Here, we present a protocol to produce 3D-bioprinted cocultures of iPSC-derived neurons and astrocytes. This coculture model, generated within a hydrogel scaffold in 96- or 384-well formats, demonstrates high post-print viability and neurite outgrowth within 7 days and shows the expression of maturity markers for both cell types.
For a cell model to be viable for drug screening, the system must meet throughput and homogeneity requirements alongside having an efficient development time. However, many published 3D models do not satisfy these criteria. This therefore, limits their usefulness in early drug discovery applications. Three-dimensional (3D) bioprinting is a novel technology that can be applied to the development of 3D models to expedite development time, increase standardization, and increase throughput. Here, we present a protocol to develop 3D bioprinted coculture models of human induced pluripotent stem cell (iPSC)-derived glutamatergic neurons and astrocytes. These cocultures are embedded within a hydrogel matrix of bioactive peptides, full-length extracellular matrix (ECM) proteins, and with a physiological stiffness of 1.1 kPa. The model can be rapidly established in 96-well and 384-well formats and produces an average post-print viability of 72%. The astrocyte-to-neuron ratio in this model is shown to be 1:1.5, which is within the physiological range for the human brain. These 3D bioprinted cell populations also show expression of mature neural cell type markers and growth of neurite and astrocyte projections within 7 days of culture. As a result, this model is suitable for analysis using cell dyes and immunostaining techniques alongside neurite outgrowth assays. The ability to produce these physiologically representative models at scale makes them ideal for use in medium-to-high throughput screening assays for neuroscience targets.
Research into central nervous system (CNS) diseases in the drug discovery industry is expanding1. However, many prevalent CNS diseases such as epilepsy, schizophrenia, and Alzheimer's disease still have no curative treatments2,3,4. The lack of effective therapeutics across CNS diseases can, at least in part, be attributed to a lack of accurate in vitro models of the brain5. This has resulted in a translational gap between current in vitro models and in vivo data and a subsequent bottleneck in research efforts.
Driven by this translational gap, there has been a significant increase in the development of novel 3D cell models within recent years, including neural organoids, neurospheroids, and scaffold-based models6. The 3D structure of these models aids in recapitulating the neural microenvironment, including biomechanical stresses, cell-cell contacts, and the brain extracellular matrix (ECM)7. The brain ECM is a dynamic element of neurophysiology that occupies the space between neural cell types, including neurons, astrocytes, oligodendrocytes, and the neurovascular unit7. Recapitulation of the brain ECM has been shown to affect neuronal morphology and neuronal firing, and many complex 3D models of the brain have demonstrated deposition of ECM proteins by neural cell types8,9,10,11. Scaffold-based models consist of mature neural cocultures suspended in a porous synthetic or biological hydrogel matrix which represents the brain ECM12. Unlike organoid and spheroid systems, scaffold-based 3D models allow the customization of ECM proteins present and have the added benefit of tunability of hydrogel stiffness to mimic biomechanical stresses13,14.
Although an overwhelming majority of 3D neural models demonstrate an increased recapitulation of the brain microenvironment, not all models are well suited for implementing drug discovery applications15. For a 3D model to be implemented into industrial processes, the system must meet throughput requirements for screening applications and have a relatively short development time16. 3D Bioprinting is a novel technology that offers the potential to create 3D scaffold-based neural models with rapid development time, increased throughput, and higher levels of precision control, alongside the removal of variability caused by human error17. This protocol presents a 3D coculture model of human iPSC-derived glutamatergic neurons and astrocytes in a hydrogel scaffold. This hydrogel scaffold contains physiologically representative bioactive peptides (RGD, IKVAV, YIGSR) and ECM proteins within a mimetic biomechanical stiffness. These full-length ECM proteins include full-length laminin-211 and hyaluronic acid, abundant in the human cortex, with a stiffness of 1.1 kPa in line with in vivo measurements18. This model is designed with practicality for drug discovery, and is created using a 3D bioprinter in a 96-well or 384-well plate format suitable for screening analysis using imaging techniques with cell dyes and antibodies, alongside neurite outgrowth assays. Cells show expression of neural cell type markers and growth of neurite and astrocytic projections within 7 days of culture. Thus, this protocol will present the methodology to develop a high-throughput 3D neural coculture model for use in drug discovery applications.
Figure 1: Illustrative overview of methodology used to 3D bioprint cocultures. Human iPSC-derived neurons and astrocytes are combined with activator and bioink solutions containing bioactive peptides and are bioprinted to hydrogel scaffolds in 96-well or 384-well formats using drop-on-demand bioprinting technology. Please click here to view a larger version of this figure.
