Improving the Predictive Value of Bioaccessibility Assays and Their Use to Provide Mechanistic Insights into Bioavailability for Toxic Metals/Metalloids – A Research Prospectus

Abstract

Widespread contamination of soil, dust, and food with toxic metal(loid)s pose a significant public health concern. Only a portion of orally ingested metal(loid) contaminants are bioavailable, which is defined as the fraction of ingested metal(loid)s absorbed across the gastrointestinal barrier and into systemic circulation. Bioaccessibility tools are a class of in vitro assays used as a surrogate to estimate risk of oral exposure and bioavailability. Although development and use of bioaccessibility tools have contributed to our understanding of the factors influencing oral bioavailability of metal(loid)s, some of these assays may lack data that support their use in decisions concerning adverse health risks and soil remediation. This review discusses the factors known to influence bioaccessibility of metal(loid) contaminants and evaluates experimental approaches and key findings of SW-846 Test Method 1340, Unified BARGE Method, Simulated Human Intestinal Microbial Ecosystem, Solubility Bioaccessibility Research Consortium assay, In Vitro Gastrointestinal model, TNO-Gastrointestinal Model, and Dutch National Institute for Public Health and the Environment bioaccessibility models which are used to assess oral absolute bioavailability and relative bioavailability in solid matrices. The aim of this review was to identify emerging knowledge gaps and research needs with an emphasis on research required to evaluate these models on (1) standardization of assay techniques and methodology, and (2) use of common criteria for assessing the performance of bioaccessibility models.

Introduction

The Earth’s crust contains nonessential and toxic metals and metalloids at concentrations ranging from low parts per billion (ppb) up to percentages levels (Koller and Saleh 2018). Natural processes such as. weathering of bedrock, water evaporation from soil and water, atmospheric deposition, and volcanic activity and anthropogenic processes such as mining and smelting, waste disposal, automobile exhaust, chemical production and application, agriculture, and manufacturing and use of consumer products have dispersed these materials widely with resulting contamination of soils, food, and water (Mohammed et al. 2011). As metals and metalloids do not degrade, there is great potential for persistent environmental pollution and bioaccumulation in food webs. It is well-established that naturally occurring arsenic (As) in the Himalayas and locales in Spain and Bosnia-Herzegovina mined over the past three millennia remain important sources of environmental pollution (Chakraborti et al. 2018; Olías and Nieto 2015; Leblanc et al. 2000; Lenoble et al. 2013). Although many studies focus on contamination of drinking water with metals and metalloids (i.e. metal(loids)) (Monteiro de Oliveira et al. 2021; Alidadi 2019; Chowdhury and Mazumder 2016; Rehman et al. 2018; Sanders et al. 2015), contamination of solid matrices such as soils, dust, and food are also of special public health concern (Mohmand et al. 2015; Zheng et al. 2013; Shi et al. 2011). Humans may be directly exposed to metal(loids) orally via ingestion of metal-contaminated food and drinking water, occupational exposures, and inhalation of dusts that may undergo mucociliary clearance and subsequent transport through the gastrointestinal tract (Briffa et al. 2020). Thus, the aim of this review was to focus on (1) evaluation of experimental approaches to assess and strengthen understanding of absolute and relative bioavailability of these materials present in solid matrices and (2) enhance performance of these experimental approaches. Information on the form and bioavailability of metal(loid)s in solid matrices provide insights into internal dosage levels and ultimately help regulators in setting appropriate exposure limits (Greim and Snyder 2018). In site-specific risk assessments, use of soil bioavailability data may be used to determine the pattern and extent of site clean-up or remediation. This application of bioavailability data might make remediation more cost-effective and efficient.

Measuring Bioavailability of Toxic Metals and Metalloids

Based upon worldwide prevalence and high potential toxicity, the World Health Organization (WHO) and the Agency for Toxic Substances and Disease Registry (ATSDR) ranked the metalloid arsenic (As), and metals cadmium (Cd), chromium (Cr), lead (Pb), and mercury (Hg) as the top 5 priority substances of public health concern (ATSDR 2017, WHO 2018). Both general and site-specific risk assessments were developed to help with management of sites contaminated with these metals and metalloids. These risk assessments are important as they provide a quantitative approach to evaluating benefits of site remediation.

An important factor in assessing risks associated with ingestion of soils is the bioavailability of contaminating metals or metalloids. Absolute bioavailability (ABA) is a measure of % of a contaminant present in ingested soil that crosses the gastrointestinal (GI) barrier, is available for systemic distribution and metabolism and varies from 0-100% (Bradham et al. 2018b). Bioavailability of a contaminating metal or metalloid is often expressed on the basis of relative bioavailability (RBA). Here, RBA of the soil contaminant is calculated as the ratio of bioavailability of the soil-borne contaminant to the contaminant ingested in a soluble, highly bioavailable form such as sodium arsenate heptahydrate or lead acetate. Deliberative and regulatory organizations commonly recommend that risk assessments for metal exposure from contaminated soils account for variation in bioavailability and use this information in risk assessment and management (Kimbrough et al. 1984; Schoof 2003; NEPI 2000; USEPA 2007a, 2017b; USDHHS et al. 2012; NRC 2003). Bioavailability measurements are in vivo measurements where studies are performed in intact animals that range in complexity from invertebrates to vertebrates, including humans. The complexity and cost of animal-based bioavailability assays may affect their use (NRC 2003).

