Montmorillonite clay-based sorbents decrease the bioavailability of per- and polyfluoroalkyl substances (PFAS) from soil and their translocation to plants

Abstract

Consumption of food and water contaminated with per- and polyfluoroalkyl substances (PFAS) presents a significant risk for human exposure. There is limited data on high affinity sorbents that can be used to reduce the bioavailability of PFAS from soil and translocation to plants and garden produce. To address this need, montmorillonite clay was amended with the nutrients carnitine and choline to increase the hydrophobicity of the sorbent and the interlayer spacing. In this study, the binding of PFOA (perfluorooctanoic acid) and PFOS (perfluorooctanesulfonic acid) to parent and amended clays was characterized. Isothermal analyses were conducted at pH 7 and ambient temperature (to simulate environmentally-relevant conditions). The data for all tested sorbents fit the Langmuir model indicating saturable binding sites with high capacities and affinities under neutral conditions. Amended montmorillonite clays had increased capacities for PFOA and PFOS (0.51 – 0.71 mol kg⁻¹) compared to the parent clay (0.37 – 0.49 mol kg⁻¹). Molecular dynamics (MD) simulations suggested that hydrophobic and electrostatic interactions at the terminal fluorinated carbon chains of PFAS compounds were major modes of surface interaction. The safety and efficacy of the clays were confirmed in a living organism (Lemna minor), where clays (at 0.1% inclusion) allowed for increased growth compared to PFOA and PFOS controls (p ≤ 0.01). Importantly, soil studies showed that 2% sorbent inclusion could significantly reduce PFAS bioavailability from soil (up to 74%). Studies in plants demonstrated that inclusion of 2% sorbent significantly reduced PFAS residues in cucumber plants (p ≤ 0.05). These results suggest that nutrient-amended clays could be included in soil to decrease PFAS bioavailability and translocation of PFAS to plants.

1. INTRODUCTION

Per- and polyfluoroalkyl substances (PFAS) are of growing concern due to the adverse effects they pose on the environment and human health. PFAS have been used in food packaging and household products like stain repellants, nonstick cookware, lubricants, and textile treatments (Trudel et al., 2008). They have also been released to the environment during fluoropolymer manufacturing or processing disposal, use of aqueous firefighting foam (AFFF), biosolid application, and landfill leachate (Teaf et al., 2019). PFAS are often referred to as “forever chemicals” as they do not easily break down in the environment. According to the National Health and Nutrition Examination Survey (NHANES), PFAS were detected in nearly all serum samples from the general population, indicating widespread exposure in the US (CDC, 2009). Adverse health effects associated with PFAS exposure include immune dysfunction, reproductive issues, liver toxicity, and various cancers (DeWitt et al., 2019; Saikat et al., 2013; Shearer et al. 2020).

Perfluorooctanoic acid (PFOA) and perfluorooctanesulfonic acid (PFOS) are two of the most commonly detected PFAS in the environment and the representative compounds used in this study. Although PFAS such as PFOA and PFOS have been voluntary phased out by most industries, they are persistent and still widely distributed in the environment. PFAS detection in water and soil has been well-documented in the literature (Eschauzier et al., 2013; Strynar et al., 2012). PFAS and PFAS precursors have been detected in levels typically < 100 ng L⁻¹ in groundwater, surface water, and drinking water and > 1000 ng L⁻¹ in areas directly affected by PFAS sources (Vo et al., 2020). PFAS were detected in 100% of soil samples from 62 locations on 7 continents, with PFOA and PFOS being the most commonly detected (up to 2.7 and 3.1 μg kg⁻¹, respectively) (Rankin et al., 2016). PFOA and PFOS have been measured at levels up to 50,000 and 373,000 μg kg⁻¹ in surface soil from U.S. Air Force AFFF-contaminated sites (Brusseau et al., 2020). Importantly, studies have shown that PFAS tend to accumulate in soil at much higher rates than in water sources. Zareitalabad et al., (2013) concluded that field-based dissociation constants suggest longer residence times of PFAS in soil than in water, which can result in increased transfer of PFAS from soil to produce. This presents an exposure risk to humans who consume produce contaminated with PFAS (Brown et al., 2020; Domingo et al., 2012; Vestergren et al., 2012). PFAS such as PFOA and PFOS have unique amphiphilic properties (hydrophobic fluoroalkyl chain and hydrophilic ionizable functional group), resulting in complex environmental behavior and difficulties in PFAS remediation of soil (Sorengard et al., 2019; Tang et al., 2010; Zhao et al., 2013). Bioavailable fractions of PFOA and PFOS in water and soil can translocate to plants such as vegetables and produce, resulting in unintended exposures to humans and animals. Lal et al. (2020) demonstrated PFOS uptake by lettuce, carrots, and tomatoes and concluded that this translocation presents a risk of PFOS exposure to humans and animals via the food chain. PFOA and PFOS have also been shown to translocate to wheat, oat, maize, potato, and ryegrass plant compartments (Stahl, 2009).

Practical strategies are needed to reduce PFAS contamination, and as most biological and chemical destruction techniques have limited efficacies (Vu & Wu, 2020), immobilization with carbon- and clay-based sorbents is a promising remediation strategy. Activated carbon (powdered and granular) is popular due to its low cost and wide applicability, and has been shown to be effective in removing PFOA and PFOS from aqueous environments (Yu et al., 2009; Carter & Farrell, 2010; Rattanaoudom et al., 2012). Activated carbon (AC) and “biochars” have also been investigated for PFAS remediation of soil (Askeland et al., 2020; Sørmo et al., 2021; Silvani et al., 2019). However, the use of clay-based sorbents to reduce PFAS bioavailability from soil and translocation to plants has not been as widely studied (Hale et al., 2017; Das et al., 2013). One study used up to 30% inclusion of a commercially available sorbent (RemBind®) and found that between 10 and 30% inclusion reduced PFOA and PFOS leaching from AFFF-contaminated soil (< 2.6 ng mL⁻¹ PFOA and < 547 ng mL⁻¹ PFOS) up to 99% (Bräunig et al., 2021). MatCARE™, another commercially-available sorbent (palygorskite-based material modified with oleylamine) retained PFOS in soil with negligible release (0.5 – 0.6%) at 10% inclusion, with the best retention in soil with lower OM, lower pH, and higher clay content (Das et al., 2013). A study focused on a firefighting training facility found that PFOS leaching was reduced 35% with 3% montmorillonite clay (Hale et al., 2017).

