Grantee Research Project Results
Final Report: Systems Engineering & Analysis for Organotypic Culture Models
EPA Grant Number: R835736C005Subproject: this is subproject number 005 , established and managed by the Center Director under grant R835736
(EPA does not fund or establish subprojects; EPA awards and manages the overall grant for this center).
Center: Mickey Leland National Urban Air Toxics Research Center (NUATRC)
Center Director: Beskid, Craig
Title: Systems Engineering & Analysis for Organotypic Culture Models
Investigators: Wikswo, John , Cliffel, David , McLean, John , Shotwell, Matt
Institution: Vanderbilt University
EPA Project Officer: Aja, Hayley
Project Period: December 1, 2014 through November 30, 2018 (Extended to November 30, 2019)
RFA: Organotypic Culture Models for Predictive Toxicology Center (2013) RFA Text | Recipients Lists
Research Category: Chemical Safety for Sustainability
Objective:
Project 5 is designed to ensure physiologically realistic function of each of the mammary, limb, fetal membrane, and liver organotypic culture models (OCMs), operating separately and in combination, and will do so by coordinating the refinement of Integrated Organ Microfluidics (IOM) modules that will provide a common platform architecture.
Conclusions:
We received and analyzed media effluent samples from the Mammary Gland OCM (Project 1), Fetal Membrane OCM (Project 3), and the Liver OCM (Project 4). Typically, effluent samples were collected from cell-loaded and cell-free bioreactors, as well as perfusion media from the syringe. For the Fetal Membrane OCM, we analyzed samples from both sides of the membrane. These samples were analyzed using reverse phase liquid chromatography (RPLC) and ion mobility-mass spectrometry (IM-MS). Project 2 elected not to conduct IM-MS analysis as it is not a milestone requirement for their OCM. Upon differential meta-analysis of the individual toxin-control data pairs, a network prediction algorithm identified potential response pathways dysregulated when cells are present or when they are exposed to a toxin or drug, clearly demonstrating the power of the OCM-RPLC-IM-MS methodology.
Project 5 has developed and delivered standalone pumps, valves, Integrated Organ Microcluidics (IOM) Perfusion Controllers (IOM-PCs, v2.0), and MicroClinical Analyzers (µCAs, v2.0) as requested by organ-chip developers to support their Organ Chip Modules (OCMs). A variety of Rotary Planar Peristaltic Micropumps (RPPMs) were in regular use by Project 1, which evaluated in detail the five-port Rotary Planar Valves (RPVs). Project 5 has made significant progress in developing multi-port RPPMs and RPVs that bring new capabilities to the pumps and valves to perfuse OCMs and reduce the cost and increase the ease of use of the hardware. We designed, fabricated, and tested 2x8-port RPVs, and eight-port RPPMs. A six-port RPPM is now being produced for the Project 1 mammary gland OCMs. VPROMPT is benefiting directly from a separately funded development of a new V3.0 motor cartridge, with circular through-plate fluidics, and an openable Puck OCM. The production of IOM fluidic chips has been streamlined with a new, soft-lithography molding process. The V3.0 IOM through-plate fluidic architecture has enabled the development of new and improved IOM fluidic chips, including Ridge, Capacitor, Push-Pull, and LoFlow RPPMs. Significant progress has been made towards full wireless control of both a Triple NEMA-8 IOM PC and a single V3.1 SmartMotor IOM Module, and the latter will be available in Q2 2020. We also have worked closely with Project 2 to develop an automated bone-construct limb-segmentation bender.
During the course of the VPROMPT Project 5 effort, VIIBRE has focused its fluidic development efforts towards the creation of compact, low-cost fluidic control components (chips, pumps, valves, enclosures, etc.) that could be produced at a sufficiently low cost to be used in very large quantities and treated as disposable. We have succeeded in simplifying the design and operation of our hardware and software, improved its reliability, and made major strides towards conversion of OCM in vitro modeling from low- to medium throughput. The vast majority of the devices presented in this report can now be obtained, at cost of production, from the Vanderbilt MicroFabrication Core.
In the design of the original 2014 proposal, we utilized staged milestones that were distributed over the project duration, with numbering that tracked the project year. Now that the entire project is complete, this staging is a distraction, so in this final summary, we group the milestones into three categories: Hardware, Software, and Measurements/Models. We provide subheadings where appropriate and concatenate the internally developed goals for the grouped milestones.