1. Bioprinting of 3D models
2. Cell culture
3. Neurite growth analysis
4. Cell viability analysis
5. Immunostaining and cell population analysis
Neurite growth analysis
In this protocol, iPSC-derived glutamatergic neurons and astrocytes were bioprinted in coculture into a hydrogel matrix using the 3D bioprinter. Over the first 7 days post-printing, cells were imaged every 12 h using a live cell microscope. Post-bioprinting, cells should have a rounded morphology and should be dispersed throughout the hydrogel matrix, gradually changing to form smaller cell clusters with few protrusions over the first few days of culture (see Supplementary Video 1 for representative healthy cell growth). By day 4, healthy cells will migrate throughout the gel to form larger clusters, which are connected through neurite outgrowths. By day 7, almost no single cells should remain, the interconnecting bundles of neurites and astrocytic projections should appear fortified, and many smaller neurite outgrowths can be seen forming from the clusters (Figure 2A). Using a series of live cell brightfield images taken over the 7-day growth period, an analysis of neurite outgrowth was performed as detailed in section 3. This analysis demonstrated that neurite outgrowth increases in a near linear fashion (R2 value = 0.84) between 12 h and 156 h (Figure 2B). During this period of neurite outgrowth, cell body clusters also increase in size (see Supplementary Video 1), which is indicative of cell migration throughout the hydrogel.
Cell viability and population ratio
In this protocol, a concentration of 20 million cells/mL, comprising 15 million neurons/mL and 5 million astrocytes/mL, is used for bioprinting the cell models. Using live cell staining with calcein-AM (live cells), ethidium homodimer-1 (dead cells) and a nuclear stain, the number of cells surviving over a 7-day period can be calculated as per section 4 (Figure 3A). Cell viability results for representative cultures are shown for day 4, where 72% ± 1% (mean ± SEM) of total cells are live and show staining for Calcein-AM, while 29% ± 2% (mean ± SEM) total cells are dead and show staining with ethidium homodimer-1 (Figure 3B). Representative images of cell staining with Calcein-AM and ethidium homodimer-1 can be seen in Supplementary Figure 1. It should be noted that cell survival values for 3D cultures cannot be directly compared to 2D cultures, as dead cells are retained in the hydrogel and will not be removed during cell feeding processes.
Using the immunofluorescent staining for Β-III tubulin and GFAP, as described in section 5 and in Figure 4, image analysis can be carried out to determine cell population ratios between neurons and astrocytes (Figure 3A). Of total cells per model in representative cultures, Β-III tubulin positive neurons represent 49% ± 3% (mean ± SEM), while GFAP positive astrocytes represent 30% ± 4% (mean ± SEM). This gives a ratio of 1:1.5, astrocytes to neurons, respectively. This leaves a remainder of 21% total cells per model, which do not stain for either cell marker. As cell viability analysis demonstrated that a mean value of 29% of cells are not viable at day 4, it is likely that these cells are dead within the hydrogel.
Expression of cell markers
Morphology of the bioprinted neurons and astrocytes was assessed through immunostaining for neuronal cell type marker (Β-III tubulin) and astrocytic marker (GFAP). In the representative cultures shown, immunostaining is localized to individual cell types, showing healthy cell morphology, with both cell types exhibiting outgrowth of cellular protrusions (Figure 4A,B). As the hydrogel and cell structures are three-dimensional, each image represents only one slice through the structure in Figure 4A,B. Figure 4C shows a merged stack of images throughout the hydrogel, demonstrating views of cell localization in the X, Y, and Z planes. Figure 4D shows immunostaining for Β-III tubulin only; highlighting finer neurite outgrowths from the cell body clusters. To further examine the phenotype of the glutamatergic neurons, immunostaining for the glutamatergic ionic receptor marker, GluR2, can be carried out. Within Figure 4E, area 4E.1 (inset) has been highlighted to show higher resolution punctate staining along the neurite bundles. This therefore, confirms that neurons in this coculture have a glutamatergic phenotype. Across all immunostaining images, immunofluorescent stained non-cell structures can be observed surrounding the cell clusters and neurites. It is likely that these structures represent debris retained within the hydrogel in combination with minor amounts of non-specific antibody binding to the hydrogel. This is expected in bioprinted cultures, as within 3D scaffold models, dead cells and debris are not removed during cell feeding. A representative negative control immunostaining image is shown in Supplementary Figure 2 for a demonstration of hydrogel non-specific binding of secondary antibodies.