Measuring the Bioaccessibility of Toxic Metals and Metalloids

In vitro bioaccessibility (IVBA) assays are sometimes employed to provide data that are utilized as a surrogate for RBA assay results (Brattin et al 2013; Diamond et al 2016). Bioaccessibility refers to the physiological solubility of metal(loids) in the gastrointestinal tract (GIT) and expressed as the ratio of the metal(loid) released from the test material and solubilized in the extraction media to the total metal(loid) in the test material. Ingested metal(loids) need to become bioaccessible in the GIT in order to be absorbed. This process may include physical transformation of metal-bearing particles such as (1) breakdown of the particle to expose metal to GIT fluids, (2) dissolution of metals, and (3) chemical transformation of dissolved metals. Bioaccessibility typically is predicted from an in vitro bioaccessibility (IVBA) assay that measures solubility of soil metals in a gastric-like (i.e. low pH) extraction medium (USEPA 2017). As a default, and in the absence of evidence to the contrary, RBA was assumed to be 100% which was a conservative estimate that may inaccurately predict the potential for exposure (NRC 2003). However, physicochemical properties of the soil-borne contaminant and the matrix commonly prevent full release of the contaminant from the soil matrix. Therefore, bioaccessibility assays were modified as an alternative empirical approach to this common assumption by focusing on the soluble fraction that was more readily available for systemic uptake (NRC 2003).

Compared to ABA and RBA assays, bioaccessibility assays cost less, take less time to perform, and reduce need for animals (Rembish et al. 2000). Bioaccessibility assays also enable easy manipulation of experimental parameters to gain a better understanding of factors that control contaminant dissolution. In addition, these assays are useful in rapid testing of potential soil amendments that reduce systemic uptake of inorganics and for identification of site-specific remediation strategies.

Use of alternative in vitro methods is supported by the USEPA, National Research Council, and international organizations and governments such as the Bioaccessibility Research Group of Europe (BARGE) and the Solubility Bioaccessibility Research Consortium (SBRC) (Kelley et al. 2002; NRC 2003; USEPA 2007a; Wragg et al. 2011; ISO 2018). The rationale for the use of IVBA assays is summarized in the following statement:

(1) the assay does not sacrifice animals; (2) the reduced cost and analysis time from use of the IVBA assay in place of in vivo RBA assays will facilitate greater numbers of samples analyzed at each site to improve representativeness; (3) regulatory acceptance of the IVBA assay would lower bioavailability assessment costs by enabling simultaneous assessments of RBA for both As and Pb using the existing standard operating protocol for the IVBA extraction protocol, which has been previously validated for assessment of RBA of Pb in soil (USEPA 2017c).

A central issue in the use of data from bioaccessibility assays to provide surrogate measurements of metal and metalloid bioavailability is the so-called validation of the bioaccessibility assay. In the broadest sense, validation of a method includes determination of its accuracy, precision, limits of detection, and quantification for analytes of interest (Peters et al. 2007). In the context of validation of bioaccessibility assays, the concept of validation is more restricted. For validation of bioaccessibility assays, prime emphasis is placed on the correlation between in vivo estimate of bioavailability of test materials obtained in an established whole animal model and in vitro estimate of bioaccessibility of test materials obtained in the model system (Hagens et al. 2009; Oomen et al. 2006; USEPA 2007b). Drexler and Brattin (2007) estimated bioaccessibility of Pb in soils obtained in a simple in vitro test system which correlated strongly with RBA obtained in immature swine, an established whole animal model. Juhasz et al (2013) suggested that imprecision in use of the term “validated” to describe the goodness of fit of data obtained in animal assays and in vitro assays is related to lack of refined criteria for evaluation of performance. These investigators emphasized that a successful model system exhibits a range of accuracy that is consistent with its intended use. Thus, for a suite of test material evaluated in both in vivo and in vitro models, estimates of bioaccessibility need to be correlated with RBA. Juhasz et al (2013) recommended criteria for acceptability of in vivo-in vitro correlation to include a) a linear relationship between in vivo and in vitro data with a correlation of coefficient (r) > 0.8 and a slope > 0.8 and <1.2, b) a within-lab repeatability of 10% relative standard deviation (RSD), and c) a between-lab reproducibility of 20% RSD.

Based upon these criteria, studies comparing RBA of As in soil using swine or mice as the test species may be described as a validation study (Bradham et al. 2013). Further evaluation of a bioaccessibility assay using a novel set of test materials for which RBA estimates are available is also part of the validation procedure for the in vitro method (Bradham et al. 2005).

Two commonly-used bioaccessibility assays, USEPA SW-846 Test Method 1340 (also known as Simplified Bioaccessibility Extraction Test or SBET) and Bioaccessibility Research Group of Europe’s Unified BARGE Method (UBM), have undergone inter- and intra-lab validation by their respective regulatory entities for predicting RBA of As and Pb in soil to support human health risk assessments (HHRA) and for which As and Pb in vitro bioaccessibility data are strongly correlated with mice and/or swine RBA (Bradham et al. 2015; Denys et al. 2012; Wragg et al. 2011). Intra-lab validation evaluates the accuracy and reproducibility of the assay over time. Intra-lab variation might reflect random error in operating procedures or systematic error resulting from changes in operating procedures. Inter-lab studies assess variation that might result from inconsistent application of the operating procedures and whether the published description of an assay is sufficiently detailed such that all aspects of the procedure may be transferred to other labs. Inter-lab evaluations commonly use a suite of test materials to determine whether assay performance is reproducible across labs.

