neuroHuMiX is an advanced gut-on-a-chip model to study the interactions of bacterial, epithelial, and neuronal cells under proximal and representative co-culture conditions. This model allows unravelling of the molecular mechanisms underlying the communication between the gut microbiome and the nervous system.
The human body is colonized by at least the same number of microbial cells as it is composed of human cells, and most of these microorganisms are located in the gut. Though the interplay between the gut microbiome and the host has been extensively studied, how the gut microbiome interacts with the enteric nervous system remains largely unknown. To date, a physiologically representative in vitro model to study gut microbiome-nervous system interactions does not exist.
To fill this gap, we further developed the human-microbial crosstalk (HuMiX) gut-on-chip model by introducing induced pluripotent stem cell-derived enteric neurons into the device. The resulting model, ‘neuroHuMiX’, allows for the co-culture of bacterial, epithelial, and neuronal cells across microfluidic channels, separated by semi-permeable membranes. Despite separation of the different cell types, the cells can communicate with each other through soluble factors, simultaneously providing an opportunity to study each cell type separately. This setup allows for first insights into how the gut microbiome affects the enteric neuronal cells. This is a critical first step in studying and understanding the human gut microbiome-nervous system axis.
The human gut microbiome plays a crucial role in human health and disease. It has been extensively studied over the past decade and a half, and its potential role in modulating health and disease is now established1. A disruption in the microbiome leading to an unbalanced microbial community (dysbiosis) has been postulated to be involved in the pathogenesis of many chronic disorders, such as obesity, inflammatory bowel disease, and colorectal cancer, or even neurodegenerative diseases such as Parkinson's disease2,3.
Although the human gut microbiome has been associated with neurological conditions, it is still unclear how the gut microbiome communicates with and affects the enteric nervous system. As the human enteric nervous system is not easily accessible for immediate study, animal models have been used in experiments so far4. However, given the apparent differences between animal models and humans5, the development of in vitro models mimicking the human gut is of immediate interest. In this context, the burgeoning and advancing field of human induced pluripotent stem cells (iPSCs) has allowed us to obtain representative enteric neurons (ENs)6. iPSC-derived ENs allow for the study of the enteric nervous system in in vitro culture models, such as cell culture inserts, organoids, or organs-on-a-chip7,8.
The human-microbial crosstalk (HuMiX) model is a gut-on-a-chip model mimicking the human gut9. The initial HuMiX model (hereafter referred to as the initial device) accommodated epithelial cells (Caco-2) and bacterial cells10,11. However, to study the gut microbiome-nervous system link, iPSC-derived ENs6 have also been introduced in the system (Figure 1). The proximal co-culture of neuronal, epithelial, and bacterial cells allows the analysis of the different cell types individually and study of the interactions between the different cell types in an environment mimicking that of the human gut.
In recent years, advances have been made in the development of models to study organs in more physiologically representative ways by using organs-on-a-chip (e.g., gut-on-a-chip) models. These models are more representative of the human gut environment due to constant nutrient supply and waste removal, as well as real-time monitoring of, for example, oxygen levels or barrier integrity8,12. These models specifically allow study of the effects of gut bacteria on host cells. However, to be able to use organs-on-a-chip to study the interrelationships between the gut microbiome and the nervous system, neuronal cells need to be integrated into such systems. Therefore, the aim of further developing HuMiX and establishing the neuroHuMiX system (hereafter referred to as the device) was to develop a gut-on-a-chip model, which includes enteric neuronal cells in proximal co-culture with gut epithelial cells and bacteria.
1. Cell culture and sorting
2. HuMiX run preparation
3. HuMiX start
4. Cell preparation and inoculation
NOTE: This section describes how to prepare the different cell types needed to inoculate the device, as well as how to inoculate them in the device in a sterile manner and without introducing air bubbles. Furthermore, it describes how to perform medium refreshment for the neuronal cells, and how to prepare medium for the bacterial culture in the device.
