Silver engineered nanomaterials and ions elicit species-specific O₂ consumption responses in plant growth promoting rhizobacteria Free

Engineered nanomaterials (ENMs), human designed and manufactured materials with at least one dimension between 1 and 100 nm,¹ are now widely utilized in an array of consumer and medical products, and industrial applications. The antimicrobial properties of silver have been exploited by humans for centuries, but silver containing ENMs (AgENMs) and other ENMs are currently entering commercial and municipal waste streams. It has been predicted the rate at which they will be deposited in terrestrial environments through biosolids amendment of agricultural soils will increase exponentially.² Debate remains regarding predicted environmental concentrations of ENMs, due in part to differences in models implemented to predict environmental concentrations and the lack of empirical data regarding them.^2–5 While biosolids amended soils are predicted to have low μg/kg to mg/kg concentrations of AgENMs in the near term,^2,5 observed concentrations of total silver in biosolids can vary greatly and have been observed as high as ∼856 mg Ag/kg dry biosolids.⁴ 

Despite the debate regarding concentrations of AgENMs in the environment, their exponential rate of release alone constitutes an urgent need to explore the response of relevant test organisms to AgENM exposure. One such group of relevant test organisms is those bacteria inhabiting the dynamic and nutrient rich 1–2 mm zone around a plant root (a.k.a. the rhizosphere) which have been shown to have significant ecological and agricultural benefits. While the effects of AgENMs (and other ENMs) on beneficial soil bacteria and other soil microbes remain largely neglected, interest is steadily growing.³ 

Most microbial nanotoxicology studies have focused primarily on “pristine” or “as manufactured” ENMs.^3,6 However, ENMs undergo various transformations during the wastewater treatment process and after application to soils which can lead to, for example, sulfidation (the formation of sulfides) and reduced toxicity of AgENMs.^6–8 Additionally, few nanotoxicological studies focus on integrative physiological responses, such as O₂ consumption. This is crucial because a range of toxic responses can occur prior to cell death, and antibiotic efficacy has recently been linked to modulation of respiration in medically relevant microbes.⁹ Respiration may also sometimes be a useful endpoint when examining metal/nanomaterial induced stress in microbes.^10–12 

The first experiment in this paper (experiment 1) demonstrates a high-throughput O₂ consumption and viability assay that was developed and tested by assessing the response of Bacillus amyloliquefaciens GB03 (formerly Bacillus subtilis GB03) to Zn²⁺, Ni²⁺, and Ag⁺ ions along with streptomycin controls (experiment 1). B. amyloliquefaciens GB03 is a relatively rapidly growing Gram-positive bacterium utilized as a fungicide¹³ which promotes plant growth.¹⁴ Silver was chosen because its toxic effects are well documented and intracellular concentrations are not tightly regulated, whereas Zn²⁺ and Ni²⁺ were chosen because they are regulated essential nutrients that are also toxic at elevated concentrations and both have been shown to display toxicity to B. subtilis (a relative of B. amyloliquefaciens) at similar concentrations under the same test conditions.¹⁵ Additionally, mechanisms governing intracellular Zn²⁺ concentrations are relatively well understood, while less is known regarding Ni²⁺ homeostasis.

In experiment 2, the method was used to evaluate O₂ consumption and viability of B. amyloliquefaciens GB03 and two additional agriculturally and ecologically relevant plant growth promoting rhizobacteria (PGPR), Sinorhizobium meliloti 2011 and Pseudomonas putida UW4, exposed to Ag⁺ ions, untransformed polyvinylpyrrolidone-coated AgENM (PVP-AgENMs), and 100% sulfidized PVP-AgENMs (sAgENMs). S. meliloti 2011 is a relatively slow growing Gram-negative bacterium capable of forming a nitrogen-fixing symbiosis with the model legume, Medicago truncatula, and related strains are known to promote lettuce growth.¹⁶ Additionally, nodulation of M. truncatula A17 by S. meliloti 2011 was inhibited by biosolids enriched with nanoscale Ti, Zn, and Ag.¹⁷ P. putida UW4 is a rapidly growing gram-negative bacterium capable of plant growth promotion and protection against many environmental/pathogen stresses, likely through modulation of plant physiology via the production of 1-aminocyclopropane-1-carboxylic acid deaminase.¹⁸ While various strains of P. putida have been the subject of many nanotoxicological studies, no study to date has reported the response of UW4 to nanomaterials.