EPA Method 1340 and UBM assays differ in complexity. EPA Method 1340 consists of one compartment representing the stomach in which a mixture of glycine, hydrochloric acid, and water simulate human stomach fluid at fasting pH 1.5 (USEPA 2017a; 2017b) (Table 1) [Table 1 near here]. Fasting pH of the gastric phase refers to the pH attained in the human stomach in absence of food. UBM is a three-compartment system including mouth, fasting stomach, and small intestine. These compartments contain a mixture of inorganic, organic, and enzymatic constituents to mimic the chemical composition of a human digestive tract (Oomen et al. 2003; ISO 2018) (Table 1). For both assays, performance was validated by USEPA and BARGE for Pb, Cd, and As in soils (USEPA 2017a; 2017b; Li et al 2016; Denys et al 2012; Juhasz et al 2010). Similarly, some investigators established linear correlations between As and Pb RBA in mice with bioaccessibility estimates using the USEPA Method 1340 for indoor dust samples (Li et al. 2014a; 2014b), in which the method is defined as the gastric phase of the SRBC method. Absence of validation for other metals and metalloids that are present in contaminated environments at concentrations above regulatory safe values (or values above which exposure poses a higher risk of adverse health effects) hinders the assays value in measurement of bioaccessibility of these contaminants. In addition, EPA Method 1340 or UBM have not been assessed for estimation of Pb and As bioaccessibility in other solid matrices such as dust and foods (Carlin et al. 2016). Expanding validation of these assays to include other matrices might establish confidence in the use of these test methods for a broader range of matrices and metal(loid)s. Broader utilization of these assays might facilitate collection of bioaccessibility data that may be employed to determine aggregate exposure to these contaminants. Identification of the limitations of these assays warrant consideration as the assays are refined to better model the phenomena that affect the dissolution which impacts uptake of metals and metalloids across the GI barrier. Areas for refinement of these assays are described in the following discussion.

Table 1:

a.) Commonly used in vitro bioaccessibility assays b.) Commonly used in vitro bioaccessibility assays

a.
Assay		Saliva/Mouth	Gastric		Intestinal Phase(s)
SW-846 Test Method 1340	Constituents (g/L)	Saliva/Mouth Phase Absent	Constituents (g/L)	glycine (30)	Constituents (g/L)	Intestinal Phase Absent
	pH		pH	1.5	pH
	Residence time		Residence time	1 hr	Residence time
	Temperature		Temperature	37°C	Temperature
	Fed or fasting		Fed or fasting	Fasting	Fed or fasting
	Solid: Liquid Ratio		Solid: Liquid Ratio	1:100	Solid: Liquid Ratio
RIVM	Constituents (g/L)	KCl (0.896), NaH2PO4 (0.888), KSCN (0.2), NA2PO4 (0.57), NaCl (0.298), NaOH (0.7), urea (0.2), amylase (0.145), mucin (0.05), uric acid (0.015)	Constituents (g/L)	NaCl (2.752), NaH2PO4 (0.266), KCl (0.824), CaCl2 (0.4), NH4Cl (0.306), 37% HCl (8.3 ml), glucose (0.65), glucuronic acid (0.000002), urea (0.085), glucosamine hydrochloride (0.33), bovine serum albumin (BSA) (1), mucin (3), pepsin (1)	Constituents (g/L)	NaCl (7.012), NaHCO3 (3.388), KH2PO4 (0.08), KCl (0.564), MgCl2 (0.05), 37% HCl (0.18ml), urea (0.1), CaCl2 (0.2), BSA (1), pancreatin (3), lipase (0.5)
	pH	6.5	pH	1.2	pH	7.8
	Residence time	5 min	Residence time	2 hr	Residence time	2 hr
	Temperature	37°C	Temperature	37°C	Temperature	37°C
	Fed or fasting	Fasting	Fed or fasting	Fasting	Fed or fasting	Fasting
	Solid: Liquid Ratio	1:15	Solid: Liquid Ratio	1:100	Solid: Liquid Ratio	1:100
SHIME	Constituents (g/L)	Saliva /Mouth Phase Absent	Constituents (g/L)	Nutrilon (15), pectin (16), mucin (8), starch (5), cellobiose (1), glucose (1), proteose peptone (2)	Constituents (g/L)	Gastric constituents + NaHCO3 (12), bovine bile (4), pancreatin (0.9)
	pH		pH	4	pH	6.5
	Residence time		Residence time	3 hr	Residence time	4-5 hrs for first intestinal phase, 20-24 hrs for each additional intestinal phase
	Temperature		Temperature	37°C	Temperature	37°C
	Fed or fasting		Fed or fasting	Fed	Fed or fasting	Fed
	Solid: Liquid Ratio		Solid: Liquid Ratio	1:2.5	Solid: Liquid Ratio	1:4
b.
Assay		Saliva/Mouth	Gastric		Intestinal Phase(s)
UBM	Constituents (g/L)	KCl (0.896), NaH2PO4 (0.888), KSCN (0.2), NA2SO4 (0.57), NaCl (0.298), 1M NaOH (1.8 mL ), urea (0.2), amylase (0.145), mucin (0.05), uric acid (0.015)	Constituents (g/L)	NaCl (2.752), NaH2PO4 (0.266), KCl (0.824), CaCl2 (0.4), NH4Cl (0.306), 37% HCl (8.3 ml), glucose (0.65), glucuronic acid (0.000002), urea (0.085), glucosamine hydrochloride (0.33), bovine serum albumin (BSA) (1), mucin (3), pepsin (1)	Constituents (g/L)	NaCl (7.012), NaHCO3 (5.607), KH2PO4 (0.08), KCl (0.564), MgCl2 (0.05), 37% HCl (0.18ml), urea (0.1), CaCl2 (0.2), BSA (1), pancreatin (3), lipase (0.5)
	pH	6.5	pH	1.2	pH	7.4
	Residence time	20 sec-5 min	Residence time	1 hr	Residence time	4 hr
	Temperature	37°C	Temperature	37°C	Temperature	37°C
	Fed or fasting	Fasting	Fed or fasting	Fasting	Fed or fasting	Fasting
	Solid: Liquid Ratio	1:15	Solid: Liquid Ratio	1:100	Solid: Liquid Ratio	1:100
SBRC	Constituents (g/L)	Saliva /Mouth Phase Absent	Constituents (g/L)	glycine (30)	Constituents (g/L)	bile (1.754). Pancreatin (0.5)
	pH		pH	1.5	pH	7
	Residence time		Residence time	1 hr	Residence time	4 hr
	Temperature		Temperature	37°C	Temperature	37°C
	Fed or fasting		Fed or fasting	Fasting	Fed or fasting	Fasting
Solid: Liquid Ratio	Solid: Liquid Ratio	1:100	Solid: Liquid Ratio	1:100
IVG	Constituents (g/L)	Saliva /Mouth Phase Absent	Constituents (g/L)	pepsin (10), NaCl (8.77)	Constituents (g/L)	bile (3.33). Pancreatin (0.35)
	pH		pH	1.8	pH	5.54
	Residence time		Residence time	1 hr	Residence time	1 hr
	Temperature		Temperature	37°C	Temperature	37°C
	Fed or fasting		Fed or fasting	Fasting	Fed or fasting	Fasting
	Solid: Liquid Ratio		Solid: Liquid Ratio	1:150	Solid: Liquid Ratio	1:150
TIM/Tiny-TIM	Constituents (g/L)	Artifical saliva and amylase	Constituents (g/L)	NaCl (0.673), KCl (0.24), NaHCO3 (0.130), CaCl di-hydrate (0.033), 0.01 M pH 7 citrate buffer, lipase (125 units), pepsin (2000 units)	Constituents (g/L)	Duodenum and jejunum: NaCl (1.25), KCl ( 0.15), CaCl di-hydrate (0.075), pancreatin (3.5), 10% w/w bile Ileum: NaCl (5), KCl (0.6), CaCl di-hydrate (0.30)
	pH	5	pH	2-5	pH	6.2-7.4
	Residence time	5 min	Residence time	1.5 hr	Residence time	4-6 hrs for 1 Tiny-TIM or 3 TIM intestinal phases
	Temperature	37°C	Temperature	37°C	Temperature	37°C
	Fed or fasting	Fed or fasting	Fed or fasting	Fed or fasting	Fed or fasting	Fed or fasting
	Solid: Liquid Ratio	1:25	Solid: Liquid Ratio	1:50	Solid: Liquid Ratio	1:50

