Testing the Efficacy of Broad-Acting Sorbents for Environmental Mixtures using Isothermal Analysis, Mammalian Cells, and H. vulgaris

Abstract

The hazards associated with frequent exposure to polycyclic aromatic hydrocarbons (PAHs), pesticides, Aroclors, plasticizers, and mycotoxins are well established. Adsorption strategies have been proposed for the remediation of soil and water, although few have focused on the mitigation of mixtures. This study tested a hypothesis that broad-acting sorbents can be developed for diverse chemical mixtures. Adsorption of common and hazardous chemicals was characterized using isothermal analysis from Langmuir and Freundlich equations. The most effective sorbents included medical-grade activated carbon (AC), parent montmorillonite clay, acid-processed montmorillonite (APM), and nutrient-amended montmorillonite clays. Next, we tested the ability of broad-acting sorbents to prevent cytotoxicity of class-specific mixtures using 3 mammalian in vitro models (HLF, ESD3, and 3T3 cell lines) and the hydra assay. AC showed the highest efficacy for mitigating pesticides, plasticizers, PAHs, and mycotoxins. Clays, such as APM, were effective against pesticides, Aroclors, and mycotoxins, while amended clays were most effective against plasticizers. Finally, a sorbent mixture was shown to be broadly active. These results are supported by the high correlation coefficients for the Langmuir model with high capacity, affinity, and free energy, as well as the significant protection of cells and hydra (p <0.05). The protection percentages in 3T3 cells and hydra showed the highest correlation as suggested by both Pearson and Spearman with r = 0.84 and rho = 0.73, respectively (p < 0.0001). Collectively, these studies showed that broad-acting sorbents may be effective in preventing toxic effects of chemical mixtures and provided information on the most effective sorbents based on adsorption isotherms, and in vitro and aquatic organism test methods.

Graphical abstract

1. INTRODUCTION

Humans and animals are continuously exposed to a multitude of substances. Most studies have focused on the safety and toxicity testing of individual chemicals, but not mixtures. Complex environmental mixtures of concern may include different classes of substances, such as pesticides, Aroclors, plasticizers, polycyclic aromatic hydrocarbons (PAHs), and mycotoxins. Pesticides are widely used in agriculture and household settings, resulting in frequent exposure to pesticide residues from food and drinking water (Damalas and Eleftherohorinos, 2011). Aroclors are complex mixtures of polychlorinated biphenyls (PCBs) that have been extensively used as dielectric fluids in transformers and heat-exchange fluids. Humans can be exposed to Aroclors by eating contaminated fish, meat, and dairy products (idhp.state.il.us). Plasticizers are used widely in medical devices, food processing and packaging, and electronics. Dietary consumption is considered as a major route of exposure to plasticizers in humans because they can partition into food and water from packaging or during food processing (Bui et al., 2016). PAHs are widespread environmental contaminants formed during incomplete combustion or pyrolysis of organic materials. The primary source of human exposure to PAHs is food, which contributes 99% (WHO, 1984). Humans and animals can also be exposed to mycotoxins via contaminated food during droughts and extended periods of heat, when fungi reach their optimal growth conditions for the production of mycotoxins.

The most effective and economical approach to remediate chemical contamination of water or other media is through sorbent filtration and purification methods; however, most sorbent studies have focused on the removal of individual chemicals, rather than complex mixtures. Also, methods to remediate contaminated food and feed, which are major routes of exposure to chemicals in both humans and animals, are lacking. Previous studies have reported the development of sorbents for binding specific chemical contaminants with high sorption efficacy. Activated carbon (AC) is known as one of the most effective sorbents. AC is widely used for water purification (Cecen and Aktas, 2011; Zhang et al., 2017); however, few studies have examined the efficacy and suitability of medical grade AC for animal and human consumption against complex chemical mixtures. Clay-derived materials, such as calcium montmorillonite, have a high affinity for binding aflatoxins and similar hydrophilic compounds; these materials are edible and safe for animal and human consumption (Phillips et al., 2019; Wang and Phillips, 2020). To develop broad-acting sorbents with high capacity for diverse environmental chemicals, a base montmorillonite clay has been activated with acid to simulate carbon structure with high surface area and porosity, and amended with organic nutrients to enhance interlayer spacing and surface hydrophobicity (Celis et al., 2007; Wang et al., 2017). These modified sorbents have been shown to bind organophilic chemicals, such as dyes, zearalenone, benzo[a]pyrene, pesticides, and PCBs (Cruz-Guzman et al., 2004; De, et al., 2009; Hearon et al., 2020; Ugochukwu and Fialips, 2017; Wang et al., 2019a; Yip et al., 2005). Therefore, their binding efficacy for an even broader range of chemical classes and mixtures was investigated.

In vitro toxicity tests have been widely applied for screening environmental chemicals and mixtures (Gibb, 2008; Krewski et al., 2010). The scientific advantages for the broad use of in vitro toxicity tests include assessing chemicals in a more time- and cost-efficient manner while providing more relevant and mechanistic insights (Novakova et al., 2020). Furthermore, to study the toxicity of complex mixtures of chemicals, in vitro cell bioassays represent sensitive and effective tools for toxicological profiling, because they can cover the combined effects of the mixture’s components and enable the prioritization of toxicity drivers (Bandele et al., 2012; Neale et al., 2015). Also, in vitro cell-based assays may serve as biological testing models to evaluate the remediation potential of sorbents for environmental chemicals (Nones et al., 2017).

H. vulgaris, a freshwater cnidarian, is an environmental model used to study the acute effects of toxicants in a living organism (Dash and Phillips, 2012). Adult hydra assays have been utilized previously to accurately predict the safety and efficacy of toxicant-binding sorbents for subsequent studies in animals and plants (Afriyie-Gyawu et al., 2005; Hearon et al., 2021; Marroquin-Cardona et al., 2009). The hydra assay has also been utilized along with an in vitro gastrointestinal model (Lemke et al., 2001) and in silico molecular dynamic simulations (Wang et al., 2019) to screen and validate sorbents.

The objective of this study was to investigate the ability of various sorbents to effectively prevent toxicity of diverse environmental chemicals and their mixtures. Sorbents were characterized using isothermal analysis based on binding capacity, affinity, heterogeneity, correlation coefficients, and free energy of sorption. Furthermore, the efficacy of sorption was characterized using 3 mammalian cell models and a hydra assay to confirm the safety of sorbent inclusion and to verify the ability of sorbents to reduce the toxicity of complex mixtures.

