How Digestive Processes Can Affect the Bioavailability of PCBs Associated with Microplastics: A Modeling Study Supported by Empirical Data

The transfer kinetics of plastic-associated chemicals during intestinal digestive processes is unknown. Here, we assessed whether digestive processes affect chemical exchange kinetics on microplastics, using an in vitro gut fluid digestive model mimicking the human upper intestinal tract. Chemical exchange kinetics of microplastics were measured for 10 polychlorinated biphenyls (PCBs) as proxies for the broad class of hydrophobic organic chemicals. Following earlier studies, olive oil was used as a proxy for digestible food, under high and low digestive enzyme activities. The micelle–water and oil–water partition coefficients of the 10 PCBs were also determined to evaluate the relative contribution of each gut component to sorb PCBs. A new biphasic and reversible chemical exchange model, which included the digestion process, fitted well to the empirical data. We demonstrate that the digestive processes that break down contaminated food can lead to a substantial increase in chemical concentration in microplastics by a factor of 10–20, thereby reducing the overall chemical bioavailability in the gastrointestinal tract when compared to a scenario without microplastics. Higher enzyme activities result in more chemicals being released by the digested food, thereby resulting in higher chemical concentrations in the microplastics. While the model-calibrated kinetic parameters are specific to the studied scenario, we argue that the mechanism of the reduced bioavailability of chemicals and the modeling tool developed have generic relevance. These digestive processes should be considered when assessing the risks of microplastics to humans and also biomagnification in aquatic food webs.

-Details of PCB congeners and concentrations in spike stock mixture and experiment set-up Lipid extraction and separation method   Table S2-Average percentages of major fatty acids found in olive oil Table S3-Analysis of covariance (ANCOVA) to analyze the influence of different levels of micelle and oil mass concentrations on the partition coefficients, Kmicelle and Koil, respectively over the log KOW (covariate) range using the Type III test Table S4-Post-hoc multiple pairwise comparisons using Tukey technique for Kmicelle.   Table S8-Kinetic rate constants of LDPE for high and low enzyme treatments Table S9-Linear regression models of k1 and k2 for high and low enzyme treatments Table S10-Analysis of covariance (ANCOVA) to analyze the influence of enzyme treatments on the transfer kinetic rate constants, k1 and k2, respectively over the log KOW (covariate) range using the Type III test.    Hawker and Connell, 1988 2 Minimum detection limit is based on the lowest detectable concentration on the calibration curve a PCB 77 and 169 had high relative standard deviations (> 15%) for the response factors of replicate measurements of the calibration standards. Therefore, they were omitted from further analysis.

Lipid extraction and separation
The method for lipid extraction was adapted from Paik et al., 2009. Two hundred µL of sample lipid emulsion was vortexed with 2 mL ternary solvent (DCM:MeOH:H2O=1:2:0.8; v/v/v) briefly. Additional dichloromethane (DCM) and MQ-water were added to the extracts to adjust the ratio of DCM:MeOH:H2O to 1:1:0.9 (v/v/v). The extracts were then separated by centrifugation at 2500 ×g for 15 mins. The upper layer was pipetted out and the remaining DCM layer was dried over anhydrous Na2SO4 and then evaporated to dryness with a gentle stream of nitrogen. The extracts were then diluted in 12 mL hexane and then stored in -80°C until further extraction was carried out.
Lipid extracts were thawed to room temperature and concentrated to 1 mL with a speed vacuum for 15 mins. The triglycerides and free fatty acids were separated based on an adapted method from Richardson et al., 2017. Briefly, 1 mL of cold distilled water was added to the concentrated lipid extracts (1 mL) and vortexed. Then 1 mL of 0.4 M NaOH in MeOH was added and vortexed for 10s. This would allow the NaOH to react with the free fatty acids to form salts. Three mL of hexane were added immediately after vortexing to separate esterified fatty acids then vortexed again. The emulsion was then left to separate for 3 min before the hexane upper phase (contains esterified triglycerides) is transferred into a clean centrifuge glass tube and dried under a gentle stream of nitrogen to 20 µL. The esterified triglycerides were then derivatized to obtain fatty acid methylated esters (FAMEs) (Richardson et al., 2017). Briefly, 20 µL of esterified triglycerides extract was mixed with 400 µL toluene, 3 mL MeOH and 600 µL of 8% HCl solution in methanol and incubated for 1.5h at 90°C. Samples were cooled down for 10 min before adding 1 mL of hexane and 1 mL of water. After phase separation, 600 µL of the hexane upper layer containing FAMEs was used for further analysis. A double internal standard spike was applied during the esterification process. Firstly, 1.6 mg of methyl pentadecanoate (C15:0) was added before the samples were heated at 90 °C. Then, 1.6 mg of methyl tridecanoate (C13:0) was added during the hexane extraction step. FAMEs were quantified by GC-FID using a Nukol column (Breuer et al., 2013;Teuling et al., 2017). The GC was calibrated using TraceCERT FAME standards purchased from Supelco. Figure S1. Schematic diagram of two-compartment low density polyethylene (LDPE) with a fast (CP,1) and slow (CP,2) reservoir and chemical interactions between LDP, water, oil and micelles (formed from sodium taurocholate (NaTC) and free fatty acids (FFA)) in the simulated gut fluid digestion assay.

