A Microfluidic Chip for Sustained Oxygen Gradient Formation in the Intestine Ex Vivo

Abstract

The oxygen gradient across the intestine influences intestinal physiology and the microbial environment of the microbiome. The microbiome releases metabolites that communicate with enterochromaffin cells, neuronal cells, and resident immune cells to facilitate the bidirectional communication across the gut-brain axis. Measuring communication between various cell types within the intestine could provide essential information about key regulators of gut and brain health; however, the microbial environment of the intestine is heavily dependent on the physiological oxygen gradient that exists across the intestinal wall. Likewise, there exist a need for methods which enable real-time monitoring of intestinal signaling ex vivo yet this remains challenging due to the inability to adequately culture intestinal tissue ex vivo while also exposing the appropriate locations of the intestine for probe insertion and monitoring. Here, we designed and fabricated a 3D printed microfluidic device to maintain the oxygen gradient across precision cut murine intestinal slices with the capability to couple to external neurochemical recording techniques. The gradient is maintained from outlets below while allowing access to the slice from above for detection with fast scan cyclic voltammetry (FSCV) and carbon-fiber microelectrodes. A series of 11 outlet ports were designed to lay underneath the slice which were connected to channels to deliver oxygenated vs. deoxygenated media. Outlet ports were designed in an oval shape where deoxygenated media was delivered to the center of the slice and oxygenated media is delivered to the outer portion of the slice to mimic the location of oxygen across the intestine. An oxygen sensitive fluorescent dye, tris (2,2’-bipyridyl)dichlororuthenium(II), was used to characterize the tunability of the gradient. Viability of the tissue was confirmed by both fluorescence microscopy and FSCV. Additionally, we measured simultaneous serotonin and melatonin signaling with FSCV in the intestine for the first time. Overall, this chip provides a significant advance in our ability to culture intestinal slices ex vivo with the added benefit of direct access for measurements and imaging.

Graphical Abstract

Here, we have developed a 3D printed microfluidic device capable of oxygen gradient formation within intestinal tissue slices ex vivo. The device is open-welled and compatible with external electrochemical recording during tissue analysis.

Introduction

Maintaining the oxygen gradient within intestinal culture enables a more physiological ex vivo analysis of live intact tissues. The intestine is a major site of importance in the body’s overall health and function, and a primary site for digestion and adsorption.¹ Unique to intestinal physiology, is a steep oxygen gradient (Figure 1B), starting with the aerobic intestinal wall then decreasing in oxygen concentration toward the intestinal villi.^1,2 This oxygen gradient is an important connection between the health of the host and the microbiota. It has been suggested that the colonization of the microbiota assists in the establishment of the oxygen gradient in the intestine after birth.² The gastrointestinal tract is host to the microbiota, which encompasses trillions of microorganisms including bacteria, viruses, and fungi.³ The microbiota has been demonstrated to contribute to the overall health of the host, with an influence on host functions such as pathogen invasion, host metabolism, and priming the immune response.³ Recreating the environment surrounding the intestine can be difficult due to the chemical complexities of its microenvironment. Intestinal slices maintain the spatial distribution of the cells involved in intestinal function and also maintain the microbiota composition and distribution. If kept in the proper environment, tissue slices could provide a powerful biological model with insights into whole organ functions. Additionally, coupling sophisticated intestinal slice culture systems with external recording techniques is challenging due to the oxygen gradient requirement; yet, would provide a significant advance in our ability to monitor chemical signaling in a physiologically-controlled intestine to yield information about critical cellular signaling. Here, we present a 3D-printed microfluidic device which enables adequately maintaining the steep oxygen gradient across murine intestinal slices ex vivo while also being open for ease of imaging and external electrochemical recording of cellular communication.

Figure 1.

Oxygen gradient demand in precision-cut intestinal slices. A) Image of a 300 μm thick intestinal slice stained with SYTO-9 to label all cells in the slice. The intestinal villi and outer intestinal wall are highlighted via an arrow and label in the image B) Schematic depicting the oxygen gradient across the intestine. The villi rich “core” region is in an anaerobic environment; whereas the outer wall is primarily aerobic.