HARDWARE
There were multiple components to the hardware effort, including the delivery of simple pumps and valves, the development of improved component production techniques, increasing the number of OCMs that can be supported by a single pump or valve, and the creation of integrated systems. Except for the final implementation of wireless motor control, which has been delayed by an unforeseen bug in the firmware of a commercial stepping-motor microprocessor, all hardware milestones have been completed.
Stand-Alone Pumps and Valves
A large quantity of stand-alone Rotary Planar Peristaltic MicroPumps (RPPMs), Rotary Planar Valves (RPVs), the associated microfluidic chips, associated support hardware, VIIBRE’s custom Automated MultiPump Experiment Running Environment (AMPERE) software, and low-cost notebook computers have been delivered to Project 1. Project 2 has utilized Project 5-developed hardware to flex collagen-chondrocyte constructs and will be utilizing automated pumps and valves in future studies. Project 3 has successfully translated Project 5 production of fetal membrane bioreactors to their laboratory. Project 4 has utilized our AMPERE and our valves in liver OCM experiments.
As our hardware systems were introduced to varied working environments, we encountered some serious, unanticipated problems with electromagnetic and electrostatic interference, crosstalk, and sensitivity to noise within the shaft encoder for the valve and the Four-Motor MicroController. We resolved these problems with a number of hardware, firmware, and software modifications. The shielded motor enclosure and commercial cables in the V3.0 motor cartridge provides a more elegant solution. We also added a registry corrector to detect and correct any static issues that cause the registry bits to change.
These noise issues and the introduction of more complex fluidic chips that support multichannel pumps and valves led to a complete reexamination of the motor cartridge. As a result of our production of hundreds of these cartridges, we have over the past several years analyzed their limitations and considered remedies. With a twelve-month contract from the Defense Thread Reduction Agency (DTRA), we developed de novo a totally enclosed, electromagnetically and electrostatically shielded motor cartridge with through-plate fluidics for our RPPM and RPV IOM PCs for the VIIBRE NeuroVascular Unit / Blood-Brain Barrier (NVU/BBB) OCM. The VPROMPT project is the immediate beneficiary of this unanticipated opportunity for hardware development, and V3.0 systems were immediately made available to VPROMPT investigators.
The greatest single improvement to the RPPM and RPV cartridges was to move from the V2.0 open-frame, naked-motor design to the fully enclosed, wipe-sterilizable V3.0 motor cartridges with through-plate fluidic chips and reduced sensitivity to electromagnetic interference. This has revolutionized device production and ease of use. We are no longer producing V2.0 motor cartridges, and whenever one comes in for service or fluidic replacement, we upgrade it to V2.5 that still has the open frame but utilizes through-plate fluidics.
The pumps currently available to project investigators and external users included VIIBRE V2.0 LoFlow RPPMs that are rated at 0-7 μl/min, and higher flow V3.0 Ridge RPPMs (0-70 μl/min), and V3.0 6- and 12-channel Spiral RPPMs. Production of the V2.0 5-port RPV was maintained until the V3.0 Ridge RPPM and the 25-port valve with through-plate fluidics were proven easier to manufacture and assemble and were found to be more useful to both Project 1 and the Cliffel group for their MicroClinical Analyzers (µCA). A number of custom valves were produced for Project 1, including a 2x8 make-before break RPV. We have preliminary designs for a 100 channel RPV that could direct the effluent from 100 OCMs to either an online µCA or mass spectrometer. These will be moving into regular production as standard units in response to demand by Project 1 and other collaborators. The significance of our V3.0 through-plate cartridge design is that the identical motor and enclosure can be used for a variety of pump and valve implementations by simply selecting the desired through-plate fluidic chip and actuator and the appropriate AMPERE software driver. The potential importance and widespread applicability of our multi-channel Spiral RPPM with zero backflow and minimal peristaltic pulsations are worthy of note.