5. Bacterial culture and inoculation
NOTE: In this study, on Day 12, a liquid culture of Limosilactobacillus reuteri strain F275 was reanimated from a glycerol stock. Depending on the needs or study designs, other bacterial species can be used.
6. HuMiX opening and sampling
NOTE: The below section describes the sampling of different cell types. For example, neuronal cell pellets are used for RNA extraction and subsequent quantitative polymerase chain reaction (qPCR), bacterial pellets for DNA extraction and 16S rRNA gene sequencing, and supernatants for enzyme-linked immunosorbent assays (ELISAs) and other assays (e.g., lactate assay).
In neuroHuMiX, we co-cultured three different cell types together-bacterial, epithelial, and neuronal cells (Figure 1). To make sure the cells were all viable, we performed different assays on the different cell types. For example, we performed CFU counts on bacterial cells, cell count and cell viability assays on the epithelial cells, while the neuronal cells were assessed via microscopic analyses.
Figure 1: Schematic representation of neuroHuMiX and its experimental setup. (A) The three chambers are held between two PC lids to keep them closed. Each chamber is filled with a specific medium for the cells grown inside. The different chambers are separated by semi-permeable membranes allowing cell communication via soluble factors passing the membranes. (B) Representation of the neuroHuMiX setup. Each chamber is connected to different media bottles. For the bacterial chamber, for the first 12.5 days, the chamber is connected to RPMI + 10% FBS, before being changed for the last 36 h to RPMI + 10% FBS + 5% MRS. Abbreviations: PC = polycarbonate; P/L/F = poly L-ornithine/laminin/fibronectin; RPMI = Roswell Park Memorial Institute cell culture medium; MRS = De Man, Rogosa, and Shapre culture medium. Please click here to view a larger version of this figure.
To determine whether the cells were appropriately attached, upon opening the devices, we assessed the formation of a cell layer on the collagen-coated membrane (Figure 6A). To make sure the cells in the device were viable, an automated cell counter count (Figure 6B) and a trypan blue exclusion assay cell count were performed (Figure 6C). The assays were performed on Caco-2 cells from three different HuMiX setups: (i) Caco-2 in culture with ENs, (ii) Caco-2 in culture with L. reuteri, and (iii) the device involving co-culture of all three cell types. Statistical testing using a one-way ANOVA did not yield any significant differences between the cell types, suggesting that the Caco-2 cells remained viable in all these initial device setups and conditions tested in this study. This underlines the fact that the bacterial density reached during the co-culture of L. reuteri and the two human cell types do not have cytotoxic effects on the human cells.
Figure 6: Assessment of Caco-2 cells on the collagen-coated membrane. (A) Layer of Caco-2 cells on the collagen-coated membrane after opening. The arrow indicates the collagen-coated membrane, which is surrounded by a dashed circle. The Caco-2 cells were growing on the spiral shape on the membrane. Cell viability of Caco-2 cells after 14 days in HuMiX. Cell counts were obtained using (B) the automated cell counter and (C) the trypan blue exclusion assay cell count. Caco-2 cell counts were determined from different culture setups in the initial device: co-culture with enteric neurons (ENs) (black), co-culture with L. reuteri (orange), and in the device (ENs and L. reuteri) (blue). A one-way ANOVA was performed, showing there is no significant difference between the different culture setups (one-way ANOVA, p = 0.1234 [ns]; error bars indicate standard error). Please click here to view a larger version of this figure.
To be able to culture L. reuteri with mammalian cells, we first optimized and adapted the culture media for use in the device. We found that a 5% mix of MRS in RPMI 1640 (supplemented with 10% FBS) was optimally suited for the growth of L. reuteri, while not being cytotoxic for the mammalian cells used in these assays. Subsequently, a CFU count was performed to assess the growth of L. reuteri when cultured in the device for 24 h. The CFU count was assessed for two different initial device setups (Figure 7)-L. reuteri co-cultured with Caco-2 and L. reuteri in the device. In both setups, the CFU counts were significantly different from the HuMiX inoculum and the harvested cells (one-way ANOVA, p = 0.0002), indicating growth of the bacterial cells inside the initial device.