Exposure Media Factors that Influence Metal/Metalloid Bioaccessibility and Bioavailability in Validated Assays

pH and Redox Status

Metal mobility in soils is strongly influenced by soil pH and redox status. Many investigators reported greater metal bioaccessibility and RBA in the gastric than in intestinal phases (pH 5.5-7) at a fasting pH of 1.2-1.5 in the gastric compartment (Pelfrêne and Douay 2018; Li et al. 2014a; Turner and Ip 2007). Redox status exerts a stronger control over release of metals in solid matrices than pH when soil pH is above 4.5-5.0 (Calmano et al. 1993). Metals typically bind to iron (III) oxide and manganese (III/IV) oxides under oxidizing conditions (He et al. 2009). Under reducing conditions, iron (Fe) and manganese (Mn) oxides undergo reductive dissolution, releasing Cd, Pb, Hg, zinc (Zn), and nickel (Ni) from the matrix (Calmano et al. 1993, He et al. 2009). As pH falls below 4.5, acidic pH enhances mobility of metals (Calmano et al. 1993). Similar trends are observed in the gastric environment where low pH facilitates metal dissolution (Bradham et al. 2011; Drexler and Brattin 2007). Mineral oxides such as Fe oxides are highly soluble at low pH, and metals adsorbed to mineral oxides display greater RBA in acidic gastric conditions (Meunier et al. 2010). Notably, different soils with the same pH do not exhibit similar bioaccessibilities; which confirms that soil pH is not the sole factor influencing bioaccessibility (Nelson et al. 2018; Calmano et al.1993).

Particle Size

In vitro bioaccessibility assays often use soil and dust size fractions below 250 μm as test materials. Particles of this size fraction often adhere to hands and may be a source of exposure for children under 3-years-old who frequently engage in hand-to-mouth activity (Xue et al. 2007). There is evidence that concentration of metals in soil particles differ as a function of particle size. In one case, Pb and Cd concentrations tripled as size fraction decreased from 5 μm to 0.3 μm (Pelfrêne and Douay 2018). This association between particle size and concentrations of metals adsorbed to soil particles might impact the degree of gastric dissolution. Overall, higher metal concentrations in fine particles of soil and dust were associated with dissolution of greater levels of Cd and Pb in the UBM gastric phase compared with the coarse particle size fraction. A similar trend was also observed with USEPA Test Method 1340 in which Pb and As concentrations rose with decreasing soil particle size (<250 μm to <150 μm) resulting in an elevation in bioaccessible concentrations of Pb and As (Karna et al. 2017). This relationship was attributed to the greater surface area to volume ratio of small particles which increases adsorption and desorption area. Notably, there were no marked differences in % bioaccessibility (i.e. ratio of bioaccessible Pb and As to total Pb and As present in the sample) between the <250 and <150 particle size fractions. The association between particle size and Pb absorption was examined in rats that consumed diets amended with differently sized particles of metallic Pb (Barltrop and Meek 1979). Both blood Pb concentrations and kidney Pb burdens were highest in mice that consumed diets that contained the smallest particles (mean diameter = 6 μm). However, these particle size-dependent effects were examined at relatively high dosage levels and not evaluated statistically. No apparent studies examined effects of particle size on RBA as measured in mouse and swine assays used to support HHRA. Thus, additional studies with better-characterized particles including metal(loid) identity, particle size, or particle composition and appropriate data analysis are needed to gain mechanistic insight into the role particle size plays in bioaccessibility or absolute and relative bioavailability of Pb, As, Cd and other metals and metalloids in which particle size was found to impact dissolution (Pelfrêne and Douay 2018; Karna et al. 2017; Barltrop and Meek 1979).

Co-Occurrence and Interactions

Many bioaccessibility and bioavailability studies focus on a single element in contaminated soils, dusts, and foods. However, investigation of co-occurring metals and minerals in environmental and food samples are more environmentally relevant because soils and dusts usually contain multiple constituents including clay, oxide, and sorbed components including Fe, Mn, and aluminum (Al), which exhibit an affinity for metal(loid) contaminants (Fergusson and Kim 1991; Sparks 2003; Bocato et al. 2019). Effects of these interactions on metal RBA have been evaluated in the UBM assay (Zhu et al. 2019). Related investigations examined metal and metalloids interaction on RBA obtained with mouse assays (Li et al. 2019).