Toxicity was assessed by first measuring O₂ consumption during exposure to metals or antibiotics followed by regrowth of the cells. Experiment 1 was performed to develop a high-throughput method of assessing O₂ consumption and viability responses (Table I). In this experiment, we examined the toxicity of Ag⁺, Zn²⁺, and Ni²⁺ to B. amyloliquefaciens. Experiment 2 was performed on a different microplate reader and the same calibrations and standard (streptomycin) was used in both experiments. In this experiment B. amyloliquefaciens, P. putida, and S. meliloti were exposed to Ag⁺, PVP-AgENMs, and sAgENMs. Highly detailed methods are available in the supplementary material.⁴⁹ 

B. amyloliquefaciens GB03, S. meliloti 2011, and P. putida UW4 cultures were made from 1:1 [glycerol:tryptone–yeast (TY) extract-culture] stocks by streaking on 1% TY agar plates and incubating at 28 °C. For each experimental repeat (block), a single colony was inoculated into TY + 50 mM 2-morpholin-4-ylethanesulfonic acid (MES) (pH = 6) medium and cultured to midlate log phase (OD_600 nm ∼ 0.6–0.86) at 28 °C. Cells were then isolated to the desired cell density (OD_600 nm ∼ 1–1.2) by washing in 50 mM MES three times before use in calibration and exposure procedures.

One hour before exposure, 50 μl of 50 mM MES was transferred to each well of an Oxoplate™ to pre-equilibrate the plate. To initiate exposure, either 100 μl of the live cell suspension or 50 mM MES was transferred to each designated well of the Oxoplate. Twenty five microliters of 8× concentrated metal solutions (in 50 mM MES) were then added to each well, followed by 25 μl of 10% glucose + 50 mM MES. The plate was then sealed with a clear, sterile polymerase chain reaction (PCR) film, and fluorescence intensity was measured over time for an O₂ sensitive indicator dye (excitation: 540 nm, emission: 650 nm) and a reference dye (excitation: 540 nm, emission: 590 nm). O₂ consumption curves were generated using the fluorescence intensity measurements and the Oxoplate supplier's instructions. After 1.5 h, cells were diluted 500× to a total of 100 μl in a clear, sterile 96 well plate. Then, 100 μl of 2× TY + 50 mM MES (pH = 6) was transferred to each well, the plate was sealed with a clear, sterile PCR film, and cell growth was monitored by measuring absorbance over time. In experiment 1, a Wallac Victor 2 microplate reader (Perkin-Elmer Life Sciences, MA) was used and spectral interference inherent to the system was overcome by generating growth curves by using a dual-wavelength method (absorbance_{450–600 nm} = absorbance_450 nm − absorbance_600 nm). In experiment 2, a Spectramax i3 microplate reader (Molecular Devices, Wokingham, UK) was used, and no spectral interference was detected, so growth curves were generated from a single absorbance wavelength (600 nm).

To assess O₂ consumption and regrowth as percentages of an unexposed control, it was necessary to calibrate the O₂ consumption and viability assays (Fig. 1). A sham exposure (without metals or antibiotics) was performed as in Sec. II B. Prior to the sham exposure, cells were diluted to known cell density proportions before being loaded into the Oxoplate. The O₂ partial pressure (in terms of percent air saturation) per well over time (1.5 h) was measured using the methods described by the Oxoplate manufacturer. O₂ consumption curves (percent air saturation over time) were generated for each well, and regression was used to determine the relationship between starting cell density and area under the O₂ consumption curves (Fig. 1).