2. MATERIALS AND METHODS

2.1. Sorbent materials

Medical grade activated carbon (AC), purity > 99%, was obtained from General Carbon Corporation (Paterson, NJ). It is labeled as a virgin powdered AC derived from a selected grade of coconut shell with 1100 m²/g surface area, 5% moisture (Generalcarbon.com), a pH of point of zero charge (pH_PZC) equal to 9.57 (Wang et al., 2020a), and a zeta potential of −31 mV (Kumar et al., 2020). Calcium montmorillonite clay was obtained from TxESI, Inc. (Bastrop, Texas) with a total surface area of approximately 850 m²/g, an external surface area of approximately 70 m²/g, a cation exchange capacity (CEC) equal to 89.2 cmol/kg, and a pH_PZC equal to 8.8 (Wang et al., 2020b). The generic formula for the clay is (Ca,Na)_0.3(Al,Mg)₂Si₄O₁₀(OH)₂·nH₂O. Its chemical characterization by X-ray powder diffraction (XRD) and thermogravimetric analysis (TGA) were previously published (Hedley et al., 2007; Marroquin-Cardona 2011). Calcium montmorillonite clay was treated with sulfuric acid at 18 normality, as previously described, to yield an acid-processed montmorillonite (APM) with a surface area equal to 1213 m²/g, and a pH_PZC of 5.29 (Wang and Phillips, 2019). In another study, this montmorillonite clay was individually amended with natural nutrients, L-carnitine and choline, at 100% CEC to develop organophilic clays (mont-carnitine and mont-choline) with enhanced d₀₀₁ interlayer spacing, and pH_PZC of 9.35 and 10, respectively (Velazquez and Deng, 2020; Wang et al., 2017; 2020). Surface morphologies of clay-based sorbents were investigated using a Tescan Vega3 scanning electron microscope (SEM) (Tescan Orsay Holding, a.s., Brno, Czech Republic) at 20 kV with a secondary electron detector at 5 kV accelerating voltage. The surface hydrophobicity of all sorbents was measured by the mass ratio of adsorption of n-heptane/water as previously described (dos Reis et al., 2016; 2018). All sorbents were sieved at 325 mesh to achieve uniform particle size equal to, or less than 44 microns.

2.2. Reagents

HPLC grade reagents were purchased from VWR (Atlanta, GA). Ultrapure deionized water (18.2 MΩ) was generated using an Elga™ (Woodridge, IL, USA) automated filtration system. Aroclors (purity > 99%) were gifts from Dr. Stephen Safe’s laboratory at Texas A&M University (College Station, TX) (Mullins et al., 1984). Analytical standards for other test chemicals,ethylenediaminetetraacetic acid (EDTA), N - tris[Hydroxymethyl]methyl - 2 - aminoethanesulfonic acid (TES), calcium chloride, gelatin from porcine skin, murine Leukemia inhibiting factor (mLIF), 2-mercaptoethanol (99%), non-enzymatic cell dissociation solution, and tetraoctylammonium bromide (TAB, 98%) were purchased from Sigma Aldrich (St. Louis, MO). Cell culture-grade dimethyl sulfoxide (DMSO, 99%) was obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Fetal bovine serum (FBS), embryonic stem cell-qualified fetal bovine serum (ES-FBS), ES cell basal medium, and Dulbecco’s modified Eagle’s medium (DMEM) were from ATCC (Manassas, VA). Penicillin-streptomycin-glutamine (100X, Pen-Strep) and MEM non-essential amino acid solutions (100X, NEAA) were obtained from Thermo Scientific (Rockford, IL). CellTiter-Glo 2.0 reagent was from Promega (Madison, WI). Pluripotent mouse embryonic stem D3 (ESD3) cell line (catalogue# ATCC®CRL 1934TM) and 3T3-L1 cell line (catalogue# ATCC®CL-173) were purchased from ATCC (Manassas, VA). Human lung fibroblasts (HLF, catalogue# CC-2512) were obtained from Lonza (Walkersville, MD).

2.3. Chemical mixtures

Class-specific chemical mixtures were prepared by dissolving pure chemicals in 100% cell culture-grade DMSO to yield an equal concentration of each chemical, except for the mycotoxins, as shown in Table S1. The pesticide mixture contained equal concentrations at 2 mg/mL of pentachlorophenol, 2,4,6-trichlorophenol and lindane (organochlorines); diazinon and glyphosate (organophosphate); linuron (urea-type); trifluralin (dinitroaniline); aldicarb (carbamate); and paraquat (bipyridyl). The most commonly found Aroclors in the environment (i.e. Aroclor 1254 and 1260) were included in the Aroclor mixture. Plasticizer precursors such as bisphenol a, bisphenol s and bisphenol f, and phthalates such as di-n-butyl phthalate (DBP) and di(2-ethylhexyl) phthalate (DEHP) were included in the plasticizer mixture. PAH mixture encompassed widely used PAHs with different numbers and arrangement of rings, including benzo[a]pyrene, pyrene, naphthalene, and benzo[b]fluoranthene. Mycotoxin mixture consisted of those considered the most toxic and most frequently occurring in crops, i.e. aflatoxin and zearalenone, at a concentration ratio of 1:6. This ratio was based on the average contamination levels of aflatoxin and zearalenone reported from worldwide surveys of finished feed (Murugesan et al., 2015).

2.4. Chemical analysis

The separation and quantification of chemicals including diazinon, glyphosate, paraquat, aldicarb, benzo[a]pyrene, linuron, pentachlorophenol, 2,4,6-trichlorophenol, lindane, trifluralin, Aroclors, aflatoxin, and zearalenone were performed as previously described (Maki et al., 2017; Wang and Phillips, 2019; 2020; Wang et al., 2019a; b; c).

The separation of naphthalene, pyrene, and benzo[b]fluoranthene were achieved on HPLC with a Waters Symmetry C18 (150 × 4.6 mm, 5 μm) at 30°C (Waters, 2008; Zhao et al., 2008). The mobile phase was acetonitrile:water (60:40, v/v) at a flow-rate of 1.2 mL/min for naphthalene and benzo[b]fluoranthene, and acetonitrile:water (90:10, v/v) at a flow-rate of 1.0 mL/min for pyrene. The injection volume was 20 μL for each sample. The column effluent for pyrene was monitored by a UV detector at 240 nm wavelength and a fluorescence detector set at an excitation wavelength of 265 nm and an emission wavelength of 394 nm. The UV detection was performed at 220 nm for naphthalene and 256 nm for benzo[b]fluoranthene based on their maximum absorption. Fluorescence detection was also performed at excitation and emission wavelength at 277 nm and 330 nm for naphthalene, and 298 nm and 436 nm for benzo[b]fluoranthene. Breeze® software (Waters, Milford, MA) was used to control the HPLC system and collect data.

Plasticizers were analyzed using a Waters Acquity® ultra performance LC/MS/MS (Milford, MA) equipped with triple quadrupole and an Acquity® BEH C18 column (2.1 × 50 mm, 1.7 μm) at 40°C (Breemen et al., 2014; PerkinElmer, 2016; Shan et al., 2014). Mobile phase contained water with 0.1% formic acid (eluent A) and acetonitrile (eluent B). For bisphenols, a gradient elution at a flow rate of 0.3 mL/min was carried out at 10% to start (eluent B), 10-100% linear gradient from 0.5 to 6 min, and hold at 100% for 0.9 min. For phthalates, the LC gradient was set as 50% eluent B (initial), 50-98% (0.06-4 min), hold at 98% (4-8 min), and 98-50% (8-10 min). A sample volume of 10 μL was used for each analysis. MS analysis was performed with an electrospray ionization interface (ESI) and operated in a negative ion mode. The spray and cone voltages were maintained at 5 kV and 50 V for bisphenols, and 4.5 kV and 20 V for phthalates, respectively. Precursor and product ions of bisphenol a, bisphenol s, bisphenol f, DBP, and DEHP were monitored on the mass spectrometer at m/z 227 to 212/133, 249 to 108.1/155.9, 199.1 to 105.1/93, 279 to 205, and 391 to 167, respectively. The source temperature was kept at 120°C. The mass spectrometer was operated under multiple reaction monitoring mode. The unit mass resolution was used for ion mass analyzers. The enhanced product ion (EPI) scan rate was 1000 amu/s, and the scan range was 106 to 396 amu. Nitrogen gas was used as the collision and curtain gas, and argon gas was used as the nebulizer and heater gas. Empower® (Waters, Milford, MA) analyst software was used to control the LC/MS/MS system and acquire the data. To minimize the contamination, only glassware was used for all the plasticizer preparation.