Improved Component Production Techniques
To support a wide variety of OCMs, VIIBRE continued to advance all aspects of the platform. In addition to the introduction of a sealed, wipe-sterilizable V3.0 motor cartridge, significant progress has been made in terms of ease of manufacture and production repeatability. These include custom tooling, vendor-fabricated assemblies, and a custom cable connector that includes a custom printed circuit board, minimizing manual soldering. The design of the V3.0 cartridge emphasizes easy, quick, and foolproof assembly, focused mainly on decreasing the number of unique parts by using off-the-shelf and combining parts as often as possible. The baseplate in which the through-plate fluidics reside is standardized and allows users to easily change from a pump to a valve or vice versa with little change in hardware. Usability was the primary driving factor behind most design decisions, including compatibility of all materials with chemicals common in collaborator labs, and an enclosure that resists fluid intrusion and protects electronics against electrostatic discharge and external electromagnetic interference. This housing also protects against manual damage, acts as a heat spreader, and provides common mount points to allow the device to be mounted to a caddy in a number of orientations or for the attachment of accessories such as an external flow meter.
To further reduce the production costs of the IOM pumps and valves, we have replaced the time-consuming, classic SU-8 photoresist-on-silicon master-mold soft-lithography process with one that is faster, easier, and lower cost, enables the production of 6 to 12 fluidic chips on a single pour of PDMS, and produces a much higher reproducibility between devices and masters than is possible with spun and photocured SU-8, for which thickness is difficult to control due to a highly temperature-dependent viscosity. Our refined manufacturing process, transferred from the V2.0 cartridge design to the new V3.0 circular, through-plate fluidic chip, begins with designing the CAD schematics for the channels. As the next step, a ball end mill in a computer numerically controlled (CNC) milling machine is used to cut channels in a sheet of acrylic to create a master mold. The master mold is used to create the lower half of the secondary mold with raised ridges that will create the channels, and a laser cutter is used to create the upper half of the mold. The secondary mold is filled with PDMS, allowed to cure slowly at room temperature to avoid distortions related to thermal expansion and contraction of the mold and the PDMS for the fluidic chip, and then opened, trimmed, and punched for tubing ports. A membrane is bonded to the back of the chip to seal the channels. After punching, the resulting circular fluidic chip is ready to use.
Extensive testing of pumps and valves produced with parametric variations of multi-device masters has allowed us to increase our production and quality control of both RPPM and RPV designs for the µCAs and OCM perfusion controllers. We now produce RPPMs that span a range of flow rate from 0.5 to 800 µL/min, and circular-segment 5-port and 25-port RPVs as needed for the integrated systems discussed below. A key feature of our mold fabrication and casting process is that we can quickly design, fabricate, and test new valve designs with a several week turnaround at a cost of one or two thousand dollars. A recent 24-port valve design minimizes series resistance by using small lines only in the valving regions, with larger channels in all supply and delivery lines.
We have begun to design a process to create cast-in-place tubing ports to eliminate the inaccuracies and tapered holes from manual port punching, but this project ended before it could benefit from this new fabrication approach. Subsequent EPA and other projects will be able to use these valves. Ultimately, when production volume justifies the expense of creating an injection mold, the cost of production of pump and valve fluidics will drop substantially.
Increasing the Numbers of OCMs Used in Parallel
The development of 25-port RPVs, 2x8 RPVs, 6- and 12-port RPPMs, the design of a 100-port RPV, and the demonstration of AMPERE code to drive them represent major steps towards medium throughput OCM studies. The forthcoming introduction of cast-in-place tubing ports, ribbon fluidics, and standardized connectors will soon make it easier to parallelize OCM measurements. The limitations of the numbers of OCMs that can operate in a single incubator will soon shift from hardware to the cost of cells, the volumes of media that have to be prepared and stored, and the technician time to load, curate, and analyze the OCMs. The further refinement of AMPERE to support automated cell-loading and analysis protocols will be another important step towards medium throughput studies. The next step, for which we are seeking funding, will be to create a robotized, 10,000-well perfusion culture system that builds upon the foundation provided by VPROMPT.
Creation of Integrated Systems
The numerous NEMA-17 multichannel pumps and valves used for Project 1 as PCs and the µCAs developed for and used by the Cliffel group are excellent examples of systems integration made possible by AMPERE. We delivered µCA systems to Projects 1 and 3, and to Project 4 at Pittsburgh (under the auspices of a different grant). Project 2 does not yet require a fluidically enabled bioreactor. µCA operation initially required our 5-port RPVs, which have been demonstrated to perform more than 30,000 full operational cycles (with 10-30 sec pauses at each valve position, 5 positions total). The μCA module has been shown to operate continuously and automatically for 18 days, requiring only the expected replacement of sensors and calibration solutions. Upgraded and new μCAs are being shipped with 25-port RPVs to allow broader ranges of calibration concentrations. We currently provide and support µCA V2.0 and V3.0 modules with clamped sensors and microfluidics, with a new, custom, screen-printed electrode (SPE) produced to our specifications by Pine Instruments that provides nine working electrodes, as well as reference and counter electrodes. We have shown how this SPE can be used to measure media conductivity as needed to correct for evaporation, and developed and calibrated sensors for glucose, lactate, glutamate, ammonia, and cholesterol.
