A microphysiological system for studying barrier health of live tissues in real time

Microfluidic chamber design

The microfluidic chamber (Fig. 1a–e), to be referred to as chamber for short, was designed using CAD (Autodesk, Inc) and fabricated using Anycubics UV sensitive resin and SLA printer. To avoid harmful effects from uncured resin, each chamber is fully cured using UV light and thoroughly rinsed with isopropyl alcohol. It is further sterilized in a low-temperature autoclave. The chamber consists of two halves assembled to hold the tissue and make connections to the integrated TEER electrodes. Each half chamber is composed of the following (Fig. 1a): the chamber body; two 1 mm thick PDMS layers for holding the electrode chip in place; one gold electrode chip; an aluminum clamp; and a printed circuit board (PCB) with spring headers that connect to the electrode chip. The fully assembled chamber (Fig. 1b) consists of two halves, the top half has spikes to hold the tissue tight when the chamber is sealed; the bottom half has corresponding openings for these spikes and is where the tissue is placed before closing the chamber. Figure 1c shows how the tissue explant positioned on the bottom half chamber, before and after the experiment. When the tissue is enclosed in the device, the chamber spikes puncture around the outer edge of the tissue but leaves the center untouched. A PDMS layer is placed over the spikes of the top chamber to create a flush seal against the tissue and prevent leaks between the chamber halves after the chamber is closed. A fully assembled chamber is shown in Fig. 1d, e in the front and the back view of the device, respectively.

Tubing is connected to each chamber half through Luer lock connectors. Media is pumped into the chamber using custom-designed syringe pumps where users can specify start and stop time points throughout the experiment period. With the tissue positioned over the circular opening between the chamber halves, a barrier is formed between the two media flows, one for the serosal side and the other for the luminal side of the tissue. The microfluidic path flows over the opening on each side exposing the tissue to the media composition. Providing balanced flow for the tissue inside the chamber is important for controlling shear stress and extending tissue viability^10,13,20. The chamber was designed using 3D fluid simulations (CFD, Autodesk Inc) to make the media flow over the tissue area with uniform velocity as possible. Figure 1f shows an example of the flow simulations performed during design.

Each chamber half has its own PCB breakout board that connects to the glass chip electrodes through gold spring headers. The spring headers are compressed against the chip during assembly. The PCB on the top half chamber has external wire connectors for connecting to the bottom halve PCB. The bottom PCB includes a card edge connector that is plugged into the top of the enclosure for the microphysiological system (see section “System Overview below”). The chamber, when plugged into the system, is oriented vertically, making the media flow in from the bottom and out above the tissue (Fig. 1f). This orientation helps push air bubbles to the top and get them pushed out of the media outlet during experiments. Air bubbles can injure the tissue and cause large deviations for the TEER measurement.

System overview

The entire electronic support system is housed in a metal enclosure (Fig. 1g, h) to shield all electronics from the external physical environment as well as EMF noise. USB ports provide user configuration and control of the experiment from the host computer. The connectors on the microfluidic chambers and the system enclosure are all universal, allowing for plug-and-play functionality. The current implementation can hold up to three chambers at a time (Fig. 1g, h). The entire system has a footprint of a typical laptop computer designed to allow it to fit into a limited environment chamber space during experiments. The supporting electronics are responsible for signal acquisition, signal conditioning and amplification, analog to digital conversion, and communication with the host computer via the USB protocol. A custom-built graphic user interface (GUI) was designed to allow user configurations and real-time control for the experiment. The TEER measurement data are acquired by the host computer and TEER results are calculated and displayed in the GUI.

Electronic circuits for TEER measurement

The block diagram of the electronic circuits to perform TEER measurement is shown in Fig. 2a. The TEER measurement is operated in the constant-current mode to avoid accidental over-current to damage electrodes and tissues inside the chamber. An FTDI module is used to provide interface between the host computer and the on-board electronics using the USB protocol. This interface allows users to control all internal enable signals from the GUI. A signal generator is used to generate a user defined AC voltage signal. This voltage signal is used to control a Howland current source (HCS), creating an AC current stimulus signal to the TEER electrodes. The HCS (Fig. 2b) is chosen because it can easily achieve high output impedance, signal-to-noise ratio (SNR), and is fully programmable through external control voltages^27,28. Up to three parallel current stimulation signals are used for the three independent chambers in the system. The read-channel (Fig. 2b) consists of a transimpedance amplifier to convert the input current signal to an output voltage, and an instrumentation amplifier to acquire the voltage response from the tissue barrier. A relay is placed in parallel with each microfluidic chamber to discharge built-up charge on the TEER electrodes when necessary. Built-up charge on the electrodes is capable of altering the DC voltage at the input to the microfluidic chamber, and therefore, has a large enough effect to shift the DC voltage towards supply rails, reducing the dynamic range of TEER measurement.