Montmorillonite clay has been shown to effectively bind a variety of environmental chemicals and reduce their bioavailability from soil (Hearon et al., 2020, 2021; Sun et al., 2014; Yang et al., 2017). Calcium montmorillonite (CM) clay can tightly bind chemicals in active interlayer surfaces as well as basal surfaces and edge sites (Gu et al., 2010; Wang et al., 2019; Orr et al., 2020). Importantly, multiple clinical trials and intervention studies have demonstrated the safety of CM clay for consumption by humans and animals (Phillips et al., 2019). Our laboratory has modified CM clays with the natural nutrients L-carnitine (CM-carnitine) and choline (CM-choline) to increase the potential to bind more lipophilic chemicals. In this study, CM-carnitine and CM-choline clays were tested for their abilities to sorb PFOA and PFOS. Natural clay minerals like the parent CM clay have hydrophilic interlayer surfaces that are intrinsically negatively charged (Bolan et al., 2021). The L-carnitine and choline modifications change CM surfaces from hydrophilic to hydrophobic, enhancing the sorption of PFAS compounds (Darlington et al., 2018). These nutrient-modified clays have been shown to effectively bind PFOA and PFOS in an aqueous environment at pH 2 (Wang et al., 2021).

The objective of this study was to evaluate binding of PFAS by carbon- and clay-based sorbents and the applicability of these sorbents in PFAS remediation of soil. This study was designed to determine the effectiveness of AC, CM, CM-carnitine and CM-choline in binding PFOA and PFOS using 1) in vitro isothermal analyses (including thermodynamics and binding mechanisms), 2) in silico molecular dynamics (MD) simulations to validate binding percentages and binding modes, and 3) in vivo Lemna minor assay, soil bioavailability and plant uptake studies to validate the potential application of these sorbents.

2. MATERIALS AND METHODS

2.1. Reagents

High pressure liquid chromatography (HPLC) grade acetonitrile and pH buffers (4.0, 7.0 and 10.0) were purchased from VWR (Atlanta, GA). PFOA and PFOS standards were purchased from Sigma-Aldrich (Saint Louis, MO). HPLC grade acetone, methanol, and water + 0.1% formic acid were purchased from Fisher Scientific (Hampton, NH). Deionized water (18.2 MΩ) was generated in the laboratory using an Elga™ automated filtration system (Woodridge, IL) and was used in all experiments.

Activated carbon (AC) derived from coconut shell (mesh size: 100 – 325; Iodine number: 1100 mg g⁻¹; bulk density: 30 – 33 lbs ft⁻³) was obtained from the General Carbon Corporation (Paterson, NJ). Calcium montmorillonite (CM) was obtained from Engelhard Corp. (Cleveland, OH). CM has an average total surface area of 850 m² g⁻¹, an external surface area of approximately 70 m² g⁻¹, and a cation exchange capacity of 97 cmol kg⁻¹ (Grant & Phillips, 1998). CM has a general formula of (Ca)_0.3(Al,Mg)₂Si₄O₁₀(OH)₂·nH₂O(Marroquin-Cardona et al. 2011). Montmorillonite clays were modified with L-carnitine and choline at 100% cation exchange capacity (CEC = 97 cmol kg^–1) as previously described (Wang et al., 2017). Briefly, calculated amounts of cations and 25 g of parent materials were added in 500 mL of 1 mM HNO₃. The suspensions were mixed and stirred for 24 h at ambient temperature, then centrifuged at 2000g for 20 min and washed with 100 mL of distilled water. The centrifugation–washing process was repeated three times. All samples were dried in a desiccator before grinding and passing through a 125 μm sieve.

2.2. In vitro studies: Isothermal adsorption analyses

PFAS stock solutions were prepared by dissolving pure crystals in deionized water (pH 7) to yield 1000 μg mL⁻¹. AC, CM, CM-choline, and CM-carnitine were used as sorbents for adsorption isotherms. Sorbents were added at 0.0005% w/vol to PFAS solutions with an increasing concentration gradient (1 to 10 μg mL⁻¹). To achieve the 0.0005% inclusion level, 10 μL of 0.5 mg mL⁻¹ clay suspension in pH 7 water was added to each sample. Sorbent suspensions were mixed vigorously during transfer to ensure equal distribution of clay to each sample. The concentration gradients of PFAS solutions were achieved by adding a calculated amount of PFAS stock solution to a complementary volume of pH 7 water in 1.5 mL centrifuge tubes to a total volume of 1 mL. Additionally, 3 controls were tested (mobile phase, PFAS solution, and 0.0005% sorbent solution). All samples were agitated at 1000 rpm for 2 hours at ambient temperature (25°C) and high temperature (37°C) to investigate the thermodynamics of the reaction. Samples were then centrifuged at 2000g for 20 minutes to separate sorbent/PFAS complexes from the solution. The supernatants were collected for analysis.