Cellular signaling in the gut occurs via neural, hormonal, gastrointestinal, microbial, and immunological signaling pathways.⁴ Maintaining this complex cellular network is important for gaining information on the extent to which these cells communicate with one another to maintain health. Using analytical methods to monitor chemical messengers in the intestine is important for advancing our understanding of how different cell types communicate with one another and how this communication functions to maintain homeostasis across the gut-brain axis. The GI tract is highly innervated both intrinsically via the enteric nervous system and extrinsically via sympathetic and parasympathetic afferent neurons.⁵ Sympathetic neurons synapse to the visceral mesentery ganglia such as the Peyer’s Patches (PP) and the mesenteric lymph nodes (MLN).^6,7 Enteric neurons innervate the intestinal wall and the entire GI tract influencing functions such as gut secretions and nutrient absorption.^8,9 Neuron projections in the gut are proposed to release a wide variety of neurochemicals and neuropeptides to regulate gut function.^6,10,11 Enterochromaffin cells are the most abundant form of enteroendocrine cells in the intestine and can synapse with efferent and afferent nerve endings.^12–14 Enterochromaffin cells contain and release approximately 90% of the body’s serotonin in response to mechanical or chemical stimuli in the gut.¹⁵ Serotonin release has thus been correlated to controlling gastrointestinal motility, and dysregulated serotonin has been linked to gastrointestinal disorders such as irritable bowel syndrome.¹⁵ Recently, an enterochromaffin cell monolayer platform was developed to enable serotonin secretion measurements from human primary intestinal cells.¹⁶ Enterochromaffin cells also release melatonin, which has been shown to influence the immune response.¹⁵ Serotonin is proposed to induce muscle contraction of the intestine while melatonin can act as a physiological antagonist of serotonin and reduce muscle contractions.¹⁵ Measurement of these key chemical messengers has been reported in tissue explants and a wealth of information has been gained based on these studies.^12,15,17 Despite these achievements, the orientation of the tissue preparation for these measurements to date have hindered the ability to accurately quantitate signaling from specific villi. Most tissue preparations involve splaying open the intestine and placing in a dish. This approach limits the ability for accurate probe insertion and therefore limits the ability to target specific subregions within the luminal space. We seek to fill this gap by not only developing a platform to better culture intestinal tissue but to have this platform designed to be compatible with precision cut intestinal slices, exposing the luminal interior of the intestine, to improve spatially-resolved measurement capability. In this work, we use fast-scan cyclic voltammetry (FSCV) at a carbon-fiber microelectrode to measure subsecond fluctuations in serotonin and melatonin signaling in the intestinal villi on-chip. FSCV is an excellent electroanalytical technique used to measure rapid neurochemical fluctuations in tissue. To date, this technique has been adapted mostly in the brain, but has also been extended to a few other peripheral organs.^7,18–20 To the best of our knowledge, this is the first time FSCV has been implemented in intestinal tissue, widening the scope of FSCV analysis.

To recapitulate the intestine ex vivo, biological models such as 2D or 3D cell culture^21,22, tissue sections^15,17,23,24, and tissue slices^25,26 have been explored. 2D cell culture enables specific subsets of cells to be probed without interference from unwanted cells;²⁷ however, the importance of the cellular distribution, morphology, and environmental conditions has become apparent in the intestine leading to the adaptation of 3D cell culture.^28,29 Prior work has demonstrated a device designed to incorporate a curved scaffolding for the cells of interest to be seeded onto: this recreates the general intestinal structure for in vitro analysis.²¹ The sponge scaffolding demonstrated that the curvature could support cell adhesion and proliferation in a culture, while also recreating the oxygen gradients present in the intestine for further analysis. 2D and 3D cell culture can provide impactful information about specific cellular interactions to gain necessary understanding of physiological processes such as the immune response. 2D intestinal crypts have been able to replicate some of the intestinal physiology allowing for compound screening in the intestine.³⁰ 3D cell cultures such as intestinal organoids have provided insights into how the shape of cells in the intestine can be essential for cellular function.³¹ Additionally, self sustaining oxygen gradient formation within 3D intestinal crypt scaffolds have been developed to study the intestinal-microbiome interface.³² Despite these major advancements, 3D culture remains limited due to the inability to account for the innervation or vascularization of the intestine. Tissue sections maintain the cellular distribution, vascularization, and innervation enabling the organ to be probed ex vivo¹⁵; however, tissue thickness affects the diffusion distance for nutrients therefore influencing the overall health of the tissue.³³ This has led to the development of precision cut tissue slices (Figure 1A) that maintain the organ architecture, cellular distribution, and neuronal innervation while providing adequate diffusional distances for nutrient penetration.^25,26 Intestinal slices have provided models for further analysis into the biotransformation of drugs²⁵ or complications such as intestinal fibrosis³⁴. Overall tissue slices provide a biological model that has the same spatial distribution as the organ in vivo, while providing more precise control of the tissue to probe specific cellular pathways important for organ function. Here we have adapted and modified existing precision cut intestinal slicing protocols in order to create a biological model compatible for FSCV recording.

This work provides a microfluidic culture platform for ex vivo intestinal slices with an open culture well design for easy coupling with external probes. The device is 3D printed and has strategically placed channels with exit ports for delivery of oxygenated and deoxygenated culture media to recreate the physiological oxygen gradient across the intestine. The size of the intestine varies depending on the area of interest and the age of the mouse; our chip is capable of tuning the oxygen gradient to accommodate changes associated with biological variability. To demonstrate the usefulness of the open culture well design, we coupled live intestinal slice culture on-chip with fast-scan cyclic voltammetry (FSCV) recording to measure subsecond fluctuations in both serotonin and melatonin signaling. To the best of our knowledge, this is the first report to use FSCV to measure subsecond fluctuations of simultaneous serotonin and melatonin signaling in the intestine. Overall, this device provides a novel culture method for intestinal slices that can not only recreate the oxygen gradient within precision cut tissue slices of the intestine but also enables external access for electrochemical recording and fluorescence imaging.