Some of these interactions might be rationalized on physical and chemical grounds. In a sorption kinetics study, addition of Mn increased the binding of As to Fe (He et al. 2009). Addition of Mn, an oxidant, converted Fe^II hydroxide to Fe^III hydroxide, a species that strongly binds to arsenate and arsenite (He et al. 2009). Arsenic sequestration by binding to Fe oxides reduced As bioaccessibility which is consistent with findings on soil As RBA obtained in a mouse model (Bradham et al. 2011). Bradham et al. (2011) observed a reduction in As, Cd and Pb RBA when these contaminants co-occurred. In mice exposed to soils spiked with As, Cd, and Pb, RBA of As was 8.9% (Ollson et al. 2017). RBA more than tripled (30%) when mice were exposed to soil spiked with As only. Similarly, RBA was lower for co-occurring Cd and Pb compared to RBA of Pb and Cd alone (Ollson et al. 2017). While soils spiked with metal(loids) offer insight into mechanisms of bioaccessibility, soil-spiked experiments are not without their limitations. It is noteworthy that the behavior of exogenously-spiked metals incorporated into the soil matrix may not resemble behavior of endogenous metal under all circumstances, thereby resulting in differential absorption and/or systemic clearance of the spiked metal(loid) compared to non-spiked metal-contaminated soils (Bradham et al. 2018a).

What Are Critical Properties of Assays?

Gastric Phase Conditions

Both validated in vitro assays, Method 1340 and UBM, commonly use a fasting gastric state condition prior to metal(loid) exposure which may affect metal speciation and bioaccessibility. The gastric phase of Method 1340 contains glycine, hydrochloric acid, and water at pH 1.5 to mimic conditions in the fasted stomach (USEPA 2017a; 2017b). The fasting stomach represents the worst-case scenario for gastric bioaccessibility of ingested metal(loids), because acidic gastric pH facilitates metal(loid) dissolution (Pelfrêne and Douay 2018; Bradham et al. 2015; Turner and Ip 2007). However, Method 1340 does not mimic the entire digestion system or account for biological processes such as gut microbiomes that are now believed to govern environmental contaminant dissolution, metabolism, and health effects (Laird et al. 2007). The Modified UBM contains multiple GI phases and addition of digestive enzymes that produce changes in the redox state of the system. α-Amylase present in the mouth phase catalyzes conversion of starches into glucose, a reducing sugar, that was found to reduce vanadium V (V⁵⁺) to more bioaccessible vanadium IV (V⁴⁺). In the gastric phase, pepsin-catalyzed degradation of bovine serum albumin generates reducing agents that reduce V⁴⁺ (Yu et al. 2019). Other UBM constituents including pancreatin, lipase, and bile in the GI phase are involved in the breakdown of the matrix and subsequent increased bioaccessibility of trivalent As (Tokalıoğlu et al. 2020). These data suggest addition of complex constituents might alter the GI fate of metal(loids).

Role of Feeding Status and Diet Composition

Characterizing the factors affecting contaminant bioaccessibility, ABA, and RBA may be reflected in in vitro systems’ capacity to mimic effects of feeding status and dietary composition on ABA and RBA obtained in in vivo models. A summary of evidence from studies of Pb absorption in humans reports that fasting enhances Pb uptake and meal consumption diminishes Pb uptake in adult humans (Vázquez et al. 2015; Maddaloni et al. 1998). Similarly, Li et al. (2018) found RBA of Pb in mice exposed to contaminated soils under fasting conditions was 1.3-3.5-fold higher than in the fed state. These differences in ABA and RBA under fed versus fasting conditions may be mediated by differential effects of food on ABA and RBA of soil-bound Pb and Pb acetate. A similar trend was also noted by Hagens et al (2009) in an investigation of soil Pb bioaccessibility in the Tiny-TIM system in which bioaccessibility was 4.5-fold higher under fasting than under fed conditions. These data aligned with human ABA and RBA data for Pb in soils.

In in vivo assays, dietary composition and feeding status might affect metal(loid) RBA and absorption. Lead RBA in mice fed high protein, high fat diets containing fat, eggs, and pork was greater than in mice fed a diet high in minerals (Li et al. 2018). Addition of cabbage to the diet elevated organic acid content and Pb RBA. Changes in levels of specific nutrients in the diet might alter RBA of Pb in the mouse model. Lowering dietary phosphate levels used in the mouse assay enhanced uptake of Pb into tissues which is indicative of increased uptake of Pb across the GI barrier (Bradham et al. 2018a). Pb and nutrient interactions that affect Pb uptake may reflect interactions in the lumen of the GI tract or competition for transporters that control uptake across the barrier. It is noteworthy that elevated Pb uptake may not result in increased RBA. In the case of Bradham et al. (2018a) while Pb uptake rose, estimated RBA for Pb in NIST 2710a standard reference soil remained unaffected.

Transporter-mediated uptake of metal(loids) across cells, including the GI epithelium, is an important site for antagonistic interactions that may affect uptake (Martinez-Finley et al. 2012). The divalent metal transporter-1 which mediates uptake of both divalent Pb and Fe ions across cell barriers and lowers Fe intake is associated with increased Pb absorption (Bressler et al. 2004). Thus, the Fe nutritional status of animals used in ABA and RBA assays may be an important experimental variable in studies of Pb ABA and RBA (Elsenhans et al. 2011). Conversely, phosphate and arsenate oxyanions compete for membrane transporters such that elevated dietary phosphate levels decreased uptake of As across the GI barrier (Gonzalez et al. 1995). Consideration should be given to modifying the conditions used in bioaccessibility assays to model the effects of changes in dietary composition on ABA and RBA.

