Recent works uncover the neuronal impact on high-grade pediatric glioma (pHGG) cells and their reciprocal interactions. The present work shows the development of an in vitro model co-culturing pHGG cells and glutamatergic neurons and recorded their electrophysiological interactions to mimic those interactivities.
Pediatric high-grade gliomas (pHGG) represent childhood and adolescent brain cancers that carry a rapid dismal prognosis. Since there is a need to overcome the resistance to current treatments and find a new way of cure, modeling the disease as close as possible in an in vitro setting to test new drugs and therapeutic procedures is highly demanding. Studying their fundamental pathobiological processes, including glutamatergic neuron hyperexcitability, will be a real advance in understanding interactions between the environmental brain and pHGG cells. Therefore, to recreate neurons/pHGG cell interactions, this work shows the development of a functional in vitro model co-culturing human-induced Pluripotent Stem (hiPS)-derived cortical glutamatergic neurons pHGG cells into compartmentalized microfluidic devices and a process to record their electrophysiological modifications. The first step was to differentiate and characterize human glutamatergic neurons. Secondly, the cells were cultured in microfluidic devices with pHGG derived cell lines. Brain microenvironment and neuronal activity were then included in this model to analyze the electrical impact of pHGG cells on these micro-environmental neurons. Electrophysiological recordings are coupled using multielectrode arrays (MEA) to these microfluidic devices to mimic physiological conditions and to record the electrical activity of the entire neural network. A significant increase in neuron excitability was underlined in the presence of tumor cells.
Pediatric high-grade gliomas (pHGG) exhibit an extended genotypic and phenotypic diversity depending on patient age, tumor anatomical location and extension, and molecular drivers1. They are aggressive brain tumors that are poorly controlled with the currently available treatment options and are the leading cause of death related to brain cancers in children and adolescents2. So, more than 80% of patients are relapsing within 2 years after their diagnosis, and their median survival is 9-15 months, depending on brain locations and driver mutations. The absence of curative treatment is the primary urge for laboratory research and highlights the immediate need for new innovative therapeutic approaches. For this purpose, patient-derived cell lines (PDCL) were developed with the hope of providing the pHGG diversity3 in two-dimensional (2D) lines and/or three-dimensional (3D) neurospheres. Nevertheless, those patient-derived in vitro cell cultures do not mimic all brain variable situations. These models do not consider the macroscopic and microscopic neuro-anatomical environments typically described in pHGG.
Usually, pHGG in younger children is mainly developing in pontine and thalamic regions, whereas adolescent and young adult's HGG concentrate in the cortical areas, especially in frontotemporal lobes1. These location-specificities across pediatric ages seem to involve different environments leading to gliomagenesis and an intricating network between tumor cells and specific neuronal activity4,5,6. Although mechanisms are still not identified, pHGG mainly develops from neural precursor cells along the differentiation trajectory of astroglial and oligodendroglial lineages. While the role of these glial lineages has been for long restricted to simple structural support for neurons, it is now clearly established that they integrate entirely into neural circuits and exhibit complex bi-directional glial-neuronal interactions able to reorganize structural regions of the brain and remodel neuronal circuitry4,7,8. Moreover, increasing shreds of evidence indicate that the central nervous system (CNS) plays a critical role in brain cancer initiation and progression. Recent works focused on neuronal activity, which seems to drive growth and mitosis of glial malignancies through secreted growth factors and direct electrochemical synaptic communications6,9. Reciprocally, high-grade glioma cells seem to influence neuronal function with an increasing glutamatergic neuronal activity and modulate the operation of the circuits into which they are structurally and electrically integrated9. So, studies using patient-derived models and novel neuroscience tools controlling neuron action demonstrated a circuit-specific effect of neuronal activity on glioma location, growth, and progression. Most of these neuronal projections involved in gliomas are glutamatergic and communicate through glutamate secretions. Specific glutamatergic biomarkers such as mGluR2 or vGlut1/2 are commonly described6.