After the O₂ consumption calibration was complete, cells were diluted 1000× and regrowth was monitored using absorbance as described in Sec. II B. Growth curves were related to starting cell density by recording the time at which cells in each well reached exponential growth and regressing this time against the starting cell density, which is similar to techniques used by others.^19,20 Because initiation of exponential growth was determined via growth curves, it was necessary to test absorbance thresholds to ensure a good relationship between starting cell density and the time at which a growth curve passes a particular absorbance threshold (Fig. 1).

In experiment 1, a randomized complete block design (RCBD) was used to evaluate the toxicity of Ag⁺ (0.5, 1, 2, 3, 4, 6, 8, and 10 μM), Ni²⁺, and Zn²⁺ (0.63, 1.3, 2.5, 5, 10, 20, 40, and 80 μM) to B. amyloliquefaciens using seven experimental blocks. A metal free control was also used across all experimental blocks. Treatments were randomized by column, and control columns were randomized by row across all experimental blocks. Technical replicates of each treatment were performed in triplicate in each experimental block. Ions were supplied as AgNO₃, ZnSO₄, or NiSO₄.

Distribution and variance of residuals were determined following linear regression using distribution plots, Q-Q plots, and Studentized residual plots. Treatment responses were normalized to metal free control wells, averaged within experiments, and the average of experimental means were compared to a hypothetical mean (H_O: μ = 100%, for viability and O₂ consumption measurements) using a two-tailed, one-sample t-test (n = 7, alpha = 0.05, assuming unequal variances). Mean O₂ consumption responses of each experiment were plotted against mean viability measurements and regression curves were fit using softmax pro 6.4 (Molecular Devices, Wokingham).

In experiment 2, an RCBD was used to assess toxicity in B. amyloliquefaciens, P. putida, and S. meliloti, with three experimental blocks and three replicates per block with randomization as in experiment 1. B. amyloliquefaciens and S. meliloti were exposed to 2.6 μM streptomycin, AgNO₃ (0.5, 1, 2, 3, 4, 5, 6, and 10 μM), 296 μM sAgENM, and PVP-AgENM (4.6, 37, 55.6, 111, 148, 185, 222, and 296 μM). The same concentrations of AgNO₃, streptomycin, and sAgENM, but 4.3, 34.5, 43, 46, 115, 144, 172, and 230 μM PVP-AgENM were used for P. putida exposures. Molarity is expressed in terms of total silver concentration. Viability and O₂ consumption estimates were normalized to the no metal controls and were found to be non-normally distributed. The data were ranked and analysis of variance was used to examine experiment by treatment interactions. Because no experiment by treatment interactions were found, observations were pooled across experiments (n = 9) and a two-tailed Wilcoxon signed rank test was used to compare the observations to a hypothetical mean of 100% viability or O₂ consumption (H₀: μ = 100%, α 0.05) using JMP^®, Version 12, SAS Institute Inc., Cary, NC, 1989-2007.

Sigmaplot was used to generate nonlinear, four-parameter dose response curves and extrapolate LC₅₀ values for viability responses to PVP-AgENM and AgNO₃. LC₅₀ values were compared using an unpaired t-test with a Bonferroni correction. Additionally, Sigmaplot was used to perform regression analyses on O₂ consumption-viability plots for interpreting O₂ consumption responses in terms of relative viable cell numbers. Dissolution measurements were used to generate predicted viability responses to the dissolved fraction of AgENMs by extrapolating responses from AgNO₃ dose response curves for the average dissolution estimate ± one standard deviation (n = 3).

