The effect of particle size on oral bioavailability and bioaccessibility of soil Ni from different sources.

The goal of the work was to contribute to a unified approach to assessing the risk to human health of soil ingestion, for contaminated sites with elevated [Ni]. Robust relationships between in vitro bioaccessibility and in vivo bioavailability of Ni in various soils, with mechanistic understanding, would enable site-specific assessments of human exposure through soil ingestion. Four soils (three ultramafic Brunisols with geogenic Ni and one Organic soil with anthropogenic Ni) were sieved into PS < 10 µm, 10-41 µm, 41-70 µm, 70-105 µm, 105-150 µm, and 150-250 µm, the [Ni]T for which ranged from 560 to 103000 mg/kg. Mass fraction-adjusted [Ni]IVBA (SBRC gastric) for each soil fraction was similar whether calculated for all particles <250 µm or <150 µm %NiIVBA ranged from 3% to 16% of [Ni]T and %NiABA (accumulated Ni in urine, kidneys, and small intestine of Sprague Dawley rats gavaged with a soil) ranged from 0% to 0.49%. The correlation between these two measurements was weak (R2 = 0.06). Multiple linear dose response relationships attributing variation in %NiABA to %NiIVBA plus soil physicochemical parameters known to influence trace element availability in soils were developed. As many soil properties measured in this study were highly correlated, ridge regression enabled a predictive relationship where the effect of each parameter was its true contribution to variation in %NiABA. Using a ridge constant (k) of 0.012, %NiABA could be predicted from %NiIVBA adjusted for soil absorptive entities (OrgC, and Fe oxides (negative coefficients)) and soil pH (positive coefficient). %NiABA predicted from this relationship was very close to 1:1 with the observed %NiABA except at the lowest observed values which were lower than predicted. This study shows that as the conditions increasingly favour soil Ni solubility, more of the Ni was bioavailable; this generalization was true regardless of particle size or soil origin.

Phytotoxicity effect concentrations (ECx) for Ce, Nd and Eu added to soil relative to total and bioaccessible soil REE concentrations, and tissue REE accumulations.

Toxicity thresholds (ECx) for radish, tomato, and durum wheat growth endpoints (shoot length, shoot mass, root length) to Ce, Nd or Eu added to a black organic soil were determined from 14-day dose-response growth assays. EC10 expressed as total soil [REE] had a more than twenty-fold range, from 337 mg/kg to more than >8000 mg/kg. Averaged over all REEs and endpoints, durum wheat was more tolerant than radish and tomato; and averaged over all endpoints, Eu appeared to be the most phytotoxic of the three REEs. Bioaccessibility of each REE was determined by extraction with 0.01 M CaCl2, which for all three REEs in this soil was quite low, <0.10% of total. However, bioaccessibility of Eu was five or six times greater than that for Ce and Nd, and thus could explain its apparently greater toxicity, namely that Eu was more likely to be accumulated at the site of toxic action in the plant. To discern inherent toxicity from enhanced bioaccumulation, concentration of each REE in root and shoot tissues was determined, for a tissue-residue approach to toxicity assessment. The EC10 expressed as tissue concentration was lower for Nd than for Ce and Eu, thus the most toxic of the three REEs. As for many cationic inorganic elements, toxicity varies with the chemistry of the exposure medium due to its effects on bioaccessibility. Simple methods to harmonize toxicity thresholds from different media enables greater integration into regulatory standards. When EC25 from this and other studies were normalized to CaCl2-extractable REE in their respective media, the range in Ce EC25 was reduced from 20-fold to 2.5-fold, and the range for Eu EC25 was reduced from 25-fold to 3-fold. This novel and low-input approach to meta-analysis of toxicity thresholds demonstrates the value of considering soil physico-chemical properties as modifiers of soil REE toxicity.

Derivation of a Ni bioaccessibility value for screening-level risk assessment of Ni substances in ingested materials including soils.