Standard concentrations of chemicals were spiked from 0.005 mg/L to 20 mg/L in the mobile phase and validated using calibration curves (Table S1). Standard solutions were spiked before and after 2 hr of agitating to determine non-specific binding. The calibration solutions were prepared daily and verified before running test samples. The linearity (r²) in all cases was 1 > r² > 0.99, as shown in Table S1. The limit of detection (LOD) of all chemicals was determined based on a signal-to-noise ratio of 3 and ranged from 32 ppt to 1 mg/L depending on the sensitivity of the detection method (Table S1).

2.5. Adsorption isotherms

The toxicant stock solutions were individually dissolved in the mobile phase summarized in Table S1 to yield final solutions equal to 6 mg/L pentachlorophenol and 2,4,6-trichlorophenol, 5 mg/L paraquat, 12 mg/L glyphosate, 12.5 mg/L lindane, 20 mg/L linuron and trifluralin, 10 mg/L diazinon, plasticizers, pyrene, benzo[a]pyrene and naphthalene, 2 mg/L Benzo[b]fluoranthene, 15 mg/L Aroclors, 5 mg/L aldicarb, 8 mg/L aflatoxin and 4 mg/L zearalenone. The final concentrations were set based on the octanol-water partitioning coefficients so that precipitation was not a factor, and the optimal ratio of toxicant/sorbent to reach saturation (equilibrium) on isotherm plots. Then a concentration of 0.002% of sorbent was added to an increasing concentration gradient (5-100%) of chemical solution. In these studies, controls consisted of untreated solution (as listed in Table S1), chemical solution without sorbent, and sorbent suspension without a chemical. The control and test groups were capped and agitated at 1000 rpm on an IKA electric shaker (VIBRAX VXR basic, Werke, Germany) for 2 hr at ambient temperature (24°C). This duration was based on preliminary data suggesting that equilibrium of the surface interaction was reached within 30 min. All samples were then centrifuged at 2000 g for 20 min to separate the toxicant/sorbent complex from solution and were detected by either ultraviolet (UV)/visible scanning spectrophotometry, HPLC, or LC/MS/MS.

The amount of the chemical removed at each data point was calculated from the concentration difference between test and control groups. These data were then plotted using Table-Curve two-dimensional (2D) and a program developed using Microsoft Excel to derive values for the variable parameters. Langmuir and Freundlich models were used to plot equilibrium isotherms from triplicate analysis based on the best fit for the adsorption data. The Langmuir isotherm describes monolayer adsorption onto a surface with a finite number of identical sites and uniform energies of adsorption. The Langmuir equation was entered as user-defined functions:

$$\mathrm{Langmuir\; model}\text{q}=Q_{\max }(\frac{K_L{C}_W}{1+K_L{C}_W})$$ (1)

q = the amount of toxicant adsorbed (mol/kg), Q_max = maximum binding capacity (mol/kg), K_L = Langmuir distribution constant (L/mol), C_w = equilibrium concentration of toxicant (mol/L). To calculate the dimensionless K_e° from K_L, the following equation (Lima et al., 2019a) was used:

$${\mathrm{Ke}}^{\circ }=\frac{K_{L}x{[\mathit{adsorbate}]}^{\circ }}{\gamma }$$ (2)

γ is the coefficient of activity, [adsorbate]° is the standard concentration of the adsorbate = 1 mol/L, K_e° is the thermodynamic equilibrium constant (Lima et al., 2019a).

The Freundlich isotherm is used to describe the adsorption characteristics for a heterogeneous surface. The Freundlich model is represented by the following equation:

$$\mathrm{Freundlich\; model}\text{q}={{\text{K}}_{\text{f}}{\text{C}}_{\text{w}}}^{1/\text{n}}$$ (3)

K_f = Freundlich distribution constant, 1/n = degree of heterogeneity.

Adsorption parameters coupled with the Gibbs free energy equation were used to calculate free energy (ΔG°):

$$\Delta \text{G}=\Delta {\text{G}}^{\circ }+{\mathrm{RTlnKe}}^{\circ }$$ (4)

R (gas constant) = 8.314 J/mol/K, T (absolute temperature) = 273 + t (°C). ΔG will be zero for an adsorption system in equilibrium. A negative ΔG° means that the adsorption is thermodynamically favorable and will go in the forward adsorption direction (K_e° > 1). If ΔG° has a positive value, the adsorption process will not be favored and the adsorption will not be significant (K _e° < 1) (Lima et al., 2019b).

2.6. Cell-based assays

Cells were first plated in 25 cm² polystyrene cell culture flasks (Corning, NY) precoated with 0.1% gelatin with different mediums according to manufacturer introductions. Cells were maintained at 37°C with 5% CO₂ in a humidified atmosphere and routinely sub-cultured every 2-3 days using non-enzymatic cell dissociation solution. In detail, ESD3 (mouse embryonic multipotent stem cell) was cultured using ES cell basal medium with addition of 15% ES-FBS, 1% 100X PenStrep, 1% Non-Essential Amino Acid, and 7.8 μg/mL of 2-mercaptoethanol. ESD3 cells were kept undifferentiated by the addition of 1000 U/mL murine Leukemia inhibiting factor. HLF (human lung fibroblast cell) and 3T3-L1 (mouse embryonic fibroblast cell) were maintained with DMEM with the addition of 10% FBS and 1% 100X Pen-Strep.

For the experiments, cells were seeded in 96-well plates (Corning, NY) with a density at 2x10⁵ cells/mL in 100 μL medium (without mLIF for ESD3 cells) in each well for 24 hr prior to treatment. Stock solutions of the mixtures in 100% DMSO were diluted in a series of 200 to 2x10⁶ times in cell-specific culture media. The final solution contained less than or equal to 0.5% DMSO, which presented no significant effects on cell growth. Chemical mixtures at different concentrations together with 50 μM tetra-octyl ammonium bromide (TAB), a cytotoxicity positive control, were shaken at 1000 rpm for 2 hr at ambient temperature and centrifuged at 2000 g for 20 min. Cell responses to different concentrations of chemical mixtures were examined to determine the most sensitive and responsive cell lines. Subsequently, chemical mixtures at 200x dilution were treated with the addition of 0.1% sorbent for 2 hr at 1000 rpm and 24°C, and centrifuged for 20 min at 2000 g. Then, 100 μL of the supernatants were collected and added to cells. After 24 hr exposure to mixtures with and without treatment, cell viability was determined based on CellTiter-Glo assay. In detail, 100 μL of CellTiter-Glo reagents were added into each well and mixed for 2 min at 600 rpm to induce cell lysis. The plates were kept at ambient temperature for 10 min to stabilize the luminescent signal. Then luminescence was recorded using FLIPR tetra (Molecular Devices, San Jose, CA).