Concentrations of stock metals were determined via inductively coupled plasma mass spectrometry (ICP-MS) analysis prior to exposure. PVP-AgENMs and sAgENMs were characterized in the exposure medium at ∼300 μM total silver and 28 °C using a Zetasizer Nano ZS (Malvern). Hydrodynamic diameter and electrophoretic mobility were estimated using direct light scattering and electrophoretic light scattering, respectively. The particles used were previously characterized and PVP-AgENMs had a primary particle size of 53–58 nm,^21,22 while sAgENMs had a primary particle size of ∼65 nm,²¹ both determined via transmission electron microscopy.

A sham exposure (without bacteria) was performed to examine the abiotic dissolution of PVP-AgENMs and sAgENMs in the exposure medium. Briefly, PVP-AgENM (29, 115, and 230 μM) and sAgENM (∼300 μM) were prepared in the exposure medium with glucose and brought to 2 ml in ultracentrifuge tubes. Samples were gently vortexed to ensure mixing and 100 μl was immediately removed for total metal analysis, before centrifuging for 90 m at 28 °C and 246 000g. According to Stoke's law, these conditions were sufficient to sediment spherical particles ≥2 nm. After centrifugation, 1 ml was removed and acidified to 1% HCl for dissolved metal analysis. Samples for total metal analysis were microwave digested at 100 °C and 400 W for 35 m (20 m ramp, 10 m hold, 5 m cooldown) in 13% HCl/40% HNO₃ before diluting 15 times in double deionized water, yielding 1% HCl/3% HNO₃. Metal concentrations were determined using an Agilent 7500 series ICP-MS (Santa Clara, CA).

In experiment 1, exposure media were modeled using Geochem-EZ (Ref. 23) to determine the free ion activities of the metals of interest (Figs. S1 and S2). It was necessary to update the ligand database to include stability constants for Zn-MES (Ref. 24) and Ni-MES complexes.^24,25 The stability constant for the Ag-MES complex was approximated with the published stability constant for Ag-taurine complexes found using the International Union of Pure and Applied Chemistry Stability Constant Database.²⁶ All stability constants were input at infinite dilution. After concentrations of exposure media constituents were entered into Geochem-EZ, pH was fixed at 6, precipitates were allowed to form, and the ionic strength was calculated using an initial “guess” of 0.01 mol, as recommended by the model developers. In experiment 2, free ion concentrations and activities of Ag⁺ from AgNO₃ or the expected dissolved silver from AgENMs were modeled using GeoChem-Ez,²³ as in experiment 1.

There were statistically significant reductions in oxygen consumption per well in the 4, 6, 8, and 10 μM Ag⁺ exposures [Fig. 2(b)]. Normalizing these responses to viable cells revealed large overall increases in oxygen consumption per viable cell in response to Ag⁺ exposure [Fig. 3(a)]. A four-parameter nonlinear curve yielded an R² = 0.895, implying O₂ consumption and viability correlate in a nonlinear manner in response to silver ions. Only ∼30% of the cells were viable after exposure to 4 μM Ag⁺ [Fig. 2(a)] and there was nearly complete mortality after exposure to 8–10 μM Ag⁺ [Fig. 2(a)]. Cell viability was reduced following exposure to all tested Ag⁺ concentrations, although the response at 2 μM was not significant (p = 0.07) [Fig. 2(a)].

Exposure of Escherichia coli to Ag⁺ concentrations similar to those in this study (<10 μM) has been shown to stimulate respiration in the presence and absence of glucose.²⁷ The authors suggest this is due to uncoupling of the electron transport chain, which may at least partly explain the respiration responses to Ag⁺ we observed. Jin et al.²⁸ found the IC₅₀ for Ag⁺ and Ag nanoparticles to be highly dependent on the presence of specific ions for both B. subtilis and P. putida, with the Ag⁺ 24 h IC₅₀ for B. subtilis ranging from <1 to ∼18 μM, Suresh et al. found ∼73 μM Ag⁺ completely inhibited growth of B. subtilis (ATCC 9372) in Luria-Bertani (LB) medium,²⁹ while Kim et al. found ∼463 μM Ag⁺ was required to completely inhibit the growth of B. subtilis (KACC10111) in LB medium.³⁰ Our results show that relatively low concentrations of Ag⁺ are toxic to B. amyloliquefaciens in the minimal exposure medium used this study [Fig. 2(a)].