PFAS were analyzed using a Waters Acquity® Ultra Performance Liquid Chromatography (UPLC)-Mass Spectrometry (MS)/MS (LC/MS-MS) equipped with a BEH C18 column (50 × 2.1 mm) following methods previously described (Liang & Chang, 2019; Roberts et al., 2017). Briefly, a gradient elution of 20 mM ammonium acetate (eluent A) and acetonitrile (eluent B) was carried out at a flow rate of 0.6 mL/min. The gradient for elution was 10% eluent B (initial), 10%−55% (0 to 0.1 min), 55%−99% (0.1 to 4.5 min), 99% (4.5 to 5 min), and 99%−10% (5 to 6.5 min). The injection volume was 10 μL for each sample. The LC/MS-MS was operated with an electrospray ionization interface (ESI) operated in negative ion mode. The spray voltage was maintained at 4.5 kV and the source temperature was kept at 450°C. The cone voltage (kV) for PFOA and PFOS was 25 and 40, respectively. The monitored precursor and product ions (m/z) for PFOA and PFOS were 413 to 369 and 499 to 80, respectively. The LC/MS-MS was operated under multiple reaction monitoring (MRM) mode. Nitrogen gas was used as the collision and curtain gas, and argon gas was used as the nebulizer and heater gas. Empower analyst software was used to control the LC/MS-MS system and acquire the data.

2.3. Data calculations and curve fitting

PFAS were detected by LC/MS-MS and quantified using standard calibration curves. PFAS concentrations in solutions were calculated from peak area at the retention times. The amount adsorbed for each data point was calculated from the concentration difference between test and control groups. The resulting data was input into a Microsoft Excel program developed in our laboratory and plotted using Table-Curve 2D to derive values for each parameter. The best fit for the data was the Langmuir model, which was then used to plot equilibrium isotherms for each analysis. The isotherm equation was entered as a user-defined function,

$$\text{Langmuir}\text{model}\text{(LM)}q=Q_{max}\left(\frac{KdCw}{1+KdCw}\right)$$

where q = toxin adsorbed (mol kg⁻¹), Q_max = maximum capacity (mol kg⁻¹), K_d = distribution constant, and C_w = toxin equilibrium concentration. Estimates for the Q_max and K_d were derived from a double logarithmic plot of the data. The plot will normally display a break in the curve, where the value of the x-axis is an estimate of K_d, and the value of the y-axis is an estimate of Q_max. The K_d value is derived from the Langmuir equation by solving for K_d:

$$K_d=\frac{q}{(Qmax-q)Cw}$$

The enthalpy (ΔH) is a parameter of the thermodynamics of the binding reaction, indicating total heat released or absorbed. It was calculated using the Van’t Hoff equation, comparing individual K_d values at two temperatures (25°C and 37°C):

$$\Delta \text{H}=\frac{-R\text{ln}\left(\frac{K{d}_2}{K{d}_1}\right)}{\left(\frac{1}{T_2}\right)-\left(\frac{1}{T_1}\right)}$$

where R (ideal gas constant) = 8.314 J/mol K, and T (absolute temperature) = 273 + t (°C).

2.4. Molecular Modeling and Molecular Dynamics Simulations

Parent CM, carnitine-modified CM, and choline-modified CM in the presence of PFOA and PFOS, independently, at neutral pH (pH 7) were computationally investigated through quintet 100 ns MD simulations. Each of the six systems (parent CM : PFOA, parent CM : PFOS, carnitine-modified CM : PFOA, carnitine-modified CM : PFOS, choline-modified CM : PFOA, and choline-modified CM : PFOS) was built and simulated analogously to our previous study investigating CM in the presence of PFAS in acidic conditions (pH 2) (Wang et al., 2021). The initial structure of the parent CM corresponded to the two-layered montmorillonite clay model at pH 7 reported in our previous study (Wang et al., 2019). This model was built using structural data extracted from the INTERFACE MD model database (Heinz et al., 2005, 2013). The d₀₀₁ spacing between the two layers was set to 21 Å (Wang et al. 2017, 2021). The two layered CM model was subsequently solvated in a 90 × 90 × 54 ų water box such that, with application of periodic boundary conditions in the MD simulations, the modeled system corresponded to an infinite number of CM layers with a d₀₀₁ spacing of 21 Å (Wang et al. 2017, 2021). The initial structure of the carnitine- and choline-modified CM was generated through a short 10 ns simulation equilibrating the modeled parent CM in the presence of carnitine or choline molecules as described in our previous study (Wang et al., 2021). Subsequently, for the parent, carnitine-modified, and choline-modified CM in the presence of PFAS, 16 copies of PFOS or PFOA, independently, were initially dispersed around the parent clay within the 90 × 90 × 54 ų water box in random configurations and orientations. The initial random configurations of all molecules for the parent and modified CM were generated from short 1 ns simulations of single molecules at infinite dilution (Wang et al., 2021). Within the simulations, the concentration of PFAS (0.055 M) was higher than experimental conditions. A higher concentration was used to enhance the statistical sampling and accelerate the potential sorption of PFAS to the clay surfaces (Wang et al., 2019, 2021; Orr et al., 2020).

Subsequently, each simulation system initially underwent energetic minimizations and a constrained 1 ns equilibration stage prior to each 100 ns MD simulation run as in our previous studies (Wang et al., 2019, 2021; Orr et al., 2020). Throughout the energy minimizations and 1 ns equilibration stage, all atoms excluding hydrogens of the clay, PFAS, and modifying molecules were constrained under 1.0 kcal mol⁻¹ Å⁻¹ harmonic constraints. Following energy minimization and equilibration, the simulation systems entered a 100 ns production stage in which all constraints were released with the exception of light 0.1 kcal mol⁻¹ Å⁻¹ harmonic constraints on aluminum atoms of the clay layers.

All MD simulations and setup were conducted in CHARMM (Brooks et al., 2009). Parameters and topologies for CM were extracted from the INTERFACE force field (Heinz et al., 2005, 2013). The PFAS and modifying molecules were all parameterized using CGENFF (Vanommeslaeghe et al., 2010; Yu et al., 2012). The CHARMM36 force field was used to model water (Jorgensen et al., 1983) and counter ions (Marchand & Roux, 1998). The temperature and pressure of the simulation systems were set to 300 K and 1 atm. Simulation snapshots were extracted in 100 ps intervals for subsequent analysis. Five 100-ns replicates of MD simulations with different initial velocities were performed for each system for an aggregate simulation time of 500 ns for each of the 6 simulation systems.