Figure 7: Limosilactobacillus reuteri CFU count of the inoculum (diluted 1:100,000) and after 24 h in HuMiX. Two different setups: Caco-2 cells in co-culture with L. reuteri and the device. A one-way ANOVA shows a significant difference (p = 0.0002 [***]) between the inoculum and the harvested cells, meaning the bacteria are growing inside HuMiX. Error bars indicate the standard error. Abbreviations: CB.HX = Caco-2 bacteria HuMiX; nHX = neuroHuMiX. Please click here to view a larger version of this figure.
To assess whether culturing the ENs within the device would alter the phenotype of the cells, the gross morphology of the ENs was observed using an inverted phase-contrast microscope. During this step, both the confluency and EN morphology were assessed. Establishment of a confluent neuronal network indicated that the cells had attached well onto the coated device's PC lid. Importantly, this highlights the notion that they grew in co-culture with Caco-2 and L. reuteri. The edge between the confluent neuronal network and the gasket-delineated spiral was clearly apparent (Figure 8).
Figure 8: Enteric neurons after 14 days of culture in the device. On the left side of the image, the neurons have grown to a confluent layer on the spiral. The edge, between the neuronal layer and the space without cells, is the edge of the spiral; magnified 10x, scale bar = 200 µm. Please click here to view a larger version of this figure.
Figure 2: Lids used in the device. Images show top (left) and bottom (right) PC lids. Each side of the PC lid is 6.4 cm. Abbreviation: PC = polycarbonate. Please click here to view a larger version of this figure.
Figure 3: Epithelial chamber gasket on bottom PC lid. Top view of the epithelial chamber gasket placed on the bottom PC lid (left), and bottom view (right) showing the alignment of the epithelial chamber gasket with the inlets and outlets of the bottom PC lid. Each side of the gaskets, as well as the PC lid, measures 6.4 cm. Abbreviation: PC = polycarbonate. Please click here to view a larger version of this figure.
Figure 4: Assembly of the device. (A) Different parts for assembling HuMiX: (1) bottom PC lid; (2) gasket with collagen-coated microporous membrane, which is placed on top of (1); (3) sandwich gasket with a mucin-coated nanoporous membrane in between and placed on top of (2); (4) top PC lid placed on top of (3). Each side of the gaskets and PC lids measures 6.4 cm. (B) All parts from (A) placed together. (C,D) Assembled device-top (left) and side (right) view. B is placed into the clamping system to close the system. (C) Each side of the top clamp measures 8 cm. Abbreviation: PC = polycarbonate. Please click here to view a larger version of this figure.
Figure 5: Parts needed for tubing line and assembled tubing line for one chamber. (A) Different parts to build a tubing line: a. pump tubing line; b. three-way stopcock; c. 40 mm needle; d. 80 mm needle; e. 120 mm needle; f. long tubing line (20 cm); g. short tubing line (8 cm); h. male Luer; i. female Luer; j. adaptor. (B) Assembled tubing line for the bacterial or epithelial chamber. For the neuronal chamber, the 120 mm needle would need to be changed to an 80 mm needle. (C) Three-way stopcock valve turned to redirect the medium flow from the device to the 'open connector' and to close the chamber. Please click here to view a larger version of this figure.
Day | 0 | 2 | 4 | 6 | 8 | 10 |
Media Composition | 100% E6 | 100% E6 | 75% E6 | 50% E6 | 25% E6 | 100% N2 |
+ LDN | + LDN | 25% N2 | 50% N2 | 75% N2 | + LDN | |
+ SB | + SB | + LDN | + LDN | + LDN | + SB | |
+ CHIR | + SB | + SB | + SB | + CHIR | ||
+ CHIR | + CHIR | + CHIR | + RA | |||
+ RA | + RA | |||||
Molecule | [concentration] | |||||
LDN | 100 nM | |||||
SB | 10 µM | |||||
CHIR | 3 µM | |||||
Retinoic Acid (RA) | 1 µM |
Table 1: Media composition.
