Dynamic silver speciation as studied with fluorous-phase ion-selective electrodes: Effect of natural organic matter on the toxicity and speciation of silver

Abstract

The widespread application of silver in consumer products and the resulting contamination of natural environments with silver raise questions about the toxicity of Ag⁺ in the ecosystem. Natural organic matter, NOM, which is abundant in water supplies, soil, and sediments, can form stable complexes with Ag⁺, altering its bioavailability and toxicity. Herein, the extent and kinetics of Ag⁺ binding to NOM, matrix effects on Ag⁺ binding to NOM, and the effect of NOM on Ag⁺ toxicity to Shewanella oneidensis MR-1 (assessed by the BacLight viability assay) were quantitatively studied with fluorous-phase Ag⁺ ion-selective electrodes (ISEs). Our findings show fast kinetics of Ag⁺ and NOM binding, weak Ag⁺ binding for Suwannee River humic acid, fulvic acid, and aquatic NOM, and stronger Ag⁺ binding for Pony Lake fulvic acid and Pahokee Peat humic acid. We quantified the effects of matrix components and pH on Ag⁺ binding to NOM, showing that the extent of binding greatly depends on the environmental conditions. The effect of NOM on the toxicity of Ag⁺ does not correlate with the extent of Ag⁺ binding to NOM, and other forms of silver, such as Ag⁺ reduced by NOM, are critical for understanding the effect of NOM on Ag⁺ toxicity. This work also shows that fluorous-phase Ag⁺ ISEs are effective tools for studying Ag⁺ binding to NOM because they can be used in a time-resolved manner to monitor the activity of Ag⁺ in situ with high selectivity and without the need for extensive sample preparation.

Graphical abstract

1. Introduction

Silver has been estimated to be released into the environment at more than 2500 tons annually (Ratte, 1999). Since ionic silver, Ag⁺, is known to be highly toxic to bacteria, the sustainable use of silver-containing products, such as silver nanoparticles, requires a thorough understanding of the environmental toxicity of Ag⁺ ions. Silver toxicity cannot be correlated to just the total silver present. Instead, the individual silver species must be considered for a correct assessment of toxicity. One of the main mechanisms of silver speciation is Ag⁺ coordination with ligands that occur naturally in the environment. For example, Ag⁺ is known to form stable complexes with Lewis bases such as amines, halides, and thiolates. Thiosulfate, sulfide, and chloride binding to Ag⁺ have been shown to reduce the toxicity of Ag⁺ (Hogstrand et al., 1996; Janes and Playle, 1995; Leblanc et al., 1984; Rose-Janes and Playle, 2000). Consequently, the formation of silver complexes depends heavily on the environmental conditions (Ratte, 1999). For a meaningful assessment of Ag⁺ toxicity, the coordinating ligands present in any particular environment must be identified, and their effect on the Ag⁺ toxicity must be characterized.

One of the most common coordinating substances in natural soil and aquatic environments is natural organic matter (NOM; also commonly referred to as dissolved organic matter, DOM). There have been several reports of heavy metal ion binding to NOM, such as for Cu²⁺, Pb²⁺, Cd²⁺, and Zn²⁺ (Di Toro et al., 2001; Hayes, 1983; Hering and Morel, 1988;Ma et al., 1998; Sikora and Stevenson, 1988). This binding results in the formation of new chemical species with altered toxicity and transport properties, which affects the bioavailability of these metals (Di Toro et al., 2001; Glover and Wood, 2005; Pagenkopf, 1983). NOM is found in environmental systems, such as surface waters, ground waters, soils, and sediments, in concentrations ranging from 1 to more than 100 mg/L (Benoit et al., 2013; Bone et al., 2012; Fabrega et al., 2009; Gatselou et al., 2014; Unrine et al., 2012). NOM results from the decomposition of plant and animal residues and is inherently a mixture of compounds without a well-defined molecular structure (Maurer-Jones et al., 2013a). NOM consists largely of humic substances (humic acids, HA, and fulvic acids, FA) but also includes non-humic substances, such as fatty acids, sterols, natural sugars, amino acids, urea, and porphyrins (Boggs et al., 1985). Humic substances have molecular weights in the range from 300 to 300,000 and have a predominantly aromatic structure. Because they have many oxygen- and nitrogen-containing functional groups, such as carboxylic, phenolic, and amino groups, they exhibit an acidic and hydrophilic character and have high metal coordinating abilities (Gatselou et al., 2014; International Humic Substances Society, 2014). Due to their abundance in the hydrosphere, biosphere, and lithosphere, and their ability to form stable complexes with metals from both natural and anthropogenic sources, humic substances are commonly used as models for studying metal and NOM speciation (Cumberland and Lead, 2009; Di Toro et al., 2001; Fabrega et al., 2009; Maurer-Jones et al., 2013a). In this study, we utilized humic substances as models for studying the effect of NOM on Ag⁺ speciation and toxicity.