Experimental

Device Fabrication and Assembly

The device is fabricated via a CADworks H-50 3D printer (CADworks3D, Ontario, CA). The file was designed using Autodesk Fusion 360 (2021), and the device was printed with the clear microfluidics resin (CADworks3D, Ontario, CA). The files were sliced into 30 μm sections using the MII Utility Shortcut V 3.27 (CADworks3D, Ontario, CA) before being sent to the printer. Once the device was printed, it was removed from the stage and cleaned with warm soapy water, the soapy water was also perfused through all channels to ensure a clean device. Once clean, the device was cured using a UV light box as per manufacturer specifications, then rinsed with water to prepare for experimentation. A 2 mm tall base piece was printed that is 24 mm wide x 50 mm long and was designed to fit within a typical perfusion chamber base (Warner Instruments) to enable easy control of the device and use under a microscope. All layers with a channel (layers 1–3) are 0.8 mm tall while layer 4 is 0.5 mm tall and layer 5 is 2 mm tall to create the culture well. All channel exit ports in the culture well are 0.5 mm in diameter, and all channels are 0.5 mm wide by 0.5 mm deep.

Preparation of 300 μm Thick Agarose Slices

Agarose slices were prepared using a Leica VT1000S vibratome (Bannockburn, IL, USA) at 300 μm thick for device characterization. 6% (w/v) low melting point agarose (Lonza, Walkersville, MD, USA) in 1X PBS (0.137 M NaCl, 2.7 mM KCL, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄) was warmed and 1 mL of the warmed agarose solution was placed into a 13 mm cylindrical mold and allowed to harden on ice. 6% (w/v) agarose was chosen due its relative stiffness, with a shear modulus predicted to be greater than 33.5 kPa³⁵, and low permeability, which is comparable to properties of intestinal tissue (shear modulus can range from 11 to 132 kPa³⁶). A 12 mm diameter section of agarose was isolated via a biopsy punch (Robbins Instrument, NJ, USA). Agarose sections were superglued to the stage of the vibratome and sliced 300 μm thick with a speed of 90 and a frequency of 3. Once prepared they were stored in a small petri dish filled with 1X PBS at 4°C for no longer than a week.

Validating Gradient Size On-Chip

A 0.1 mg/mL solution of fluorescein (Sigma-Aldrich) in 1X PBS was prepared and used to validate gradient formation on-chip. Fluorescein was delivered to the middle delivery ports (Figure 2D), and PBS delivered to the outer delivery ports (Figure 2D), and waste pulled from the middle ports. The waste ports in between the two delivery ports function to help facilitate gradient formation on-chip. A plain agarose slice was placed into the culture well and a 11 mm diameter stainless steel washer (Fastenal) was placed on top of the agarose slice to serve as an anchor. Fluorescence images of the gradient created within a slice were taken using a Zeiss AxioZoom microscope (Carl Zeiss Microscopy, Germany) with a Axiocam 506 mono-camera and a GFP filter cube (Zeiss filter set # 38). Images were taken every 10 second for 1 min, every min for 9 min, then every 10 min for 50 additional min. A total combined flow rate of 75 μL/min was applied in two different flow rate combinations: 7.5 μL/min fluorescein + 67.5 μL/min 1X PBS, and 15 μL/min fluorescein + 60 μL/min 1X PBS using a peristaltic pump (Ismatec Regalo ICC, Cole-Palmer, USA). The waste was pulled at a rate of 75 μL/min. Using FIJI, the distance of a line vs the fluorescence intensity along that line was plotted in Figure S1A and depicts the line drawn across the center of the culture well from one end of the washer to the other. From this data, the distance of the fluorescent dye at half of the maximum intensity was calculated as the spread of the fluorescence gradient within the tissue (Figure S1B).

Figure 2.

Microfluidic chip to sustain oxygen gradient delivery to an intestinal slice. A) Image of 3D printed device incorporated into a commercial perfusion chamber base. B) Brightfield image of the culture well showing the channels and their respective delivery ports. Scale bar on image. C) 3D model of the assembled device. D) Side-view schematic showing fluidic delivery to an intestinal slice. Waste inlet not visible at this cross-section. E) All layers of the chip exploded to depict the delivery channels and their respective delivery ports and dimensions. Layer 1 (bottom layer) delivers the deoxygenated media to the middle of the culture well. Layer 2 has ports enabling delivery of the deoxygenated media and a channel which facilitates the removal of waste from the culture well. Layer 3 has ports to facilitate delivery of deoxygenated media and the removal of waste and a channel system that enables delivery of oxygenated media to the culture well. Layer 4 has exit ports for all channels that facilitate the delivery of the respective media or removal of waste from the culture well. Layer 5 is the thickest layer to create a 2 mm deep culture well.