Interestingly, despite their molecular heterogeneity, pediatric and adult high-grade gliomas show a typical proliferative response to glutamatergic neuronal activity and other secreted factors such as neuroligin-3 or BDNF (brain-derived neurotrophic factor)4,6,10,11,12,13. In cortical regions, pediatric and adult HGGs can induce neuronal hyperexcitability through an increased glutamate secretion and inhibit GABA interneurons leading to gliomas associated with epileptic network activity14,15. On top of that, neural circuits can be remodeled by gliomas pushing specific neurological tasks, for instance, language, and can requisition additional organized neuronal activity9.
Based on this rationale, advancing the understanding of bidirectional communications between glioma cells and neurons must be fully elucidated and integrated with the early stages of in vitro pHGG approaches. Such innovative modeling is crucial in understanding and measuring the neuronal electrical activity impact during drug testing and anticipating pHGG response into brain circuitry. Recent developments in neuroscience tools, such as microfluidic devices and pHGG research works, are the bed to develop new modeling approaches and be able now to integrate brain microenvironment in in vitro pHGG models3,16,17,18,19. Coupled with electrophysiological recordings using multielectrode arrays (MEA), microfluidic devices20,21,22 offer the possibility to mimic physiological conditions while recording the electrical activity of the entire neural network and extract network connectivity parameters under several conditions. This device23,24 allows first the precise deposition of cells in a chamber directly on MEA. This technology enables the control of cell seeding density and homogeneity on MEA and the fine control of media exchange, which is a critical step for human neural progenitor differentiation directly into devices. Moreover, the present deposition chamber can be seeded with multiple cells at different time points.
So, this study aimed to develop a functional in vitro model co-culturing human Pluripotent Stem (hiPS)–derived cortical glutamatergic neurons and pHGG-derived cells into microfluidic devices and recording their electrical activity to evaluate electrical interactions between both cell populations. First, hiPS-derived cortical glutamatergic neurons were obtained and characterized in microfluidic devices at different stages of culture [day 4 (D4), as hiPS cells, and day 21 (D21) and day 23 (D23), as glutamatergic matured neurons]. For the second step of co-culture, two pHGG models were used: commercialized pediatric UW479 line and pHGG cells initiated from a patient tumor (BT35)3, bearing an H3.3 K27M driver mutation. Finally, we performed electrophysiological recordings of glutamatergic cells at D21 before pHGG cell seeding and D23 after 48 h of co-culture into the same microfluidic device. The interactions between glutamatergic neurons and pHGG cells were characterized by a significant increase in the recorded electrophysiological activity.
For this protocol, the accreditation number related to the use of human materials is DC-2020-4203.
1. Microfluidic device fabrication, preparation and treatment
2. Cell preparation and seeding in the microfluidic device
3. Co-culture protocol
4. Electrophysiological recording
5. Electrophysiological data processing
Before studying electrical interactions between glutamatergic neurons and glioma cells, hiPS-derived cortical glutamatergic neurons were characterized to validate the feasibility of culturing them in microfluidic devices (Figure 1A). Their characterization was assessed using Nestin, Sox2, mGlurR2 (metabotropic Glutamate Receptors 2), and vGLUT1 immunostaining, represented in Figure 1A(2-7). As Nestin is an intermediate filament protein required for survival, renewal and mitogen-stimulated proliferation of neural progenitor cells, it is thus expressed in undifferentiated CNS cells during development. Sox2 is a member of the SRY-like HMG-box gene (SOX) transcription factor family involved in the regulation of embryonic development, and it is predominantly expressed in immature and undifferentiated cells of the neural epithelium in CNS. mGluR2 is widely distributed throughout the brain, with high expression in the cortex and