Analytical methods that have been used in NOM/Ag⁺ speciation studies have been based on ion exchange equilibrium (Chen et al., 2012, 2013), equilibrium dialysis (Angel et al., 2013; Boggs et al., 1985; Chen et al., 2012), atomic absorption and emission spectroscopy, mass spectrometry (Benoit et al., 2013; Hou et al., 2013; Unrine et al., 2012), ion-selective potentiometry (Benoit et al., 2013; Gunsolus et al., 2015; Maurer et al., 2012; Sikora and Stevenson, 1988), and the assessment of Ag⁺ and Ag⁺-NOM complex toxicity towards organisms (Chen et al., 2013; Fabrega et al., 2009). Several reports suggest that NOM samples from various sources decrease the toxicity of Ag⁺ to various organisms (Brauner and Wood, 2002; Gao et al., 2012; Glover et al., 2005a; Glover and Wood, 2005; Kimet al., 2013; Rose-Janes and Playle, 2000; VanGenderen et al., 2003; Wirth et al., 2012). This effect is usually attributed to Ag⁺ binding to NOM, which lowers the free Ag⁺ activity and, thereby, mitigates the ability of silver to act at the sites of toxic action in organisms (Glover and Wood, 2005; Janes and Playle, 1995). On the contrary, some NOM samples were reported to have no significant effect on the toxicity of Ag⁺ to multiple organisms, and in these cases Ag⁺ binding to NOM was concluded to be insignificant (Chen et al., 2013; Fabrega et al., 2009). Surprisingly, there has been no report to date that quantitatively investigates the correlation between the extent of Ag⁺ binding to NOM and Ag⁺ toxicity. Clearly, to investigate this correlation, it is advantageous to use techniques that directly probe Ag⁺ speciation without the added complexity introduced by the choice of organism, cell culture medium, and the type of toxicity assay as is necessary in an indirect toxicology assessment.

A challenge in direct Ag⁺ speciation studies is distinguishing different silver species, i.e., Ag NPs, free Ag⁺, and Ag⁺-NOM complexes. Except for ion-selective potentiometry, all the techniques mentioned above lack this ability. To account for this lack of selectivity, specific silver species are usually isolated by several sample preparation steps, e.g., by using molecular cut-off filters (Hou et al., 2013) or centrifugation (Hou et al., 2013). Unfortunately, this sample preparation can introduce further complexity and potential errors in measurements and the interpretation of results. Such complications include silver adsorption to sample containers and interference from positively charged complexes (in the case of the ion exchange equilibrium method). Moreover, these methods cannot be used for in situ or kinetic studies due to the long analysis time resulting from the need for sample preparation (e.g., analysis times are approximately 2 h for ion exchange equilibrium methods (Chen et al., 2012) and 3 days for equilibrium dialysis (Chen et al., 2012)). Even though binding of Ag⁺ to NOM has been studied for more than a decade, and several hypotheses about its kinetics have been proposed, the kinetics of this reaction have not been investigated directly, possibly due to the lack of appropriate methodology (Glover et al., 2005a; Ma et al., 1998; Maurer et al., 2012).

Potentiometry with ion-selective electrodes, ISEs, offers selective and sensitive in situ Ag⁺ detection, requires no substantial sample preparation, is non-destructive, has fast response times, detects only non-complexed ions, and can be used for speciation and kinetics studies (Buhlmann and Chen, 2012). There have been few literature precedents with use of commercially available solid-state ISEs to study Ag⁺ speciation (Benoit et al., 2013; Maurer-Jones et al., 2013b;Maurer et al., 2012; Peretyazhko et al., 2014; Sikora and Stevenson, 1988), possibly due to the common issue of solid state ISE biofouling in biological samples (biological molecules such as proteins adsorb strongly through sulfur groups to silver halide and sulfide electrodes) (Buhlmann et al., 1998; Chang et al., 1990; Kulpmann, 1989; Park et al., 1991). ISEs with polymeric sensing membranes suffer less from adsorption, but extraction of lipophilic biological interferents into their sensing membranes can still causing biofouling of these ISEs (Frost and Meyerhoff, 2002; Ward et al., 2003). In this work, we used ionophore-doped ISEs with fluorous sensing membranes that are less susceptible to biofouling effects. Fluorous phases prepared from perfluorocarbon derivatives have low polarity and polarizability, are both hydrophobic and lipophobic (i.e., alkanes are not miscible with perfluoroalkanes), limit extraction of lipophilic interferents into the sensing membrane, and thus are less susceptible to biofouling than other polymeric membrane ISEs (Boswell et al., 2005). Moreover, fluorous-phase Ag⁺ ISEs offer exceptional Ag⁺ selectivity due to the non-coordinating and poorly solvating properties of the fluorous phase. They also exhibit fast response times (less than 1 s), making them a unique tool for environmental Ag⁺ speciation studies (Boswell and Buhlmann, 2005; Lai et al., 2010; Maurer-Jones et al., 2013b). We used these sensors to study open questions regarding the interaction of Ag⁺ and NOM, specifically the kinetics of Ag⁺ and NOM binding and the correlation between the extent of Ag⁺ binding to NOM and the resulting Ag⁺ toxicity. While the current study focuses on Ag⁺ binding to NOM, the effect of NOM on the toxicity of silver nanoparticles is also crucial for a thorough risk assessment of silver-containing products and was addressed in parallel work (Gunsolus et al., 2015).