Figure 2: Glutamatergic neurons and astrocytes were bioprinted into the hydrogel matrix using the bioprinter and were imaged every 12 h using a brightfield microscope. (A) An example brightfield image taken of cell cultures during analysis. The image represents time point 156 h, and the scale bar represents 400 µm. (B) The average length of neurite outgrowths (µm) from the cultures measured using the NeuronJ package for ImageJ. Each data point is n = 3 neurites, and data is shown as mean ± SEM. Please click here to view a larger version of this figure.
Figure 3: A concentration of 20 million cells/mL of activator solution was used for bioprinting the cell models. (A) Cell viability was calculated using live/dead cell dyes (Calcein-AM and ethidium homodimer-1, respectively). Values show 72% ± 1% (mean ± SEM, n = 3) of total cells per well are live and 29% ± 2% (mean ± SEM, n = 3) of cells are dead of total cell population per well at Day 4. Values shown represent mean ± SEM. (B) Cell populations percentage of neurons and astrocytes per well were calculated through image analysis of staining shown in Figure 4. Neurons represent the percentage of cells staining positive for Β-III tubulin at Day 7 (49% ± 3%, mean ± SEM, n = 3), while astrocytes represent the percentage of cells staining positive for GFAP at Day 7 (30% ± 4%, mean ± SEM, n = 3). Values shown represent mean ± SEM. All imaging for calculations shown in Figure 3 was performed on a confocal imaging system, and all analyses were performed on the image analysis platform and GraphPad Prism as per methods. Please click here to view a larger version of this figure.
Figure 4: Expression of neural cell type markers in 3D bioprinted cocultures of glutamatergic neurons and astrocytes at day 7. (A,B) Immunofluorescent staining of neuronal marker Β-III tubulin and astrocyte marker GFAP, imaged on an inverted microscope platform at 10x magnification. Scale bars represent 100 µm. (C) Immunofluorescent staining of neuronal marker β-III tubulin and astrocyte marker GFAP co-stained with Hoechst, shown in XYZ plane view, imaged on a confocal imaging system at 10x magnification. Created on the image analysis platform. The scale bar represents 100 µm. (D) Immunofluorescent staining of neuronal marker β-III tubulin co-stained with Hoechst, imaged on a confocal imaging system at 20x magnification. Scale bar represents 100 µm. (E) Immunofluorescent staining of glutamatergic marker GluR2 co-stained with Hoechst, imaged on an inverted microscope platform at 10x magnification. Box 3E.1 shows highlighted areas of GluR2 staining. The scale bar represents 100 µm. Please click here to view a larger version of this figure.
Supplementary Video 1: Glutamatergic neurons and astrocytes were bioprinted into the hydrogel matrix using the bioprinter and were imaged every 12 h using a brightfield microscope. Video of brightfield images taken of cell cultures during analysis, time points are indicated in the bottom right corner, and scale bars represent 400 µm. Please click here to download this File.
Supplementary Figure 1: Example images of live/dead analysis of bioprinted neurons and astrocytes on Day 4. Calcein-AM stain shown in green (488 nm), and ethidium homodimer stain shown in red (647 nm). The image is shown in XYZ plane view, created on the image analysis platform . The scale bar represents 100 µm. (A) Imaging was carried out using a confocal imaging system at 4x magnification. (B) Imaging was carried out using a confocal imaging system at 10x magnification Please click here to download this File.
Supplementary Figure 2: Example of negative control image after immunofluorescent staining. Primary antibodies were omitted, and green (488 nm) and red (647 nm) secondary antibodies were used as per immunostaining protocols. The image is shown in XYZ plane view, created on the image analysis platform. The scale bar represents 100 µm. Imaging was carried out using a confocal imaging system at 10x magnification. Please click here to download this File.
The need for accurate models of the CNS has never been higher, and limitations of two-dimensional (2D) traditional cell culture models have driven a generation of complex CNS models in recent years19. However, many complex 3D models that represent interactions between neural cell types and the ECM have limitations that would prevent the application of these models in industrial processes6,20,21. In this protocol, we develop a 3D coculture model of human iPSC-derived neurons and astrocytes, which aims to resolve some of these limitations using 3D bioprinting technology to create a bioactive hydrogel scaffold in 96-well and 384-well formats.