2.5. Structural Analysis of PFAS Sorption to Montmorillonite Clays within Molecular Dynamics Simulations

In-house FORTRAN programs were used to identify interactions between the PFAS and the parent CM, carnitine-modified CM, or choline-modified CM. A PFAS molecule was defined as interacting with the clay if the distance between any pair of atoms between the PFAS molecule and the clay was within 3.5 Å; such interactions were considered as “self-binding” interactions, in all sets of simulation systems comprising CM, carnitine-modified CM, and choline-modified CM. A carnitine or choline molecule was considered to be a modifying carnitine or choline if the distance between any pair of atoms between the carnitine or choline and the clay surface was within 3.5 Å. A PFAS molecule was defined as interacting with a modifying carnitine or choline, if the distance between any pair of atoms between the PFAS molecule and the modifying carnitine or choline was within 3.5 Å; such interactions were considered as “assisted-binding” interactions. It is worth noting that a PFAS molecule could be bound to the clay through both self- and assisted-binding interactions simultaneously within the simulation systems of the carnitine- or choline-modified CM such that a PFAS molecule interacts concurrently both with the clay and a modifying molecule; such interactions were considered as assisted-binding interactions only (to avoid double counting and to also highlight the importance of carnitine and choline in the modified CM).

To identify the predominant binding modes of the PFAS binding to parent, carnitine-modified, and choline-modified CM, we performed a statistical analysis on the interactions between the PFAS and the clay layers within their respective simulations. Binding modes of the PFAS and the modifying molecules were classified based on decomposition of the molecules into functional groups. Self-binding interactions with unmodified or modified clay were clustered into different binding modes depending on which functional groups of the toxic compounds, independently or in combinations, were binding to the clay layer. Assisted-binding interactions with modified CMs were classified into assisted-binding modes based on the functional groups of the modifying molecule, independently or in combinations, binding to the clay layer as well as the functional groups of PFAS and modifying molecules, independently or in combinations, interacting with each other based on the 3.5 Å cutoff.

2.6. In vivo studies: Lemna minor assay

The efficacy and safety of sorbents was predicted for a living organism using a freshwater aquatic plant (Lemna minor). This well-established toxicity assay was used to screen and validate sorbent ability to protect against PFAS toxicity before more extensive soil and plant studies were conducted. L. minor were obtained from Aqua L’Amour (Elk Grove, CA). L. minor were kept under a growth light (8:16 light: dark cycle) in ambient temperature using Steinberg nutrient medium (3.46 mM KNO₃, 1.25 mM Ca(NO₃)₂, 0.66 mM KH₂PO₄, 0.072 mM K₂HPO₄, 0.41 mM MgSO₄, 1.94 μM H₃BO₃, 0.63 μM ZnSO₄, 0.18 μM Na₂MoO₄, 0.91 μM MnCl₂, 2.81 μM FeCl₃, 4.03 μM EDTA; pH 5.5 ± 0.2) (Drost et al., 2007).

L. minor studies were conducted based on the OECD Test Guidelines for Testing Chemicals, Lemna sp. Growth Inhibition Test (OECD, 2002). Experiments were conducted in 24-well plates with 2 mL of test solution and three 2-frond L. minor. Plants were monitored and photographed over the 7-day exposure period. Frond numbers were recorded and surface areas were calculated using ImageJ (Abramoff et al., 2003). Frond surface areas were used to calculate the growth rate (μ):

$$\text{μ}=\frac{\text{ln}(At2)-\text{ln}(At1)}{t2-t1}$$

where A_t1 and A_t2 are the surface areas (mm²) calculated at day t₁ and day t₂, respectively. The growth rate inhibition was calculated by comparing the growth rates of samples and controls:

$$\%\text{inhibition}=100\times \left(1-\frac{\text{μ}sample}{\text{μ}control}\right)$$

(Drost et al., 2007). L. minor were homogenized in 90% acetone using a 150 Homogenizer and 7 mm Generator Probe (Fisher Scientific, Waltham, MA) and placed in a freezer for 72 hours to extract chlorophyll (Su et al., 2010; Taraldsen & Norberg-King, 1990). Chlorophyll a and b concentrations were measured using a UV-Visible spectrophotometer at wavelengths of 663 and 644 nm, respectively. Total chlorophyll (a and b) was calculated using Beer’s Law:

$${\text{C}}_{\text{a}+\text{b}}=\left(\frac{A_a}{{\epsilon }_a\times b}\right)+\left(\frac{A_b}{{\epsilon }_b\times b}\right)$$

Where C is the total chlorophyll concentration (chlorophyll a + chlorophyll b), A_a and A_b are the absorbance values measured at 663 nm and 644 nm, respectively, ε_a and ε_b are the absorption coefficients for chlorophyll a and chlorophyll b (87.67 L g cm⁻¹ and 51.36 L g cm⁻¹, respectively) and b is the length of the light path (Jeffrey & Humphrey, 1975).

2.7. Soil studies

Garden soil (with a composition including compost, processed forest products, sphagnum peat moss, a wetting agent, and fertilizer containing 0.09% total nitrogen, 0.05% available phosphate, and 0.07% soluble potash) was obtained from The Scotts Miracle-Gro Company (Marysville, OH) and compost (with a composition including aged forest products, sphagnum peat moss, perlite, sandy loam, and fertilizer containing 0.30% total nitrogen, 0.45% available phosphate, and 0.05% soluble potash, and 1.00% calcium derived from fish emulsion, crab meal, shrimp meal, earthworm castings, bat guano, kelp meal and oyster shell) was obtained from Foxfarm Soil & Fertilizer Company (Humboldt County, CA). Soil and compost were air-dried and sieved through a 1 mm screen before use. Each 1 g sample in a disposable culture tube was spiked with 2 mL of 1.0 μg mL⁻¹ PFAS/acetone solution and thoroughly mixed to ensure even distribution of PFAS. Spiked soil samples were used in proof-of-concept experiments to provide a controlled environment that would facilitate sample replication and measurement of PFAS concentrations. Samples were left uncapped in a fume hood for 3 days allowing evaporation of the acetone. Sorbents were added at 2% w/w to soil and compost samples and thoroughly mixed. Samples were hydrated by adding 4 mL of distilled water and then slowly agitating at 200 rpm for 24 hours. A soil extraction method (Huset & Barry, 2018) with modification was used to extract PFOA and PFOS from soil/compost samples. Briefly, 4 mL of methanol + 1% NH₄OH was added to each sample before agitating at 1000 rpm for 1 hour. Samples were centrifuged at 2000g for 20 minutes and the supernatants were transferred to new culture tubes. The supernatants were then passed through Strata C18-E (55 μm, 70A) columns preconditioned with 2 mL methanol and 2 mL water. Extracts were analyzed for PFAS using LC/MS-MS. Calibration curves were conducted for each group of extracts to ensure linearity of peak concentrations and consistency of the extraction method. Controls for each group of extracts included blank soil, soil spiked with PFAS, and soil with sorbent. Peak areas of PFAS from samples that included sorbents were compared to the PFAS control samples to determine the percent reduction in PFAS bioavailability from soil.