Our µFs, developed with support from AstraZeneca, NIH/NCATS, and CN Bio Innovations, were not needed by VPROMPT investigators. We produced, with other support, two 96-channel µFs and eight 24-channel µFs, and hence these systems are available to future EPA investigators. We have developed a prototype monolithic integrated triple NEMA-8 module that includes on its input side a 5-port input RPV for reagent/media/toxin selection; a 100 µl/min RPPM for toxin delivery; and a 24-port output RPV for bioreactor selection, with a symmetric, RPV-RPPM-RPV system for sample withdrawal.
We have recognized that four of the 24-channel µFs that can be created with a 5-port RPV, an RPPM, and a 24-port RPV can be used to individually address each well and insert in a 24-transwell plate, and enable automatic measurement of the TransEndothelial Electrical Resistance (TEER) of each insert membrane. This concept will be tested as soon as we have an appropriate application.
We have designed several iterations of a smaller Version 2.0 µCA, internally referred to as the NanoClinical Analyzer (nCA) due to its 250 nl sensor chamber volume. This analyzer version, in combination with the use of PEEK tubing, would allow us to use a total sampling volume of approximately 5 μL. Work stopped on this project for lack of a VPROMPT project needing this specific technology.
While the technical capabilities for integrated systems developed under Project 5, for example controlled media recirculation, timed gradual replacement, and programmed delivery of drugs and toxins to match pharmacokinetic (PK) profiles, exceeded the needs of Projects 2, 3, and 4, these technologies are now readily available to EPA and other investigators through the Vanderbilt MicroFabrication Core (VMFC), with transactions handled by the Agilent iLab core management system. Our 2x8 valve is ideal for an OCM bioreactor with an on-line µCA to return the sampled media back to the organ bioreactor reservoir, thereby allowing more frequent measurements of cellular bioenergetics without depleting the local media reservoirs.
Wireless Control of OCMs
At the beginning of the VPROMPT project, we were evaluating a prototype wireless-control triple-NEMA-8 module. As a major step towards full wireless control, we simplified the firmware for the 3xNEMA-8 hardware as compared to the firmware for the Four-Motor MicroController, which was developed over the preceding six years. We simplified the command set so that commands with lower-case letters correspond to reading the values to the parameters, and upper-case letters correspond to writing the values to the parameters. With the introduction of the V3.0 NEMA-17 motor cartridge, we realized that there was room inside the housing to include a commercially available Mechaduino microprocessor motor controller. The higher torque provided by the NEMA-17 motors over the NEMA-8 led us to defer completion of the triple NEMA-8 PC until after the end of VPROMPT. We made great strides towards completion of a wireless NEMA-17 V3.5 motor cartridge, but we will not be able to implement this technology until we identify and correct a bug in the commercial Mechaduino microprocessor firmware, a straightforward but somewhat tedious process. We hope to be able to deliver V3.5 wireless motor cartridges in Q2 2020. We have prototyped a wireless power system for integrated IOM systems that uses Qi wireless chargers at each workstation and incubator, and an incubator-rated, compact, sealed, gel, lead-acid battery that will allow any of our integrated, multi-OCM instruments to be transported from cell culture hood to incubator to microscope and back without interrupting fluid flow. We expect this module will be available in late 2020. These will represent yet another significant advance in the OCM state-of-the-art.