2.7. H. vulgaris assay

H. vulgaris were obtained from Environment Canada (Montreal, Qc) and maintained in hydra medium (consisting of 4 mg/L EDTA, 115 mg/L TES, 147 mg/L calcium chloride adjusted to pH 6.9-7.0) at 18°C. Using a hydra classification method, the morphology of hydra was rated over time as an indicator of solution toxicity. The morphological scoring of hydra was objective and repeatable (Dash and Phillips, 2012). The hydra response was scored after exposure to various concentrations of design mixtures in the classes of pesticides, Aroclors, plasticizers, PAHs, and mycotoxins. The assay included closely monitoring mature and non-budding hydra at 0, 4, 20, 28, 44, 68, and 92 hr, without changing solutions during testing. Hydra medium with 1% DMSO was included as a control. The concentration of the design mixture that resulted in 100% mortality of hydra in 92 hr was defined as the minimum effective dose (MED) and was further tested on hydra with sorbent inclusion. Individual sorbents and a sorbent mixture of an equal amount of AC, APM, and amended clays were added to the mixture solutions at 0.1% inclusion rate. All solutions were mixed at 1000 rpm for 2 hr and centrifuged at 2000 g for 20 min prior to exposure of hydra in Pyrex dishes. Three hydra colonies in each group were exposed to 4 mL of test medium at 18°C. The average score for each group was used to determine the toxicity rating at each time point.

2.8. Statistical analysis

Negative controls (cell culture medium, vehicle, or hydra medium with 1% DMSO) and a positive control (TAB) were included in all cell-based (independently performed) and hydra-based assays in triplicate. One-way analysis of variance (ANOVA) was used to analyze the statistical significance using GraphPad Prism 8.2.1 (San Diego, CA), where p < 0.05 represents significant difference. Protection efficiencies were calculated by the change of toxicity of mixtures after sorbent treatment. Both Pearson and Spearman correlations were reported to compare the results from in vitro cell models and the in vivo H. vulgaris assay.

3. RESULTS AND DISCCUSSION

3.1. Adsorption analysis

Equilibrium isotherms were used to describe adsorption isotherms for each test sorbent. Adsorption parameters, coupled with the Gibbs free energy equation, were used to calculate free energy (ΔG° in kJ/mol). Thermodynamic free energy indicated the spontaneity of the adsorption process. Table 1 summarized the binding parameters described by the best fit model of individual chemicals on various sorbents. All of the adsorption isotherms fit either the Langmuir or the Freundlich model with r² > 0.80.

Table 1.

Summary of chemical binding parameters and correlation coefficients from adsorption isotherms