The methodology for developing these models has been simplified through the plate map design software, auto-generated print protocols, and guided print process from the bioprinter. However, due to the sensitive nature of the sensitive iPSC-derived cell types used in this protocol care should be taken with the following critical steps in thawing and culture. Firstly, the inclusion of ROCK inhibitor (ROCKi) has several benefits throughout the bioprinting process and during early culture. Cell thawing is a critical point in which the neurons can experience a stress response, and improper thawing protocols can decrease the chances of survival22. It is typically recommended to thaw cells, add media, and raise cells to incubator temperature as efficiently as possible23. However, during the bioprinting process described in this protocol, it is necessary that neurons and astrocytes are resuspended in an activator solution rather than media, and cells will not be raised above room temperature until the end of the print run (up to 30 minutes post-thaw). Thus, the addition of ROCKi to the media immediately after thawing and including this during the two centrifugation steps (steps 2.1–2.7 and 1.3.15-1.3.20) is imperative to inhibit cell stress pathways, which would result in lower viability levels24. Furthermore, ROCKi has been shown to promote neurite outgrowth and improve neuronal maturation25. Thus, ROCKi supplementation is continued for 48 h post-bioprinting. However, it is imperative to remove ROCKi supplementation after 48 h to ensure complete wash-off during the subsequent media changes before cells are used for assay.
A further step that requires critical attention is during post-print media addition and media changes (steps 2.8-2.13). The bioprinted hydrogel scaffold has an equivalent biomechanical stiffness of only 1.1 kPa, equivalent to grey matter. As described in step 2.10, it is critical to pipette gently into the side of the well during media addition and aspiration to prevent disturbance. This is of particular relevance to 384-well plates, where the gel level takes up a higher proportion of total well volume. This method should also be used in 2D control wells to prevent edge lifting of cells and shearing of neurite outgrowths. The authors would also like to highlight the importance of sterile technique within the bioprinter, which should be treated with equivalent caution to that of a biosafety cabinet used for iPSC-derived cell cultures. This includes sterile filtering 70% EtOH and dH2O used in the greenlighting and printing procedures, keeping lids on the cartridges and plates while moving hands in and out of the bioprinter, and decontaminating surfaces inside the bioprinter with 70% ethanol wipes before and after printing.
The bioprinted hydrogel scaffold, formed from bioink and activator solutions, selected to develop this model is selected from a range of bioinks and activator solutions developed by Inventia Life Science for use within the RASTRUM bioprinter. Laminin and hyaluronic acid were identified as molecules of relevance to iPSC-derived neuronal maturation due to their role in axonal guidance, synapse formation, and formation of the perineuronal net26,27. Furthermore, a biomechanical stiffness of 1.1 kPa was selected, as lower-density hydrogels have been shown to enable better neurite outgrowth from neurons12. If modifications are made to the protocol by using neurons and astrocytes that have been differentiated in-house or from a different commercial supplier, it would be recommended to do a matrix selection test to determine the most supportive hydrogel scaffold15. Furthermore, cell density may also need to be optimized if changes to cell sources are made to ensure optimal viability and prevent hydrogel overcrowding. For all modifications and troubleshooting related to the bioprinter function, the authors recommend contacting manufacturers and referencing manufacturer protocols.
The CNS contains a broad range of neuronal subtypes and glial cells, all of which exist in different brain niches and have specific roles contributing to neural function28. Within the context of this broad scope, this model represents only the two most abundant cell types (astrocytes and excitatory glutamatergic neurons). Important cell types such as microglia, oligodendrocytes, and blood-brain barrier-forming endothelial cells are omitted from this system. The inclusion of microglia could be of relevance in focus on neuroimmune interactions, and oligodendrocytes could be of interest in diseases that affect central myelination. In addition to their role in pathology, cells such as blood-brain barrier-forming endothelial cells excrete drug-metabolizing enzymes, which could affect the use of this model for pharmacokinetic assays29. A further limitation of the model may be the ratio of astrocytes to neurons; the ratio of astrocytes to neurons varies greatly between brain regions, with suggested values of between 1:1 and 1:330,31. This model has an approximate ratio of 1:1.5 astrocytes to neurons; thus, this model might not be of relevance to model brain areas where astrocytes are more abundant, such as in white matter areas30.