Validating Oxygen Concentration Gradient On-Chip

An oxygen sensitive fluorescent dye tris(2,2 bipyridyl)dichlororuthenium (II) hexahydrate (RTDP) (Sigma Aldridge) was used to correlate fluorescence intensity to oxygen concentration. A 0.1 mg/mL solution of RTDP in 1X PBS was prepared, and separated into two aliquots. One aliquot was oxygenated by bubbling 5% CO₂ with 95% O₂, and the other was deoxygenated by bubbling N₂ gas into solution. The deoxygenated buffer was delivered to the middle delivery ports and the oxygenated solution was delivered to the outer delivery ports while waste was pulled from the middle ports. Fluorescence images of the gradient within an agarose slice were taken with a Zeiss AxioZoom microscope (Carl Zeiss Microscopy, Germany) with an Axiocam 506 mono-camera and the Zeiss filter cube set #40. The scaling per pixel was 6.48 × 6.48 microns and the bit depth was 14-bit. Images were taken every 10 seconds for 1 min, every min for 9 min and every 10 minutes for 50 minutes. A total combined flow rate of 75 μL/min was applied in two different flow rate combinations, 7.5 μL/min fluorescein + 67.5 μL/min 1X PBS, and 15 μL/min fluorescein + 60 μL/min 1X PBS using a peristaltic pump (Ismatec Regalo ICC, Cole-Palmer, USA). The solutions were allowed to bubble with their respective gas for 30 minutes before calibration. After 30 minutes, a dissolved oxygen probe (Hamilton Co.) was used to determine the dissolved oxygen concentration of each buffer before experimentation began and validate that the oxygen level remained constant for the entirety of the experiment. After the oxygen concentration was evaluated with the probe, fluorescence images were taken of each solution in the device separately starting with the oxygenated solution. These measurements were used to calibrate the dye off-chip, so that oxygen concentration could be estimated within the slice. The device was first filled with the deoxygenated and oxygenated solutions. Once the culture well was full, waste removal was initiated, and any bubbles present in the channels were removed via a 1 ml plastic syringe (BD Plastics) and needle (Hamilton Co.). Once bubbles were removed, the device was allowed to perfuse for 20 minutes before experimentation. After 20 minutes images were taken every 10 seconds for 1 minute, every minute for 9 minutes, and every 10 minutes for 50 minutes for the total analysis time of one hour. A line scan of the image at each time point is taken as mentioned above (Figure S1) to plot the distance vs fluorescence intensity. Fluorescence intensity or mean gray value of the solution can be correlated to oxygen concentration using the Stern-Volmer equation. $\frac{I_0}{I}=k_q\left[O_2\right]+1$

The $\frac{I_0}{I}$ is calculated using the two solutions mentioned above, the I_o is used as the intensity of the deoxygenated solution and the I as the intensity of the solution in question. To obtain the $k_q$, the oxygen concentration of each solution (found with the probe mentioned above) are plotted vs the $\frac{I_0}{I}$ values. The $k_q$ is the slope of the line created from the two points. The I_o value and the slope are then used to calculate the oxygen concentration at each point along the culture well with the Stern-Volmer equation.

Preparation of Murine Intestinal Slices

All animal procedures described were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Cincinnati and performed in accordance with the Guide for the Care and Use of Laboratory Animals (The Guide) by the National Research Council. Female C57/BL6 mice (Jackson Laboratory, USA) aged 4–12 weeks were housed in a vivarium and provided with food and water ad libitum. On the day of the experiment, mice were anesthetized with isoflurane (Henry Shrein, USA) and immediately euthanized by cervical dislocation. The duodenum and the ileum were separated from the rest of the intestine with scissors and placed in ice cold Krebs buffer. The fat was separated from the intestinal piece in the buffer and the intestinal piece was sectioned into 5 to 8 mm sections. 6% low melting point agarose was placed into the centrifuge tube molds (Figure S2), allowed to cool slightly, then the intestine pieces were placed into the agarose. Intestinal sections were strategically placed in the molds to ensure a cross-section displaying the interior of the intestine is achieved when sliced. The molds were then placed on ice and allowed to harden. Once hardened, a scalpel was used to remove the embedded intestine piece, the pieces of agarose were cut to the same height and superglued onto the specimen disk. The specimen disk was then anchored into the buffer tray with slightly oxygenated ice cold Krebs Buffer, and sliced 300 μm thick using a Leica VT 1000S vibratome. Slices were transferred using a round brush (Michaels, USA) into a six-well culture plate filled with Krebs buffer. After slicing, the plate was placed into a 37°C humidified sterile incubator with 5% CO₂ and the slices were allowed to recover for 1 hour prior to experimentation.

Fluorescently-Labelled Tissue Slices

Intestinal slices were stained with a bacterial viability kit (ThermoFischer, USA) to assess the viability of bacteria in the tissue slice during both on- and off-chip culture. A 3.34 mM SYTO-9 and 20 mM propidium iodide solution were prepared as per manufacturer specifications then separated into 12 μL aliquots and stored in the freezer for future use. On the day of the experiment, the aliquot was thawed and diluted to 4 mL with Krebs buffer, separately 0.5 mL of the solution was then pipetted into 8 wells of a 24 well plate. The slices were then placed into the wells with the solution and allowed to incubate in the dark at room temperature for 20 minutes. After the incubation period the slices were rinsed with Krebs buffer and imaged with the Zeiss AxioZoom microscope and using the GFP and DS Red filters (used above) in a small petri-dish (AE Bios, Ohio, USA) filled with Krebs buffer.