2.8. Plant uptake studies

Organic cucumber seeds were obtained from Ferry Morse (Norton, MA). Time course and sorbent inclusion studies were conducted using a 1:1 mixture of the soil and compost used in soil bioavailability studies. Each planter contained 10 g of soil/compost hydrated with 20 mL water. Cucumber seeds were planted with at least 2 seeds per planter to ensure an adequate number of viable sprouts. Planters were monitored and watered throughout the study to ensure adequate hydration. Seeds were allowed to germinate and sprout for 7 days before planters were moved to hydroponic systems containing 2 L nutrient solution (potassium nitrate, calcium nitrate, monopotassium phosphate, potash, magnesium sulfate, and trace elements at 2.9 – 3.2 EC and 6.3 pH) on day 7. Hydroponic systems were used to simulate flood scenarios during which roots might be exposed to higher levels of water and potentially higher levels of contaminants. Cucumber roots had significantly grown into the nutrient solutions after 7 days. For each study, ten experimental groups (soil/compost control, chemical control with 1 μg mL⁻¹ PFAS, 4 sorbent controls at 2% in base media, and 4 treatment groups of 2% sorbent and 1 μg mL⁻¹ PFAS) were set up with six planters in each group. Experimental groups were administered 500 mL of nutrient solution containing 1 μg mL⁻¹ for chemical controls, 2% sorbent for sorbent controls and 1 μg mL⁻¹ plus 2% sorbent for treatment groups. After a 7-day exposure period, planters were removed from hydroponic systems. Roots and sprouts were separated, washed thoroughly with DI water to remove soil, and air-dried before dry weights were measured. Samples were then ground to a powder-like texture before extraction.

Extraction methods described by Huset & Barry (2018) were followed with modification. Briefly, 5 mL of methanol + 1% NH₄OH were added to each sample in disposable culture tubes. Samples were homogenized for 5 min using a 150 Homogenizer and 7 mm Generator Probe (Fisher Scientific, Waltham, MA). The tubes were capped and agitated at 1000 rpm for 20 min and then centrifuged at 2000g for 20 min. The supernatants were passed through Strata C18-E (55 um, 70A) columns preconditioned with 2 mL methanol and 2 mL water. 2 mL of the extracts were added to autosampler vials and analyzed using LC-MS/MS.

2.9. Statistical analysis

A one-tailed t-test was used to determine statistical significance. Each experiment was conducted in triplicate to derive means and standard deviations for the following: 1) Q_max and K_d values from isothermal analyses, 2) growth parameters from the lemna assay, 3) concentrations of PFAS in soils and plants, and 4) root weight and sprout length and weight. These were compared using a Tukey test. T-values and degrees of freedom were used to determine the p-value. Results were considered significant at p≤0.05.

3. RESULTS & DISCUSSION

3.1. Adsorption isotherms

Adsorption isotherms for PFOA and PFOS were conducted at pH 7 to simulate environmentally-relevant conditions (Figure S1). Isothermal data was generated with TableCurve 2D software to derive sorbent-PFAS binding parameters. All isotherms fit the Langmuir equation (r² > 0.95), which was used to determine Q_max (capacity) and K_d (affinity) values for PFOA and PFOS binding to AC, CM, CM-carnitine and CM-choline (Table 1). The resulting Q_max and K_d values along with the curved shape of the Langmuir plot indicated saturable binding of PFOA and PFOS to active surfaces of all tested sorbents. Importantly, the modified CM-carnitine and CM-choline clays had significantly higher capacities for both PFOA and PFOS than AC and CM (**p < 0.01). The L-carnitine and choline modifications change the active surfaces of the parent montmorillonite clay from hydrophilic to hydrophobic, which allows them to attract more lipophilic moieties like the PFOA carboxylic functional group and the PFOS sulfonic functional group.

Table 1.

Q_max and K_d values for all four sorbents binding PFOA and PFOS derived from isotherms conducted at 25°C and 37°C.

Chemical	Sorbent	Qmax (25° C)	Qmax (37° C)	Kd (25° C)	Kd (37° C)	ΔH
PFOA	AC CM CM-carnitine CM-choline	0.39 0.37 0.51 0.61	0.42 0.58 0.99 1.09	1.86E+05 2.34E+05 1.63E+05 1.41E+05	1.80E+05 9.12E+04 5.28E+04 7.14E+04	−2.06 −60.23 −72.24 −43.52
PFOS	AC CM CM-carnitine CM-choline	0.38 0.49 0.71 0.63	0.47 0.59 0.98 0.93	3.51E+05 1.99E+05 2.05E+05 3.40E+05	2.67E+05 9.34E+04 8.56E+04 1.06E+05	−36.68 −48.47 −55.79 −74.18

To determine the enthalpies of the binding reactions between PFAS and sorbents, isotherms were also conducted at 37° C (Figure S2). The K_d values for isotherms conducted at 25° C and 37° C were compared using the Van’t Hoff equation to calculate enthalpies (ΔH). Determining the ΔH of sorbent-chemical binding can provide insight into the strength of reactions and potential mechanisms (i.e. chemisorption or physisorption). The lower ΔH values for PFOA and PFOS binding to AC surfaces may indicate weaker bonds such as hydrogen bonds, dipole-dipole interactions, and Van der Waals attractions (i.e., physisorption). The negative enthalpy values (Table 1) indicated exothermic reactions for PFAS binding to clay-based sorbents, and the high absolute values of ΔH (more than 20 kJ mol⁻¹) indicated chemisorption mechanisms (tight binding) of PFOA and PFOS. These results suggest that the PFAS-sorbent complexes are stable and PFOA/PFOS would not be readily dissociated from clay sorbent surfaces.