hippocampus. Usually, it defines the presynaptic cells with their neuronal excitability and synaptic transmission (e.g., the glutamatergic neurons). vGLUT1 is also used as a glutamatergic biomarker. Therefore, Figure 1 focuses on glutamatergic neurons with Nestin and Sox2 expressions at D4 and D21 to validate their progressive differentiation into the culture conditions (Figure 1A(2-4). Nestin positive cells' percentage decreased from 5.4 ± 0.4% at D4 to 0.69 ± 0.3% at D21 (p < 0.01, Figure 1A5). Similarly, confirming the differentiated state of glutamatergic neurons, the percentage of Sox2 positive cells decreased from 12.2 ± 2.1% at D4 to 5.5 ± 1.7% at D21 (p < 0.05, see Figure 1A6). The expression of mGluR2 was detected in 45.4 ± 4.7% of glutamatergic cells at D21 (data not shown), and a larger vGLUT1 immunostaining was observed during glutamatergic differentiation at D21. These results demonstrated that the proportion of neural progenitors remained low and that most cells were well differentiated after a 21-day culture in the microfluidic devices. In Figure 1B, a general protocol of the multistep co-cultures and electrophysiological recording describes the UW479 or BT35 cell seeding in the microfluidic devices at D21. Those cell lines (Figure 1C) were already described and are comparable to the previous observations made by us3,19. In the end, two electrophysiological recordings were obtained: one at D21 before co-culturing and the other at D23 after 48 h of pHGG cell line seeding.
For understanding and following the mobility and viability of BT35, UW479, and glutamatergic neurons during co-cultures, microscopic assessments were regularly performed during the electrophysiological recording period from D21 to D23 and are presented in Figure 2. The microscopic images revealed a progressive extension and particular distribution of glutamatergic neurons across microfluidic devices (Figure 2A). The glutamatergic cells form some aggregates progressively in a similar way in co-culture and when cultured alone until D23 (control experiment). In Figure 2B,C, when adding BT35 and UW479, it was observed that the floating cells present immediately after glioma cells' seeding at D21 progressively disappeared until D23 and became adherent pHGG cells into the device. This was particularly notable for the UW479 line.
Moreover, there was no increase in floating cells after all electrophysiological recordings at D23 (data not shown). Nevertheless, the percentages of viable cells in Figure 2D were estimated at 83.02% and 85.16% for BT35 and UW479, respectively. Both line viabilities were comparable and underlined the fact that patient-derived cell lines can be used appropriately. The technical preparation of microfluidic devices coupled to MEA and cell manipulations does not alter their adhesion capacities. It does not seem to induce cell death or modify their microscopic aspects. BT35 and UW479 cells seem to map the exact locations of glutamatergic neurons in the device, migrating closely to the excitatory neurons.
Figure 3 describes as an example UW479/glutamatergic neuron co-cultures with their spike detection along recording time in Figure 3A and its raster plot at D23 in Figure 3B, showing increased electrical activity when adding pHGG cells. Figure 3C presents the temporal array-wide firing rate in BT35/glutamatergic neuron co-cultures. The differences between the control experiment (culture of glutamatergic neurons) and co-cultures of glutamatergic neurons with pHGG cells are significant when recording D23 electrical activity, as shown in Figure 3D, which indicates an impact of pHGG on neuron excitability.
Components | Seeding medium | D4 medium | D7 medium | D11 and onward medium |
DMEM/F-12 Medium | 0.5x | 0.25x | 0.125x | Ø |
Neurobasal Medium | 0.5x | 0.25x | 0.125x | Ø |
BrainPhys Medium | Ø | 0.5x | 0.75x | 1x |
SM1 Supplement | 1x | 1x | 1x | 1x |
N2 Supplement-A | 1x | 1x | 1x | 1x |
Ala-Gln (GlutaMax) | 0.5 mM | 0.5 mM | 0.5 mM | 0.5 mM |
BDNF | 10 ng/mL | 10 ng/mL | 10 ng/mL | 10 ng/mL |
GDNF | 10 ng/mL | 10 ng/mL | 10 ng/mL | 10 ng/mL |
TGF-β1 | 1 ng/mL | 1 ng/mL | 1 ng/mL | 1 ng/mL |
Geltrex | 30 µg/mL | Ø | Ø | Ø |
Seeding Supplement | 1x | Ø | Ø | Ø |
Day 4 Supplement | Ø | 1x | Ø | Ø |
Table 1: Culture media composition for hiPS-derived glutamatergic neurons.