Fast-Scan Cyclic Voltammetry at Carbon-Fiber Microelectrode Recording in a Perfusion Chamber

FSCV at carbon-fiber microelectrodes was used to monitor simultaneous serotonin and melatonin release in the intestine while cultured in a standard tissue perfusion chamber. Slices were placed into the prefusion chamber while oxygenated Krebs buffer was flowed through the chamber using a peristaltic pump, similar to traditional tissue perfusion experiments.^7,37 Cylindrical carbon-fiber microelectrodes were constructed by vacuum aspirating a single 7 µm T-650 carbon-fiber (Mitsubishi Rayon Co, LTD, Tokyo Japan) through a 1.2 mm x 0.68 mm x 4 in glass capillary (A-M Systems, Sequium WA), vertically pulling the capillary into two (PE-22 Electrode Puller, Narishige Scientific, Tokyo Japan), followed by trimming the fiber to approximately 150 µm from the end of the glass seal using a microscope. The electrode was dip coated with 5 % Nafion dispersion in methanol (Ion Power, Inc.) for 30 seconds then baked for 45 minutes to evaporate any remaining ethanol from the solution, as previously described, to improve serotonin detection in tissue³⁸. The electrode was then backfilled with 1M KCl to create an electrical connection and strategically implanted into the center of the intestinal villi. Our lab previously developed a waveform to enable co-detection of melatonin with serotonin, with limited fouling, using FSCV³⁹ and this was used to monitor these neurochemicals in the intestine. The triangular waveform scans from 0.2 V to 1.3 V at a scan rate of 600 V/s and a frequency of 10 Hz. Measurements were made against a Ag/AgCl reference electrode using a Dagan Potentiotstat and a 5 MΩ headstage (Pine Instruments, Durham, NC). Any non-faradaic charging currents were removed through background subtraction.

Spontaneous, unstimulated melatonin and serotonin release was detected at the microelectrode for 30 minutes during perfusion. Spontaneous transients were analyzed as previously reported in other organs.⁷ The amplitude of each transient was converted to concentration by using the slope of the calibration curve for each analyte. The slope for serotonin at Nafion-coated electrodes was 10.66 nA/μM while melatonin was 6.60 nA/μM (Figure S3). Any current smaller than 0.5 nA, and duration less than 0.5 s were excluded due to difficulty in accurate quantitation. The time between events, the inter-event time, was calculated to evaluate the frequency of transients during the monitoring time. The time between individual serotonin and melatonin events was also calculated to determine how these neurochemicals correlated in the intestine.

Fast-Scan Cyclic Voltammetry at Carbon-Fiber Microelectrodes Recording In Intestinal Slices On-Chip

Neurochemical recording using FSCV at carbon-fiber microelectrodes was done in intestinal slices on-chip to validate that the slices remained functional in the chip environment. Oxygenated and deoxygenated buffer solutions were prepared as mentioned above and perfused through their respective channels in the device. Once the device was full and the equilibration period was finished, slices were centered in the culture well of the chip and anchored with a washer to ensure the slice did not move during implantation and fluid delivery. The experiment and analysis was done exactly as described above.

Statistics

All values are reported as mean ± standard error of the mean (SEM) unless otherwise noted. All statistics were performed in GraphPad Prism 9 (GraphPad Software, Inc, La Jolla, CA, USA). The “n” represents the number of slices unless otherwise noted.

Results and Discussion

Device Design

We developed a platform which promotes easy access to an intestinal slice for neurochemical recording and imaging, while the physiological oxygen gradient is maintained. The 3D printed device is comprised of 5 layers: the culture well layer, the delivery port layer, the oxygenation delivery layer, the waste layer, and the deoxygenation delivery layer (Figure 2). A side-view schematic shows how deoxygenated vs. oxygenated media is delivered to intestinal slices ex vivio (Figure 2D). The channels on all layers are 0.5 mm wide by 0.5 mm deep, and all exit ports are 0.5 mm in diameter while the inlets/outlets to the delivery channels are 1.57 mm in diameter to accommodate the tubing. The culture well in Layer 5 (Figure 2E) is 11.5 mm in diameter by 2 mm deep to accommodate the diameter of the intestinal slice embedded in agarose. The exit port layer (Layer 4) has strategically placed ports to either deliver media or pull media from the culture well. The oxygenated buffer channel layer (Layer 3) has a keyhole channel with ports in the middle placed for the other delivery channels. The waste layer (Layer 2) has an oval keyhole channel with strategically placed ports for the deoxygenated media delivery. Finally, the deoxygenated delivery channel (Layer 1) is connected to an oval in the middle of the device that is 1 mm wide by 3.3 mm long with a depth of 0.5 mm.

Flow rate of delivery and waste removal was controlled on-chip to create an adequate gradient (Figure 2C). Waste was pulled from the middle ports at a rate of 75 μL/min using a peristaltic perfusion pump, while delivery into the device culture well was achieved via two separate delivery channels with a total combined flow rate of 75 μL/min. The flow rates of the deoxygenated media and the oxygenated media were varied in either a 1:4 or a 1:9 ratio to tune the size of the gradient produced. For the characterization experiments, 6% w/v agarose slices were used as a semi-transparent porous matrix to enable visualization of fluorescent media through the slice for adequate characterization. A healthy intestine functions as a barrier^40,41, with low permeability, and therefore it was also critical to use a high weight percent agarose to best match the intestinal microenvironment. To mimic other more permeable tissues, 2 % w/v agarose is often used.^42–44 Although we recognize that 6% w/v agarose is not an exact match to the intestinal barrier, it serves as a suitable model for gradient formation analysis. The 10 mm diameter slice was placed over the ports and secured with a washer to prevent movement of the slice during experimentation. This approach has been done previously in organs including the brain and lymph node.^44–46