The objective of the present study was to derive a Ni bioaccessibility value for screening-level risk assessment of Ni substances in ingested materials including soils where multiple Ni substances are expected but not definitively identified. Broad ranges of Ni mass loading and dissolution time of a simple gastric assay were applied to pure Ni substances (removing the confounding factors of soil constituents on dissolution), thus broadening the applicability of the conclusions. The data were also used to support current knowledge of 'read across' for Ni substances. Release of Ni from pure manufactured Ni substances (Ni metal, NiO, NiSO4, Ni3S2, and NiS) was determined relative to Ni mass and substance surface area loading. Mass loadings ranged from 0.33 to 20.0 g Ni per L of 0.15 M HCl, and dissolution time ranged from 1 to 168 h. Proton exhaustion was indicated only at the highest loading (20 g/L) of NiO and Ni-M. Dissolution of substances other than NiSO4 was most likely limited by formation of intermediate products at the particle surface or particle agglomeration, impeding access to the principal Ni substance. The bioaccessibility of Ni for these substances was consistent with previously published data: substances other than NiSO4 were < 48% bioaccessible for a variety of gastric assays, which is much lower than all data for NiSO4, the usual reference substance. Thus, we suggest that Ni bioaccessibility data from gastric assays that are most relevant to human exposure can be relied upon to develop scientifically sound screening-level human health RA decisions for Ni contamination in soils and sediments in the absence of detailed Ni speciation.

Gastric bioaccessibility is a conservative measure of nickel bioavailability after oral exposure: Evidence from Ni-contaminated soil, pure Ni substances and Ni alloys.

Oral bioaccessibility (BAc) is a surrogate for the bioavailability (BAv) of a broad range of substances, reflecting the value that the approach offers for assessing oral exposure and risk. BAc is generally considered to have been validated as a proxy for oral BAv for the important soil contaminants Pb, Cd, and As. Here, using literature data for Ni BAc and BAv, we confirmed that Ni BAc (gastric only, with HCl mimicking stomach conditions) is a conservative measure of BAv for the oral exposure pathway. Measured oral BAv of Ni in soil was shown to be 50-100 times less than the simplest oral BAc estimates (%BAv = 0.012(%BAc) - 0.023 (r = 0.701, 95%CI [0.456, 0.847], n = 30)) in rats, demonstrating a significant conservatism for exposure assessment. The relationship between the oral BAv and BAc of nickel sulfate hexahydrate (NSHH) was comparable to that of soil, with measured oral BAv of NSHH (1.94%) being a small fraction of NSHH gastric BAc (91.1%). BAc and BAv reflect the underlying Ni speciation of the sample, with the bioaccessible leaching limits being represented by the highly soluble Ni salts and the poorly soluble Ni monoxide, and the environmental (e.g. soil properties) or gastric (e.g. food present) conditions. BAc has potential utility for chemical classification purposes because pure Ni substances can be grouped by %BAc values(using standardized methodologies for the relevant exposure routes), these groupings reflecting the underlying chemistry and speciation of the samples of substances tested here, with 0.008% %BAc for alloys (SS304, SS316, Inconel, Monel), <1% in green NiO and Ni metal massives, 0.9-23.6% for Ni powders, 9.8-22.7% for Ni sulfides, 26.3-29.6% for black oxidic Ni, and 82-91% for the soluble Ni salts. Oral BAc provides realistic yet conservative estimates of BAv for the hazard classification and risk assessment of Ni substances.

Modeling phytoremediation of aged soil Ni from anthropogenic deposition using Alyssum murale.

Several field-scale phytoextraction scenarios were created in a greenhouse study to investigate the feasibility of using Alyssum murale, to remediate three types of industrially Ni-contaminated soil (heavy clay, sand, organic muck) from Port Colborne, Ontario. The observed distribution of Ni mass between soil and aboveground vegetation was used in STELLA modeling software to predict timelines for the target soil Ni concentration, namely 1200 mg Ni/kg. Alyssum murale grown in sand would have a relatively constant pool of Ni available for plant uptake, which would not be the case for plants grown in organic muck and heavy clay. The maximum Ni extraction (%, plant Ni mass/soil Ni mass) was achieved in A. murale grown in unfertilized clay soil at the higher irrigation rate. Using these data, the STELLA model predicted that 246 years would be required to reduce soil Ni concentration in the most efficient combination of treatments to the remediation target. In addition, hypothetical A. murale Ni extraction in plant-soil systems optimized by manipulating soil chemistry and physical attributes, were modeled. The most optimized A. murale plant-soil systems for Ni extraction would require 9 years to achieve the same reduction, and it is not clear that this optimization can be achieved in the field. This study showed that phytoremediation using A. murale is not likely a time-sensitive approach for these soils.