3.2. Major Binding Sites & Mechanisms

The binding propensities of PFOA and PFOS to CM, CM-carnitine, and CM-choline were independently computationally assessed through calculations within their respective simulations. Our computational analysis showed that both PFOA and PFOS frequently bound to CM and that both carnitine and choline modifications improved the binding of both PFAS. Additionally, CM-carnitine appeared to be more effective at binding PFOS while CM-choline appeared to be more effective at binding PFOA.

For CM, the binding percentages of PFOA and PFOS onto parent CM were 25±1% and 27±3%, respectively (Figure 1). Both PFOA and PFOS predominantly bound to the parent CM interlayer through hydrophobic interactions with the fluorinated carbon chains. Specifically, in 62% and 51% of binding instances, PFOA and PFOS interacted with parent CM at the terminal end of their fluorinated carbon chains, respectively (Figure 2A & 2B). In 24% and 40% of binding instances, PFOA and PFOS interacted with parent CM through their entire fluorinated carbon chain, respectively (Figure 2A & 2B). The predominance of PFOA and PFOS binding to parent CM through their hydrophobic tails suggests that hydrophobic interactions are key components of the binding mechanism for both PFAS to CM, which agrees with previous studies (Zhang et al., 2019).

Figure 1.

Overall binding percentages of CM (solid red bars), CM-carnitine (dotted red bars and green bars), and CM-choline (striped, red bars and blue bars) for PFOA and PFOS. Red bars correspond to self-binding in which the PFAS molecule is bound directly to the clay (without contacting any modifying carnitines or cholines), green bars correspond to assisted-binding in which the PFAS molecule is bound to CM-carnitine through a modifying carnitine, and blue bars correspond to assisted-binding in which the PFA molecule is bound to CM-choline through a modifying choline. Error bars correspond to the standard error across the quintet MD simulations.

Figure 2.

Molecular graphics images of representative prominent binding modes for (A) PFOA and (B) PFOS binding to CM, (C) PFOA and (D) PFOS binding to CM-carnitine, and (E) PFOA and (F) PFOS binding to CM-choline. Zoomed in images of the prominent binding modes are encircled by dotted lines and are reoriented to facilitate the comparison of different binding modes.

Both carnitine and choline improved the binding of PFAS to CM (Figure 1). For CM-carnitine, the overall binding percentages of PFOA and PFOS onto CM-carnitine were 36±2% and 39±2%, respectively. For CM-choline, the binding percentages of PFOA and PFOS onto CM-choline were 38±2% and 34±2%, respectively. The higher binding percentages of the PFAS onto CM-carnitine and CM-choline compared to parent CM were attributed to the PFAS binding to modifying carnitine or choline through assisted-binding (Figure 1, green and blue bars).

For CM-carnitine and CM-choline, the modifying molecules primarily bound to the clay surface through their positively charged quaternary ammonium groups. Carnitine bound to the surface through its quaternary ammonium group in over 90% of its binding instances, and choline bound to the surface through its quaternary ammonium group in over 80% of its binding instances. For CM-carnitine assisted-binding, PFOA frequently bound to CM-carnitine through hydrogen bond formation between the hydroxyl group of carnitine (Figure 2C & 2D). In this binding mode, PFOA simultaneously bound to clay interlayers through its fluorinated carbon chain (Figure 2C & 2D). PFOA also frequently participated in assisted-binding to CM-carnitine through its fluorinated carbon chain (Figure 2C & 2D, 35% of assisted-binding instances). For the assisted-binding of PFOS, PFOS predominantly bound to CM-carnitine by forming hydrogen bonds to hydroxyl group of carnitine with its sulfonate group (Figure 2C & 2D, 47% of assisted-binding). Additionally, PFOS participated in assisted-binding to CM-carnitine through its fluorinated carbon chain (Figure 2C & 2D, 23% of assisted-binding). For CM-choline assisted-binding, PFOA predominantly bound to CM-choline through hydrogen bond formation between the hydroxyl group of choline and its carboxyl group (Figure 2E & 2F, 48% of assisted-binding). PFOA also participated in assisted-binding to CM-choline through its fluorinated carbon chain (Figure 2E & 2F, 36% of assisted-binding). For the assisted-binding of PFOS, PFOS predominantly bound to CM-choline by forming hydrogen bonds to the hydroxyl group of carnitine with its sulfonate group (Figure 2E & 2F, 61% of assisted-binding). Additionally, PFOS participated in assisted-binding to CM-choline through its fluorinated carbon chain (Figure 2E & 2F, 28% of assisted-binding).

The predominant binding modes of PFOA and PFOS to CM or CM-carnitine uncovered in this study at pH 7 are largely the same to the predominant binding modes uncovered in our recently published study at pH 2, simulating the conditions of the stomach (Wang et al., 2021). However, because the carboxyl group of carnitine was considered to be protonated at pH 2, the carboxyl or sulfonate groups of PFOA or PFOS, respectively, also formed hydrogen bonds with the protonated carboxyl group of carnitine in addition to the hydroxyl group of carnitine (Wang et al., 2021). Our previous computational study did not investigate PFOA or PFOS binding to CM-choline.