Factors Influence Performance of Non-Validated Models for Measuring Bioaccessibility and RBA of Metal(loid)s

Overview

Assays, developed in the past 25 years and for which additional reporting of validation studies is needed including the Solubility Bioaccessibility Research Consortium (SBRC) assay, TNO intestinal model (Tiny-Tim), and Dutch National Institute for Public Health and the Environment (RIVM), and the Simulated Human Intestinal Ecosystem (SHIME®) (Wragg and Cave 2003; Oomen et al. 2006; Bradham et al. 2018b) have (1) filled gaps in existing knowledge, (2) identified new factors that have the potential to impact gastric dissolution of inorganics, (3) in some cases achieved accurate prediction of contaminant ABA and RBA after oral exposure, and (4) addressed some of the limitations inherent in validated assays such as: Method 1340, UBM, and animal models. SHIME® was used to measure bioaccessibility of toxic metals and metalloids. This system was originally developed for use in the European food and drug industries for assessing food and drug metabolism in the GIT (Van de Wiele et al. 2015). SHIME® is a 5-compartment GI model system consisting of a stomach, small intestine, and ascending, transverse, and descending colon. The latter three compartments are inoculated with human fecal bacteria to mimic human gut microbiome. The gut microbiome is defined as “the characteristic microbial community occupying a reasonably well-defined habitat [the gut] which has distinct physicochemical properties [and thus] refers not only to the microorganisms involved but also encompasses their theatres of activity” (Whipps 1997). The use of SHIME® has reached beyond the European food and drug industries and extended to investigations on the gastric fate of inorganic contaminants detected in food and soils at each phase of digestion. SHIME® also enables investigation of the relationship between the gut microbiome and contaminant fate, a topic which has remained unexamined in validated bioaccessibility models. Although bacteriological functionality within the simulated colon was validated via metabolic byproducts for nutrition studies, SHIME® data have not been examined against in vivo results and the microbiome community has not been fully characterized (Molly et al. 1994).

Despite these limitations, this model has given new insight into lesser-known factors affecting bioaccessibility and contaminant transformation. In two SHIME® studies, bioaccessibility of soil-bound As and mine tailings ranging from 15.5-30,000 mg total As/kg sample across both studies were greater in the colon (1.7-5.8-fold greater for soils and 1.6-3.5-fold higher for mine tailings) than in the small intestine (Laird et al. 2007; Yin et al. 2017a). Inorganic arsenic (iAs) also underwent speciation transformations (Yin et al. 2017a) (Figure 1) [Figure 1 near here]. The bulk of As^V was reduced to the more toxic As^III species and further transformed into several methylated species. These processes were significantly decreased in the absence of the intestinal microbiome. Similar changes in transformation of pentavalent soil As to trivalent and methylated species in the colon were also observed in other SHIME® studies (Van de Wiele et al. 2010; Yin et al. 2016). Results of a separate SHIME® study suggests the degree of As methylation differs between adults and children. A SHIME® model inoculated with fecal bacteria from an adult exhibited As methylation in the colon that was 3-fold higher than the SHIME® colon inoculated with fecal bacteria from a 6-year-old child. SHIME® inoculated with a child’s fecal microbes displayed significantly higher levels of As^III (Yin et al. 2017b). These investigators concluded that adult gut microbiota was significantly involved in As transformation. These results point to the potential for age-dependent variability in toxic metal metabolism and enhanced risk of toxic metal exposure for young children (Yin et al. 2017b).

Figure 1.

Scheme for production of methylated oxy- and thio-arsenicals by the microbiota of the gastrointestinal tract. Arsenite, areduced oxy-arsenical (iAs^III) is methylated to monomethylated (MMAs^III) and demethylated (DMAs^III)products. Oxidation of these arsenicals yield arsenate (iAs^V) and monomethylated (MMAs^V) and demethylated (DMAs^V) species that are converted to thioarsenicals containing one to four groups.

Investigators have also employed SHIME® and other non-validated bioaccessibility and RBA models (Table 1) (Oomen et al. 2002; 2003; Van de Wiele et al. 2015; Wragg and Cave 2003; Bradham et al. 2018b) to identify relationships between diet, microbe presence and activity, GI properties, and inorganic bioaccessibility after simulated exposure to metals bound to food matrices. Rice is a staple diet in countries worldwide and in 2019 global rice consumption was 499.2 million metric tons (USDA Foreign Agricultural Service 2019). Rice cereal, an important global food crop, ranks second in global cereal production (FAOSTAT 2014). Several studies examined bioaccessibility of metal contaminants (primarily As) in rice, since uptake of toxic metals by rice grown in water-inundated fields may pose an elevated risk of metal exposure to consumers (Alava et al. 2013; Praveena and Omar 2017; Sharafi et al. 2019; Sun et al. 2012). Praveena and Omar (2017) and Sharafi et al. (2019) used the Netherlands National Institute for Public Health and the Environment (RIVM) bioaccessibility model consisting of an oral cavity, stomach, and small intestine to determine bioaccessibility of a suite of metals including but not limited to, As, Pb, Cd, Cr, copper (Cu), and/or Al detected in several varieties of rice. RIVM is a chemical simulation of the upper digestive tract that lacks a large intestine and its simulated microbiome, unlike SHIME® which includes a large intestine phase and intestinal microbiome. Sharafi et al. (2019) and Praveena and Omar. (2017) examined inorganic fate in each digestive phase and after passage through all digestive phases, respectively. Sharafi et al (2019) found the stomach phase contained the highest levels of bioaccessible As, Cd, and Pb across several varieties of rice due to the low pH of the stomach environment (pH 2-3). Both Sharafi et al (2019) and Praveena and Omar (2017) reported differential bioaccessibility of multiple elements with Cu and Al exhibiting greater bioaccessibility than Cd and Cd displaying greater bioaccessibility than As and Pb. In contrast to these investigations Sun et al (2012) reported that iAs bioaccessibility in a SHIME® study was highest in the colon followed by intestinal and gastric phases, respectively, upon exposure to contaminated rice. In the colon phase, iAs was transformed into the toxic monomethylarsonous acid (MMA^III) species, likely via microbial transformation.