SOFTWARE
VIIBRE’s Automated MultiPump Experiment Running Environment (AMPERE) software operates on a low-cost notebook computer and drives VIIBRE’s custom Four-Motor MicroController. With funding from EPA, NIH, and the Defense Threat Reduction Agency (DTRA), we have continued over the past five years to refine the software and made almost innumerable improvements to AMPERE and the microcontroller firmware it controls, many of which are invisible to the user but provide a more robust, error-free operation. We have instituted version-tracking and bug-reporting software that allows us to continue to test and refine AMPERE as we receive user feedback. We have shown that AMPERE is well suited for fully automated perfusion and testing of organ bioreactors using VIIBRE PC and µCA modules. AMPERE enables users to build custom perfusion and testing protocols and can trigger the customized CHI-1440 multipotentiostat to collect measurements while conducting automated calibrations and sample delivery for the μCA. We have worked continuously through the project to improve new releases based on user feedback. We developed an AMPERE wrapper that controls 6- and 24-motor, 24- and 96-channel µFs to provide timed toxin exposure, including synthesis of pharmacokinetic (PK) exposure profiles. Stress testing of AMPERE at Vanderbilt and AstraZeneca showed that AMPERE and a single notebook computer could successfully handle twenty-four motors simultaneously (sixteen RPPMs and eight 25-port RPVs plugged into six four-motor controllers connected to the notebook computer by a USB hub).
Multiple additions to AMPERE make the software and pump and valve hardware easier to use. Valves now have customizable port names to select positions rather than motor angles. The Controller Manager uses a discovery functionality to automatically detect types of controllers connected to the notebook computer. We improved the Timer Widget’s look and functionality. The development of the Well Plate Tool has simplified the operation of the MicroFormulator system. It provides biologist-friendly commands to drive each OCM or well with a different pharmacokinetic profile without having to configure multiple devices together. The Schedule View has given the user a timeline interface which improves the clarity of the operation of hardware in relation to each other over time. Quick View buttons and layout options have been added to easily configure the screen layout. A toolbar has been added to quickly use common activities on a particular device type. Loading experiments can now be reconfigured with different controllers. We added a Calendar to schedule experiments to run at certain times of the week. In order to simplify the use of automated pumps and valves in organ-chip experiments, we have created a “Bench” tool, which provides an intuitive interface for AMPERE users to build protocols to connect pumps, valves, experiments, and reservoirs.
There have also been improvements to the accuracy and efficiency of the microcontroller firmware that operates in the Four-Motor MicroController. We added a time tracker to smooth the stepping of a pump to achieve a more accurate RPM over time. We reworked the Tare procedure for valves. We mitigated issues with the stepper motor taking the incorrect number of steps. We added the ability to “blink” a port in AMPERE to help with hardware identification. We also added the ability to modify the default direction of the motor. As discussed under Hardware, we are putting into place software features to support the wireless Triple NEMA-8 and single V3.5 NEMA-17 SmartMotors. We have developed firmware to drive the radio of the Single V3.5 NEMA-17 IOM. We have also created a Wireless Manager in AMPERE that can detect the radios and set up network Transmission Control Protocol (TCP) communication.
We have developed an AMPERE AutoCalibration Tool for fully automated, error-correcting, long-duration, gravimetric calibration of coupled RPPM-RPV modules, such as with the µF, and created interfaces for AMPERE to control an IDEX HPLC valve used by Project 1, multiple Sensirion flow meters and an electronic digital scale for pump calibration, and a CCD inspection camera to coordinate the measurements of pump transients with actuator position.
MEASUREMENTS/MODELS
We received and analyzed media effluent samples from the Mammary Gland OCM (Project 1), Fetal Membrane OCM (Project 3), and the Liver OCM (Project 4). Typically, effluent samples were collected from cell-loaded and cell-free bioreactors, as well as perfusion media from the syringe. For the Fetal Membrane OCM, we analyzed samples from both sides of the membrane. These samples were analyzed using reverse phase liquid chromatography (RPLC) and ion mobility-mass spectrometry (IM-MS). Project 2 elected not to conduct IM-MS analysis as it is not a milestone requirement for their OCM.
For the Liver OCM, our metabolic secretion baseline signature revealed multiple significantly different metabolite compounds. Pairwise comparisons were prioritized using the following significance criteria (p ≤ 0.05 and a fold change of ≥ 2 or ≤ 1/2). Using this analysis, 116 metabolite compounds were secreted (or not consumed) and 106 were consumed (or not secreted) when comparing the cellularized Liver OCMs vs. the cell-free Liver OCMs or media (syringe). These significantly changing metabolite compounds unique to the cellularized bioreactor, either secreted or consumed, represent the metabolic secretion baseline signature for the Liver OCM. Tentative structural identifications of the baseline signature metabolic compounds have been obtained and can be used for later targeted analysis and untargeted comparison of toxicant-challenged Liver OCMs. We then demonstrated metabolic changes in a toxicant-challenged Liver OCM device via IM-MS. Influx and efflux media samples chosen for the analysis included 220 µM (10x Cmax) tolcapone, 40 µM (10x Cmax) entacapone, and vehicle controls.