Chemical	Sorbents	Binding model	r2	Binding parameters
Pentachlorophenol	AC	Langmuir	0.88	Qmax=0.31 mol/kg; KL=6.20E5 L/mol; ΔG°=−32.9 kJ/mol
	Montmorillonite	Freundlich	0.86	Kf=2.43E4; 1/n=1.30
	APM	Langmuir	0.85	Qmax=0.24 mol/kg; KL=1.17E6 L/mol; ΔG°=−34.5 kJ/mol
	Mont-carnitine	Freundlich	0.84	Kf=2.56E4; 1/n=1.35
	Mont-choline	Freundlich	0.82	Kf=1.23E4; 1/n=1.10
2,4,6-trichloriphenol	AC	Langmuir	0.89	Qmax=0.37 mol/kg; KL=8.80E5 L/mol; ΔG°=−33.8 kJ/mol
	Montmorillonite	Freundlich	0.86	Kf=1.19E2; 1/n=0.63
	APM	Langmuir	0.95	Qmax=0.25 mol/kg; KL=8.07E5 L/mol; ΔG°=−33.6 kJ/mol
	Mont-carnitine	Langmuir	0.83	Qmax=0.09 mol/kg; KL=4.00E5 L/mol; ΔG°=−31.8 kJ/mol
	Mont-choline	Langmuir	0.84	Qmax=0.13 mol/kg; KL=6.57E5 L/mol; ΔG°=−33.1 kJ/mol
Lindane	AC	Langmuir	0.95	Qmax=0.83 mol/kg; KL=7.78E4 L/mol; ΔG°=−27.8 kJ/mol
	Montmorillonite	Freundlich	0.85	Kf=6.20E4; 1/n=1.41
	APM	Langmuir	1	Qmax=0.53 mol/kg; KL=1.31E5 L/mol; ΔG°=−34.8 kJ/mol
Diazinon	AC	Langmuir	0.99	Qmax=0.63 mol/kg; KL=1.02E6 L/mol; ΔG°=−34.2 kJ/mol
	Montmorillonite	Langmuir	0.93	Qmax=0.19 mol/kg; KL=3.61E6 L/mol; ΔG°=−37.3 kJ/mol
	APM	Langmuir	0.94	Qmax=0.47 mol/kg; KL=1.57E6 L/mol; ΔG°=−35.2 kJ/mol
	Mont-carnitine	Langmuir	0.96	Qmax=0.25 mol/kg; KL=3.41E5 L/mol; ΔG°=−37.1 kJ/mol
	Mont-choline	Langmuir	0.89	Qmax=0.34 mol/kg; KL=8.43E5 L/mol; ΔG°=−33.7 kJ/mol
Glyphosate	AC	Langmuir	0.90	Qmax=0.55 mol/kg; KL=5.64E5 L/mol; ΔG°=−32.7 kJ/mol
	Montmorillonite	Langmuir	0.95	Qmax=0.32 mol/kg; KL=2.44E5 L/mol; ΔG°=−30.6 kJ/mol
	APM	Langmuir	0.96	Qmax=0.43 mol/kg; KL=1.83E5 L/mol; ΔG°=−29.9 kJ/mol
	Mont-carnitine	Langmuir	0.85	Qmax=0.40 mol/kg; KL=1.59E6 L/mol; ΔG°=−35.3 kJ/mol
	Mont-choline	Langmuir	0.80	Qmax=0.42 mol/kg; KL=3.88E5 L/mol; ΔG°=−31.8 kJ/mol
Linuron	AC	Langmuir	0.93	Qmax=0.17 mol/kg; KL=3.88E4 L/mol; ΔG°=−26.1 kJ/mol
	Montmorillonite	Freundlich	0.97	Kf=2.15E5; 1/n=1.40
	APM	Freundlich	0.97	Kf=2.67E4; 1/n=1.16
	Mont-carnitine	Freundlich	0.99	Kf=1.60E4; 1/n=1.26
	Mont-choline	Freundlich	0.98	Kf=4.32E3; 1/n=1.15
Paraquat	AC	Langmuir	0.92	Qmax=0.45 mol/kg; KL=7.66E5 L/mol; ΔG°=−33.5 kJ/mol
	Montmorillonite	Langmuir	0.87	Qmax=0.29 mol/kg; KL=2.80E6 L/mol; ΔG°=−36.6 kJ/mol
	APM	Langmuir	0.93	Qmax=0.24 mol/kg; KL=5.11E6 L/mol; ΔG°=−38.1 kJ/mol
	Mont-carnitine	Langmuir	0.83	Qmax=0.38 mol/kg; KL=8.07E6 L/mol; ΔG°=−39.3 kJ/mol
	Mont-choline	Langmuir	0.93	Qmax=0.22 mol/kg; KL=7.35E6 L/mol; ΔG°=−39.0 kJ/mol
Aldicarb	AC	Langmuir	0.94	Qmax=0.49 mol/kg; KL=3.07E5 L/mol; ΔG°=−31.2 kJ/mol
	Montmorillonite	Freundlich	0.95	Kf=3.34E5; 1/n=1.24
	APM	Langmuir	0.88	Qmax=0.24 mol/kg; KL=2.97E6 L/mol; ΔG°=−36.8 kJ/mol
	Mont-carnitine	Langmuir	0.97	Qmax=0.28 mol/kg; KL=1.76E6 L/mol; ΔG°=−35.5 kJ/mol
	Mont-choline	Langmuir	0.89	Qmax=0.26 mol/kg; KL=1.83E6 L/mol; ΔG°=−35.6 kJ/mol
Trifluralin	AC	Langmuir	0.88	Qmax=0.31 mol/kg; KL=6.20E5 L/mol; ΔG°=−38.6 kJ/mol
	Montmorillonite	Langmuir	0.89	Qmax=0.06 mol/kg; KL=6.70E4 L/mol; ΔG°=−27.4 kJ/mol
	APM	Langmuir	0.93	Qmax=0.15 mol/kg; KL=3.23E4 L/mol; ΔG°=−25.6 kJ/mol
	Mont-carnitine	Freundlich	0.95	Kf=8.87E4; 1/n=1.26
	Mont-choline	Freundlich	0.91	Kf=4.16E5; 1/n=1.42
Bisphenol A	AC	Langmuir	0.99	Qmax=0.52 mol/kg; KL=8.85E5 L/mol; ΔG°=−33.8 kJ/mol
	Montmorillonite	Langmuir	0.84	Qmax=0.25 mol/kg; KL=9.98E5 L/mol; ΔG°=−34.1 kJ/mol
	APM	Langmuir	0.99	Qmax=0.26 mol/kg; KL=1.44E6 L/mol; ΔG°=−35.0 kJ/mol
	Mont-carnitine	Langmuir	0.95	Qmax=0.29 mol/kg; KL=7.78E5 L/mol; ΔG°=−33.5 kJ/mol
	Mont-choline	Langmuir	0.90	Qmax=0.24 mol/kg; KL=2.10E6 L/mol; ΔG°=−35.9 kJ/mol
Bisphenol S	AC	Langmuir	0.92	Qmax=0.17 mol/kg; KL=2.37E5 L/mol; ΔG°=−30.6 kJ/mol
	Montmorillonite	Freundlich	0.94	Kf=2.90E3; 1/n=1.00
	APM	Freundlich	0.81	Kf=1.40E2; 1/n=1.90
	Mont-carnitine	Freundlich	0.99	Kf=4.53E4; 1/n=1.31
	Mont-choline	Freundlich	0.99	Kf=1.59E4; 1/n=1.20
Bisphenol F	AC	Langmuir	0.98	Qmax=0.39 mol/kg; KL=3.51E4 L/mol; ΔG°=−25.8 kJ/mol
	Montmorillonite	Freundlich	0.85	Kf=1.26E4; 1/n=1.17
	APM	Langmuir	0.89	Qmax=0.13 mol/kg; KL=5.35E4 L/mol; ΔG°=−26.9 kJ/mol
	Mont-carnitine	Freundlich	0.84	Kf=3.69E3; 1/n=1.08
	Mont-choline	Freundlich	0.98	Kf=1.35E5; 1/n=1.37
DBP	AC	Freundlich	0.88	Kf=6.83E2; 1/n=0.95
	Montmorillonite	Langmuir	0.85	Qmax=0.01 mol/kg; KL=2.31E5 L/mol; ΔG°=−30.5 kJ/mol
	APM	Freundlich	0.97	Kf=2.71E4; 1/n=1.25
	Mont-carnitine	Langmuir	0.96	Qmax=0.03 mol/kg; KL=2.24E5 L/mol; ΔG°=−30.4 kJ/mol
	Mont-choline	Langmuir	0.93	Qmax=0.05 mol/kg; KL=1.46E5 L/mol; ΔG°=−29.4 kJ/mol
DEHP	AC	Freundlich	0.97	Kf=1.12E2; 1/n=0.71
	Mont-carnitine	Freundlich	0.84	Kf=1.91E4; 1/n=1.22
	Mont-choline	Freundlich	0.95	Kf=1.25E3; 1/n=0.90
Benzo[a]pyrene	AC	Langmuir	0.93	Qmax=0.69 mol/kg; KL=2.39E6 L/mol; ΔG°=−36.3 kJ/mol
	Montmorillonite	Freundlich	0.85	Kf=8.91E2; 1/n=0.97
	APM	Langmuir	0.90	Qmax=0.23 mol/kg; KL=2.41E6 L/mol; ΔG°=−36.3 kJ/mol
	Mont-carnitine	Langmuir	1	Qmax=0.09 mol/kg; KL=8.31E4 L/mol; ΔG°=−28.0 kJ/mol
	Mont-choline	Langmuir	0.96	Qmax=0.09 mol/kg; KL=9.01E4 L/mol; ΔG°=−28.2 kJ/mol
Pyrene	AC	Langmuir	0.98	Qmax=2.07 mol/kg; KL=5.33E4 L/mol; ΔG°=−26.9 kJ/mol
	Montmorillonite	Langmuir	0.95	Qmax=1.77 mol/kg; KL=1.08E5 L/mol; ΔG°=−28.6 kJ/mol
	APM	Langmuir	0.98	Qmax=1.14 mol/kg; KL=7.82E4 L/mol; ΔG°=−27.8 kJ/mol
	Mont-carnitine	Langmuir	0.94	Qmax=1.91 mol/kg; KL=8.46E4 L/mol; ΔG°=−28.0 kJ/mol
	Mont-choline	Langmuir	0.92	Qmax=1.29 mol/kg; KL=2.61E5 L/mol; ΔG°=−30.8 kJ/mol
Naphthalene	AC	Freundlich	0.92	Kf=5.46E2; 1/n=0.78
	Montmorillonite	Langmuir	0.94	Qmax=0.45 mol/kg; KL=8.90E3 L/mol; ΔG°=−22.5 kJ/mol
	APM	Freundlich	0.98	Kf=3.43E5; 1/n=1.48
	Mont-carnitine	Freundlich	0.88	Kf=6.36E2; 1/n=0.87
	Mont-choline	Langmuir	0.81	Qmax=0.18 mol/kg; KL=1.55E4 L/mol; ΔG°=−23.8 kJ/mol
Benzo[b]fluoranthene	AC	Langmuir	0.94	Qmax=0.38 mol/kg; KL=1.34E7 L/mol; ΔG°=−40.5 kJ/mol
	Montmorillonite	Freundlich	0.93	Kf=6.48E3; 1/n=1.17
	APM	Freundlich	0.95	Kf=7.28E6; 1/n=1.62
	Mont-carnitine	Freundlich	0.87	Kf=7.04E2; 1/n=0.84
Aflatoxin	AC	Langmuir	0.89	Qmax=0.43 mol/kg; KL=7.06E5 L/mol; ΔG°=−33.3 kJ/mol
	Montmorillonite	Langmuir	0.86	Qmax=0.37 mol/kg; KL=1.17E6 L/mol; ΔG°=−34.5 kJ/mol
	APM	Langmuir	0.95	Qmax=0.34 mol/kg; KL=1.17E6 L/mol; ΔG°=−34.5 kJ/mol
	Mont-carnitine	Langmuir	0.94	Qmax=0.39 mol/kg; KL=1.40E6 L/mol; ΔG°=−34.9 kJ/mol
	Mont-choline	Langmuir	0.94	Qmax=0.29 mol/kg; KL=4.52E5 L/mol; ΔG°=−32.1 kJ/mol
Zearalenone	AC	Langmuir	0.84	Qmax=0.82 mol/kg; KL=9.44E5 L/mol; ΔG°=−34.0 kJ/mol
	Montmorillonite	Freundlich	0.90	Kf=2.91E4; 1/n=1.13
	APM	Langmuir	0.95	Qmax=0.28 mol/kg; KL=1.70E6 L/mol; ΔG°=−35.4 kJ/mol
	Mont-carnitine	Freundlich	0.97	Kf=1.85E3; 1/n=0.84
	Mont-choline	Freundlich	0.96	Kf=2.18E4; 1/n=1.08