Sustained Delivery of Fluorescein Gradient

The size of the gradient is imperative to delivering the correct oxygen concentration to the physiologically relevant portion of the intestinal slice. Fluorescein (0.1 mg/mL in PBS) was delivered to the deoxygenated media ports, while PBS (non-fluorescent) was delivered to the oxygenated media ports in order to observe and quantitate the control of fluid flow within the slice. Agarose was used as a sufficient porous matrix due to its optical clarity which enables ease of analysis of fluorescent gradients. The weight percent, 6 %, was chosen due to the relative stiffness and low permeability of intestinal tissue.⁴⁵ The flow rates were varied in a 1:4 (15 μL/min fluorescein + 60 μL/min buffer) or 1:9 (7.5 μL/min fluorescein + 67.5 μL/min buffer) ratio for a total flow rate of 75 μL/min. At a flow rate of 75 μL/min in a cylindrical channel (the culture well) the shear stress was calculated as 0.00011 dyne/cm² using $\tau =\frac{4\eta Q}{r^3\pi }$ where η is the dynamic viscosity of PBS buffer $\left(0.13\frac{g}{cm \cdot s}\right)$, Q is the flow rate $\left(0.00125\frac{c{m}^3}{s}\right)$, and r is the radius of the culture well (0.575 cm). The size of the gradient was calculated by plotting the intensity vs. distance of a line drawn through the center of the middle of the washer on the slice (Figure S1). The maximum fluorescence intensity was divided in half, then the intersection of this value and the distance vs intensity trace was found at each end of the bell curve. The difference between these points equated to the overall spread of the fluorescent signal. The average gradient size over the course of 60 minutes (Figure 3B) for the 1:9 ratio was 1181.4 ± 8.0 μm and was not significantly different at each time point (One-Way ANOVA with Bonferroni post-tests, p = 0.3709, n = 7). While the average gradient for the 1:4 ratio was 1398.2 ± 7.9 μm and was also not significantly different at each time point (One-Way ANOVA with Bonferroni post-tests, p = 0.4050, n = 5). The lack of difference in the gradient over time indicates stability of the gradient within the slice over time. Because this chip was designed to incorporate FSCV recording of neurochemicals and a typical experimental time frame for slices with FSCV is 45 min¹⁰, we analyzed the extent to which the gradient was maintained for an hour. To analyze the gradient over time, time lapse images were taken every 10 s for 1 min, every 1 min for 9 mins, and every 10 mins for 50 mins (Figure 3A). The difference between the two flow rate combinations was found to be significantly different indicating a stable tunability of the gradient (Unpaired t-test, p < 0.0001). This experiment demonstrated that a sustained gradient delivery could be tuned to the size of the slice being examined.

Figure 3.

Fluorescent gradient is maintained and tunable on-chip. A) Images at 0, 10, and 60 minutes at varied flow rate combinations. The 1:4 ratio represents 15 μL/min fluorescein and 60 μL/min PBS; a larger gradient size was observed when fluorescein was delivered slower, but the gradient remained stable over time. All time points from 0 – 30 minutes were non-significantly different (One-Way ANOVA with Bonferroni post-tests, p > 0.9999). The difference between 40 and 60 minutes and 50 (One-Way ANOVA with Bonferroni post-tests p = 0.4462) and 60 minutes were found to be non-significantly different (One-Way ANOVA with Bonferroni post-tests p = 0.4461) The 1:9 ratio represents 7.5 μL/min fluorescein and 67.5 μL/min PBS delivery. B) Distance of gradient vs time for every 10 s for 1 minute, every minute for 9 minutes, and every 10 minutes for 50 minutes for a total time of 60 minutes. The 1:4 ratio (n = 5) produced an average gradient distance of 1398.2 ± 7.9 μm while the 1:9 ratio (n = 7) produced an average distance of 1181.4 ± 8.0 μm. The flow rate combinations were found to be significantly different from each other (Unpaired t-test, p < 0.0001).

Oxygen Gradient Concentration Visualization

Quantifying the oxygen concentration along the gradient is imperative to ensure the viability of the intestinal tissue is maintained. An oxygen fluorescent dye, tris(2,2’-bipyridyl)dichlororuthenium(II) was used to visualize and quantitate the oxygen concentration delivered within the slice. In the presence of oxygen, the fluorescence of this dye is quenched. Therefore, when delivering deoxygenated media to the center of the slice, the center of the slice should be fluorescent while the outside edges should not. Because this data was imaged from the top of the slice, and delivery occurred from underneath, we anticipate that the estimated oxygen concentration values are within the slice, and not just below it. The size of the gradient was calculated the same as above, while the fluorescence intensity can be correlated to oxygen concentration using the Stern-Volmer equation (see methods). The Stern-Volmer equation is based on the intensity of fluorescence without a quencher present (I₀) and in this case, that is the deoxygenated solution. The intensity in the presence of a quencher (I), and [O₂] is the concentration of the quenching species (oxygen). Finally, the k_q value is the quenching rate coefficient and is calculated for each day the experiment was conducted from the standard solutions. The experiment was performed the same as with fluorescein. The fluorescence intensity along the distance of the culture well was examined (similar to Figure S1). The intensity at each spot along the gradient was used as the “I value” for that point. From this, the oxygen concentration can be back calculated at each point along the gradient as mentioned above. The distance of the gradient was calculated, the gradient size for the 1:4 flow rate ratio was 1629 μm ± 7.9 (Figure 4A). The 1:9 flow rate ratio gradient size was 1123 μm ± 6.8 (Figure 4A). These values are similar to the fluorescein gradient achieved above. Overall, this experiment demonstrated that a sustained oxygen gradient could be delivered to a tissue slice.