3.3. Lemna minor assay

The Lemna minor assay is widely accepted and has been extensively studied as a relevant method to identify and characterize potential hazards of chemicals and toxicity on vegetative growth. In this study, we expanded on this traditional application to include sorbent safety and efficacy testing. These studies were important in evaluating sorbent safety to a plant organism and the sorbent’s ability to protect the plants from chemical toxicity.

Dose-response experiments (0 to 100 μg mL⁻¹) were conducted and 100 μg mL⁻¹ PFOA and 100 μg mL⁻¹ PFOS were identified as the concentrations that significantly reduced L. minor growth over the 7-day exposure period (53% and 57% reduction in frond surface area with 100 μg mL⁻¹ PFOA and 100 μg mL⁻¹ PFOS, respectively). These exposure levels also decreased chlorophyll content by 42.1% and 37.0% for PFOA and PFOS, respectively. The phytotoxicity of PFOA and PFOS (at similar and higher concentrations to those used in this study) has been seen with various plant species (Zhao et al., 2013). These exposure concentrations were used in subsequent sorbent studies, in which sorbents were tested at various inclusion levels (0.05 to 1.0%) to protect against chemical toxicity. Optimal protection against PFAS toxicity was achieved with 0.1% sorbent inclusion. Figure 3A and 3B show increased L. minor surface areas with 0.1% sorbent inclusion compared to PFOA and PFOS controls. With CM-carnitine and CM-choline inclusion, L. minor surface area increased by 134% and 129% compared to PFOA control, respectively, and by 152% and 164% compared to PFOS controls, respectively. Additionally, chlorophyll content (chlorophyll a and b) significantly increased in the sorbent treatment groups compared to the PFOA (Figure 3C) and PFOS (Figure 3D) controls. Chlorophyll content increased from 1.0 g L⁻¹ in the PFOA exposure group to 6.27 g L⁻¹ and 6.5 g L⁻¹ in the CM-carnitine and CM-choline treatment groups, respectively. Chlorophyll content increased from 1.62 g L⁻¹ in the PFOS exposure group to 6.04 g L⁻¹ and 4.02 g L⁻¹ in the CM-carnitine and CM-choline treatment groups respectively. These studies indicated that the modified clays were consistently more effective than the parent CM clay, similar to what was seen in the isothermal analyses. Importantly, these screening toxicity and sorbent efficacy studies were important in validating the isothermal results in vivo before conducting more extensive hydroponic plant studies.

Figure 3.

Frond surface area of Lemnaminor over the 7-day period exposed to 100 ppm PFOA (A) and 100 ppm PFOS (B) and protection of lemna with 0.1% sorbent inclusion. Chlorophyll content of lemna exposed to PFOA (C) and PFOS (D) measured at the end of the exposure period.

3.4. Soil studies

Soil studies were conducted to determine the efficacy of sorbents in reducing PFAS bioavailability from soil. Sorbent inclusion was higher in this study than the isotherms and L. minor assay due to the organic matter (OM) content of soil and compost. Doses of up to 30% have been used in media with high OM content (Bjerkli, 2019; Bräunig et al., 2021). Preliminary studies in our laboratory that evaluated the efficacy of various sorbent inclusion levels (0.1 – 5%) showed 2% inclusion of AC, CM, CM-carnitine, and CM-choline was optimal. Soil and compost samples (1 g) were spiked with 1 μg g⁻¹ PFAS. This was chosen as the exposure level because it is higher than typical environmental levels (Strynar et al., 2012; Rankin et al., 2016), lower than levels detected at contaminated sites (Brusseau et al., 2020), and comparable to concentrations used in studies focused on translocation of PFAS from soil to plants (Lal et al., 2020; Zhao et al., 2013; Garcia-Valcarcel et al., 2014). Figure 4 shows reduced bioavailability of PFOA and PFOS from spiked soil with inclusion of AC, CM, CM-carnitine, and CM-choline compared to bioavailability from spiked soil without sorbents. Results showed the modified CM-carnitine and CM-choline sorbents reduced PFOA bioavailability from soil by 58.0% and 57.9%, respectively, and from compost by 58.5% and 62.8%, respectively. CM-carnitine and CM-choline reduced PFOS bioavailability from soil by 77.5% and 79.4%, respectively, and from compost by 60.8% and 60.5%, respectively. The modified clays were consistently more effective than the parent CM clay. We expect that at lower and more environmentally-relevant doses of PFAS than used in this study, CM and modified-CM sorbent efficiencies would be higher.

Figure 4.

Soil studies showing significant reduction of PFOA (A,B) and PFOS (C,D) bioavailability from soil with 2% inclusion of AC, CM, CM-carnitine and CM-choline. Amended clays were consistently more effective than the parent clay.

3.5. Plant uptake studies

Plant uptake studies were important in evaluating translocation of PFAS to cucumber sprouts and sorbent efficacy in reducing PFAS translocation. Cucumbers were chosen as the model plant in this study because PFAS compounds, especially PFOA and PFOS, tend to accumulate in fruit vegetables and have been shown to have high transfer rates from water and soil (Lechner & Knapp, 2011; Liu et al., 2019). Time course studies were conducted to evaluate PFOA and PFOS translocation to cucumber roots, stems, and leaves over 21-day exposure periods. Cucumber sprouts were exposed to PFOA or PFOS in both the soil and the nutrient solutions of hydroponic systems. Dry weights for cucumber sprouts exposed to PFOA measured for roots (Figure 5A), stems (Figure 5B) and leaves (Figure 5C) were not significantly different than controls. Cucumber components were extracted and PFOA was measured in roots (Figure 5D), stems (Figure 5E), and leaves (Figure 5F). Over the 21-day exposure period, PFOA concentrations increased from 10.1 to 18.0 μg/g in roots, from 1.2 to 8.9 μg/g in stems, and from 1.7 to 22.5 μg/g in leaves. As indicated by the proportions of total PFOA in the cucumber components, PFOA began to accumulate in the leaves by the end of the exposure period (Figure 5G). Dry weights for cucumber sprouts exposed to PFOS measured for roots (Figure 6A), stems (Figure 6B) and leaves (Figure 6C) were also not significantly different than controls. Over the 21-day exposure period, PFOS concentrations increased from 12.1 to 27.0 μg/g in roots (Figure 6D), decreased from 15.2 to 13.7 μg/g in stems (Figure 6E), and increased from 1.4 to 21.3 μg/g in leaves (Figure 6F). Similar to PFOA, PFOS began to accumulate in the leaves at the end of the 21-day exposure period (Figure 6G). We expect that this accumulation would increase if the plants continued to grow past the 21-day exposure. Because PFOA and PFOS exist in ionized forms at pH 7, potential mechanisms behind the translocation of PFOA and PFOS to cucumber sprouts could be facilitated diffusion and active transfer (Zhao et al., 2013).