The HCS (Fig. 2b) uses an ultra-low offset voltage operation amplifier and consists of five high precision film resistors and a single feedback capacitor for bandwidth control. High output impedance is achieved by resistor matching. The response TEER voltage from the chamber is read by an instrumentation amplifier (INA) with low input bias current. The INA gain is controlled by the resistor \({{{\rm{R}}}}_{{{\rm{g}}}}\). The response TEER current from the chamber is read using a transimpedance amplifier (TIA), with the current gain set by \({{{\rm{R}}}}_{{{\rm{gain}}}}\). An analog-to-digital converter (ADC) is used to convert the amplified analog outputs from the read-channel to a digital signal that is sent to the host computer via the USB port. Due to different operating voltages of different components along the signal chain to achieve the required output dynamic range, electronic level shifting is required at various points in the system. They are performed by operational amplifiers where the voltage gain is controlled by on board resistors and DC shift is adjusted using a digital potentiometer. The digital potentiometer is calibrated before each measurement using a binary search algorithm to find the smallest DC offset current. The input stimulus was also designed to have a high and low current setting to maximize the system’s output dynamic range.

The supporting electronics are partitioned into two separate PCB boards. The main control board (Fig. 2e) contains the external power supply, USB connectors, microcontroller, ADC, signal generators, level shifters for control signals, and connectors to the TEER acquisition board. The TEER acquisition board (Fig. 2d) contains the HCSs, the read channels for each chamber, the calibration digital potentiometers, and the connections for the card edge connectors. Even though the system allows three chambers to be used at a time, the system architecture was designed to be scalable to allow future expansion to accommodate more chambers and sensors.

Electrode design and manufacturing

The electrodes are manufactured in gold on a glass substrate. Each electrode chip consists of two gold (Au) electrodes to allow 4-point measurement. The electrode chip (Fig. 2c) was fabricated on a 25 × 25 mm glass substrate through an in-house photolithography, deposition, and lift-off process. The mask was designed using AutoCAD software (Autodesk, Inc.) and manufactured by Artnet Pro (San Jose, CA). The full photolithography steps are described previously²⁹.

Chamber sterilization

To prevent infection during experiments, all components of the microfluidic chamber (chamber body, glass electrode chip, PDMS layers, PCBs, tubing, and Luer locks) were put through the first round of sterilization protocol:

1. 1.

   20-min bath in 1:10 bleach to water ratio.

2. 2.

   10-min soapy water bath inside of ultrasonic cleaner.

3. 3.

   Next, thoroughly rinse with DI water.

4. 4.

   45-min bath in 70% ethanol.

5. 5.

   Finally, thoroughly rinse with DI water and let air dry.

After the first round of sterilization, the chamber was fully assembled with metal screws and clamps that have been autoclaved (30-min, gravity cycle). After all chambers are assembled, the chambers went through low temp gas sterilization and are kept in a sealed bag before use.

Animals, tissue collection, and media preparation

In all experiments, male C57BL/6 background mice aged 3–4 months were used. Mice were kept on a 12 h light/dark cycle with access to standard chow and water ad libitum. Animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC) at Colorado State University under United States Department of Agriculture (USDA) guidelines.

Mice were deeply anesthetized with isoflurane and terminated via decapitation to prepare for tissue collection. The intestines were removed and immediately placed in 4 °C 1x Krebs buffer (in mM: 2.5 KCl, 2.5 CaCl₂, 126 NaCl, 1.2 MgCl₂, 1.2 NaH₂PO₄). To prevent contractions during dissection, the Krebs buffer contained 1 µl 1 mL⁻¹ nicardipine (Sigma Aldrich, St. Louis, MO), an L-type calcium ion channel blocker. Colon was then dissected to remove any remaining mesentery. For experiments in which muscle was removed, a 26 G needle was used to gently tease away the muscle layer on the mesenteric edge of the tissue. Tissue was then cut longitudinally using angled vascular scissors to form flat pieces of tissue around ~5 mm.