Figure 4.

The oxygen concentration is controlled on-chip. A) Distance of gradient (μm) vs time for both 1:4 (n = 6) and 1:9 (n = 5) flow rate combinations demonstrate that the gradient size is maintained and controllable over time. The average gradient size for the 1:4 ratio is 1629 ± 7.9 μm. The average gradient size for the 1:9 ratio is 1123 ± 6.8 μm. B) Distance (μm) vs. Oxygen Concentration (mg/L) for the 1:4 flow rate combination. Gray lines depict the intersection of the distance vs. intensity and the intensity half max at each end of the bell curve. C) Distance (μm) vs. Oxygen Concentration (mg/L) for the 1:9 flow rate combination.

Viability of Intestinal Slices

The viability of the intestinal slice is important to ensure physiologically relevant biology can be explored. Not only is the viability of the organ important but the viability of the bacteria within the intestine is also important to ensure proper function. To ensure viability of the intestinal slices, the slices were stained with a combination dye of SYTO-9 (green, Figure 5) and propidium iodide (red, Figure 5). The SYTO-9 stains all gram-negative and gram-positive bacteria while the propidium iodide stains all cells with a compromised membrane. The images in Figure 5 show an intestinal slice after an hour of traditional culture (incubator) which was used as a live control, and after 30 minutes in ethanol for a negative control (n = 4). A separate set of slices were stained and imaged after an hour on-chip to ensure the slices remained viable. One hour was chosen due to the necessity of keeping the intestinal tissue viable throughout the length of a typical voltammetry experiment (described below) and to confirm that the tissue is viable for the length of time used to confirm oxygen gradient sustainability. SYTO-9 staining remained intense for both on- and off-chip culture conditions, while PI staining remained relatively minimal. Conversely, slices that were purposefully killed demonstrated a greater degree of PI staining, indicating more cellular death, as expected. Overall, this demonstrated that slices remain viable over the course of an hour experiment on-chip. Future work could investigate longer-term culture to enable a variety of differing applications of the platform.

Figure 5.

Intestinal slices remain similarly viable on-chip as compared to an hour of traditional culture in an incubator. Slices are stained with SYTO-9 (green) a stain for all gram negative and gram positive bacteria, and Propridium Iodide (red) a stain for all bacteria with a compromised membrane. The intensity of SYTO-9 stain remains bright for all conditions, while the intensity of the PI stain remains low indiciating that the cells are mostly viable both on- and off-chip. (A-C) Representative images of a slice after an hour of traditional culture methods: A) image of entire slice B) Image of the center of the intestinal slice C) image of the villi of the intestinal slice D) Image of slice after an hour on-chip E) Image of slice after being soaked in ethanol for 1 hour. Slices purposefully killed demonstrate more intense PI staining indicating more cellular death.

Neurochemical Detection of Serotonin and Melatonin

The open culture well design of the chip is important for facilitating easy access to the organ for neurochemical recording with external electrodes. It has been demonstrated previously with amperometry that spontaneous serotonin and melatonin signaling from the enterochromaffin cells occurs rapidly and frequently.¹⁵ Real-time serotonin and melatonin co-detection was performed using fast-scan cyclic voltammetry (FSCV) at carbon-fiber microelectrodes as previously demonstrated by our lab (Figure S3)³⁹. This experiment provides further validation that our tissues are not only viable, but functioning on the device. Incorporating electrodes into the device isn’t amenable to traditional FSCV recording due to the need to implant electrodes within the tissue several times in several different spots to locate groups of signaling cells. Likewise, moveability of the electrode within the slice is important for control over where the measurement is being taken. Because of this, we did not incorporate pre-fabricated electrodes into this device, and left the culture well open to enable coupling to traditional electrochemical recording techniques in tissue. A concentration curve ranging from 0.05 μM, 0.1 μM, 0.5 μΜ, 1 μΜ, 5 μM, 10 μM, 20 μM, to 100 μM of both melatonin and serotonin was collected to enable estimation of the relative concentration of each neurochemical event in the tissue (Figure S3B and E). Our lab previously developed a waveform which enables co-detection of both melatonin and serotonin with FSCV and this waveform was adopted here to measure these two analytes in the intestine³⁹. The waveform scans from 0.2 V to 1.3 V and back at 600 V/s at a 10 Hz waveform application frequency. Prior work has suggested significantly improved detection of serotonin at Nafion-coated carbon fiber microelectrodes and therefore, we also adapted this approach to increase our probability of detection.³⁸ Serotonin’s oxidation peak occurs at 0.7 V (Figure S3C) and melatonin’s primary oxidation peak is at 1.0 V (Figure S3F) at this waveform. The sensitivity to the analyte can be determined through the slope of the linear regression of the concentration curve. For serotonin, the slope at the “melatonin waveform” was 10.63 nA/uM (r² = 0.9998, n = 5); whereas, the slope for melatonin was 6.60 nA/uM (r² = 0.9918, n = 5) .