2. Experimental section

NOM samples: Suwannee River humic acid II, SRHA (Cat. No. 2S101H), Suwannee River fulvic acid II, SRFA (Cat. No. 2S101F), Pony Lake fulvic acid, PLFA (Cat. No. 1R109F), Pahokee Peat humic acid standard, PPHA (Cat. No. 1S103H), and Suwanee River Aquatic NOM, SRNOM (Cat. No. 2R101N) were purchased from International Humic Substances Society, IHSS (St. Paul, MN). The fabrication and calibration of fluorous-phase Ag⁺ ISEs was reported previously (Maurer-Jones et al., 2013b) and is discussed in the Supporting information along with a description of the data analysis methods. Buffer preparation and toxicity assessments are also explained in detail in the Supporting Information. All the solutions were prepared with deionized water (18 M Ω cm specific resistance, EMD Millipore, Burlington, MA). For preparation of the pH = 6.0 buffer, 0.100 M NaCH₃CO₂ and 0.006 M CH₃CO₂H were mixed at room temperature, followed by adjustment of the pH by addition of aliquots of NaOH or CH₃CO₂H. The components of the pH=7.5 buffer were 0.028M K₂HPO₄ and 0.015M KH₂PO₄. The pH of the solution was adjusted to 7.5 by addition of aliquots of KOH. The pH = 9.0 buffer contained 0.087 M NaHCO₃ and 0.044 M Na₂CO₃, and the pH was adjusted by addition of aliquots of NaOH. The HEPES buffer with pH = 7.5 was prepared by dissolving 0.20 M HEPES, 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid, in deionized water, and the pH of the solution was adjusted to 7.5 by addition of aliquots of KOH. The MOPS buffer with pH = 7.5 was prepared by dissolving 0.30 M MOPS, N-morpholino-3-propanesulfonic acid, in deionized water, and the pH of the solution was adjusted to 7.5 by addition of aliquots of KOH.

The components of the phosphate (pH=7.5), carbonate (pH=9.0), and acetate (pH = 6.0) pH buffers were chosen to interact minimally with Ag⁺ and thus minimize the interference with respect to Ag⁺ binding to NOM. The latter was assessed by measurements of the potentiometric response to Ag⁺ by fluorous-phase Ag⁺ ISEs in deionized water and in pH buffers. The response in deionized water and all the pH buffer solutions were very similar, which confirmed that Ag⁺ did not bind significantly to the pH buffer components (less than 15 mV and 2 mV change in the intercept and slope of the calibration equation, respectively). In the case of Ag⁺ binding to the pH buffer components, the calibration equation would have shifted to lower emf values because of the lowering of the concentration of free Ag⁺ ions as a result of complexation (Maurer-Jones et al., 2013b), which must be prevented to avoid errors and an unrealistic evaluation of Ag⁺ binding to NOM. Silver toxicity to the test organism, Shewanella oneidensis MR-1, was assessed by evaluating bacterial membrane integrity after exposure to Ag⁺ using the LIVE/DEAD BacLight Viability Kit (Product L-7012, Life Technologies).

3. Results and discussion

3.1. Ion-selective electrodes

The electrical potential of an ISE is measured with respect to a reference electrode and is referred to as emf (see Fig. 1). At a constant temperature, the emf increases linearly with the logarithm of the Ag⁺ activity. For example, at 20 °C, a 10-fold increase in the activity of Ag⁺ results in a 58.2 mV increase in the emf (Buhlmann and Chen, 2012; Lindner et al., 1981; Yajima et al., 1997). The fluorous-phase Ag⁺ ISEs were calibrated by addition of aliquots of concentrated AgCH₃COO (aq), followed by measurements of the emf. As predicted by theory, a linear relationship between the emf and Log c (Ag⁺)was observed for solutions with a fixed ionic strength, where activity coefficients are assumed to be constant (see Fig. 1). The experimentally obtained emf data can be easily converted to Ag⁺ concentrations using the calibration equations. The inherent response time of an ionophore-based ISE for the target ion is determined by ionic redistribution across the nanometer-sized charge separation layer at the interface of the sample and the ISE sensing membrane. In a typical experiment, the response time of the ISE measurement is, therefore, determined by how quickly an old sample can be replaced by a new one and not by a property of the electrode itself. In this work, all solutions were stirred, resulting in response times of less than one second (see Fig. 1). The detection limit of the fluorous-phase Ag⁺ ISEs used in this work was 0.05 μM. This is not an inherent limitation of these ISEs and with proper optimization, detection limits as low as 4.0 × 10⁻¹¹ M have with been achieved with fluorous sensing membranes (Lai et al., 2010). It should be noted that ISEs selectively detect un-complexed (“free”) Ag⁺. This allowed us, in previous work, to utilize fluorousphase Ag⁺ ISEs to quantify the Ag⁺ speciation in bacterial growth media and show that, in cell culture media that are rich in coordinating ligands, less than 5% of the silver is in the free Ag⁺ form (Maurer-Jones et al., 2013b). We also showed that these sensors can be used for dynamic monitoring of Ag⁺ release from silver nanoparticles in the presence of interfering capping agents such as trisodium citrate (Gunsolus et al., 2015; Maurer-Jones et al., 2013b). That work suggested that these sensors would very likely also be useful analytical tools for probing Ag⁺ binding to NOM.

Fig. 1.