Towards an exposure narrative for metals and arsenic in historically contaminated Ni refinery soils: Relationships between speciation, bioavailability, and bioaccessibility.

Archived soils contaminated with Ni, Cu, Co, and As from legacy operations of a nickel refinery at Port Colborne, Ontario, Canada were speciated using mineral liberation analysis. Four Ni minerals were identified as fingerprint compounds of the historical refinery emissions. Cu and Co were present in solid solution in these minerals due to their presence in the refinery's feed. The highest concentrations of Ni, Cu, Co, and As in these soils were 18,553, 1915, 196, and 79mg/kg, respectively, these elevated contaminant concentrations attesting to the importance of incidental soil ingestion to the oral exposure pathway in Port Colborne. The in vitro gastric bioaccessibility (BAc) was determined for these contaminants, as was in vivo oral bioavailability (BAv), using a mass balance approach in male Sprague-Dawley rats. In spite of the elevated soil concentrations of Cu, the BAv of this physiologically important metal could not be distinguished from that in commercial rat chow, suggesting low potential for exposure. Co and As also had low apparent BAv (<2%). For Ni, baseline oral BAv of naturally sourced dietary Ni was found to be approximately 2%, as was the oral BAv of Ni from nickel sulfate hexahydrate. The mass balances of NiSO4·6H2O were fully accounted-for in urine and feces after a single gavage dose, indicating little to no organ incorporation from this highly soluble salt. Therefore, the urinary estimates of Ni BAv for these soils were assumed to represent true BAv despite variable fecal recoveries. The high Ni concentrations enabled BAc-BAv relationships to be developed for these contaminated soils. For absolute bioavailability (ABA) and relative bioavailability (RBA) the relationships were: ABA=0.0116(BAc)-0.0479 and RBA=0.5542(BAc)-2.2817. These findings will advance the development of robust exposure narratives for soil metal contamination in Port Colborne and elsewhere.

Bioaccessibility estimates by gastric SBRC method to determine relationships to bioavailability of nickel in ultramafic soils.

Frameworks for human health risk assessment often include the opportunity to correct the estimate of exposure for bioavailability, which could be predicted from bioaccessibility. Lead and As are the only metallic elements for which bioavailability and bioaccessibility have been correlated across a spectrum of mineralogy and particle types. The objective of the present study is to correlate in vivo bioavailability with ex vivo bioaccessibility for elevated Ni in soils of ultramafic origin and explore attribution of any variation in this correlation to mineralogical characterization of the Ni. Ultramafic soils were field collected in British Columbia, CA. Rietveld quantitative X-ray diffraction was used for the characterization and quantification of crystalline materials containing Ni. Bioaccessible Ni was determined using the in vitro method developed by the Solubility/Bioaccessibility Research Consortium. Bioavailable Ni was determined by gavage dose of the soils to Sprague-Dawley rats. Urine and feces were collected every 24 h. At the end of 72 h, the animals were humanely sacrificed using carbon dioxide as per the approved animal care protocol. All organs were harvested, washed and preserved. Fecal elimination of gavaged Ni ranged from 35 to 95% including positive control. Relative bioavailability (RBA) ranged from 5 to 18%. In vitro bioaccessibility (IVBA) of soil Ni ranged from 0 to 17%; IVBA explained 86% of the variation in RBA. Normalizing both axes to soil olivine accounted for an additional 10% of the variation in RBA. For risk assessment of Ni contaminated soils, IVBA would be a useful and cost effective tool in estimating exposure of mammals through ingestion of soil particles, with some additional benefit of considering Ni mineralogy.

The influence of physicochemical parameters on bioaccessibility-adjusted hazard quotients for copper, lead and zinc in different grain size fractions of urban street dusts and soils.