There was a decrease in O₂ consumption per well in response to all Zn²⁺ concentrations tested [Fig. 2(b)]. The O₂ consumption-viability plots showed increased O₂ consumption when percent viability was low and decreased O₂ consumption when viability was high [Fig. 3(b)]. A log-log regression was used to generate a nonlinear curve (R² of 0.838), supporting the notion that a nonlinear relationship exists between O₂ consumption and viability responses to Zn²⁺. A statistically significant decrease in viability was induced by 0.63 and 10–80 μM Zn²⁺, with viability gradually declining with increasing [Zn²⁺] [Fig. 2(a)]. We found only ∼15% of B. amyloliquefaciens cells exposed to 80 μM Zn²⁺ remained viable after exposure [Fig. 2(a)].

Previously, 1.8 mM Zn²⁺ was found to completely inhibit growth of B. subtilis WT, while 1 mM Zn²⁺ led to only a 20% reduction in optical density (600 nm).¹⁵ Additionally, Kim and An³¹ found colony forming units (CFU) counts of B. subtilis KACC10111 were significantly reduced (∼40% of the control) in response to 1 mM Zn²⁺. Van Nostrand et al.³² found a 64% reduction in growth of Burkholderia cepacia PR1₃₀₁ exposed to 3.8 mM Zn²⁺ at pH 6. It is also worth noting that a meta-analysis examining the toxicity of nanoparticles and associated ions to environmentally relevant test organisms estimated the average MIC of Zn²⁺ to be ∼459 μM.³³ Relative to these studies, we observed responses to low Zn²⁺ concentrations in B. amyloliquefaciens.

While a gradual decline in O₂ consumption per well was observed in response to 2.5–80 μM Ni²⁺ [Fig. 2(b)], the O₂ consumption-viability plots showed high variability in O₂ consumption in terms of viability [Fig. 3(c)]. In fact, a log –log regression showed a nonlinear correlation between O₂ consumption and viability, but an R² value of 0.301 indicates this is a relatively weak correlation. Statistically significant declines in cell viability were observed in response to 0.63 μM Ni²⁺ and for 10–80 μM Ni²⁺ [Fig. 2(a)].

Gaballa and Helmann¹⁵ reported 1.8 mM Ni²⁺ or Zn²⁺ were required to completely inhibit growth of B subtilis WT. We also found responses to Zn²⁺ and Ni²⁺ were similar in terms of O₂ consumption per well and viability in B. amyloliquefaciens (Fig. 2); however, the O₂ consumption-viability plots show a difference in the respiration response induced by these metals (Fig. 3). Modeling the exposure solutions showed, at equal total molar concentrations, the estimated concentration and activity of free Ni²⁺ were greater than that of Zn²⁺, due Zn-MES interactions (Fig. S1). The higher free ion activity of Ni²⁺ may be partially responsible for the difference we found in O₂ consumption-viability relationships compared with Zn²⁺; however, the relationship between O₂ consumption and viability was weakly correlated under Ni²⁺ exposure and not Zn²⁺, suggesting a different biological response to the metals.

When compared with the no-metal/antibiotic control, 2.6 μM streptomycin significantly decreased the number of viable cells (∼40%), while having no significant influence on O₂ consumed per well (Fig. 2). Expressing the respiration response relative to viable cells shows elevated O₂ consumption per cell of B. amyloliquefaciens exposed to streptomycin (Fig. 3). This response is of interest because accelerated respiration in response to sublethal antibiotic concentrations has recently been shown to potentiate bactericidal lethality of a range of antibiotics,⁹ select for antibiotic resistant bacteria, or increase mutagenesis.³⁴ 