Other studies have delved more deeply into how diet, the gut microbiome, and biochemical processes within the body are occurring simultaneously to alter As bioaccessibility and transformation. Arsenic bioaccessibility in a SHIME® model inoculated with fecal bacteria from human volunteers who consumed either a high fat/high protein diet or a low fat/low protein/high carbohydrate diet was greatest in the stomach phase where pH was the lowest (Alava et al. 2015). Arsenic bioaccessibility significantly declined as it progressed through the small intestine and colon phases. In the colon stage, significant levels of iAs were transformed into MMA^III and MMMTA^V for all diets. Overall, As dissolution was highest in the low fat/low protein diet SHIME® than in the high fat/high protein diet SHIME®. The high fat/high protein diet SHIME® had greater levels of methylated and thiolated methylarsenicals compared to the low fat/low protein diet SHIME®. A similar trend was detected by Alava et al. (2013) in a separate study in which a simulated GI model (In Vitro Gastrointestinal model, IVG) consisting of gastric and microbe-free intestinal phases and fed a fat source (butter) was exposed to As‐contaminated rice. Alava et al (2013; 2015) concluded fat interacted with bile salts which bind methylated arsenicals generated from iAs and diminished its bioaccessibility. Fat also altered the composition of the colon microbiome, favoring bacteria that more easily emulsified lipids especially in the presence of bile salts (Alava et al. 2015). In the SHIME® and IVG models, thiolated and methylated As species adsorbed to pepsin and iAs interacted with Fe found in bile salts. These interactions promoted precipitation of the complexes and reduced intestinal organic and inorganic As bioaccessibility in the intestine (Alava et al. 2013). Taken together, results of these studies suggest multiple factors affect bioaccessibility of As. Studies which incorporate multiple simulated GI compartments and/or gut bacteria enable investigators to identify pathways for metabolism of iAs and the role of metabolism in bioaccessibility. Without these evolving simulated systems, researchers and risk assessors may be unaware of the roles of these variables in the fate of metal(loid) contaminants.

Harmonization of SHIME® and Related Assays for Risk Assessment Applications and Characterization of Mechanisms of Bioaccessibility

Overview

SHIME® and related assays provide a promising tool to investigate the role of the gut microbiota in the absorption of metals and metalloids. By incorporating compartments that represent the large intestine and its associated microbiota, these models provide a means to examine the role of these microorganisms in bioaccessibility of metals and metalloids. The following paragraphs summarize the areas in which additional research is needed to validate or evaluate performance of these models for prediction of RBA.

Standardization of assay techniques and methodology

Validated bioaccessibility models have standardized procedures for assay methodology, data collection and analysis, and interpretation. Standardization ensures inter- and intra-lab reproducibility and enables comparison of data from studies based upon common protocols. Lack of standardization increases variability among studies. Inter-study variability might arise from differences among samples or variations in test procedures. Differences among samples tested in different labs might be minimized by common collection and processing of samples with a consistent chain of custody that ensures provenance and history of a tested sample be documented. Differences in test procedures may be minimized by development of detailed operating procedures that fully describe all materials and methods and strict adherence to these protocols. Routine evaluation of test materials with certified values for analytes of interest such as Standard Reference Materials from the National Institute of Standards and Technology might be employed to track intra- and inter-lab variability.

A challenge to standardization is determining “ideal” conditions or conditions that may result in accurate results across a variety of sample matrices and inorganic contaminants. Test conditions often differ among and between different assays performed in different labs (Yu and Yang 2019; Van de Wiele et al. 2010; Laird et al. 2007) (Table 1). These include (1) duration of sample incubation in each GI phase, (2) selection and concentrations of reagents used to make each type of GI phase, (3) number and types of phases (mouth/saliva, gastric, multi-compartment colons) used, (4) selection of dietary inputs for maintaining microbes, and (5) bacteria inoculation methods (Table 1).

In the case of SHIME ® studies, standardizing gut bacteria communities presents a unique challenge, because these highly diverse microbial communities are in flux until the system stabilizes as bacteria-generated metabolic byproducts become less variable over time. Selective pressures such as microbiota interactions, aging of the in vitro system, environmental conditions, competition for nutrients, and inability to culture select bacteria in the lab result in a fluctuating and evolving bacteria community (Justice et al. 2017) that has not been well-characterized despite common use of this in vitro system. Simply controlling overall bacterial concentration and volume of the inoculum may not dampen the force of these selective pressures. Further research may enable researchers to characterize these pressures and their relationship with contaminant fate in the GI tract. In addition, fecal bacteria are primarily used to inoculate the system due to the ease of fecal collection. However, cecal bacteria may be a better representation of the bacteria colonizing the mammalian colon. The cecum is the site of bacteria propagation and metabolism of various nutrients by a different community of bacteria than the community characterized in feces (Gu et al. 2013; Marteau et al. 2001). Differences in microbial community structure between fecal and cecal sources may be reflected in differential dissolution and transformation of metals. Developing methods and coming to a consensus on the use of those methods enables researchers to achieve optimal inter- and intra-lab assay performance for all commonly used assays.