Spontaneous serotonin and melatonin signaling at a Nafion-coated carbon-fiber microelectrode was detected from the villi of intestinal slices both on-chip and off-chip (Figure S4, n = 3). The intestinal slices were perfused in a traditional perfusion chamber as a control to compare to neurochemical release detected on-chip. Example cyclic voltammograms (CVs) for co-detection of 1 μm serotonin (oxidation peak at 0.7 V) and 5 μm melatonin (oxidation peak at 1.0 V) in a flow cell (Figure 6A) and in tissue slices (Figure 6B) are shown to demonstrate (1) that these two molecules can be co-quantitated with FSCV and (2) confirms the identity of these peaks in tissue. On average, for a 15 minute monitoring period, 70.2 ± 3.7 total transients were observed with an average inter-event time of 7.1 ± 0.8 s between both molecules off-chip. Of the 70 events, on average 34.9 ± 4.5 transients were serotonin and 35.6 ± 1.7 transients were melatonin (Figure 7G and H). Low nanomolar events were observed with an average 18 ± 0.6 nM for serotonin and 11.0 ± 0.4 nM for melatonin (Figure 7C and D). The duration of events were rapid, with each serotonin event lasting approximately 1.5 ± 0.1 s for serotonin and 1.5 ± 0.1 s for melatonin (Figure 7A and B). The average inter-event time between serotonin events was 14.9 ± 1.0 s and 14.0 ± 1.0 s for melatonin. Comparatively 34.5 ± 2.5 serotonin transients and 36.5 ± 2.0 transients were observed from intestinal slices on-chip (n = 2) for a 15-minute monitoring period (Figure 7G and H). The number of serotonin (unpaired t-test, p = 0.6855) and melatonin transients (unpaired t-test, p = 0.8698) on and off chip were found to be non-significantly different. Upon evaluation a total serotonin concentration of 8.4 ± 0.4 nM (unpaired t-test, p < 0.0001) with a duration of 1.2 ± 0.1 s (unpaired t-test, p = 0.2533) was measured, the concentration was significantly different, yet the duration was not. Likewise, the average concentration of 7.1 ± 0.4 of melatonin (unpaired t-test, p < 0.0001) was observed with an average duration of 1.3 ± 0.1 s (unpaired t-test, p = 0.3839). The concentration was significantly different between conditions, yet the duration was not.. The inter-event time between serotonin transients on-chip is 15.3 ± 1.6 which is non-significantly different from detection off-chip (unpaired t-test, p =0.8456). Similarly, the inter-event time between melatonin transients on-chip is 14.0 ± 1.4 which is also non-significantly different from transients observed off-chip (unpaired t-test, p = 0.9367). Despite differences in concentration observed for each analyte between on and off-chip, the total number of events, how long they are lasting, and how frequent they are did not change. Small changes in concentration between different tissues is expected due to slight differences in electrode placement and the inability to accurately sample from the same number of cells each experiment. Overall, this data suggests that (1) FSCV can be used to monitor these two molecules simultaneously in intestinal tissue and that (2) the tissue remains functional on-chip.

Figure 6.

Representative cyclic voltammograms (CVs) for co-detection of serotonin and melatonin in flow cell and ex vivo. A) Current (nA) vs. Voltage (V) of 1 μm serotonin and 5 μm melatonin detected simultaneously in a traditional flow cell. B) Current (nA) vs. Voltage (V) of serotonin and melatonin detected simultaneously in an intestinal slice.

Figure 7.

Serotonin and melatonin transients observed were statistically similar comparatively between the culture on and off chip. A) Average melatonin transient duration on (n = 2) and off chip (n = 3) were not significantly different (unpaired t-test, p = 0.3839) B) Average serotonin transient duration on and off chip were not significantly different (unpaired t-test, p = 0.2533) C) Average melatonin concentration on and off chip was significantly different (unpaired t-test, p < 0.0001) D) Average serotonin concentration on and off chip was significantly different (unpaired t-test, p < 0.0001) E) Average melatonin inter-event time on and off chip were not significantly different (unpaired t-test, p = 0.9367) F) Average serotonin inter-event time on and off chip were not significantly different (unpaired t-test, p = 0.8456) G) Average number of melatonin transients observed on and off chip were not significantly different (unpaired t-test, p = 0.8698) H) Average number of serotonin transients observed on and off chip were not significantly different (unpaired t-test, p = 0.6855)

Conclusion

Overall, we demonstrate an open microfluidic culture device that recreates the physiological oxygen gradient in intestinal slices. The tunability and size of the gradient is controllable on-chip providing the ability to accommodate slices of varying sizes. As a proof of concept, the open well device was paired with FSCV monitoring to demonstrate the ability to measure neurochemical signaling in intestinal slices on-chip. Transient serotonin and melatonin release was observed in ex vivo intestinal slices cultured in a standard well. These events were compared to events measured while maintaining the oxygen gradient on-chip and we observed no significant differences demonstrating that our chip is just as functional as a traditional perfusion chamber. This platform provides a novel culture platform for live intestinal slices, with the ability to image or probe the slice during experimentation. We demonstrate the use of this device for maintaining oxygen in intestinal tissue; however, the design could be implemented for a variety of other tissue culture applications which necessitate the need to create a physiologically- relevant gradient. Overall, this chip will significantly advance our ability to culture precision cut intestinal slices and will widen the experimental capabilities of monitoring intestinal tissue ex vivo.