For the Mammary Gland OCM, to prioritize compounds from this untargeted analysis, we used significance criteria of Anova p≤0.5 and a fold change of ≥2, resulting in 197 metabolite compounds that were upregulated and 188 that were downregulated in cellularized bioreactors vs. media (syringe) or the cell-free device vs. media (syringe). These significantly different metabolite compounds unique to the cellularized bioreactor, either secreted or consumed, represent the metabolic secretion baseline signature for the Mammary Gland OCM. Tentative structural identifications of the baseline signature metabolic compounds could be used for later targeted analysis and untargeted comparison of toxicant-challenged bioreactors.
We received and analyzed effluent samples for the Fetal Membrane OCM to establish metabolic secretion baselines for both upper and lower chambers: cell-free media, cellularized bioreactor effluent samples, before and after a 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD, i.e., dioxin) toxin exposure, as well as infusion media (syringe). We created a global view of the metabolic profiles by heat mapping the data to demonstrate unique differences between the four sample groups, for both upper and lower chambers, and to illustrate distinct metabolic profiles for the media passing through an empty device vs. when cells are present or exposed with the toxin TCDD. Using RPLC and IM-MS, we identified significantly different metabolite compounds for each pairwise comparison grouping common and unique metabolites. A large number of changing metabolite compounds that met our selection criteria (p-value ≤ 0.05, fold-change (FC) ≥ |2|) were unique and specific to the cellularized Fetal Membrane OCM and represent the metabolic secretion baseline signature. A network prediction algorithm was used to high-light potential response pathways dysregulated when cells were present or when they were exposed to the TCDD toxin, for both the upper and lower chambers of the bioreactor. Upon differential meta-analysis of the individual data pairs, a network prediction algorithm has been used to highlight potential response pathways dysregulated when cells are present or when they are exposed to the TCDD toxin, for both the upper and lower chambers of the bioreactor, demonstrating our ability to identify significant response pathways based on the network activity prediction analysis for the upper chamber of the Fetal Membrane OCM when cells are present or exposed with the toxin TCDD. A similar analysis was performed for the lower chamber.
The MicroClinical analyzer (µCA) was used to analyze a complete set of samples to identify the metabolic changes associated with hepatotoxicity following exposure of the Liver OCM to rosiglitazone, a known hepatotoxic pharmaceutical. A low interferent osmium-based glucose sensor was used to analyze these Liver OCM samples. The primary hepatocytes were treated with either rosiglitazone or a vehicle. In recognition that the degree of hepatotoxic response may depend upon the state-of-health of the liver, we examined the media from Liver OCMs from control cells from a liver that had a benign tumor, and diseased cells from a liver that had cancer and had been given a single treatment of a chemotherapeutic. We found that rosiglitazone primarily affects diseased hepatocytes to decrease cellular glucose metabolism, diseased hepatocytes show increased lactate production when compared to healthy hepatocytes, and, when treated with rosiglitazone, diseased cells showed a significant decrease (p<0.05) in lactate production as well as possible increased lactate consumption. These studies demonstrated the ability of OCM-µCA studies to identify toxin-related metabolic changes in cells. Ultimately, this would be done on-line, with a valve directing media from multiple OCMs to a single µCA.
From this work, we recommended a uniform protocol for determining AC50s and maximal responses based on non-linear least squares regression methods. We have discussed potential methodologies for clustering metrics and for constructing meaningful prediction models.
We collaborated with researchers at EPA’s National Center for Computational Toxicology on cell-agent-based models of how chemicals disrupt particular events during embryogenesis.