Representative calibration curve of a fluorous-phase Ag⁺ ISE. (A) Experimental Setup. (B) Red arrows indicate additions of AgCH₃COO aliquots to the measuring solution. The emf of the fluorous-phase Ag⁺ ISE increases after each rise in Ag⁺ concentration. For better visualization, only a snapshot of the addition experiment is shown. (C) The linear relationship between the emf and Log c_Ag+ can be used as the calibration equation for converting emf values to Ag⁺ concentrations. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

3.2. Interference of the sample matrix on Ag⁺ binding to NOM

NOM has both acidic and basic functional groups and, upon introduction into a solution, can affect the pH, which will influence the strength of Ag⁺ binding to NOM. For a meaningful evaluation of the extent of Ag⁺ binding to NOM, it is, therefore, important to choose pH-buffered test solutions that are representative of environmental samples. There have been several reports of silver speciation in silver nanoparticle solutions as well as of Ag⁺ binding to NOM that described the use of pH buffer components such as N-morpholino-3-propanesulfonic acid (MOPS) or 4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid (HEPES) (Angel et al., 2013; Chen et al., 2012, 2013; Choi and Cui, 2012; Mumper et al., 2013; Piccapietra et al., 2012; Sun et al., 2005; Zhan et al., 2007). HEPES and MOPS were recommended because they were reported not to bind several heavy metal ions such as copper, cadmium, and zinc (Chen et al., 2012; Mash et al., 2003; Poulson and Drever, 1996; Soares et al., 1999). Because HEPES and MOPS contain amino groups (see structural formulas in Scheme S1 in the SI), and because Ag⁺ is well known to form stable complexes with amines, we suspected that HEPES and MOPS form complexes with Ag⁺, considerably complicating any speciation studies. The complexation of HEPES and MOPS with Ag⁺ was confirmed by monitoring the emf of fluorous-phase Ag⁺ ISE. Immediately after addition of HEPES and MOPS to a 5.0 μM Ag⁺ solution, the emf decreased, indicating a decrease in the concentration of free Ag⁺ as a result of Ag⁺ binding to the buffer species (see Fig. 2A). To quantify the extent of HEPES and MOPS complexation with Ag⁺, HEPES, MOPS, and phosphate buffers (the latter has a low tendency to coordinate with Ag⁺) with pH=7.5 and an ionic strength of 0.1Mwere prepared, and potentiometric responses to Ag⁺ in these pH buffers as well as in deionized water were measured (see Fig. S1). Because of Ag⁺ binding to the pH buffer components, the calibration curves shifted to lower emf values when the emf was plotted versus the total silver concentration in the calibration solutions (see Fig. 2B) (Maurer-Jones et al., 2013b). Use of the calibration curve obtained in deionized water for comparison (Fig. S1) shows that only 10% of the Ag⁺ in the MOPS buffer and less than 1% of the Ag⁺ in the HEPES buffer is in its free (uncomplexed) form. In contrast, almost 99% of the Ag⁺ in the phosphate buffer is in the free and non-complexed form(see Fig. 2C). This is of great importance because HEPES and MOPS have been used as pH buffer components in a number of Ag⁺ speciation and mechanistic studies without consideration of their high tendency for coordination with Ag⁺ (Chen et al., 2012, 2013; Mumper et al., 2013; Zhan et al., 2007).

Fig. 2.

Interference of buffer components on Ag⁺ binding to NOM at pH 7.5, as studied by fluorous-phase Ag⁺ ISEs. (A) Concentrated HEPES (shown in red) and MOPS (shown in black) were added to 5.0 μM Ag⁺ solutions to reach a final concentration of the buffer of 0.05 M. (B and C) Aliquots of AgNO₃ were added (shown by asterisks) to deionized water (black), phosphate buffer (pH=7.5, red), MOPS buffer (pH=7.5, green), and HEPES buffer (pH=7.5, blue), while monitoring the emf. The quickness of the emf response is illustrated by the emf time trace shown in Panel A. Panel C shows the concentration of free Ag⁺ in the buffer solutions as a function of the total silver concentration in solution, as determined from the calibration curve in deionized water. Panel D illustrates the interference of HEPES when studying Ag⁺ binding to NOM. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).

Fig. 2D illustrates the extent of HEPES interference in Ag⁺ binding to NOM, where Pony Lake fulvic acid (PLFA) was added to 5.0 μM Ag⁺ in either HEPES buffer (which competes with NOM in binding to Ag⁺) or phosphate buffer (which does not interfere with binding). Both buffers had the same pH (7.5). Upon addition of PLFA to 5.0 μM Ag⁺ in the phosphate buffer, the emf is significantly decreased by more than 20 mV (which corresponds to a 50% decrease in the free Ag⁺ concentration due to Ag⁺ binding to NOM), but a similar addition of PLFA to 5.0 μM Ag⁺ dissolved in HEPES buffer resulted in no detectable emf change (two-tailed t test, p > 0.05). This can be explained by considering that 99% of the silver is bound to the HEPES buffer, making it impossible for the PLFA to compete with HEPES to form Ag⁺-NOM complexes in a significant amount.