Adult Neurobasal media was custom made in house with 2% B27 supplement (ThermoFisher scientific, Waltham, WA), 4 mM glucose, 3% 1 M HEPES buffer (Sigma Aldrich, St. Louis, MO), without phenol red. To help maintain the gut microbiome, luminal media contained 0.4 mg ml⁻¹ inulin (soluble fiber) and 0.5 M sodium sulfite (oxygen scavenger) to decrease oxygen levels¹⁷. The serosal media had ambient levels of oxygen creating an oxygen gradient across the tissue which we previously demonstrated¹⁵ is necessary for preservation of microbiome. After 24 h, luminal media for control tissue was not changed. Treatment group luminal media contained 5.80*10⁻² U of broad spectrum bacterially sourced collagenase (Worthington Biochemical, Lakewood, NJ) or was treated with hydrochloric acid (HCl) to acidify the pH to 2. After completion of experiments, 0.05 M phosphate buffered saline (PBS) containing 0.5% cetylpyridinium chloride (CPC) was gently pipetted onto the tissue to preserve the mucus layer¹². The tissue was then gently removed from the device and placed in 4% paraformaldehyde (PFA) containing 0.5% CPC at 4 °C for 24 h. Tissue was stored in PBS at 4 °C until sectioning.

Tissue sectioning and histochemistry

Detailed methodology can be found in our previous publication¹². Briefly, 1–3 mm sections of colon were submerged in agarose until polymerization. Tissue was then cut on a vibrating microtome (VT100S; Leica microsystems, Wetzlar, Germany) at a thickness of 50 µm. For lectin and immunohistochemistry, sections were first washed in 1x PBS, then incubated in 0.1 M glycine followed by PBS washes and incubated in 0.5% sodium borohydride followed by PBS washes. Sections were then blocked in PBS with 5% normal goat serum (NGS; Lampire Biological, Pipersville, PA), 1% hydrogen peroxide, and 0.3% Triton X (TX). Next, sections were placed in PBS containing 0.3% TX and 5% NGS with the appropriate lectin or antibody for 2 days. The lectin used was Ulex Europaeus Agglutinin I conjugated to Rhodamine (UEA-1; Vector Labs) at a concentration of 0.125 µg mL⁻¹. Primary antibodies used were anti-claudin1 (Invitrogen) 1:200 and anti-peripherin (Sigma-Aldrich) 1:300. After lectin or primary antibody incubation, sections were washed in PBS with 1% NGS. Sections incubated in primary antibody were then incubated with PBS containing 0.02% TX and Alexa Fluor 594 conjugated to secondary antibodies specific to the species of the primary antibodies at a 1:500 dilution. Finally, sections were washed in PBS, mounted on slides, and cover slipped. Images were taken using a Zeiss LSM800 upright confocal laser scanning microscope and a 20x (W Plan-Apochromat 20X/1.0 DIC Vis-ir ∞ /0.17) objective or an Olympus BH2 brightfield microscope.

TEER calculation

Processing of TEER signals involves conditioning steps to reduce noise and other artifacts. The conditioned signals were further processed by applying a curve fitting algorithm to obtain the magnitude and phase of the voltage and current response signals. The impedance magnitude (\({|Z|}\)) and phase difference (\({\theta }_{{\text{diff}}}\)) can then be calculated using Eqs. (1) and (2). Where Av_current and Av_voltage are the current and voltage gain values and \({\theta }_{{\text{current}}}({{\rm{deg}}} )\) and \({\theta }_{{\text{voltage}}}({{\rm{deg}}} )\) are the current and voltage respective phase values.

The magnitude and phase values are determined for each frequency to obtain the impedance spectrum of the tissue sample, commonly referred to as electrical impedance spectroscopy (EIS). Due to its versatility of revealing impedance information across a wide range of frequencies, EIS is a widely-used technique to discover the impedance characteristics of tissue/cell-culture samples in Ussing Chambers, organ-on-a-chip devices, and well inserts^{13,18,23,27,30,31,32,33,34,35,36}.

The sinusoidal curve fitting is necessary to further reduce noise and unwanted artifacts in the acquired TEER signal as it is illustrated in Fig. 3a–c. The smoothed signal can provide more accurate magnitude and phase values for the subsequent TEER calculation (Fig. 3c). The curve fitting algorithm also provides drifting correction to the acquired voltage response signal. Drifting of the response voltage signal is caused by offset DC current from the Howland circuit. This offset DC current builds up charge on the serial capacitance associated with electrode’s double layer capacitor, resulting in a constant rate increasing (or decreasing) of the DC voltage at the voltage electrodes from the chamber. This effect can be seen in Fig. 3a. The time dependent DC shift of the sinusoidal signal in Fig. 3a needs to be leveled before the sinusoidal curve fitting algorithm can be applied to obtain its magnitude and phase. This is done by subtracting a 1^st-order polynomial function from the acquired (drifted) voltage signal as illustrated in Fig. 3b.