We observed mortality of P. putida in the Oxoplate when OD_600 nm was 1.191 (before dilution in the Oxoplate); this was the greatest cell density used in the initial calibration curves, so this point was removed from the curve (Table S1). At cell densities similar to those used in the exposure conditions, P. putida exhibited no significant change in cell viability (Table S1). S. meliloti showed modest levels of growth during the O₂ consumption assay. We found strong correlations between starting cell density and time to OD_600 nm of 0.09 in viability assays, and between starting cell density and the area under the curve in O₂ consumption assays (Table S2). Hydrodynamic diameter and electrophoretic mobility of PVP-AgENMs and sAgENMs were stable in the exposure medium over the course of the study, and in the absence of bacteria (Table II). Dissolution of sAgENM was minimal (0.14% ± 0.02), while dissolution of PVP-AgENM was moderate (6.15% ± 0.23, error is one standard deviation).

In terms of viability, PVP-AgENMs were less toxic than ions on a molar basis across all organisms (Fig. 4). Others found ∼65 μM PVP-AgENM (10 nm) was required for 50% reduction in viability of S. meliloti 1021, whereas we observed an LC₅₀ for AgENM of 207 μM in S. meliloti 2011.³⁵ The large difference in primary particle size used between the studies could explain the differences in observed LC₅₀'s (10 nm vs ∼60 nm; smaller nanoparticles are generally more toxic); however, strain specificity in AgENM responses may occur even in the same bacterial species³⁶ and the exposure medium used was different which can influence the toxicity of ENMs.³⁷ 

The differences in CFUs/ml across test organisms (Table S3) does not completely explain the differences in LC₅₀'s observed. For instance, S. meliloti and P. putida were both exposed to metals at higher CFUs/mL than B. amyloliquefaciens, yet P. putida and B. amyloliquefaciens exhibited similar LC₅₀'s for AgNO₃ and PVP-AgENM (Table III). B. subtilis, which is closely related to B. amyloliquefaciens, was generally found in previous studies to be more susceptible than P. putida when exposed to uncapped AgENMs; however, viability responses depended on the presence of specific ions (i.e., Ca²⁺, Mg²⁺, Na⁺, K⁺, SO₄²⁻, Cl⁻, HCO₃⁻).²⁸ A study using a comparable exposure procedure showed ∼0.23 μM citrate capped-AgENM (10 nm) was sufficient for nearly complete mortality of B. subtilis KACC10111 after 1–2 h of exposure.³⁰ Another study showed ∼300 μM AgENM (32 nm) yielded a ∼50% reduction in light emission of P. putida BS566::luxCDABE after 1.5 h exposures.³⁸ 

B. amyloliquefaciens and P. putida both exhibited significant decreases in % viability starting at 1 μM AgNO₃ and 37 μM PVP-AgENM, with significant reductions continuing with increased concentrations (Fig. 4). Interestingly, both P. putida and B. amyloliquefaciens viability was reduced by ∼30% in response to ∼300 μM sAgENM exposure, while there was no significant effect on S. meliloti viability (Fig. 4). The observation that S. meliloti is generally more resistant to Ag⁺, PVP-AgENMs, and sAgENMs has environmental and agricultural implications. In a previous study, we found that amending soil with Ag, Ti, and Zn ENM-containing biosolids significantly inhibited nodulation of M. truncatula A17 by S. meliloti 2011 but did not reduce S. meliloti populations.¹⁷ Results from this study and others³⁹ suggest silver induced reduction in bacterial viability is likely not the mechanism of toxicity or nodule inhibition.