When the hazard quotient for ingestion (HQI) of a trace element in soil and dust particles is adjusted for the element's bioaccessibility, the HQI is typically reduced as compared to its calculation using pseudo-total element concentration. However, those studies have mostly used bulk particles (<2 mm or <250 µm), and the reduction in HQI when expressed as bioaccessible metal may not be similar among particle size fractions, the possibility probed by the present study of street dusts and soils collected in Tehran. The highest Cu, Pb and Zn near-total concentrations occurred in the finest particles of dusts and soils. Bioaccessible concentrations of Cu, Pb and Zn in the particles (mg kg-1) were obtained using simple bioaccessibility extraction test (SBET). The bioaccessibility (%) did not vary much among near-total concentrations. In the bulk (<250 µm) sample, the bioaccessible concentration of Cu and Pb increased as the pH of sample increased, while Zn bioaccessibility (%) in the bulk particles was influenced by organic matter and cation exchange capacity. X-ray diffraction identified sulfide and sulfate minerals in all of the size-fractionated particles, which are insoluble to slightly soluble in acidic conditions and included most of the Cu and Pb in the samples. The only Zn-bearing mineral identified was hemimorphite, which would be highly soluble in the SBET conditions. The calculated HQI suggested potential non-carcinogenic health risk to children and adults from ingestions of soils and dusts regardless of particle size consideration, in the order of Zn > Pb ≥ Cu. The HQI calculated from near-total metal was not much different for particle size classes relative to bulk particles; however, the bioaccessibility percent-adjusted HQI for Pb was higher for the smaller particles than the bulk. This work is novel in its approach to compare HQI for a bulk sample of particles with its composite particle size fractions.

In vitro estimates of bioaccessible nickel in field-contaminated soils, and comparison with in vivo measurement of bioavailability and identification of mineralogy.

This study determined nickel (Ni) bioaccessibility in weathered smelter-contaminated soils, separately for particle-sized fractions using two in vitro methods: simulated gastrointestinal digestion (PBET) and PBET followed by absorption by Caco-2 cells. Relative bioavailability of Ni in soils was determined in vivo using rats, validating in vitro estimates; a mineralogical basis of variation in bioavailability/bioaccessibility among soils was explored. In vitro assays identified the same difference in bioaccessibility for Ni among particle size fractions. PBET estimates were more precise, thus likely to be more useful in discerning differences among soils. In vivo bioavailability for Ni was below limit of detection for the small soil particles, and 31% and 56% for the larger particles. The relative bioavailability calculated from this work suggests that risk from ingesting Ni-contaminated soils could be overestimated by between 2- and 50-fold if the estimates of exposure are not adjusted for the lower bioavailability of weathered Ni originating from smelter emissions. The overestimation that would occur by using total Ni is greatest for the particle size that is most likely to adhere to the hands of children, demonstrating the importance of particle-size separation of soils for bioavailability determination and risk assessment.

The uptake and metabolism of benzo[a]pyrene from a sample food substrate in an in vitro model of digestion.

Food ingestion is the major route of exposure to many hydrophobic organic contaminants (HOCs) such as benzo[a]pyrene (BaP). It has been proposed that food-bound HOCs may become bioavailable after their mobilization by gastrointestinal fluids. The purpose of this research was to measure the uptake efficiency of [(14)C]-BaP bound to skim milk powder using an in vitro model of gastrointestinal digestion followed by sorption to human enterocytes (Caco-2 cells). Neutralization of intestinal fluids released [(14)C]-BaP into the soluble fraction. Ageing of benzo[a]pyrene onto skim milk for 6 months significantly decreased the mobilized fraction but did not affect the amount of benzo[a]pyrene taken up into Caco-2 cells. Hence, significant differences in aqueous phase concentrations may not always be reflected in significant differences in uptake. We obtained evidence that the digestion/uptake of skim milk lipids is accompanied by the diffusive uptake of BaP (the fat flush hypothesis) as trans-cellular transfer of BaP was favoured in the apical to basolateral direction. These data support the theory that non-polar substances including HOCs are preferentially transferred from the lumen into the bloodstream and provide indirect evidence that the uptake is related to the fugacity gradient created by the unidirectional uptake of dietary lipids.

Benzo[a]pyrene bioavailability from pristine soil and contaminated sediment assessed using two in vitro models.