The TEER value of an epithelial barrier is the resistance of the transcellular and paracellular pathways combined. However, the TEER values obtained from Eqs. (1) and (2) include additional impedance such as the electrode double-layer capacitance and the media bulk resistance^3,10,34. In order to obtain the actual TEER values associated with the epithelial barrier, baseline TEER measurements were performed for each experiment to capture the medial bulk resistance. The final TEER value of interest was obtained by subtracting the baseline TEER values from the acquired TEER values. It should be noted that the magnitude \({|Z|}\) used for TEER measurements should be at an appropriate frequency not too low where the impedance of the electrode double-layer capacitance dominates, and also not too high where the epithelial layer is shorted by its parallel capacitance, this can be deduced from the equivalent circuit of the epithelial barrier¹⁰. From the impedance spectrum of the tissue measured with this device, it was found that this value is close to 5 kHz.

The TEER value can also be calculated by finding the DC response from a square wave. Since the microfluidic chamber system is also capable of producing a square wave stimulus signal, the TEER using the square waveform stimulus was also calculated. This value shows the pure resistance of the tissue barrier. Figure 3d, e shows a set of TEER values obtained using a 5 kHz sinusoidal stimulus vs. a square wave stimulus. It was found that the TEER values obtained using the square waveform are lower (7.2% on average, \({{\rm{n}}}=10\)) than those obtained using the 5 kHz sinusoidal waveform by a constant margin. This is due to the fact that the relatively fast transitions in the square waveform stimulus were able to reduce the effect of the double layer capacitance associated with the electrodes on TEER magnitudes compare to that from the 5 kHz sinusoidal stimulus. If the sinusoidal stimulus frequency is decreased, then the effect of the double layer capacitance is more pronounced, making the TEER value increase as the input frequency decreases. Figure 3e confirms this by showing the percent difference between the sinusoidal and square wave increases as the frequency of the sinusoidal stimulus decreases. When the stimulus frequency reaches 3 kHz and above, the difference between sinusoidal and square wave data flattens out, indicating that the stimulus frequency is now high enough to bypass the double layer capacitance.

Experiment and measurement procedure

After all tissue explants were cut and prepared according to the protocol in section “Animals, Tissue Collection, and Media Preparation”, the explants were loaded into the microfluidic chamber, one by one. First, the explants were placed on the bottom half chamber and then gently flattened out using forceps, careful not to touch the luminal side and damage the mucosa. After the tissue was flattened and centered over the holding cavity (Fig. 1c) on the bottom half chamber, the top chamber was slid down the metal screw guides to secure the tissue in place and create a tight seal. The chamber was then tightened using wing nuts and inserted into the card edge connector on the metal enclosure of the system (Fig. 1g). Next the inlet and outlet tubing were connected. The media outlet tubes fed into empty glass bottles as a way to determine whether even media outlet from each side of the tissue was achieved during experiments. To remove any air bubbles in the chamber the media was purged into the chamber at an increased rate (25,000 μL h⁻¹) for 45 seconds at the start of each experiment. After the initial purge, the media flow rate was reduced to 250 μL h⁻¹ throughout the experiment. The chambers, media, and the system enclosure are all kept in an incubator set to 37 °C.

To evaluate the live tissue viability, tissues were kept inside of the device for up to 72 h with fresh control media being pumped through continuously. Tissue viability was assessed by examining the tissue with histochemical methods and searching for characteristics of healthy tissue. During TEER experiments tissues were kept under control conditions for the first 24 h. After this calibration period, tissues were exposed to different media compositions, either collagenase or acidic media (discussed in section “Animals, Tissue Collection, and Media Preparation”) for the next 24 h. Over the course of the full experiment TEER measurements are performed every 2 h, creating a timeline of the full 48 h experiment. For each TEER measurement the input AC current magnitude was set at 85\(\,{{\rm{\mu }}}{{\rm{A}}}\) and the frequency was swept from 12 Hz to 5 kHz at 20 different frequency points. At each measurement point, the TEER was measured using the sinusoidal waveform stimulus as well as the square waveform stimulus. After the TEER experiments the tissue follows the same protocol as the tissue viability experiments, this protocol is outlined in section “Tissue Sectioning and Histochemistry”.

For experiments comparing the reduction in TEER after being exposed to collagenase or acidic media the reduction is measured from after the calibration period (~24 h) to the end of the experiment (48 h). Only TEER values from the 24 h – 48 h mark are considered for the experiment because the first 24 h is considered a good calibration period for the tissues. We found that TEER values found during the first 24 h provide sporadic results as the tissue reaches equilibrium in the device environment around 24 h as the TEER signal settles. The TEER experiments were not considered past 48 h because of the degradation witnessed in the low pH exposed tissue and the same method relying on sectioning and imaging tissues after the experiments was used for viability for consistency.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.