Other protocols to develop 3D bioprinted coculture models have been published in recent years. A publication by Sullivan et al., 2021, presented a 3D bioprinted neural model using iPSC-derived neural progenitor cells, which demonstrates high post-print viability and enhancement of neuronal function compared to 2D cultures32. However, in this protocol, neural progenitor cells were used as a cell source and were maintained in culture for 4 weeks. In this protocol, commercially available iPSC-derived glutamatergic neurons and astrocytes were used. This allows a 3D network of co-cultured cells to be established in as little as 7 days; as demonstrated by neurite growth analysis, neurite outgrowth begins within 24 h and continues in a linear fashion throughout the 156 h period for which cell growth was monitored. The rapid establishment of these networks can be partly attributed to the use of glutamatergic neurons that use optimized doxycycline-inducible gene expression of NGN2, which shows expression of mature neuronal subtype markers within 7 days, even in 2D culture33. The shortening of this growth period using this technique is important to implementing models within the biopharmaceutical industry, as assay development requires rapid turnaround and development of cell models15.
In conclusion, this model shows potential for a 3D model of neurons and astrocytes, which is established quickly and conveniently for screening purposes. Future applications for this model type could be for drug discovery efforts across different CNS diseases, with the opportunity to expand to different diseases using patient or gene-edited disease iPSC lines. Furthermore, the use of doxycycline-inducible NGN2 expression iPSC-derived glutamatergic neurons allows cells to reach maturity in less time, which can be utilized for developing models of the aging brain for neurodegeneration research. This system could also be expanded through the use of additional cell types in coculture, including microglia and oligodendrocytes.
The authors have nothing to disclose.
The authors would like to thank Alex Volkerling, Martin Engel, and Rachel Bleach for their assistance in developing the protocol and feedback on the manuscript.
2-mercaptoethanol | Thermofisher | 31350010 | |
384-well plate | PerkinElmer | 6057300 | |
96-well plate | PerkinElmer | 6055300 | |
Activator fluid F299 | Inventia Life Science | N/A | |
Activator fluid F3 | Inventia Life Science | N/A | |
B27 (50x) minus Vit A | Thermofisher | 12587010 | |
Bioink fluid F261 | Inventia Life Science | N/A | |
Bioink fluid F32 | Inventia Life Science | N/A | |
Doxycycline hyclate | Sigma Aldrich | D5207 | |
GlutaMAX (100x) | Thermofisher | 35050061 | |
Goat anti-mouse IgG H&L Alexa Fluor 647 | Abcam | ab150115 | |
Goat anti-rabbit IgG H&L Alexa Fluor 488 | Abcam | ab150077 | |
Hoechst | Abcam | ab228551 | |
Human BDNF Recombinant Protein | Thermofisher | PHC7074 | |
Human NT3 Recombinant Protein | Thermofisher | PHC7036 | |
iCell Astrocytes | Fujifilm CDI | 1434 | |
INCell Analyser 6500HS | Molecular Devices | N/A | high content imaging system |
Incucyte S3 | Sartorius | N/A | |
ioGlutamatergic Neurons (Large vial) | Bit.bio | e001 | |
Laminin (1 mg/mL) | Sigma Aldrich | L2020 | |
Live/dead kit (Calcein-AM, Ethidium homo-dimer-1) | Invitrogen | L3224 | |
Mouse anti-BIII tubulin NL637 conjugated | R&D systems | SC024 | |
Neurobasal media | Thermofisher | 21103049 | |
Normal Donkey Serum | Abcam | ab7475 | |
NucBlue Live (Hoechst 33342) | Thermofisher | R37605 | |
Opti-MEM | Thermofisher | 11058021 | |
Paraformaldehyde | Sigma Aldrich | P6148 | |
PEI 50% in H2O | Sigma Aldrich | 181978 | |
Pierce Borate Buffer 20x | Thermofisher | 28341 | |
Prism | GraphPad | Data analysis software | |
Rabbit anti-ionotropic glutamatre receptor 2 (GluR2) | Abcam | ab206293 | |
RASTRUM(TM) Bioprinter | Inventia Life Science | N/A | Bioprinter |
RASTRUM(TM) Bioprinter Cartridges | Inventia Life Science | N/A | Bioprinter Cartridges |
RASTRUM(TM) Targeting plate | Inventia Life Science | N/A | Targeting plate |
Rho kinase (ROCK) inhibitor | Abcam | ab120129 | |
Sheep anti-GFAP NL493 conjugated | R&D systems | SC024 | |
Signals Image Artist | PerkinElmer | N/A | Image analysis platform |
Triton X-100 | Thermofisher | HFH10 | |
Zeiss Axio Observer | Zeiss | N/A | Inverted microscope platform |