Determination of distribution coefficients Kmicelle and Koil
According to chemical mass conservation, the mass of chemicals residing in the three phases of the system (i.e., POM, water and micelle/oil) remains equal to the initial mass of chemicals at all times.
Here, we demonstrate the chemical mass balance for the micelle-water experiment: where is the total initial mass of PCB in the system, is the chemical mass on the POM passive sampler, is the chemical mass in the water phase and is the chemical mass in the micelle compartment. All chemical mass are in µg. Each of the terms in eq. S1 are divided by the volume of the system and this yields: Where each concentration term is in µg/L.
The partition equilibrium constant is: where [ ] is the mass concentration of POM in the system in kg/L (mass of POM divided by the total volume of the system) and KPOM is in L/kg. Hence, Similarly, is: Where [micelle] is the mass concentration of micelle in the system in kg/L, (L/kg) is the partition coefficint of micelle with water. Therefore, substituting eq. (S4) and (S5) into (S2): Rearranging and substituting Similarly, for the oil-water partition coefficient experiment, the can be calculated from the following equation:

S7
Biphasic sorption on microplastic model (based on Mohamed Nor and Koelmans, 2019) The exchange of chemicals between the fast and slow reservoir of the plastic is modelled as: where C1 * and C2 * are concentrations of contaminants in the fast and slow reservoirs of the polymer (µg/kg) respectively, f1 is the fast reservoir fraction of the total bound mass of chemical (dimensionless), k1 is the sorption rate constant (d -1 ), k2 is the desorption rate constant (d -1 ) and k3 is the intra-polymer rate constant (d -1 ), [P] is the plastic mass concentration (kg/L).

Auxiliary equations
Relationship between log KPOM and log KOW (Hawthorne et al., 2009): Percentage reduction in chemical bioavailability: × 100 (S12) Where (0) and ( ) is the total concentration in the system (µg/L) at time 0h or t, * (0) and * ( ) is the PCB concentration in LDPE (µg/kg) at time 0h or t, and [P] is the LDPE mass concentration in the system (kg/L).

Molecular weight of olive oil:
= 3 × . + 38.049 (S13) Where 38.049 g/mol is the the weight of the glycerol backbone and the average MWFFA is 279.33 g/mol based on the percentage of fatty acids shown in Table S3.

Partition coeffcient of LDPE with water
For units, please refer to previous section.

Calculation of chemical contribution from small MP particles (<10 µm)
MP in the environment or in diet components relevant for human consumption typically follow a power law: 'Abundance= b*size^(-alpha)' with alpha (i.e., the power law parameter) typically having a value between 1.5 and 3 (average power law = 2.5) . The average size of polydisperse particles 1 -10 µm ( , ) is calculated with : In which is the power law slope 'alpha' set at 2.5, xUL is the upper limit of the range (here 10 µm) and xLL is the lower limit of the default MP size range, set at 1 µm.
Equation S15 then yields an average size of MP particles in the range 1 to 10 µm, of 2.12 µm. The same calculation can be done for the other particles in a default MP size range i.e., from 10 to 5000 µm, which yields an average size of polydisperse particles between 10 and 5000 µm of 28.7 µm. Assuming uniform distribution of HOCs in the polymer, the chemical fraction present in the 1 -10 µm particle fraction can be calculated from the relative particle volumes in these fractions, which scale to size with a power of three, i.e. fraction of chemical in the potentially translocatable fraction of particles is 2.12 3 /(2.12 3 + 28.7 3 ) = 4×10 -4 . This assumes all < 10 µm MP particles pass the gut lining, while this is actually around 0.3% (Mohamed Nor et al., 2021). This reduces the chemical fraction available for translocation to 4×10 -7 of the total chemical mass present in ingested MP.
Given that the complementary fraction of particles that stays in the gut to be egested (i.e., 1 -4×10 -7 = 0.9999996, or 99.99996 %), is subjected to the same 10 -20 fold increase in chemical concentration, the net effect of MP mediated chemical transfer due to translocation will be overwhelmed (more than undone) by the attenuation of biomagnification and removal via egestion.
Note that using another alpha value, e.g. 2 or 3, does not affect this conclusion. Furthermore, extending the definition of MP to submicron particles (nanoplastics) reduces the average size of the particles < 10 µm even further, which further reduces the contribution via these small particles.   Figure S4. PCB concentrations in the LDPE compartment (µg/kg) over time (h) for high and low enzyme treatment and control (no PCBs spiked) of each replicate system. Solid lines represent the fitted models for the high enzyme treatment whereas the dashed lines represent the fitted models for the low enzyme treatment. Each colour represents a different replicate system.   Figure S6. Percentage increase in LDPE after 72h of lipase digestion. S20 Figure S7. Percentage reduction in bioavailability of each PCB congener in high and low enzyme treatments respectively (based on empirical data: plastic is removed from system at every sampling timepoint) S21 Figure S8. Percentage reduction in bioavailability of each PCB congener in high and low enzyme treatments respectively (based on simulated data: plastic remains constant in gut over time) Figure S9. Percentage distribution of the PCB18 (top) and PCB156 (bottom) in each compartment of the system (i.e., water, plastic, micelle and lipids) at 0h and 72h after digestion for high enzyme (left panel) and low enzyme (right panel) treatment in the experiment.
S23 Figure S10. Percentage distribution of the PCB18 (top) and PCB156 (bottom) in each compartment of the system (i.e., water, plastic, micelle and lipids) at 0h and 72h after digestion for high enzyme (left panel) and low enzyme (right panel) treatment in the environmentally realistic scenario that the plastic remains constant over time.