There is a growing interest in the use of OCMs to evaluate chemical toxicity in human cells cultured in appropriate 3D heterotypic environments; however, these microsystems are often fabricated from PDMS, which has high affinity for small hydrophobic molecules. To understand how this affects a toxicity assay, we have modeled chemical adsorption onto PDMS surfaces and the transport of potential toxicants in organ-on-chip microsystems using computational fluid dynamics. The goal of this model is to predict the cultured cells’ actual time-dependent toxicant exposure. The parameters for our modeling effort were extracted from experiments in which solutions were placed in contact with PDMS surfaces for variable durations and depletion of a specific chemical from solution was measured using UV-vis spectroscopy. The seven selected potential toxicants showed a wide range of PDMS-binding behavior spanning high- and low capacity and reversible and irreversible binding. We determined that timing is critical for delivery of toxicants that reversibly bind to PDMS in order to avoid over- or under-dosing cells. For such toxicants, a bolus dose at the inlet may translate into an extended exposure for cells in the device due to delayed release of the chemical from PDMS surfaces. In addition, for toxicants with strong affinity to PDMS, the actual exposure may be an order of magnitude less than the nominal inlet concentration. The model can be used in a forward direction to properly evaluate toxicity by accurately quantifying cellular exposures or in a backward direction to achieve desired exposures by optimizing inlet dosing strategies.
In moving towards prediction models that link the outputs of OCM assays to toxicity in humans, we tested outputs from the Liver OCM for 15 drugs for their correlation with the frequency of adverse liver function tests reported in prescription databases. We found a high correlation (coefficient β = 0.816) by simply comparing the clinical frequency against the Number of Adverse Responses in a Liver OCM (NARLOCM; Figure P5-27). Digging deeper, we find that NARLOCM is highly correlated not only with clinical adverse responses, but also with ToxCast results, specifically the total number of “hits” for each chemical against HepG2 cell assays. There was even a correlation with the total number of hits against all ToxCast assays (not surprising, given the high correlation between total hits and HepG2 hits). We evaluated NARLOCM for correlations with individual ToxCast assays, but having a limited test set of just 15 chemicals (not all of which are in ToxCast) was insufficient to construct any more detailed assay-specific model. Nonetheless, our preliminary work shows that OCM results can certainly add value to ToxCast results and can be predictive of human clinical outcomes.
Journal Articles on this Report : 2 Displayed | Download in RIS Format
Other subproject views: | All 38 publications | 8 publications in selected types | All 8 journal articles |
---|---|---|---|
Other center views: | All 169 publications | 57 publications in selected types | All 56 journal articles |
Type | Citation | ||
---|---|---|---|
|
Miller DR, McClain ES, Cliffel DE. Electrochemical Microphysiometry Detects Cellular Glutamate Up-take. Journal of The Electrochemical Society 2018;165(12):G3120-G3124. |
R835736 (Final) R835736C005 (2018) R835736C005 (Final) |
Exit Exit Exit |
|
Rogers M, Sobolik T, Schaffer DK, Samson PC, Johnson AC, Owens P, Codreanu SG, Sherrod SD, McLean JA, Wikswo JP, Richmond A. Engineered microfluidic bioreactor for examining the three-dimensional breast tumor microenvironment Biomicrofluidics 2018;12(3):034102. |
R835736C005 (Final) |
Exit Exit |
Supplemental Keywords:
Microfluidic pumps and valves, integrated organ microfluidics.Relevant Websites:
Vanderbilt Institute for Integrative Biosystems Research and Education Exit
The Vanderbilt-Pittsburgh Resource for Organotypic Models for Predictive Toxicology (VPROMPT) Exit
Progress and Final Reports:
Original AbstractMain Center Abstract and Reports:
R835736 Mickey Leland National Urban Air Toxics Research Center (NUATRC) Subprojects under this Center: (EPA does not fund or establish subprojects; EPA awards and manages the overall grant for this center).
R835736C001 Mammosphere Bioreactor For Life-Stage Specific Toxicology
R835736C002 Organotypic Culture Model to Analyze DevelopmentalLimbMalformationsResulting from Toxicant/Teratogen Exposure
R835736C003 Validating a fetal membrane on a chip model for characterizing
reproductive toxicant exposure risks
R835736C004 Organotypic Liver Model for Predictive Human Toxicology and Metabolism
R835736C005 Systems Engineering & Analysis for Organotypic Culture Models
The perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.
Project Research Results
- 2018 Progress Report
- 2017 Progress Report
- 2016 Progress Report
- 2015 Progress Report
- Original Abstract
8 journal articles for this subproject
Main Center: R835736
169 publications for this center
56 journal articles for this center