ , a DCFH2-DA derivative dye, Newport Green TM (Thermo Fisher) was used. 39 Bacterial culture were adjusted to OD600 0.2 in minimal medium with dextrose, and incubated at 37 o C for 1 hour in the dark. After washing off the excessive dye and resuspending in fresh medium, the cells were exposed to LiCoO2 suspension for 30 minutes. After washing and resuspending in fresh medium, OD600 and fluorescence intensity was recorded (505 nm / 535 nm).
All intracellular fluorescence assays were first normalized over cell density (F/OD600) to correct for any discrepancies in the number of bacterial cells. The corrected fluorescence intensities were then normalized over control to yield a "fold change".
Bacterial double-strand DNA breakage upon LiCoO2 exposure as an indication for DNA damage has been studied using the single-cell gel electrophoresis method, as previously described. 11,42 Briefly, 10 µL of diluted LiCoO2 exposed cells at OD600 0.05 were mixed in 100 µL Low Melting Agar, and placed on a FLARE TM Slide (Trevigen, MD). Slides were incubated at 4 ᵒC for 10 minutes to solidify, followed by the addition of a second LMA gel layer containing a 0.5 % lysozyme solution. Once the second layer is solidified, the slide was incubated at 37 ᵒC for 30 minutes in the dark, followed by overnight immersion in a lysing solution (2.5 M NaCl, 100.0 mM EDTA, 10.0 mM Tris -HCl, 1% sodium N-lauryl sarcosine, 0.6% Triton® X-100, pH 10.0), and an enzyme digestion solution (2.5 M NaCl, 10.0 mM EDTA, 10.0 mM Tris-HCl, and 0.5 mg/mL proteinase K, pH 7.4) at 37 °C for 2 hours. Slide then underwent electrophoresis in a chilled opaque electrophoresis tank with sodium acetate -Tris electrophoretic buffer at pH 9.0 at 12 V for 30 minutes in the dark. The slide was then dehydrated in a sequence of solutions of 1 M ammonium acetate in ethanol (20 minutes), absolute ethanol (30 minutes), and 70% ethanol (10 minutes), and dried in ambient air for 5 minutes. Lastly, the slides were rehydrated in 20 µL of DMSO solutions (5% DMSO, 10 mM Na2H2PO4) and stained with 20 µL of 1 µM YOYO-1 dye in 5% DMSO and imaged with a fluorescence microscope (λex = 491 nm, λem = 509 nm). Positive controls have been previously conducted using kanamycin and a cationic polymer, poly(allylamine hydrochloride).
To study the changes in gene expression levels in B. subtilis upon exposure to different age and concentrations of LiCoO2 suspensions or to corresponding amount of Co 2+ released, bacterial cultures were harvested at mid-log phase and adjusted to OD600 of 0.2 in the minimal growth medium, and incubated with various nanoparticle or ion treatments at 37 o C for 30 minutes. After incubation, the exposed bacterial cells were harvested by spinning and flash-freezing, and stored in -80 o C until RNA extraction. Four biological replicates were collected for the controls and each treatment groups.
A detailed description for RNA extraction is provided in the ESI. The same qPCR procedure has also been previously reported. 42 Briefly, to synthesize complementary deoxyribonucleic acid (cDNA) following Invitrogen's protocols and the iCycler base module of an iQ5 Multicolor Real-Time, extracted RNA was mixed with Master Mix 1 (random primers (Invitrogen, 48190-011) and dNTP (Invitrogen, 18427013)) for 5 minutes at 65 o C, then chilled on ice for 1 minute. Then Master mix 2 (5x-First Strand Buffer, Dithiothreitol, RNaseOUT TM recombinant ribonuclease inhibitors (Invitrogen, 10777019), and Superscript RT III reverse transcriptase (Invitrogen, 18080-044)) was added to the reaction in an iQ5 Multicolor Real-Time PCR Detection System at 25 ᵒC for 5 minutes, 50 ᵒC for 60 minutes, 70 ᵒC for 15 minutes for random primer extension. The resulting cDNA samples were characterized on a NanoDrop UV-vis spectrometer and stored at -20 ᵒC until qPCR analysis.
Target genes relevant to oxidative stress, metal homeostasis and DNA repair have been selected for B. subtilis. Table S1 in ESI provides detailed information regarding the functions of the genes and the sequences of the primers. For qPCR using an iQ5 real-time PCR detection system (Bio-Rad Laboratories) with SYBR Green for the fluorescent intercalating dye (iTaq™ Universal SYBR® Green Supermix, Bio-Rad), manufactural protocol was followed. cDNA, primers and SYBR Green Supermix (Bio-Rad) were combined to react in a qPCR 96-well plate. The reaction started at 95 ˚C for 10 minutes to denature the DNA, followed by 40-times temperature cycles of amplification of 15 s at 95 ˚C and 30 s at 60 ˚C. The gene expression data was processed by normalizing the raw threshold cycle (Ct) numbers against the output of a housekeeping gene, arsR. Each qPCR reaction was duplicated, and each gene analysis included four biological replicates.
Nanomaterial characterizations and quantitative analysis (e.g. ICP-MS) were carried out with three technical replicates and three analytical replicates. Biological exposure, intracellular ROS, viability, and gene expression experiments throughout the study were carried out at least in triplicates (n 3). Depending on the parameters Please do not adjust margins Please do not adjust margins compared in each analysis, different statistical tests were used. Details of the analysis are provided in each figure caption.
Bacterial cells used for single cell gel-electrophoresis analysis were from 4 biological replicates of nanoparticle or ion exposure. DNA tail length data from different replicates were deemed identical to be combined only when the tail lengths analysis from the control groups were not tested to be statistically different (p > 0.05). Merged DNA tail length data was tested for normality using the D'Agostino & Pearson normality test first, followed by non-parametric Kruskal-Wallis tests with Dunn's multiple comparisons test.