Coordinating ligands such as HEPES and MOPS are not present in authentic environmental samples. To prevent errors in the evaluation of Ag⁺ binding to NOM, such buffers should be avoided. Herein, we will utilize a potassium phosphate buffer with an ionic strength of 0.1 M to minimize buffer artifacts. Use of a fixed ionic strength ensures that the activity coefficients are approximately constant and that the interaction of Ag⁺ and the NOM is not simply electrostatically driven but is the result of specific metal ligation to functional groups of the NOM (Sikora and Stevenson, 1988). Note that the phosphate buffer is representative of real-life conditions since most high concentration components of real-life samples (i.e., specifically, Na⁺, K⁺, ${\text{HCO}}_3^{-},{\text{SO}}_4^{2-},{\text{NO}}_3^{-}$, F⁻, Mg²⁺, and Ca²⁺) will not directly interfere with Ag⁺ speciation (Hem and Geological Survey (U.S.), 1985; Maurer-Jones et al., 2013b). However, chloride, which occurs in natural water supplies in high concentrations, will compete with NOM to bind to Ag⁺ and will affect the extent of Ag⁺ binding to NOM binding. Therefore, we excluded Cl⁻ from our test matrix, facilitating the investigation of the extent of Ag⁺ binding to NOM and its correlation to protective effects of NOM against Ag⁺ toxicity.

3.3. Binding of Ag⁺ to NOM

The concentration of NOM in natural environments ranges from 0.1 to 100 mg/L, with a mean value of 45 mg/L (Boggs et al., 1985; Gatselou et al., 2014; McDonald et al., 2004; Nagao et al., 2003; Robards et al., 1994). Thus, to cover the most environmentally relevant NOM concentration range, we looked at 50 mg/L NOM in this study. We considered five different NOM isolates: Suwannee River humic acid II (SRHA), Suwannee River fulvic acid II (SRFA), Pony Lake fulvic acid (PLFA), Pahokee Peat humic acid standard, PPHA, and Suwanee River Aquatic NOM (SRNOM). To quantify Ag⁺ binding to NOM, fluorous-phase Ag⁺ ISEs were placed in a solution of 5.0 μM AgCH₃COO with a pH buffered to 7.5 and a fixed ionic strength. Aliquots of concentrated solutions of NOM were added to this solution, and the emf of the ISE was monitored continuously. The Ag⁺ concentration of 5 μM was chosen to mimic the concentration of Ag⁺ released from 1–5mg/L AgNPs (12-nm-diameter) (Maurer-Jones et al., 2013b). Results are depicted in Fig. 3A. When Ag⁺ binds to NOM, the concentration of free Ag⁺ is reduced. Consequently, a decrease in the emf is observed. Stronger binding of Ag⁺ to NOM results in a larger decrease in the concentration of free Ag⁺ ions and, therefore, a greater decrease in the emf values. Addition of SRFA, SRNOM, SRHA, PPHA, and PLFA resulted in 8.5 ± 0.2%, 11.1 ± 2.1%, 15.4 ± 2.0%, 40.4 ± 1.8%, and 57.7 ± 0.2% decreases in the free Ag⁺ concentration, respectively (all changes were statistically significant, as assessed by the two-tailed t test, p < 0.05). It follows that the ability of different NOM types to bind Ag⁺ falls in the sequence:

$$\text{PLFA}>\text{PPHA}>\text{SRHA}>\text{SRNOM}>\text{SRFA}.$$

Fig. 3.

Ag⁺ and NOM binding and its effect on Ag⁺ toxicity: (A) Shown are the emf values measured with fluorous-phase Ag⁺ ISEs in 5.0 μM AgNO₃ followed by 51.0 mg/L addition of (indicated by asterisks) of SRFA (black), SRHA (green), SRNOM (purple), PPHA (blue), or PLFA (red). For better visualization, the emf traces are shifted vertically relative to one another. (B) Aliquots of AgNO₃were added (indicated by asterisks) to the Ag⁺ and NOM solution to readjust [Ag⁺] to5.0μM. (C) Viability of S. oneidensis MR-1 after 30minute exposure to Ag⁺ and NOM solutions with specified identities and concentrations. Error bars represent the standard deviation of three biological replicates. Statistical significance was determined using the paired t-test. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).

Based on our findings, we conclude that fulvic acid does not bind Ag⁺ more or less strongly than humic acid, but it is noticeable that NOM types that are rich in nitrogen show higher binding of Ag⁺. Specifically, the nitrogen contents of PLFA and PPHA of 6.5% and 3.7%, respectively, are much higher than for SRHA, SRFA, and SRNOM, which contain 0.7%, 1.2%, and 1.1% nitrogen, respectively (International Humic Substances Society, 2014). Most of the nitrogen in these NOM samples was shown to be in the form of amides, aminoquinones, amino sugars, and heterocyclic nitrogen structures, which can all bind Ag⁺(Thorn and Cox, 2009). Moreover, NOM also contains sulfur-containing functionalities in the form of exocyclic and heterocyclic sulfur and sulfoxide, sulfone, sulfonate, and sulfate groups (Manceau and Nagy, 2012). It has been shown that the majority of sulfur in SRHA, SRFA, and PLFA is in low oxidation states as exocyclic and heterocyclic sulfur (Manceau and Nagy, 2012), which binds strongly with Ag⁺. Therefore, the high Ag⁺ binding ability of PLFA is also the result of its sulfur content (3.0%), which is much higher than in the case of SRFA and SRHA (0.5%).