Coll et al.⁴⁰ reported predicted no effect concentrations of AgENMs in soil ranging from 40 μmol AgENM/kg to 116 μmol AgENM/kg. Others have found 45 μmol sAgENM/kg (supplied with biosolids amendment) significantly inhibited potential ammonium oxidation activity, with nearly 60% inhibition after 140 days.⁴¹ In the European Union, biosolids amended soils could have AgENM concentrations of approximately 9.3 μmol AgENM/kg by 2020.⁵ While it is difficult to directly compare results from liquid culture systems to findings and predictions in soils, we detected responses to AgENMs in liquid cultures that are within the same order of magnitude of predicted near-term environmental concentrations of AgENMs in biosolids amended soils (when comparing mol/kg to molarity).

Predicted free Ag⁺ activities were higher in PVP-AgENM treatments than AgNO₃ treatments, which induced similar reductions in viability (Fig. S3). Accordingly, abiotic dissolution of PVP-AgENM generally over-predicted observed reductions in viability, but correlated well with the observed mortality in P. putida (Fig. 4). Abiotic dissolution is only an estimate of the dissolved fraction present during exposure; however, the results suggest the <2 nm fraction likely explains most of the observed response to PVP-AgENM exposures, which agrees with a growing body of literature.^37,42

Abiotic dissolution estimates under-predicted sAgENM viability responses in B. amyloliquefaciens and P. putida, implying particle-specific toxicity for these bacteria (Fig. 4). Previously, Colman et al.⁷ observed shifts in microbial community structure, decreased microbial biomass, and increased N₂O production in long-term terrestrial mesocosms treated with AgENM containing biosolids slurries relative to nano-free slurry treatments. The ENM were sulfidized to some degree and the effects of AgENMs were similar or greater than those observed with AgNO₃ treatment. This finding, along with ours, shows sulfidation of AgENMs may result in nanospecific delivery of ions leading to prolonged toxicity, likely due to enhanced dissolution at the nano–bio interface. Dissolution at the nano–bio interface has been correlated with toxicity of many ENMs.³⁷ Additionally, sAgENMs were found to induce phytotoxic effects in the crop plants, Vigna unguiculata L. Walp. and Triticum aestivum, that were only observed after weeks of exposure,⁴³ and AgENM enriched biosolids were found to inhibit mycorrhizal-plant associations, including sAgENMs at environmentally relevant concentrations.⁶ 

Streptomycin reduced cell viability in S. meliloti and B. amyloliquefaciens by ∼40%–45%, while P. putida showed a 10% reduction in viability (Fig. 4). The similarities in viability reduction between S. meliloti and B. amyloliquefaciens along with the minimal reduction in viability of P. putida shows any observed differences in toxicity between the species are not likely purely driven by the differences in cell densities used in this study.

O₂ consumption/well of P. putida suggests no treatment effect [Fig. 5(c)]; however, O₂ consumption-viability plots show that oxygen consumption increased per viable cell under most treatment conditions [Fig. 6(c)]. While it was possible to fit four-parameter curves to B. amyloliquefaciens O₂ consumption/well measurements [Fig. 5(a)], meaningful four-parameter curve fits for the response in the other organisms were not possible (Fig. 5). The results suggest trends in O₂ consumption in response to AgNO₃ and PVP-AgENM were specific for the different PGPR tested; with the differences being pronounced when O₂ consumption was normalized to viable cell estimates (Fig. 6).

B. amyloliquefaciens generally showed similar increases in O₂ consumption in response to AgNO₃ and PVP-AgENM exposure, with the effect being exaggerated at <50% viability [Fig. 6(a)]. The trends in O₂ consumption-viability plots of B. amyloliquefaciens exposed to AgNO₃ were similar to those from experiment 1, highlighting the reproducibility of the methods across detection platforms. PVP-AgENM and ion exposure of S. meliloti also yielded increased O₂ consumption at viable cell densities <50%, but decreased O₂ consumption was observed when % viability was ∼50%–80% [Fig. 6(b)]. PVP-AgENM and ion exposure of P. putida yielded increased O₂ consumption at all viable cell densities <100% [Fig. 6(c)]. There was increased O₂ consumption in all PGPR exposed to sAgENM, although the responses were more variable in S. meliloti [Fig. 6(b)].