Media | Components (concentrations listed in Table of materials) | Volume (mL) |
N2 media (50 mL) | DMEM-F12 | 48 |
N2 Supplement | 0.5 | |
L-Glutamine | 0.5 | |
Penicillin/Streptomycin | 0.5 | |
NEAA | 0.5 | |
N2B27/ENS Media (50 mL) | Neurobasal | 48 |
N2 Supplement | 0.5 | |
L-Glutamine | 0.5 | |
Penicillin/Streptomycin | 0.5 | |
B27-A | 0.5 |
Table 2: Media recipes.
Sterilization Temperature (°C) | 116 |
Sterilization Time (min) | 20 |
Dry Time (min) | 10 |
Pulses | 3 |
End Temperature (°C) | 99 |
Table 3: HuMiX autoclave run.
Rotations per minute (rpm) | Average flow rate (µL/min) |
0.5 | 13 |
2 | 79 |
5 | 180 |
Table 4: Flow rates of the peristaltic pump.
It is now established that the human gut microbiome influences the host's health and disease. Despite the knowledge suggesting the importance of our microbiome, especially in neurological disorders such as Alzheimer's or Parkinson's disease3,13, it remains largely unknown how the gut microbiome interacts with the enteric nervous system, and subsequently, with the brain.
A representative model to study the interactions between the gut microbiome and the nervous system has thus far been unavailable. Studies regarding the gut-brain axis have traditionally been performed using murine models13. Mice and humans share 85% of their genomic sequences14, but there are significant differences to consider when comparing mice to humans. Regarding the gut, it is important to note that, compared to humans, mice are exclusively herbivores. As a result, their gastrointestinal tract differs in length and characteristics, such as the 'gastric emptying'14. Murine brains also show important differences, whereby the overall structure between mice and humans are different15. Importantly, humans have longer cell cycle times of neural progenitors15. Consequently, it is important to develop representative models that include human-derived cells, including intestinal and neuronal cells5. In this context, the development of more reproducible research viain vitro models reduces the need to use animal models and improves reproducibility.
neuroHuMiX is an advanced version of the previous HuMiX model9. HuMiX is a gut-on-a-chip model allowing proximal and representative co-cultures of epithelial and bacterial cells. Cell-cell communication is possible through the proximal co-culture and diffusion of secreted factors and metabolites via semipermeable membranes. However, to expand the utility of the initial device to study the human gut environment, the introduction of an additional cell type is required. To address this, neuroHuMiX, developed with the introduction of iPSC-derived ENs, enables a proximal co-culture of bacteria, intestinal epithelial cells, and ENs. The resulting in vitro model allows us to address questions regarding the human gut microbiome in relation to the human nervous system. Co-culturing different cell types, especially co-cultures of mammalian cells and bacteria, has several challenges, including the loss of viability, poor adhesion, and overall loss in confluence16. Here, we have demonstrated that within this device, we are able to co-culture three different cell types within the same system while keeping the cell viability high.
A critical step in the protocol is to ensure confluency of the neuronal cells-80%-90% cell confluency and viability-before inoculating into the device. Since it is not possible to assess the cell growth during the run, it is of utmost importance to ensure the cells are confluent and growing well before introducing them in the model. While this may be a limiting factor, the overall viability and confluency observed within the device is generally high.