Our observations are consistent with findings of Sikora and Stevenson, who concluded that amine and thiol functional groups are mainly responsible for Ag⁺ complexation while oxygen-containing functional groups have only minor effects on Ag⁺ binding to NOM (Boggs et al., 1985; Glover et al., 2005b; Sikora and Stevenson, 1988). This conclusion is also consistent with representative stability constants for oxygen, nitrogen, and sulfur containing functional groups. For example, the stability constants K₁₁, for the 1:1 Ag⁺ complexes (Sikora and Stevenson, 1988) are 4.4 with the oxygen ligand acetate, 2.1 × 10³ for ammonia, 2.5 × 10² for NEt₃, 1.2 × 10² for pyridine, and 6.6 × 10⁸ for the sulfur ligand thiosulfate (Harris, 2003; Skoog et al., 1997). Further research with techniques such as X-ray photoelectron spectroscopy (XPS) or X-ray absorption near edge structure (XANES) spectroscopy could be used to study the identity of the functional groups involved in the binding process (Boudou et al., 2008; Manceau and Nagy, 2012).

Observing a relatively low Ag⁺ binding extent for Suwannee River humic and fulvic acids can explain the discrepancies among published reports on the extent of Ag⁺ binding to Suwannee River NOM isolates (Chen et al., 2012, 2013; Gao et al., 2012; Glover and Wood, 2005; Hou et al., 2013; VanGenderen et al., 2003). Differences in the experimental conditions used in those studies, such as different pH values, affect the extent of binding and cause contradicting conclusions. To understand the extent of the pH effect, we quantified Ag⁺ binding to SRFA and SRHA at pH = 6.0, 7.5, and 9.0 (see Fig. S5). Interestingly, at pH = 6.0, no significant change in the free [Ag⁺] can be detected upon SRFA or SRHA addition to Ag⁺ solutions (two-tailed t test, p > 0.05), which confirms that Ag⁺ does not bind to SRFA and SRHA at acidic pH. Increasing the pH to 7.5 and 9.0 results in 9 ± 2% and 40 ± 1% Ag⁺ binding to SRFA, and 16 ± 1% and 55 ± 1% binding to SRHA, respectively (Fig. S5). (Increases in extent of Ag⁺-NOM binding are significant as assessed by the two-tailed t test, p < 0.05.) Clearly, the pH at which binding occurs plays a critical role in the extent of Ag⁺ and NOM binding when the NOM binds Ag⁺ weakly.

Fig. 3 also illustrates the kinetics of Ag⁺ binding to NOM. Even though binding of Ag⁺ to NOM has been studied for more than a decade, and several hypotheses about its kinetics have been proposed, the kinetics of this reaction has not been investigated directly, possibly due to the lack of proper methodology (Glover et al., 2005a; Ma et al., 1998; Maurer et al., 2012). The fast response time of fluorous-phase Ag⁺ ISEs (<1.0s) allows real-time detection of Ag⁺ and makes it possible to directly observe the kinetics of Ag⁺ binding to NOM. On one hand, after additions of SRHA, SRFA, SRNOM, or PPHA to a Ag⁺ solution, the emf decreased in less than 1 s, indicating a fast decrease in the free Ag⁺ concentration as the result of fast Ag⁺ binding to NOM. No further changes were observed over the following 24 h, showing that equilibrium was reached very quickly.

On the other hand, after addition of PLFA to 5.0 μM AgCH₃COO, the emf did not stabilize after the initial very quick decrease, but continued to drift even after 24 h, albeit at a decreasing rate (see Fig. S2). Stepwise addition of PLFA and alteration of the pH of the solution did not eliminate this drift (see Fig. S3). To confirm that this behavior was indeed caused by Ag⁺ binding to PLFA and not by an artifact of the ISE measurement, the fluorous-phase Ag⁺ ISEs were recalibrated after exposure to PLFA, confirming that the fluorous-phase Ag⁺ ISEs were still fully functional and that the calibration curve was valid throughout the experiment. Moreover, when fluorous-phase Ag⁺ ISEs were inserted into a solution of Ag⁺ and PLFA that had been preequilibrated for 24 h, the measured emf did not show any drift, confirming that the observed emf drifts after PLFA addition to the Ag⁺ solution were indeed caused by a chemical transformation in the solution (see Fig. S4) and not by some unexplained effect of PFLA on the response of the ISE.

An explanation for the slow decrease in the free silver ion activity after the addition of PLFA to 5.0 μM AgCH₃COO is given by the reduction of Ag⁺ to Ag by PLFA as the reducing agent. Indeed, the formation of Ag NPs as the result of Ag⁺ reduction by NOM in environmentally relevant conditions at ambient laboratory temperature and lighting was shown by Akaighe and co-workers (Akaighe et al., 2011). Under their experimental conditions, Ag⁺ reduction by NOM took, depending on the NOM source and concentration, up to several days before the formation of Ag NPs was visually noticed (Akaighe et al., 2011). Slow Ag⁺ reduction by NOM under environmentally relevant conditions was also reported by F. Maurer and co-workers (Maurer et al., 2012). In our study, formation of Ag NPs through reduction of Ag⁺ by PLFA was also confirmed using transmission electron microscopy and dark-field microscopy with hyperspectral imaging (see Figs. S7 and S8).