Streptomycin exposure increased O₂ consumption in all PGPR, though the effect was small in P. putida (Fig. 6). The stimulatory effects of streptomycin on bacterial respiration was observed long ago.⁴⁴ In our studies, it was a useful treatment for observing increased O₂ consumption, comparing responses across detection platforms, and examining species-specific responses to a medically relevant antibiotic (Fig. 6).

ENMs containing CeO₂, Fe₃O₄, and SnO₂ have been shown to increase respiration as assessed by the metabolic quotient (qCO₂ = basal CO₂ evolution rate/microbial biomass-C) in ENM treated soils in a temporal and spatial (soil horizon) dependent manner.¹² Others have reported decreased respiration in response to Li_xNi_yMn_zCo_1-y-zO₂ containing ENMs in Shewanella oneidensis MR-1; however, it is unclear if this is a species-specific response and if respiration responses were normalized for viability responses.⁴⁵ 

In our study, silver ions and PVP-AgENMs elicited similar respiration responses, which were unique in each of the studied PGPR species. This result, in agreement with others,³⁶ illustrates the importance of examining the ENM responses of specific bacterial species/strains under identical exposure conditions. Determining the exact biological mechanism of the observed respiration responses was beyond the scope of this study; however, our results are promising for researchers wishing to probe the mechanisms of metabolic responses of bacteria to ENMs, metal ions, and antibiotics.

In experiment 1, we developed a high-throughput method of examining O₂ consumption and viability responses in aerobic bacteria. To demonstrate why it is important to account for viable cell numbers in respiration-based toxicity assays, we performed a series of experiments using a variety of substrates pertinent to medical, food, and environmental sciences. The O₂ consumption-viability plots revealed metal specific O₂ consumption responses, with Ag⁺ and streptomycin inducing more pronounced increases in O₂ consumption compared with Zn²⁺ or Ni²⁺. This approach is similar to the metabolic quotient commonly used in ecological research,⁴⁶ but different from the bacterial growth efficiency metric previously devised to interpret growth responses in terms of growth and respiration.⁴⁷ 

In experiment 2, the toxicity of the metals was Ag⁺ > PVP-AgENMs > sAgENMs, from most to least toxic. S. meliloti had the highest LC₅₀ for Ag⁺ and PVP-AgENMs, while P. putida and B. amyloliquefaciens exhibited similar LC₅₀'s. Toxicity of PVP-AgENMs was likely driven by abiotically dissolved silver, while sAgENM toxicity is likely driven by interactions with the cell surface or extracellular milieu. The observed metal and antibiotic specific O₂ consumption responses in B. amyloliquefaciens were consistent across two different detection platforms. This finding displays the utility of normalizing respiration responses to viable cell estimates or biomass, which has also been observed by others.^11,12,48 The metal induced species specific respiration responses we observed show the significance of examining toxicity of ENMs and metals to isolated bacterial species/strains. Given the role of microbial respiration in antibiotic efficacy, the distinct % O₂ consumption-% viability patterns observed in the PGPR in response to Ag⁺ and PVP-AgENMs have implications with respect to agricultural application of biosolids and microbial amendments. ENM accumulation in biosolids amended soils is occurring exponentially; concurrently, there has been an increased utilization of microbial amendments for biocontrol of pathogens/pests and to enhance agricultural production. Thus, it is crucial to understand how the distinct respiration responses observed in this study translate to observed toxicity in more realistic scenarios where PGPR and plants are coexposed to AgENMs in test plots.

This research was funded by a grant from the U.S. Environmental Protection Agency's Science to Achieve Results program, the Transatlantic Initiative for Nanotechnology and the Environment (Grant No. RD834574). Any opinions, findings, conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the EPA. This work has not been subjected to EPA review and no official endorsement should be inferred. Assistance regarding experimental design and statistical analyses was received from Edward Roualdes and Catherine Starnes, both at the University of Kentucky.