The device is connected via tubing lines to a peristaltic pump. Each cell chamber has its specific tubing line. The tubing comprises a pump tubing that allows the use of a peristaltic pump for the perfusion of medium, as well as tubing connecting the pump tubing to the device and tubing connecting the device to the outflow/waste bottles. Sampling ports are included before and after the device, to allow the inoculation and sampling of outflow medium. Each chamber can be connected to a different medium, allowing the best culture conditions for each individual cell type. Each chamber can be opened or closed depending on the specific needs for medium supply. In the device, the neuronal chamber stays closed for most of the experiment, while the bacterial and epithelial chambers are open all the time, meaning they get fresh medium throughout the whole experimental run. To make sure the medium is flowing without interruption, it is crucial to not have any air left in the tubings, connectors, or in the device. Therefore, it is important to first let the devices run for a few minutes at the priming step. This often resolves the issue. If not, one of the other lines that are dropping can be closed for a short amount of time by closing the three-way stopcock of the outflow. This redirects the medium to the line with the air bubble, thus resolving the issue by pushing the bubble outward through the tubing.
For any cell culture experiment, the medium is a key component, where each cell type has its respective medium. In a co-culture setup, the medium needs to be compatible not only for the cell type growing in it, but also for the other cell types within the co-culture. This is no different for the device, which poses an additional challenge as we have three different compartments with three different cell types inside-bacterial, epithelial, and neuronal cells. We have, however, shown that by modifying the bacterial media-with the addition of 5% MRS to RPMI 1640 with 10% FBS-all cell types, in particular bacterial and epithelial cells, can be successfully co-cultured within the system. However, in the device, different cell types are co-cultured in proximity, and are hence not in direct contact with one another. Even though this is not fully representative of the direct contact between cells in the human gut, and therefore a limitation, the proximal and representative co-culture condition is a strength for downstream analyses. Soluble factors exchange between the different chambers and cell types; hence, the cells are still interacting with each other. Additionally, the fact that the cell types can be harvested and analyzed separately allows us to study the effect of a healthy and/or diseased microbiome on different cell types (including neuronal cells) and thereby determine/retrieve cell type-specific read-outs. Another limitation is that the morphology of the cells cannot be followed-up during the experimental run, as the device can only be opened and the cells checked at the end of each experiment.
To our knowledge, neuroHuMiX is the first gut-on-a-chip model including ENs. This is a step toward elucidating the communication between the gut microbiota and the enteric nervous system. It is a model allowing investigation of the interplay between a bacterial species, an epithelial layer, and ENs. Its design allows us to study the exchange of soluble factors secreted by the different cell types and their effect on one another. Going forward, it would be important to not only have iPSC-derived ENs, but also iPSC-derived epithelial cells inside the device, to transition the device into a personalized model. Importantly, this personalized model could be used to test pre-, pro-, and synbiotics10,11 and potentially develop personalized screening and therapeutic approaches17. Personalized neuroHuMiX could eventually shed light on the 'dark matter' of the human gut microbiome and its interactions with the nervous system along the gut microbiome-nervous system axis, paving the way for therapeutic assessment and interventions.
We can conclude that being able to have a gut-on-a-chip including the enteric neuronal system is crucial to progressing in the study and understanding of interactions along the gut microbiome-nervous system axis. NeuroHuMiX allows us to study the effects of bacterial species on host cells and provides us with a good basis to improve the model even further in an even more physiologically representative way.
The authors have nothing to disclose.
The authors would like to thank Dr. Jared Sterneckert for providing us with the cells from the K7 line. We also want to thank the long-standing collaborators Dr. Frederic Zenhausern and Matthew W. Barret from the University of Arizona for their assistance with the engineering aspects. We would also like to acknowledge Dr. Valentina Galata for her help in designing the schematic representation of neuroHuMiX. This project has received funding from the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation program (grant agreement 863664). Figure 1 was partially created with Biorender.com.