Reduction of Ag⁺ by NOM also appears to be consistent with findings by Glover et al., who observed that longer incubation of Ag⁺ and NOM before addition to Daphnia magna (a freshwater flea) resulted in lower Ag⁺ toxicity (Glover et al., 2005a). Glover et al. explained the time-dependent toxicity of Ag⁺ in the presence of NOM by the hypothesis of slow kinetics of Ag⁺ binding to NOM. However, these authors provided no evidence to exclude alternative explanations, such as the reduction of Ag⁺ by NOM. Based on the findings from our study and evidence from other studies employing a variety of experimental techniques, we believe that binding of Ag⁺ and NOM is fast and that the reduction of Ag⁺ by certain types of NOM in environmentally relevant conditions has slow kinetics and contributes to gradual changes in [Ag⁺] in NOM-containing media.

An alternative explanation for the slow changes in the free Ag⁺ activity could be gradual changes in the macromolecular structure of NOM and, concomitantly, changes in the NOM’s ability to bind Ag⁺. The conformation of humic substances depends on physicochemical parameters such as pH, ionic strength, and the composition of the samples (Baalousha et al., 2006), which were all kept constant throughout our experiments with a phosphate pH buffer. Whereas conformational changes of NOM induced by iron complexation have been reported recently (Baalousha et al., 2006; Nuzzo et al., 2013), evidence that those conformational changes have a major effect on the activity of free iron does not exist. The slow conformational changes induced by the presence of iron may be preceded by comparatively fast binding of iron to NOM. In the case of Ag⁺, no reports of such conformational changes of NOM have been made to date.

3.4. Effect of NOM on Ag⁺ toxicity

NOM has been reported to reduce Ag⁺ toxicity to large and microorganisms by binding to Ag⁺ and reducing the concentration of free Ag⁺ (Brauner and Wood, 2002; Kim et al., 2013; Rose-Janes and Playle, 2000; VanGenderen et al., 2003; Zhang et al., 2012). After quantifying the extent of Ag⁺ binding to different NOM isolates, we were able to assess the validity of the commonly held belief that NOM reduces Ag⁺ toxicity to organisms by binding to Ag⁺. We employed a bacterial model (S. oneidensis MR-1) to study the effect of NOM on silver toxicity. S. oneidensis MR-1 is a facultative anaerobe and also a metal reducing bacterium. This respiratory diversity allows it to survive in a variety of locations, e.g., in freshwater, saltwater, and sediments, making it a relevant model to assess the environmental impact of silver-containing products that may leach silver into a variety of natural environments (Hau and Gralnick, 2007). We assessed changes in membrane integrity of S. oneidensis MR-1 (using the LIVE/DEAD BacLight Viability Kit, see Supporting Information) after exposure to Ag⁺ with and without NOM, using experimental conditions similar to those used for studying the extent of Ag⁺ binding to NOM (pH 7.5, potassium phosphate buffer, 5.0 μM Ag⁺, and 51 mg/L NOM). It should be noted that, under nutrient-poor conditions like those employed in these experiments, NOM can serve as a source of nutrition for bacteria and thereby increase their viability. We observed significantly higher cell viability (two-tailed t test, p < 0.05), presented as the ratio of cells identified to be live vs. dead, in the presence of 51 mg/L PLFA, SRFA, SRHA, PPHA, and SRNOM(with no silver present) than in the absence of NOM (see Fig. 3C). Exposing the bacteria to 5.0 μM Ag⁺ reduced the live to dead cell ratio from 3.2 (0.0 μM Ag⁺ control) to less than 1.0 (see Fig. 3C), confirming that Ag⁺ is toxic to the bacteria at this concentration. Addition of 51 mg/L NOM induced no significant changes in the toxicity of 5.0 μM Ag⁺ to the bacteria in the case of SRFA, SRHA, and SRNOM (shown in purple in Fig. 3C). Our finding is consistent with precedent studies that observed no significant effect of SRHA on the toxicity of Ag⁺ to Pseudomonas fluorescens (Fabrega et al., 2009), Chlamydomonas reinhardtii (Chen et al., 2013), and Pseudokirchneriella subcapitata (Chen et al., 2013). As discussed in the previous section, SRNOM, SRFA, and SRHA have a low affinity for Ag⁺, and less than 20% of the Ag⁺ is bound to NOM present in a concentration of 51 mg/L. Given that these NOM types only decrease the free [Ag⁺] slightly (see Fig. 3A), we attribute their insignificant effects on the Ag⁺ toxicity to S. oneidensis MR-1 to their low affinity for Ag⁺. Even PPHA with its slightly higher Ag⁺ binding affinity (upon binding to Ag⁺ reduces its concentration by 50%) did not improve the viability of S. oneidensis MR-1. This confirms that due to overall weak Ag⁺- NOM binding, the NOM-induced changes in Ag⁺ concentration do not necessarily result in significant effects on the Ag⁺ toxicity to the S. oneidensis MR-1.