Q_max, binding capacity; K_L, binding affinity; r², correlation coefficients; 1/n, degree of heterogeneity; K_f, Freundlich distribution constant; ΔG°, free energy

The binding interactions of individual pesticides including pentachlorophenol, 2,4,6-trichlorophenol, lindane, diazinon, glyphosate, linuron, trifluralin, aldicarb, and paraquat on the surfaces of five sorbent materials (AC, parent calcium montmorillonite, acid-processed montmorillonite, and montmorillonite amended with L-carnitine or choline), are shown in Table 1. For all pesticides, AC showed the highest binding in terms of: 1) good fit to the Langmuir model based on correlation coefficients, 2) highest binding capacities compared to all other sorbents, 3) high binding affinity, and 4) high absolute values of free energy of adsorption. The Langmuir isotherm describes monolayer adsorption onto a surface with a finite number of identical sites and uniform energies of adsorption. It indicates that all adsorption sites have equal adsorbate affinity and that adsorption at one site does not affect adsorption at an adjacent site, which results in saturable and homogeneous binding sites on the surface of AC. The high K_L values derived from the Langmuir model reflected a preference for the chemicals to be bound on AC surfaces rather than dissolved in the solution (Deutsch, 1997).

Contrary to AC, pesticide binding on montmorillonite surfaces, except for glyphosate and paraquat, showed a good fit for the Freundlich model describing heterogeneous binding. This difference was possibly due to the hydrophilicity of glyphosate (logP = −4.0) and paraquat (logP = −4.5), facilitating their attraction to the interlayer surfaces of parent montmorillonite. These surfaces are hydrophilic as suggested by less adsorption of n-heptane versus adsorption of water with a ratio of 0.41 (Figure 1A). Although the binding of organophilic diazinon and trifluralin on montmorillonite also showed a Langmuir trend, its binding capacities were relatively low compared to AC and other montmorillonite-based sorbents. This result indicated that montmorillonite was more efficient in adsorbing hydrophilic chemicals, while partitioning activity may be the major binding mode for hydrophobic chemicals. The activation of clay by acid enhanced the binding performance for APM, which showed the second-highest binding capacity for organochlorines (pentachlorophenol, 2,4,6-trichlorophenol, and lindane), organophosphates (diazinon and glyphosate), carbamate (aldicarb) and dinitroaniline (trifluralin). Compared to parent clay, APM showed decreased swelling in water, increased surface area by almost 50%, and decreased concentrations of trace metals, including framework aluminum, and interlayer calcium and sodium (Wang and Phillips, 2019). Furthermore, SEM results show that the acid treatment modified the original layered (flaky) structure of the parent clay (Figure 1B) to heterogeneous surfaces with irregular layers and enhanced porosity (Figure 1C and 1D). These properties of APM have been shown to enhance its porosity and toxicant binding activity. Although nutrient-amended clays were less effective in binding pesticides as the broad-acting AC and APM, their binding efficacy was enhanced for certain pesticides, compared to the parent clay. For example, the adsorption of 2,4,6-trichlorophenol and aldicarb changed from Freundlich partitioning activity on parent montmorillonite to Langmuir activity with saturable binding sites, capacities, increased affinities and free energy of adsorption. Diazinon and glyphosate showed Langmuir binding on homogeneous montmorillonite surfaces, and amendments with carnitine and choline further increased the clay’s binding capacities (Q_max = 0.25 and 0.34 versus 0.19 mol/kg for diazinon, and Q_max = 0.40 and 0.42 versus 0.32 mol/kg for glyphosate). This enhanced binding may be a result of interlayer carnitine and choline that contain hydrophobic surfaces which attract more n-heptane than water, with ratios equal to 1.03 and 1.26, respectively (Figure 1A); it may also be due to increased interlayer spacing as suggested by SEM (Figure 1E and 1F) and XRD (Celis et al., 2007; Velazquez and Deng, 2020).

Figure 1.

Ratio of the adsorption capacities for water and n-heptane onto sorbent materials (A). SEM of montmorillonite (B), APM (C, D), mont-carnitine (E), and mont-choline (F).

The adsorption data for all bisphenols on AC was best fit by the Langmuir model, indicating the presence of homogeneous and saturable binding sites on AC surfaces. The binding capacities of bisphenol a, bisphenol s, and bisphenol f on surfaces of AC were equal to 0.52, 0.17, and 0.39 mol/kg, respectively, indicating that AC was the most effective sorbent for reducing concentrations of bisphenols in the solution. Contrary to bisphenols, the adsorption of phthalates showed more efficient sorption onto nutrient-amended clays, based on the Langmuir model and high Q_max for DBP, and high affinities for DEHP. This enhanced binding onto surfaces of amended clays possibly resulted from hydrogen bonding between phthalates and quaternary ammonium on organic cations, as described by molecular dynamics simulations (Orr et al., 2020).

Frequently detected PAHs containing different ring numbers were included in the isothermal analysis. Among all tested sorbents, AC remained the best sorbent for most of the hydrophobic PAHs with large ring numbers (i.e., benzo[a]pyrene, pyrene, and benzo[b]fluoranthene). The adsorption of these PAHs onto AC fit the Langmuir model and showed the highest Q_max, K_L, and ΔG° values, indicating that AC provided sufficient and homogeneous binding sites facilitating the sufficient and tight binding of these PAHs. However, the binding of naphthalene with only 2 benzene rings showed the highest attraction to parent montmorillonite with a Q_max value equal to 0.45 mol/kg, while all other sorbents showed Freundlich binding trends for naphthalene with relatively low affinity values. This unique binding of naphthalene possibly resulted from its relatively low hydrophobicity (logP = 3.3), compared to other highly hydrophobic PAHs, and thus facilitated its attraction to the more hydrophilic surfaces on parent montmorillonite.

Aflatoxins are the most potent and widely distributed mycotoxins occurring in food and feed. Parent montmorillonite clay binding of aflatoxin showed high capacity, affinity, and free energy as described by the Langmuir model. Furthermore, montmorillonite clay was shown to be effective in reducing aflatoxin biomarkers and residues when a low level of montmorillonite was included in the diet of animals and humans (Phillips et al., 2019). Other sorbents also maintained similar binding of aflatoxins. Parent montmorillonite clay was not a good binder for zearalenone, but AC and APM showed saturable isotherms with high binding capacities, affinities and free energy of adsorption. This result suggests that AC and APM can serve as effective mycotoxin binders, especially for the more hydrophobic zearalenone.

3.2. Cell-based assays

Chemicals from different classes, including pesticides, Aroclors, plasticizers, PAHs, and mycotoxins, representing common environmental pollutants, were combined into class-specific mixtures and tested in a set of in vitro cell models. Cytotoxicity of each mixture was evaluated using a wide range of concentrations with a series of dilutions of 200x to 2x10⁶x from the stock solution, and exposed to various cell models. The concentration-response curves for HLF (human lung fibroblasts), ESD3 (mouse embryonic multipotent stem cell), and 3T3 (mouse embryonic fibroblast cell line) in Figure 2 showed that all design mixtures had adverse effects on cell viability, indicating that these cells are susceptible to the toxicity of design mixtures. Specifically, ESD3 and 3T3 cell lines showed high susceptibility to several mixtures (especially pesticides and plasticizers) at concentrations equal to 200x dilution from stock solutions, presented as 100% dilution factor on the x-axis. This resulted in greater than 50% cytotoxicity compared to vehicle control (adjusted to 100%). HLF showed more resilience (25% cytotoxicity) to mixtures at the same concentrations. Cytotoxicity of TAB was also evaluated in each experiment as a positive control with consistent EC₅₀ values to ensure the reproducibility of the assay (Figure S1).