A major route of exposure to hydrophobic organic contaminants (HOCs), such as benzo[a]pyrene (BaP), is ingestion. Matrix-bound HOCs may become bioavailable after mobilization by the gastrointestinal fluids followed by sorption to the intestinal epithelium. The purpose of this research was to measure the bioavailability of [14C]-BaP bound to pristine soils or field-contaminated sediment using an in vitro model of gastrointestinal digestion followed by sorption to human enterocytes (Caco-2 cells) or to a surrogate membrane, ethylene vinyl acetate (EVA) thin film. Although Caco-2 cells had a twofold higher lipid-normalized fugacity capacity than EVA, [14C]-BaP uptake by Caco-2 lipids and EVA thin film demonstrated a linear relationship within the range of BaP concentrations tested. These results suggest that EVA thin film is a good membrane surrogate for passive uptake of BaP. The in vitro system provided enough sensitivity to detect matrix effects on bioavailability; after 5 h, significantly lower concentrations of [14C]-BaP were sorbed into Caco-2 cells from soil containing a higher percentage of organic matter compared to soil with a lower percentage of organic matter. The [14C]-BaP desorption rate from Caco-2 lipids consistently was twofold higher than from EVA thin film for all matrices tested. The more rapid kinetics observed with Caco-2 cells probably were due to the greater surface area available for absorption/desorption in the cells. After 5 h, the uptake of BaP into Caco-2 lipid was similar in live and metabolically inert Caco-2 cells, suggesting that the primary route of BaP uptake is by passive diffusion. Moreover, the driving force for uptake is the fugacity gradient that exists between the gastrointestinal fluid and the membrane.

Mobilization of chrysene from soil in a model digestive system.

Accurate estimates for the oral bioavailability of hydrophobic contaminants bound to solid matrices are challenging to obtain because of sorption to organic matter. The purpose of this research was to measure the bioavailability of [14C]chrysene sorbed to soil using an in vitro model of gastrointestinal digestion and absorption to a surrogate intestinal membrane, ethylene vinyl acetate (EVA) thin film. The [14C]chrysene moved rapidly from soil into the aqueous compartment and reached steady state within 2 h. Equilibrium was reached in the EVA film within 32 h. Aging the spiked soil for 6 or 12 months had no effect on chrysene mobilization. This was supported by the finding that the data best fit a one-compartment model. Despite significant decreases in [14C]chrysene mobilization when water or nonneutralized gastrointestinal fluids were used in place of the complete medium, the equilibrium concentration of [14C]chrysene in EVA film remained the same in all conditions. Thus, the driving force for uptake was the fugacity gradient between the aqueous phase and the EVA film. Cultured human enterocytes (human colorectal carcinoma cell line [Caco-2 cells]) had a higher lipid-normalized fugacity capacity than EVA film, but the elimination rate constants were the same, suggesting that the rate was controlled by the resistance of the unstirred aqueous layer at the membrane-water interface.

Oral bioavailability of glyphosate: studies using two intestinal cell lines.

Glyphosate is a commonly used nonselective herbicide that inhibits plant growth through interference with the production of essential aromatic amino acids. In vivo studies in mammals with radiolabeled glyphosate have shown that 34% of radioactivity was associated with intestinal tissue 2 h after oral administration. The aim of our research was to investigate the transport, binding, and toxicity of glyphosate to the cultured human intestinal epithelial cell line, Caco-2, and the rat small intestinal crypt-derived cell line, ileum epithelial cells-18 (IEC-18). An in vitro analysis of the transport kinetics of [14C]-glyphosate showed that 4 h after exposure, approximately 8% of radiolabeled glyphosate moved through the Caco-2 monolayer in a dose-dependent manner. Binding of glyphosate to cells was saturable and approximately 4 x 10(11) binding sites/cell were estimated from bound [14C]. Exposure of Caco-2 cells to > or =10 mg/ml glyphosate reduced transmembrane electrical resistance (TEER) by 82 to 96% and increased permeability to [3H]-mannitol, indicating that paracellular permeability increased in glyphosate-treated cells. At 10-mg/ml glyphosate, both IEC-18 and Caco-2 cells showed disruption in the actin cytoskeleton. In Caco-2 cells, significant lactate dehydrogenase leakage was observed when cells were exposed to 15 mg/ml of glyphosate. These data indicate that at doses >10 mg/ml, glyphosate significantly disrupts the barrier properties of cultured intestinal cells.