Antibodies | Stock concentrations | Working dilutions |
Nestin | 0.5 mg/mL | 1:500 |
(1 µg/mL) | ||
Sox2 | 1 mg/mL | 1:500 |
(2 µg/mL) | ||
mGluR2 | 0.2 mg/mL | 1:50 |
(4 µg/mL) | ||
vGlut1 | 0.25 mg/mL | 1:50 |
(5 µg/mL) | ||
Donkey anti-rabbit IgG H&L (AF 647) | 2 mg/mL | 1:1000 to 1:2000 |
(1 to 2 µg/mL) | ||
Donkey anti-mouse IgG H&L (AF 555) | 2 mg/mL | 1:1000 |
(2 µg/mL) |
Table 2: List of immunofluorescent antibodies.
Figure 1: Cell characterization and co-culture protocol of glutamatergic neurons and high-grade pediatric glioma (pHGG) lines. (A) Prerequisite microscopic and immunofluorescent characterization of glutamatergic neurons. (1) Microscopic pictures at day 21 (D21) and day 23 (D23) of glutamatergic neurons. (2) Nestin labeling in green and DAPI in blue at D4 (upper row) and D21 (bottom row). (3) Sox2 labeling in green and DAPI in blue at D4 (upper row) and D21 (bottom row). (4) mGluR2 labeling in green and DAPI in blue at D21. (5 and 6) Statistical comparisons between Nestin and Sox2 labeling at D4 and D21 of culture, demonstrating cell differentiation. (7) vGlut1 staining in green and DAPI in orange at D21. (B) General protocol for electrophysiological recording to study interactions when co-culturing glutamatergic neurons and glioma cells on devices. (C) Prerequisite microscopic aspects of BT35 and UW479 cell lines. Please click here to view a larger version of this figure.
Figure 2: Microscopic images of glutamatergic and glioma cells cultured in microfluidic devices coupled to multielectrode arrays (MEA). Glutamatergic neurons are differentiated and cultured alone in the microfluidic device until day 21 (D21). (A) For this experimental control, only the medium was added to glutamatergic cells at D21, while either BT35 (B) or UW479 (C) cell lines were added in their respective conditions. Images at D22 and D23 were performed before the electrophysiological records for each condition. All the images were obtained using a classical phase-contrast microscope. (D) Tumor cell viability for BT35 and UW479 at D23 was estimated using an automated count with image analysis software. Please click here to view a larger version of this figure.
Figure 3: Electrophysiological recording. (A) Spike detection along recording time in UW479/glutamatergic neuron co-cultures. (B) An example of the computed raster plot at D23 represents the events detected for each active electrode as a function of time in UW479/glutamatergic neuron co-cultures. (C) Temporal array-wide firing rate (or instantaneous firing-rate) of all the active electrodes recording electrical activity in BT35/glutamatergic neuron co-cultures, allowing to monitor synchronicity of the neural network activity. (D) Bar charts of mean firing rate in the control condition (e.g., the culture of glutamatergic neurons) and when adding tumoral UW479 cell line (in green) and BT35 cell line (in red). Data are shown as mean ± SEM. (* p < 0.05). Please click here to view a larger version of this figure.
Supplemental Figure S1: 3D (3-dimensional) representation of SU-8 molds. (A) Creation and picturing of SU-8 molds with AutoCAD software. (B) Picture of the SU-molds with the two layers of photoresist structures and four holes of 3 mm width. