In this protocol, we aim to describe a reproducible method for combining dissociated human pluripotent stem cell derived neurons and astrocytes together into 3D sphere cocultures, maintaining these spheres in free floating conditions, and subsequently measuring synaptic circuit activity of the spheres with immunoanalysis and multielectrode array recordings.
A barrier to our understanding of how various cell types and signals contribute to synaptic circuit function is the lack of relevant models for studying the human brain. One emerging technology to address this issue is the use of three dimensional (3D) neural cell cultures, termed 'organoids' or 'spheroids', for long term preservation of intercellular interactions including extracellular adhesion molecules. However, these culture systems are time consuming and not systematically generated. Here, we detail a method to rapidly and consistently produce 3D cocultures of neurons and astrocytes from human pluripotent stem cells. First, pre-differentiated astrocytes and neuronal progenitors are dissociated and counted. Next, cells are combined in sphere-forming dishes with a Rho-Kinase inhibitor and at specific ratios to produce spheres of reproducible size. After several weeks of culture as floating spheres, cocultures ('asteroids') are finally sectioned for immunostaining or plated upon multielectrode arrays to measure synaptic density and strength. In general, it is expected that this protocol will yield 3D neural spheres that display mature cell-type restricted markers, form functional synapses, and exhibit spontaneous synaptic network burst activity. Together, this system permits drug screening and investigations into mechanisms of disease in a more suitable model compared to monolayer cultures.
Astrocytes are a highly abundant glial cell type within the central nervous system (CNS) with a variety of functional responsibilities beyond structural support. Through secretion of soluble synaptogenic factors and extracellular matrix (ECM) components, astrocytes aid in the establishment and clustering of mature synapses during development1. They also play a critical role in maintaining the health and plasticity of synapses through extracellular signaling2,3,4,5, and contribute to long-term stability of homeostatic environments by regulating extracellular potassium and glutamate, as well as the secretion of energy substrates and ATP6,7,8. Finally, they can contribute to neurotransmission by influencing extrasynaptic currents9, and can indirectly influence activity through other cell types such as promoting myelination10. Importantly, because abnormality or dysfunction of astrocytes can lead to many neurodevelopmental syndromes and adult neuropathology, there is an obvious need to include astrocytes alongside neurons within engineered neural networks in order for an improved model of the endogenous brain environment. An integral characteristic of astrocytes is their ability to form dynamic interactions with neuronal synapses1,11,12. In the absence of glia, neurons form a limited number of synapses, which in general also lack functional maturity13.
Human astrocytes display morphological, transcriptional, and functional characteristics — such as increased size and complexity of branching, as well as species-specific genes — that are not recapitulated in rodents12,14,15. As a result, studies utilizing human pluripotent stem cell (hPSC)-derived neural cells have become widely accepted as a means of examining CNS-related diseases in vitro while developing novel therapies, injury models, and culture paradigms16,17. Furthermore, hPSCs permit the study of human synapse formation and function without the need for primary tissue18,19.
A barrier to our understanding of how various cell types and signals contribute to synaptic circuit function is the lack of relevant models of the human brain. There is a need for an appropriate platform to recapitulate its synaptic networks with high fidelity and reproducibility. Recently, interest has emerged in the production of 3D culture systems (broadly known as 'organoids,' 'spheroids,' or 'mini brains')20 to model complex three-dimensional (3D) structures at the cellular and macro levels. 3D culture systems retain ECM and cell-cell interactions that are normally absent or limited during typical 2D coculture paradigms21,22. An abundance of techniques exist for culturing 3D neural spheroids23,24,25; however, many require lengthy culture periods (months to years) for spontaneous development and layer preservation, with the user exhibiting very little control over the output.
Here, we illustrate a systematic method to rapidly and consistently bioengineer neural interactions among multiple cell types (pre-differentiated neurons and astrocytes) derived from hPSCs by assembling cells into sphere cocultures ('asteroids')26 that recapitulate human-specific morphological complexities in 3D. This high-density neural system generates evenly-dispersed neural subtypes that take on mature properties over time and can be screened or assayed in a high-throughput manner. We demonstrate for the first time that human astrocytes induce synaptic network burst activity in these 3D cocultures. In addition, this protocol is easily adaptable to generate spheres of different sizes, to utilize cells specified to different regional identities of the CNS, and to study interactions of multiple other cell types as desired.
1. Cell Culture and Reagent Preparation
NOTE: The protocols in this section are written in the order in which they appear in the differentiation protocol (section 2). See the Table of Materials for materials and catalog numbers.
2. Generation of Neural Subtypes from Human Induced Pluripotent Cells (hPSCs)
NOTE: All cell cultures should be maintained in an incubator with 5% CO2 at 37 °C. These cultures are maintained at room oxygen levels, though lower levels may be utilized.
3. Preparation and Maintenance of 3D Sphere Cocultures
4. Measurement of Live Synaptic Physiology with Multielectrode Arrays (MEAs)
5. Measurement of Synaptic Density with Immunocytochemistry
When performed properly, this protocol will produce defined populations of functional cocultures of astrocytes28,33,34 and neurons35 generated from hPSCs (Figure 1A-1C), as detailed previously26 and described here in steps 2.1–2.2. This stepwise procedure, with the use of microwell plates, is expected to yield 3D neural spheres of consistent size and shape (step 3.3; Figure 1D) with evenly dispersed cells and without significant signs of cell death. A range of starting cell densities, with a desired ratio of cell types, will produce spheres of varying sizes.In the absence of microwell plates, cells will combine to form significantly larger and nonuniform aggregates whose diameters (> 2 mm) surpass the limit of diffusion (Figure 2A-2B). It is anticipated that the use of a spinner flask or equivalent bioreactor will maintain uniformity among spheres and reduce fusion (step 3.4; Figure 2C). However, while spinner flasks allow for culture of spheres for weeks, their use is not necessary.
To stabilize cocultures for electrophysiological analysis, ECM-mimicking substrates allow spheres to readily adhere while maintaining their 3D structure (Figure 3A, 3A'). During recordings, healthy spheres of iNeurons placed on MEAs will spontaneously elicit voltage spikes greater than ± 40 µV with consistent firing frequencies (steps 4.1–4.2; Figure 3B-C, 3F). Coculture spheres are expected to display greater network connectivity with the presence of hAstros, resulting in an increase of synchronous network bursts of spikes (Figure 3B'-3C'). The application of CNQX35,39,40, a postsynaptic AMPA receptor antagonist, reduces network burst synchrony in coculture spheres (Figure 3D'-3E').
Neural cell-type restricted protein markers visually demonstrate the evenly dispersed arrangement and maturity of astrocytes and neurons within 3D spheres. A maximum projection of astrocyte morphology and branching is shown in Figure 4A in a representative 3D sphere with hAstros sparsely labeled with membrane-bound GFP. Mature hAstros express markers including GFAP (Figure 4B), S100B, and Glt134, whereas hNeurons and iNeurons express microtubule-associated protein 2 (MAP2; Figure 4C) and tubulin beta 3 (Tuj1/TUBB3). Finally, though iNeurons express pre- and post-synaptic proteins including Synapsin 1 (Syn1) and Homer or PSD95, respectively, synaptic density is significantly enhanced by the presence of hAstros in coculture (Figure 4D)26. If available, the alternate use of brain clearing techniques and light sheet microscopy will enable rapid imaging of intact spheres.
Taken together, the expression of mature neural markers along with spontaneous electrical activity confirm the success of the protocol detailed above in producing 3D cocultures of functional synaptic microcircuits.
Figure 1: Stepwise depiction of the differentiation and formation of 3D neural spheres derived from hPSCs. (A) Timeline of key steps in the protocol. (B) Pure populations of neural cells can be generated from human induced pluripotent stem cells (hPSCs). For the generation of astrocyte progenitors (dotted green box), see step 2.1 as well as Ref28. (C) Inducible neurons (iNeurons; see section 2.2) generated from transgenic hPSCs via induced overexpression of neurogenin 2 demonstrate neuronal morphology on 2D ECM (day 7) and are positive for MAP2 (inset). (D) Spheres removed from microwell plates (see steps 3.1–3.3) demonstrate consistent size for high-throughput screening. Spheres may be cultured in a spinner flask bioreactor (inset; see step 3.4) if desired to prevent fusion. Scale bar = 50 µm (C, inset). Scale bars = 200 µm (A, D). Please click here to view a larger version of this figure.
Figure 2: Systematic and reproducible generation of bioengineered 3D neural spheres. (A) Microwell (µ-well) culture plates are used to form 3D neural spheres of desired density and neuron-to-astrocyte ratios in a systematic and user-specified manner, yielding consistent size and shape. Images are shown 1 day after sphere formation. Scale bars = 200 µm. (B) A range of starting cell densities (from 4 x 103 to 2 x 104 cells per microwell) produces spheres of varying sizes. In the absence of microwell plates, hPSCs combine to form significantly larger and nonuniform aggregates whose diameters surpass the limit of diffusion after a week (n = 6-14 spheres per group and time point). (C) iNeuron spheres cultured in a spinner flask at 80 rpm exhibited less fusion and thus were significantly smaller, more uniform, and exhibited greater circularity compared to those in stationary culture over a period of three weeks (n = 6-51 spheres per group and time point). Plots represent mean ± SD; * indicates significance (p < 0.05) between groups, determined using two-tailed t-tests. Please click here to view a larger version of this figure.
Figure 3: Representative live synaptic physiology of neural spheres on multielectrode arrays (MEAs). (A, A') MEAs are utilized to measure live synaptic physiology of neural spheres. Spheres of iNeurons (A) or cocultures of iNeurons and hAstros (A') can be cultured on MEA surfaces with ECM-mimicking substrates (see step 4.1). (B, B') Raster plots of representative spontaneous electrical activity measured across all electrodes (vertical axis) during a recording window (horizontal axis). After 4 weeks of culture in neurophysiological basal medium41, iNeuron spheres (B) displayed spontaneous spikes, while coculture spheres (B') revealed increased network bursts of synchronous firing patterns (orange arrows). (C, C') Histogram of spikes measured during MEA recording windows from representative electrodes. (D, D') Raster plots of spontaneous electrical activity measured across all electrodes after the application of 50 µM CNQX, an AMPA receptor antagonist (same time scale as B, B'). CNQX eliminated the presence of synchronous bursts of spikes observed in cocultures (D'). (E, E') Histogram of spikes measured during MEA recording windows from representative electrodes after 50 µM CNQX application (same time scale as C, C'). (F) Map of spike tracings of iNeuron spheres from all electrodes over time (left). Demonstrative single electrode displaying multiple clustered traces over time (middle); individual spike traces can be sorted and analyzed individually (right). Please click here to view a larger version of this figure.
Figure 4: Characteristic immunohistochemistry of synaptic microcircuits. (A) A membrane-bound GFP (mGFP) reporter is useful for examining the arrangement and morphology of hAstros in 3D neural spheres. (B) hAstros differentiated from hPSCs are GFAP-positive. (C) Spheres containing hAstros and iNeurons can be visualized and analyzed using cell type-restricted protein markers such as GFAP and MAP2. Scale bars = 50 µm. (D) Representative images of pre- and post-synaptic densities (Syn1 and PSD95, respectively) in spheres of iNeurons without (left) and with (right) cocultured hAstros. A significantly increased density of colocalized Syn1 and PSD95 was observed in coculture spheres compared to iNeuron spheres at day 35, demonstrating the ability of hAstros to induce synapse formation (n = 3 independent replicates each; data reprinted from Ref26 with permission). Scale bars = 10 µm. Plot represents mean ± SEM; significant differences (* indicates p < 0.05; ** indicates p < 0.01) between groups were determined using two-tailed t-tests. Please click here to view a larger version of this figure.
In this protocol, we describe a systematic method for the production of 3D spheres of neural cocultures. The spheres are composed of astrocytes and neurons, which are derived independently from hPSCs. Though not the focus of this protocol, the generation of pure populations of astrocytes from hPSCs28 is a critical step and can be technically challenging if performed without prior experience. This first step in the generation of these synaptic microcircuits should be performed with meticulous timing and attention to detail. A limitation of the use of hPSC-derived astrocytes is the lengthy differentiation process; however, the production of large quantities of cells that can be frozen for future use and thawed at the time of experimentation (see step 2.1.5) eliminates the need to begin the process from the initial hPSC stage. Though the iNeurons generated from transgenic hPSCs are synaptogenic and have the ability to exhibit spontaneous postsynaptic electrical currents, both characteristics are notably enhanced by the presence of glia12,35— including the hAstros detailed in this protocol. Thus, it should be noted that the choice of cell type, number, and developmental stage or maturity is a critical component of this protocol, and adjustments may result in variations in electrical activity or synapse visualization. However, this protocol also permits manipulation of cell density or ratios in a manner that reflects the flexibility of possible conditions or outcomes from the individual researcher.
As described above, we take advantage of bioengineering tools to produce an alternative to neural organoid methods42,43, with the caveat that this system does not recapitulate the self-organization and layering phenotypes of organoids that may be desired20,44. The simultaneous use of microwell culture plates26 and a spinner flask bioreactor45,46 ensure reproducibility, thus streamlining not only the production but also the examination and analysis of neural microcircuits with a variety of cell culture assays. It should be emphasized that although the use of a spinner flask or equivalent bioreactor is optional, a stationary culture of spheres in close contact may result in their fusing together, thus limiting nutrients and oxygen beyond the limit of diffusion in the center of spheres47. Notably, the use of 3D printed mini-bioreactors have been reported as a higher throughput approach compared to large bioreactors46.
Traditional tools to measure synaptic formation include physiological methods such whole-cell patch clamping, MEAs36,48,49, and calcium imaging50. Here, we choose to describe the use of MEAs as they provide a simple and high-throughput examination of electrical activity in the spheres on a brief time scale. However, other techniques may also be utilized.
A limitation of the use of microwell plates in this protocol is the maximum number of spheres (~300) and cells per sphere (~2 x 104) that are permitted in each well. Alternative sphere formation tools may be used for an increased number; however, a higher number of cells will be required at the starting point. The 24-well format was chosen here for its ability to generate larger spheres that can be handled for analysis and visible by eye, while limiting cell death within the center of the spheres. Overall, the addition of this step to the protocol will ensure a robust and scalable production of uniform 3D spheres.
Our protocol also boasts the flexibility of adjusting the number and ratio of each cell type as well as the possibility of introducing other cell types into the 3D spheres in a controlled, defined manner. The use of region-specific neuronal and astrocyte subtypes permits the study of synaptic microcircuits of different regions of the central nervous system. These coculture spheres can also be fused together51 to study cellular migration and long range signaling. The addition of oligodendrocytes, endothelial cells, or microglia could prove these 3D spheres to be an informative brain model with increased complexity, as a promising future direction for this method. Finally, in addition to disease and injury modeling, this system permits the study of therapeutic approaches, such as cellular replacement therapy or drug development, to enhance neuroregeneration.
The authors have nothing to disclose.
We would like to thank Dr. Erik Ullian (UCSF) for intellectual input on the design of these procedures, Dr. Michael Ward (NIH) for technical advice on iNeuron differentiation, and Saba Barlas for preliminary image analysis.
6 well plate | Fisher Scientific | 08-772-1B | |
15 ml conical tubes | Olympus Plastics | 28-101 | |
Accutase | Sigma | A6964-100ML | Detachment solution |
AggreWell plate | Stemcell Technologies | 34850 | |
Anti-Adherence Rinsing Solution | Stemcell Technologies | 7010 | Prevent cell adhesion to microwell plates |
Anti/anti | Thermofisher | 15240062 | |
B27 | Thermofisher | 17504044 | Media Supplement |
BrainPhys neuronal medium | Stemcell Technologies | 5790 | Neurophysiological basal medium alternative |
Circular glass coverslips | Neuvitro | GG-12-oz | |
Cryostor CS10 | Stemcell Technologies | 7930 | Cryopreservation medium with 10% DMSO |
DMEM/F12 | Thermofisher | 10565-042 | With GlutaMAX supplement |
DMH-1 | Stemcell Technologies | 73634 | HAZARD: Toxic if swallowed. Working concentration: 2 uM |
Donkey serum | Lampire Biological Laboratories | 7332100 | Working concentration: 5% in primary blocking buffer, 1% in secondary blocking buffer |
Doxycycline Hydrochloride (Dox) | Sigma | D3072-1ml | HAZARD: Toxic for pregnant women. Working concentration: 2 ug/mL |
Epidermal growth factor (EGF) | Peprotech | AF-100-15 | Working concentration: 10 ng/mL |
Fibroblast growth factor-2 (FGF) | Peprotech | 100-18B | Working concentration: 10 ng/mL |
Fluoromount-G mounting solution | Southern Biotech | 0100-01 | |
Glass slides | Fisherbrand | 22-037-246 | |
Goat serum | Lampire Biological Laboratories | 7332500 | Working concentration: 5% in primary blocking buffer, 1% in secondary blocking buffer |
Hemacytometer or automatic cell counter | Life Technologies | AMQAX1000 | |
Heparin | Sigma | H3149-50KU | Working concentration: 2 mg/mL |
Magnetic plate | DLAB | 8030170200 | |
Matrigel membrane matrix | Corning | 354230 | ECM coating solution. Working concentration: 80 ug/ml. Prepare on ice and ensure that pipettes, tubes, and media are pre-chilled. |
MEA 2100 System | Multichannel Systems | MEA2100 | |
Mounting solution | |||
N2 | Thermofisher | 17502048 | Media Supplement |
OCT | Tissue-Tek | 4583 | Tissue embedding solution for cryosectioning |
Pap Pen (Aqua Hold) | Scientific Device Laboratory | 9804-02 | |
Paraformaldehyde (PFA) | Acros Organics | 169650025 | HAZARD: Toxic if inhaled. Working concentration: 4% in PBS |
Phosphate buffered saline (PBS) | Stemcell Technologies | CA008-300 | |
Poly-l-ornithine (PLO) | Sigma | P3655-100MG | Working concentration: 0.5 mg/mL |
Rectangular glass cover slips | Fisherfinest Premium Superslip | 12-545-88 | |
ReLeSR | Stemcell Technologies | 5872 | Detachment and passaging reagent |
Rho-Kinase Inhibitor Y27632- (Y) | Tocris | 1254 | Working concentration: 10 uM |
SB431542 | Stemcell Technologies | 72234 | Working concentration: 2 uM |
Spinner flasks | Fisher Scientific | 4500-125 | |
Sucrose | Fisher Chemical | S5-3 | Working concentration: 20% or 30% in PBS |
T25 Culture Flask | Olympus Plastics | 25-207 | Vented caps |
T75 Culture Flask | Olympus Plastics | 25-209 | Vented caps |
Terg-A-zyme | Sigma | Z273287-1EA | Detergent. Working concentration: 1% |
TeSR-E8 basal medium | Stemcell Technologies | 5940 | Human pluripotent stem cell (hPSC) medium |
TeSR-E8 supplements | Stemcell Technologies | 5940 | Supplements for human pluripotent stem cell medium |
TritonX-100 | Sigma | X100-500ML | Detergent for cell permeabilization. Working concentration: 0.25% in blocking buffer |
Trypan blue | Invitrogen | T10282 | |
Antibodies | |||
AlexaFluor 488 | Thermofisher | A-11029 | Secondary antibody |
AlexaFluor 594 | Thermofisher | A-11037 | Secondary antibody |
Ezrin | Thermofisher | MA5-13862 | Primary antibody; astrocytes perisynaptic |
GFAP | Chemicon | MAB360 | Primary antibody; astrocytes |
GFP | Aves | GFP-1020 | Primary antibody; astrocytes |
Glt1 | Gift from Dr. Jeffrey Rothstein | n/a | Primary antibody; astrocytes |
Homer | Synaptic Systems | 160 011 | Primary antibody; neurons, post-synaptic |
MAP2 | Synaptic Systems | 188 004 | Primary antibody; neurons |
PSD95 | Abcam | ab2723 | Primary antibody; neurons, post-synaptic |
S100 | Abcam | ab868 | Primary antibody; astrocytes |
Synapsin 1 | Synaptic Systems | 106 103 | Primary antibody; neurons, pre-synaptic |
TuJ1/β3-tubulin (TUBB3) | Covance | MMS-435P | Primary antibody; neurons |