Figure 5.

Time course plant study showing uptake of PFOA by cucumber sprouts. Root dry weight (A), sprout dry weight (B) and total sprout length (C) were measured over the duration of the 21-day exposure period. PFOA concentrations were measured in roots (D), stems (E), and leaves (F). Proportions of PFOA in the sprout compartments suggest accumulation of PFOA in the leaves by the end of the exposure period.

Figure 6.

Time course plant study showing uptake of PFOS by cucumber sprouts. Root dry weight (a), sprout dry weight (b) and total sprout length (c) were measured over the duration of the 21-day exposure period. PFOS concentrations were measured in roots (d), stems (e), and leaves (f). Proportions of PFOS in the sprouts compartments suggest accumulation of PFOS in the leaves by the end of the exposure period.

Sorbent inclusion studies were conducted to determine sorbent ability to reduce PFAS translocation to cucumber sprouts. Root, stem, and leaves dry weights were measured for all groups at the end of the exposure period. In the roots (Figure 7A), stems (Figure 7B), and leaves (Figure 7C) of cucumbers in the PFOA study, the CM-carnitine control group had significantly higher growth compared to the blank group (p < 0.05). L-carnitine has been shown to be important in fatty acid metabolism and cell homeostasis in plants as well as promoting cell division and plant growth (Jacques et al., 2018; Oney-Birol, 2019). These positive effects on plant growth could explain the increased growth seen in the CM-carnitine control group. At 2% sorbent inclusion, all tested sorbents were able to significantly reduce PFOA translocation to cucumber roots, stems, and leaves (Figure 7D, 7E, & 7F). CM averaged 51.2% reduction for all plant compartments, while CM-carnitine and CM-choline averaged 63.4% and 64.7%, respectively. For PFOS, there were no significant differences between exposures groups for root (Figure 8A), stem (Figure 8B), and leaves (Figure 8C) dry weights. Clay-based sorbents were highly effective, reducing PFOS residues by an average of 50.1%, 69.9%, and 67.2% for CM, CM-carnitine, and CM-choline, respectively (Figure 8D, 8E, & 8F). The modified CM clays were more effective in reducing PFOA and PFOS translocation to cucumber sprouts than the parent CM clay. This increased efficacy along with the beneficial growth effects of L-carnitine suggest that modified CM clays could present a valuable soil amendment strategy for PFAS remediation and protection of plants.

Figure 7.

Plant study showing translocation of PFOA to cucumber sprouts and reduction of PFOA translocation with sorbents. Root (A), stem (B), and leaf (C) dry weights were measured for all exposure groups. Sorbents significantly reduced PFOA residues in cucumber roots (D), stems (E), and leaves (F) (**p < 0.01; *p < 0.05).

Figure 8.

Plant study showing translocation of PFOS to cucumber sprouts and reduction of PFOS translocation with sorbents. Root (A), Stem (B), and leaf (C) dry weights were measured for all exposure groups. Sorbents significantly reduced PFOS residues in cucumber roots (D), stems (E), and leaves (F) (**p < 0.01).

AC was also able to significantly reduce PFOA and PFOS residues in cucumber sprouts (averaged 69.9% and 74.2% for PFOA and PFOS, respectively). AC and other carbon-based sorbents have been widely used and shown to be effective for contaminant immobilization in soil. However, it is important to note that some studies have suggested that AC can negatively impact soil and plant health. AC has been shown to reduce microbial denitrification and nutrient availability (Bonaglia et al., 2019). AC can also reduce macrofauna living in sediment and decreased survival and growth rates of benthic invertebrates (Janssen & Beckingham, 2013; Samuelsson et al., 2015). To reduce these effects and to potentially enhance plant health, modified clays alone or in combination with AC could enhance PFAS remediation from soil and benefit plant growth (Cornelissen et al., 2011). Studies have shown that combinations of AC and clay-based sorbents can enhance binding efficacy of contaminants such as metals and can increase nitrogen availability (Hao et al., 2019; Zhu et al., 2019). Based on our research, the use of nutrient-modified CM clays, alone or in combination with AC, could be a better strategy than AC alone to decrease chemical bioavailability from soil and translocation to plants. Future studies that include legacy PFAS, short-chain PFAS, and PFAS precursors will facilitate the further evaluation of these novel sorbents.

CONCLUSIONS

Adsorption with modified CM clays presents a promising strategy for PFAS remediation of soil and reducing PFAS translocation to produce. The efficacy of this technique is supported by in vitro, in vivo, and in silico results that show sorbents can tightly bind PFOA and PFOS and can reduce bioavailability from soil and translocation to cucumber sprouts. CM-carnitine and CM-choline were consistently more effective in binding PFOA and PFOS than the parent CM clay due to increased hydrophobicity of clay surfaces. Importantly, these sorbents are environmentally-friendly, energy-efficient, and practical. Additional studies are needed to investigate the field efficacy and long-term potential of these sorbents in reducing PFAS bioavailability. Future studies including AC, CM, modified CM clays, and mixtures of these sorbents will investigate the effects of sorbents on nutrient utilization in plants and the mitigation of other environmentally relevant chemicals following natural disasters, chemical spills, and emergencies.