Use Common Criteria for Model Validation

Once assays are standardized and validated, confidence in the use of this data as surrogates for bioavailability estimates enables investigators to test the predictive capability of in vitro models for a variety of different matrices. For bioaccessibility assays predicting Pb and As bioavailability in soils and soil-like materials, USEPA validation criteria include: (1) a thorough description of the test protocol, (2) test performance correlating the assay test method endpoint with the animal model, (3) determining variability within the test, (4) measuring inter- and intra-lab assay variability,(5) performing the assay using reference materials, and (6) comparing assay results with that of validated assays the candidate assay may replace (USEPA 2009; 2017c). The assay also needs to be time and cost-efficient, allow researchers to limit animal use, and results need to undergo independent scientific review. All international regulatory bodies do not require in vivo in vitro correlation of data (IVIVC). International harmonization and IVIVIC might strengthen the case for using data generated from the use of these models in development of HHRA and for characterizing mechanisms of bioaccessibility.

Focus on Defining and Addressing Data Gaps for Models

Gender, life stage, stress status, and diet affect the composition of the gut microbiome (O’Toole and Claesson 2010; Alava et al. 2015; Marzorati et al. 2017; Bailey et al. 2011). However, few studies examined whether these differences affect dissolution, transformation, and absorption of inorganic contaminants. Data suggest that the microbiome of young children, stressed individuals, males and individuals on high fat and protein diets may enhance exposure to microbiota-transformed toxic metal species and enhance potential for negative health effects (Chi et al. 2017; Alava et al. 2015; Yin et al. 2017b; Wang et al. 2018). Notably, 95% (or 159 out of 168 products) of baby food were reported to contain one or more toxic metals of which 88% (or 148 products) lack FDA-enforceable limits or guidance (Houlihan and Brody 2019). Foods tested included infant formula, teething foods, cereal, fruit juices, rice-based snacks, fruits, and vegetables. Childhood exposure to toxic metals through diet pose a serious area of concern due to daily consumption of metal(loids) that are known to exert deleterious effects on child growth and development (Rehman et al. 2018; Kordas et al. 2007). Further investigation is needed to support these findings and identify what aspects of microbiome communities enhance susceptibility to contaminant exposure in sensitive groups. Investigating how food matrix properties such as ingredients, cooked versus uncooked, pureed versus solid, method of food preparation or metal binding affinity to food matrix affect gastric release of toxic metals may allow us to predict if there is a significant exposure risk to ingesting metals that are in foods one may eat daily.

Prior exposure to As is also an area of research that has been largely ignored. Populations in Argentina exposed to high levels of As in their natural environments for generations have developed resistance to the toxic effects via greater expression of As methyltransferase polymorphisms responsible for As metabolism (Schlebusch et al. 2015). It remains unknown how long this evolution takes to appear in a population, whether these populations possess different gut microbiomes than individuals living outside of the contamination zone, and if these microbes play a contributing role in As metabolism. For communities with recent exposure to inorganic contaminants, studies have not examined whether the gut microbiome has adapted to prior exposure and if these adaptations impact contaminant bioavailability with successive exposures. Addressing these unknowns in future research might enable researchers and assessors to understand whether different communities and/or populations are likely to be exposed to high levels of and adversely impacted by contaminant exposure based upon the aforementioned factors.

Use of in vitro models to develop strategies to reduce uptake of toxic metals

In vitro bioaccessibility models have primarily been employed to assess and understand risk of exposure to contaminants of concern. However, a standardized model with a high level of predictive capability may also be used to identify and/or design soil amendments or dietary supplements that may reduce toxic metal(loid) bioavailability. In vitro bioaccessibility assays show promise in the successful identification of potential remediation strategies that might reduce toxic metal bioavailability. Soils amended with phosphoric acid or Triple Super Phosphate exhibited lower Pb RBA than non-amended soils within a modified- SBRC system (Kastury et al. 2019; Juhasz et al. 2016). Fe, Al, and titanium dietary supplementation decreased the bioaccessibility of As in contaminated foods in an in vitro GI model (Clemente et al. 2016). However, presence of salts such as sulfates and silicates reduced the efficacy of Fe and Al supplements. The results of these studies correlated with data from animal models (Clemente et al. 2019). Additional research might build certainty in selection and practical use of substances that reduce or prevent exposure to Pb, As, and other inorganic pollutants of concern.

Concluding Remarks

Standardized in vivo bioavailability models and validated bioaccessibility models that are used to predict relative and absolute bioavailability of ingested metal(loid)s rely on a set of standardized procedures including a fasted state, narrow ranges of life stages, and specific gastric phase(s). Lack of variability in these standardized test conditions does not enable one to account for variabilities in exposure conditions and human sensitivities when metal(loid)s are ingested orally. Consequently, validated IVBA models may lose predictive value under conditions outside of the assays, thereby limiting applicability of the results to humans. The studies discussed in this prospectus underscore how variations in feeding status fed versus fasted states, life stage, diet, fate of metal(loid)s in GI phases, particle size, exposure media composition, and gut microbiome may reveal an underestimation or overestimation of the bioavailable fractions of metal(loids) as observed in standardized assays under more restrictive assay conditions. This prospectus is a framework for the use of these promising research findings to define future experimental directions. More comprehensive exploration of the mechanisms affecting oral bioavailability of metal(loids) would advance our understanding of bioaccessibility and bioavailability of inorganic contaminants in solid matrices and improve predictive capability of IVBA assays across a wider range of conditions. Specifically, this research effort might promote development of a new suite of in vitro methods that complement and extend on-going work on bioavailability of metals. Current investigation efforts are beginning to explore these topics, but a more structured approach may ensure that this work provides new tools for research. Major challenges to this investigation are selecting and prioritizing metals and metalloids of interest, and standardizing and validating in vitro assays with multiple compartments, especially those that incorporate a compartment containing gut microbiota. Because food may be an important source of exposure to some toxic metals including As, in vitro assays need to be standardized and validated for other test matrices including food. Broader use of in vitro assays with a range of sample matrices may provide data on aggregate exposure from all relevant sources. Addressing these challenges and developing and refining bioaccessibility tools may encourage consideration of this broader range of data when performing site-specific risk assessments that might be used to determine efficient, cost-effective site clean-up or remediation.