Grantee Research Project Results

Final Report: Measuring Toxicokinetics for Organ-on-Chip Devices

EPA Grant Number: R840031
Title: Measuring Toxicokinetics for Organ-on-Chip Devices
Investigators:
Institution:
EPA Project Officer:
Project Period: August 1, 2020 through May 6, 2025
Project Amount: $790,352
RFA: Advancing Toxicokinetics for Efficient and Robust Chemical Evaluations (2019) RFA Text |  Recipients Lists
Research Category: Chemical Safety for Sustainability

Objective:

As a way to reduce the need for animal testing, EPA and other regulatory agencies have funded investigations of organotypic culture models and organ-on-chip devices. These new approach methodologies place multiple human cell types in appropriate 3D geometries under continuous microfluidic perfusion to better approximate in vivo cellular microenvironments – and thus yield more predictive responses to potential toxicants. Nonetheless, translating organ-on-chip results to predict human health effects still requires in-vitro-to-in-vivo extrapolation. Such extrapolation is always difficult, but becomes even more complicated for organ-on-chip devices because their high surface-to-volume ratios and permeable materials such as polydimethylsiloxane (PDMS) can sequester hydrophobic compounds. Using results from these devices thus requires two calculations: (1) from nominal inlet concentration to in-device cellular dose; and (2) from that dose to equivalent organismal exposure. The latter has been the subject of decades of work, but the former is just beginning to be explored. Our primary objective is to establish methods, measurements and models for the toxicokinetics of PDMS-based organ-on-chip devices.

Summary/Accomplishments (Outputs/Outcomes):

None of our goals or hypotheses were modified from the original application. As proposed and pursued, this project had three major Aims:

1. Assess how chemical-PDMS interactions and in-device toxicokinetics impact toxicity research using organ-on-a chip devices.
   1. Measure the parameters describing PDMS interactions and transport kinetics for a few target chemicals.
   2. Investigate how these parameters are changed by mitigation strategies, e.g., PDMS annealing or plasma treatment.
2. Measure the parameters describing PDMS interactions and transport kinetics for a larger list of 48 chemicals.
3. Develop a quantitative structure-property relationship (QSPR) model to predict chemical-PDMS interaction parameters to enable prospective toxicokinetic modeling for any chemical.

By the end of this project, we completed Aims 1 and 2, including the extension of Aim 1b to test additional mitigation strategies such as replacing PDMS with a styrene-ethylene-butylene-styrene (SEBS) co-polymer or adding a carrier protein (e.g., bovine serum albumin) to the aqueous media. When we proceeded towards Aim 3, using the data from Aims 1 and 2 to train a QSPR model, we found that the model had limited predictivity. The training data was insufficient to build a robust model.

The outputs/outcomes from this project include the following:

1. We have developed calibration protocols based on UV/Vis spectroscopy and multi-wavelength Partial Least Squares Regression for non-destructive measurements of the concentration of a chemical remaining in solution. We have used these protocols to assess limit-of-detection for cuvette-based and ATR-crystal-based measurements.
2. We have conducted disk-soak and diffusion-through-membrane experiments, and we have developed nonlinear regression methods for simultaneously fitting these experimental results to models that return well-constrained estimates of the relevant parameters: PDMS-to-water partition coefficient, K; and diffusion constant in PDMS, Dp.
3. We have validated methods for conducting disk-soak and diffusion-through-membrane experiments for chemicals with limited water solubility. Experiments can be done in 30% to 70% water-DMSO mixtures and the results extrapolated to pure aqueous solution using a log-linear relationship for the PDMS-to-water partition coefficient, K.
4. We have used disk-soak experiments for indole, which has high affinity for and high diffusion through PDMS, to test how PDMS interaction parameters vary among PDMS lots and how they are altered by annealing PDMS.
5. We have used disk-soak and diffusion-through-membrane experiments to evaluate several mitigation strategies for reducing chemical sequestration by PDMS.
   1. We found that replacing PDMS with an alternative SEBS co-polymer does not reduce partitioning of hydrophobic compounds into the polymer, but it does strongly reduce diffusion of these compounds deeper into the polymer bulk.
   2. We found that addition of 5-10% (w/v) of bovine serum albumin to an aqueous solution can strongly reduce partitioning of hydrophobic chemicals into PDMS. Lower albumin concentrations (1% or less), which are more consistent with the amounts used in cell culture media, have limited impact.
6. We have conducted experiments in which fluorescent dyes were observed via confocal microscopy as they diffuse out of a solution-filled channel and into a surrounding block of PDMS.
   1. We validated that the diffusion constants estimated from these experiments match those from our disk-soak and diffusion-through-membrane experiments.
   2. We also found that many dyes exhibit anomalous sub-diffusive behavior in PDMS. We have characterized this sub-diffusive behavior in terms of time-fractional diffusion and shown its key implications for chemical crosstalk between nearby channels in PDMS-based devices.
7. We have used the above experimental protocols and simultaneous regression techniques to complete measurements of PDMS interaction parameters and in-PDMS diffusion coefficients for 41 additional chemicals of interest. This set of chemicals covers a wide range of physico-chemical properties.
   1. We found that the hormones progesterone and estrogen significantly interact with PDMS, which has important implications for reproductive toxicology experiments in organ-on-chip systems. We measured their PDMS interaction parameters had modeled their distribution under realistic flow and microchannel geometries.
8. We have developed partial-differential-equation models of partitioning into PDMS and diffusion through PDMS to predict chemical distribution within the user-defined geometry of specific microfluidic devices at different flow rates. We confirmed that these models can accurately predict chemical loss under continuous flow through a microfluidic channel in PDMS.

Journal Articles:

No journal articles submitted with this report: View all 5 publications for this project

Progress and Final Reports:

Original Abstract

2021 Progress Report

2022 Progress Report

2023 Progress Report

2024 Progress Report