Figure 2.

Concentration-response curves of design mixtures in HLF, ESD3, and 3T3 (top panel). Mixtures with a series of dilutions at 200x to 2x10⁶x from stock solutions are presented as 100% to 0.01% dilution factor on the x-axis. Cell viability of mixtures at 200x dilution (100% dilution factor) with comparison to the vehicle control (VEH) (bottom panel) (# p < 0.05).

Following the concentration-responses of different mixtures in each cell line, the effects from the addition of different sorbents were evaluated. Individual sorbents including AC, montmorillonite, APM, mont-carnitine, mont-choline, and a sorbent mixture were suspended in cell medium at 0.1% inclusion rate (1 mg/mL). As shown in Figure S2, all sorbents at 0.1% inclusion were not toxic to cells. To investigate the binding efficacy of sorbent treatment, each sorbent was included at 0.1% with the mixtures at 200x dilution and agitated for 2 hr. Supernatants with remaining unbound toxicants were exposed to cells for 24 hr before cell viability was detected by luminescence, as shown in Figure S3. Protection percentage of sorbent treatment in each cell model was calculated as the ratio of reduced toxicity in sorbent treatment groups versus the total toxicity of design mixtures (Table 2).

Table 2.

Protection of cells from chemicals by sorbents and a sorbent mixture

HLF
	Pesticides	Aroclors	Plasticizers	PAHs	Mycotoxins
AC	48±4.6%**	20±8.5%	27±5.0%*	27±17%	74±15%**
Montmorillonite	0%	69±11%**	1.4±4.8%	38±19%	22±15%
APM	29±5%**	98±40%*	18±5.8%	25±3.9%*	86±7.9%**
Mont-carnitine	0.67±6.5%	20.14±32%	6.7±10%	20±19%	75±22%*
Mont-choline	0.36±9.0%	31±30%	55±6.5%**	1.4±10%	56±11%*
Sorbent mixture	24±1.9%**	62±17%*	32±2.0%**	5.5±20%	19±2.7%
ESD3
	Pesticides	Aroclors	Plasticizers	PAHs	Mycotoxins
AC	70±4.0%**	6.8±13%	26±9.5%*	59±9.4%**	67±4%
Montmorillonite	12±6.6%	17±80%	9.07±6.26%	52±22%*	9.2±8.5%
APM	11±8.4%	54±116%	11.05±2.3%**	48±17%*	12.0±15%
Mont-carnitine	14±6.9%*	2.7±17%	12.97±3.9%*	15±4.6%	6.0±21%
Mont-choline	21±7.2%*	2.6±22%	16.76±4.9%*	30±4.7%	31±7.5%
Sorbent mixture	56±4.6%**	7.3±50%	24±2.5%**	72±7.8%**	33±19%
3T3
	Pesticides	Aroclors	Plasticizers	PAHs	Mycotoxins
AC	63±1.8%**	16±4.8%	44±1.5%**	66±1.2%**	78±2.0%**
Montmorillonite	16±4.5%**	40±0.50%**	13±4.5%**	19±7.1%*	37±7.2%*
APM	32±0.77%**	62±3.0%**	34±4.1%**	29±0.43%**	87±4.4%**
Mont-carnitine	17±2.6%**	0.11±8.0%	3.3±5.2%	9.6±3.1%**	5.1±10%
Mont-choline	25±2.9%**	6.7±9.6%	53±5.1%**	23±2.3%**	15±16%
Sorbent mixture	88±1.0%**	6.2±17%	50±1.3%**	25±1.7%**	20±9.5%

Data presented as average ± standard deviation.

^*p<0.05

^**p<0.01

For the pesticide mixture, AC showed the most significant enhancement in cell viability (Figure S3) with a protection efficacy (Table 2) of 48%, 70%, and 63% in HLF, ESD3, and 3T3, respectively (p < 0.01). This is consistent with isothermal data that AC was the most efficient sorbent for all pesticides with the highest binding capacity and affinity derived from the Langmuir model. APM also showed significant protection against pesticides, especially in HLF and 3T3 (p < 0.01), correlating with its high binding efficacy in isotherms from individual pesticides. Compared to the very limited protection from parent montmorillonite against the pesticide mixture, amended montmorillonite clays showed increased binding efficacy in terms of higher viability and protection percentage in all 3 cell lines. This enhancement possibly resulted from the higher binding capacity for diazinon and glyphosate and tighter binding for trichlorophenol and aldicarb, based on their isotherms. The sorbent mixture containing an equal ratio of AC, APM, and mont-choline, displayed the highest protection percentage equal to 88% (p < 0.01) in 3T3, which was higher than the individual sorbent component. This result supports the possibility that a sorbent mixture can be delivered as a treatment strategy during high levels of exposures.

For the Aroclor mixture, APM was shown to be the most effective sorbent resulting in the highest cell viability and protection percentage of 98% for HLF, 54% for ESD3, and 62% for 3T3. This result is consistent with previous studies that APM showed the highest binding capacity, affinity and enthalpy for coplanar and non-coplanar PCB congeners, and that low inclusion levels of APM in a shellfish diet could significantly reduce PCB uptake and residues in oysters (Wang and Phillips, 2020). Parent montmorillonite also showed high protection against Aroclor toxicity with 69%, 17%, and 40% in HLF, ESD3, and 3T3 cells, respectively. This result is supported by the Langmuir binding of individual PCBs onto montmorillonite with higher Q_max values than AC and other clays, and its moderate reduction of PCB residues in oysters. Additionally, AC, mont-choline, and the sorbent mixture were consistently effective in binding the plasticizer mixture and reducing its toxicity in all 3 cell lines. This is supported by isotherms for individual plasticizers, suggesting that AC was the best sorbent for bisphenols with high Q_max and K_L values; while, mont-choline showed the highest affinities for phthalates. For the PAH mixture, AC consistently delivered significantly enhanced cell viability and protection at 27%, 59%, and 66% in HLF, ESD3, and 3T3, respectively. This is possibly due to its high binding capacity for benzo[a]pyrene, pyrene, and benzo[b]fluoranthene, based on Langmuir isotherms. Montmorillonite clay also delivered significant protection against PAHs at 38% in HLF and 52% in ESD3, possibly because it was the best sorbent for naphthalene. The mycotoxin mixture mainly contained zearalenone due to its more prevalent presence in crops and feed products. As expected from isotherm studies, AC and APM showed the highest protection against the mycotoxin mixture in all 3 cell lines. Specifically, the cell viability of AC and APM treatment against mycotoxins showed no significant difference from the vehicle control group in HLF.

Despite distinctions and variations in the 3 cell lines, AC consistently showed the highest protection against pesticides, plasticizers, PAHs, and mycotoxins, while APM was the best sorbent for pesticides, Aroclors, and mycotoxins. Additionally, parent montmorillonite clay moderately increased cell viability in response to Aroclors and PAHs, and choline-amended montmorillonite was the most efficient sorbent against plasticizer toxicity. This protection in cells agrees with the binding parameters derived from the isotherms of individual chemicals. The cell assay results suggest that sorbents included in the therapeutic treatment can be adjusted based on exposures to class-specified chemical mixtures.