Even though PPHA and PLFA showed similar affinities for Ag⁺, i.e., 50% binding for PPHA and 60% binding for PLFA, only the addition of PLFA improved the viability of cells exposed to 5.0 μM Ag⁺ (the live to dead ratio increased from less than 1.0 to 2.0, two-tailed t test, p < 0.05), while PPHA had no significant protective effect against Ag⁺ toxicity. The latter can be explained based on the gradual decrease in [Ag⁺] in the presence of PLFA. Immediately following NOM addition to Ag⁺, the [Ag⁺] was observed to be similar for both PLFA and PPHA (2–3 μM Ag⁺). However, during the time that the bacteria were incubated with the Ag⁺- and NOM-containing solution, [Ag⁺] gradually decreased to less than 1.0 μM, likely due to Ag⁺ reduction by PLFA, as explained in the previous section, whereas [Ag⁺] remained approximately constant in PPHA-containing solutions. Even though PLFA and PPHA showed similar initial extents of Ag⁺ binding, PLFA reduced the total effective exposure of the cells to Ag⁺, resulting in higher cell viability. (Note that 1.0 μM Ag⁺ cannot be confirmed to be toxic to the bacteria, while 2.5 μM Ag⁺ reduces the viability of S. oneidensisMR-1, as shown in Fig. S8). Given that PPHA, SRFA, SRHA, and SRNOM all have different binding abilities but exhibit similar effects on Ag⁺ toxicity to S. oneidensis MR-1, our results demonstrate that the effect of NOM on bacterial cell viability does not necessarily correlate with the extent of Ag⁺ binding to NOM, but strongly depends on free Ag⁺ concentration. To confirm this, after addition of PLFA or PPHA to 5.0 μM Ag⁺, which significantly reduced the free Ag⁺ concentration due to binding to NOM, we adjusted the free Ag⁺ concentration to 5.0 μM by adding aliquots of AgNO₃ to the solution while monitoring the emf using fluorous-phase ISEs (see Fig. 3B). AgNO₃ was added until the ISE indicated that the concentration of free Ag⁺ had again reached 5.0 μM, as was the case before NOM addition. As shown in Fig. 3C, a similar toxicity was found for solutions that contained 5.0 μM free Ag⁺ and no added NOM as for solutions that contained 5.0 μM free Ag⁺ and 51.0mg/L PPHA or PLFA, despite the fact that the latter solutions contained 3.5 and 5.3 μM complexed silver, respectively.

These findings are important since the protective ability of NOM against Ag⁺ toxicity is usually attributed to direct binding of Ag⁺ to NOM whereas other forms of speciation, such as NOM-induced Ag⁺ reduction and nanoparticle formation, are often ignored. This study provides evidence that the protective ability of NOM against Ag⁺ toxicity results both from Ag⁺ binding to NOM and NOM-induced Ag⁺ reduction, and also shows that the lack of NOM effect on Ag⁺ toxicity does not exclude the possibility of Ag⁺ binding to NOM.

4. Conclusions

This work has demonstrated that fluorous-phase Ag⁺ ISEs are effective tools for the dynamic investigation of Ag⁺ binding to NOM as they can be used to monitor the in situ activity of Ag⁺ in a time-resolved manner with high selectivity and without the need for substantial sample preparation. The extent of Ag⁺ binding to NOM was quantified using these sensors, showing the following trend for Ag⁺ binding capacities of different NOM types:

$$\begin{array}{l}\text{Pony}\text{Lake}\text{fulvic}\text{acid}>\text{Pahokee}\text{Peat}\text{humic}\text{acid}\text{standard}>\text{Sumannee}\text{River}\text{humic}\text{acid}\text{II}\\ >\text{Suwanee}\text{River}\text{Aquatic}\text{NOM},\text{SRNOM}>\text{Suwannee}\text{River}\text{fulvic}\text{acid}\text{II},\text{SRFA}.\end{array}$$

We showed fast kinetics for Ag⁺ binding to NOM and slow kinetics for the reduction of Ag⁺ by certain types of NOM in environmentally relevant conditions. This is the first report on the kinetics of Ag⁺ binding to NOM with time resolution of less than a second. Moreover, we showed that pH affects the extent of Ag⁺ binding to NOM, where higher pH results in stronger binding. Studies of Ag⁺ speciation should ensure that buffer components are selected to avoid unwanted complexation with Ag⁺. We showed that buffer compounds such as HEPES and MOPS should be excluded from Ag⁺ speciation studies since they form stable complexes with Ag⁺ and interfere with NOM binding to Ag⁺. Lastly, we showed that the ability of NOM to protect against Ag⁺ toxicity does not directly correlate with the extent of Ag⁺ binding to NOM. Other aspects of silver speciation, such as NOM-induced Ag⁺ reduction and nanoparticle formation, also affect the observed toxicity.

HIGHLIGHTS.

• Ag⁺ binding to natural organic matter (NOM) studied time-resolved and in situ

• Quantification of Ag⁺ binding to NOM and impact of NOM on Ag⁺ toxicity

• Fast kinetics of Ag⁺ binding to NOM and strengths of binding to different NOMs

• Silver speciation affects effect of NOM on Ag⁺ toxicity

Appendix A. Supplementary data

Supporting Information. Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Supplementary data associated with this article can be found, in the online version, at doi: http://dx.doi.org/10.1016/j.scitotenv.2015.07.151.