2-Mercaptoethanol | Sigma Aldrich | 10712 | |
Aeration cannula (length: 1.10 diameter: 30 mm) | VWR (B.Braun) | BRAU4190050 | |
Agar-agar | Merck Millipore | 1.01614.1000 | |
Aluminium Crimp | Glasgerätebau Ochs | 102050 | |
Ascorbic acid | Sigma Aldrich | A4544 | |
B-27 Supplement Minus Vitamin A (50x) | Gibco | 12587-010 | |
Bacterial Cell Membrane, pore size: 1 µm | VWR (Whatman) | 515-2084 | |
Caco-2 cells | DSMZ | ACC169 | |
Cell Counter & Analyzer CASY | OMNI Life Sceince | ||
CHIR | Axon Mechem BV | CT99021 | |
Collagen I, Rat Tail | Invitrogen | A1048301 | |
Costar 6-well Clear Flat Bottom Ultra-Low Attachment Plates | Corning | 3471 | |
Difco Lactobacilli MRS Broth | BD Biosciences | 288130 | |
Discofix 3-way stopcock | B. Braun | BRAU40951111 | |
DMEM/F12, no glutamine | Thermofisher Scientific | 21331020 | |
Dulbecco's Phosphate-Buffered Saline, D-PBS | Sigma Aldrich | 14190-169 | |
Essential 6 Medium | Thermofisher Scientific | A1516401 | |
Essential 8 Medium | Thermofisher Scientific | A1517001 | |
Female Luer Lock to Barb Connector | Qosina | 11733 | |
FGF2 | R&D Systems | 233-FB | |
Fibronectin | Sigma Aldrich | F1141 | |
Foetal Bovine Serum, FBS | Thermofisher Scientific | 10500-064 | |
GDNF | PeproTech | 450-10 | |
Human Cell Membrane, pore size: 50 nm | Sigma Aldrich (GE Healthcare) | WHA111703 | |
HuMiX Gasket Collagen | Auer Precision | 216891-003 | |
HuMiX Gasket Sandwich Bottom | Auer Precision | 216891-002 | |
HuMiX Gasket Sandwich Top | Auer Precision | 216891-001 | |
iPSC | Max Planck Institute for Molecular Biomedicine | K7 line | |
L-Glutamine (200 mM) | Gibco | 25030081 | |
Laminin from Engelbreth-Holmswarm | Sigma Aldrich | L2020 | |
LDN193189 | Sigma Aldrich | SML0559 | |
Limosilactobacillus reuteri | ATCC | 23272 | |
Live/Dead BacLight Bacterial Viability kit | Thermofisher Scientific | L7012 | |
Male Luer with Spin Lock to Barb | Qosina | 11735 | |
Marprene tubing (0.8 mm x 1.6 mm) | Watson-Marlow | 902.0008.J16 | |
Matrigel hESC-qualified matrix | Corning | 354277 | |
Mucin, from porcine stomach | Sigma Aldrich | T3924 | |
N2 Supplement (100x) | Gibco | 17502048 | |
NEAA | Thermofisher Scientific | 11140050 | |
Needle (length: 120 mm; diameter: 0.80 mm) | B.Braun (color code: green) | 466 5643 | |
Needle (length: 40 mm; diameter: 0.70 mm) | Henke Sass Wolf (color code: black) | 4710007040 | |
Needle (length: 80 mm; diameter: 0.60 mm) | B.Braun (color code: blue) | 466 5635 | |
Neurobasal Medium | Gibco | 21103049 | |
PE/Cy7 anti-human CD49d antibody | Biolegend | 304314 | |
Penicillin-Streptomycin | Sigma Aldrich | P0781 | |
Peristaltic pump | Watson-Marlow | 205CA | |
Poly-L-ornithine Hydrobromide | Sigma Aldrich | P3655 | |
Polycarbonate lids (HuMiX) | University of Arizona | HuMiX 1.0 / 2.0 | |
Retinoic Acid | Sigma Aldrich | R2625 | |
RLT Buffer (RNeasy Minikit) | Qiagen | 74104 | |
RPMI 1640 Medium | Thermofisher Scientific | 72400-021 | |
SB431542, ALK inhibitor | Abcam | ab120163 | |
Serum bottles | Glasgerätebau Ochs | 102091 | |
Syringe | BD Biosciences | 309110 | |
Trypsin-EDTA solution | Sigma Aldrich | T3924 | |
Y-27632 Dihydrochloride | R&D Systems | 1254 |