3.3. H. vulgaris assay

H. vulgaris is very sensitive to environmental toxicants and has been widely used as an indicator of toxicity. Design mixtures at a concentration gradient were dissolved in hydra medium with 1% DMSO and exposed to hydra for 92 hr. As shown in the left panel in Figure 3, the MEDs for pesticides, Aroclors, plasticizers, PAHs, and mycotoxins were equal to 2 mg/L, 80 mg/L, 1 mg/L, 30 mg/L, and 6 mg/L zearalenone + 1 mg/L aflatoxin, respectively. Therefore, mixtures at MEDs were treated with 1% of each sorbent, including AC, montmorillonite, APM, carnitine and choline amended montmorillonite, and the sorbent mixture (Figure 3). Protection percentage was calculated by the same equation in the cell assay as the ratio of reduced toxicity versus the total toxicity of the mixture (Table 3).

Figure 3.

Hydra toxicity from design mixtures at different concentrations (left panel) and protection against toxicity of design mixture MEDs by 0.1% sorbents (right panel). Hydra medium with 1% DMSO was included as a control. Morphological scores were determined at 0, 4, 20, 28, 44, 68, and 92 hr, and presented as average ± standard deviation (n=3). A, activated carbon; B, montmorillonite; C, APM; D, mont-carnitine; E, mont-choline; F, sorbent mixtures.

Table 3.

Protection of hydra from chemicals by sorbents and a sorbent mixture

	Pesticides	Aroclors	Plasticizers	PAHs	Mycotoxins
AC	90%**	14%**	30%*	90%**	100%**
Montmorillonite	30%**	38±5.8%*	10%	43±5.8%**	15±7.1%
APM	67±5.8%**	67±5.8%**	15±7.1%	53±5.8%**	100%**
Mont-carnitine	33±5.8%*	24±5.8%**	3.3±5.8%	33±5.8%*	17±5.8%*
Mont-choline	20%*	14%**	65±7.1%**	20%*	17±5.8%*
Sorbent mixture	100%**	29%**	30%*	53±5.8%**	97±5.8%**

Data presented as average ± standard deviation.

^*p < 0.05;

^**p < 0.01.

Among individual sorbents, AC and APM at 0.1% w/v inclusion displayed the highest protection of hydra that reduced toxicity of the pesticide mixture by 90% and 67%, respectively (p < 0.01) (Table 3). This result is consistent with adsorption isotherms of individual pesticides and in vitro cell assays where AC and APM showed the highest binding parameters and cell viability. Importantly, the sorbent mixture showed the most significant protection (100%) from the toxicity of the pesticide mixture, supporting the possibility of using sorbent mixtures for high-level exposures. Based on the isothermal and cell assay results, APM was the most effective binder for PCB congeners and Aroclors, followed by the parent montmorillonite clay. The hydra assay was consistent with these findings, where APM and montmorillonite delivered the most significant reduction of the Aroclor toxicity by 67% and 38%, respectively (p < 0.05). For the plasticizer mixture, mont-choline showed the highest protection of hydra with 65% reduction in toxicity (p < 0.01), as supported by its high binding capacity and affinity for phthalates and high cell viability. AC at the same inclusion rate also showed a high reduction rate of 30% against plasticizers (p < 0.05), corresponding to its high binding of bisphenols and high protection in 3 cell models. For the PAH mixture, AC delivered a significant reduction of toxicity by 90%, which was higher than any other sorbents (p < 0.01). This is possibly due to its high binding efficacy for PAHs with large ring numbers, as suggested by the isotherms. Consistent with isothermal and cell viability results, AC, APM, and the sorbent mixture were the most effective binders for the mycotoxin mixture, and significantly reduced toxicity in the hydra assay (p < 0.01). This result supports their high binding of both aflatoxin and zearalenone shown in the isotherms. Overall, the protection efficacy and reduction of toxicity in the hydra assay agree with isothermal parameters and cell assay results.

3.4. Correlation between in vitro and in vivo testing systems

In vivo and in vitro testing strategies were used to evaluate the toxicity of class-specific chemical mixtures and the protective effects of different sorbent treatments. All sorbents tested in this study showed no adverse effects on cells and hydra since no significant difference in cell viability and hydra morphology, compared to negative controls, were shown. To investigate the correlation and consistency in the different testing systems, protection efficacy in HLF, ESD3, 3T3, and hydra were calculated based on the changes of cell viability or hydra morphology, with and without, sorbent additions. The correlation between the in vitro cell models and in vivo hydra dataset was further compared using both Pearson and Spearman methods. In the heatmap (Figure 4), the correlation analysis indicated that each testing model was positively associated. The protection percentages in 3T3 and hydra models showed the highest correlation as suggested by both Pearson and Spearman with r = 0.84 and rho = 0.73, respectively, (p < 0.0001). Although ESD3 and HLF were less correlated to the hydra, with correlation coefficients of approximately 0.5, they still showed statistical significance (Figure S4). It is possible that the use of highly correlated in vitro and in vivo models can facilitate the evaluation of chemical mixtures and predict the efficacy of various remediation strategies.

Figure 4.

Correlation of sorbent treatment efficacy in tested organisms including in vitro cell models and the in vivo hydra model. Heatmaps show the overall correlations using Pearson (A) and Spearman (B) methods.

4. CONCLUSION

In this study, we characterized various parameters including surface capacity, binding affinity, heterogeneity of binding sites, and free energy of sorption derived from Langmuir or Freundlich models for representative environmental chemicals. The most effective and broad-acting sorbents were identified and selected based on isothermal results. Furthermore, the protective efficacy of broad-acting sorbents against potential toxicity of class-specific chemical mixtures was tested using in vitro mammalian models including HLF, ESD3, and 3T3 cell lines and an in vivo H. vulgaris assay. Results demonstrated that AC showed the highest efficacy for pesticides, plasticizers, PAHs, and mycotoxins. Parent montmorillonite clays were more effective for hydrophilic chemicals due to their ability to hydrate and their surface polarity. Acid-processed montmorillonite clay (APM) effectively protected against pesticides, Aroclors, and mycotoxins, possibly due to its multi-structural composition and diverse sites. Nutrient-amended clays were shown to be the most effective against plasticizers due to their increased interlayer spacing and their acquired hydrophobicity. The reduction of toxicity to cells and hydra was supported by their high binding activity in isotherms, suggesting positive correlations between adsorption isotherm parameters and reduction of toxicity in vitro and in vivo. Due to the difference in tested sorbents for different chemical classes, a sorbent mixture containing AC, APM, and mont-choline was also tested and this mixture showed broad efficacy. The combination of isothermal, cell-based, and hydra results suggest that edible sorbent inclusion in food and drinking water may be an effective mitigation strategy against the hazardous effects of environmental mixtures. Importantly, the positive correlation between the in vitro 3T3 cell model and the in vivo H. vulgaris model provides a novel method to predict the best sorbents for environmental chemicals and mixtures. Collectively, the highly correlated isothermal analysis, mammalian cells, and hydra assay determined the affinity of class-specific mixtures for various sorbents and provided insights into binding mechanisms.

Highlights.

• Broad-acting and edible sorbents for humans and animals were investigated

• Class-specific mixtures were PAHs, pesticides, Aroclors, plasticizers and mycotoxins

• Binding efficacy was tested using Langmuir/Freundlich, in vitro and in vivo models

• Adsorption isotherms, mammalian cells, and the hydra assay showed high correlation

• Medical grade carbon and nutrient-/acid-treated montmorillonite were broadly active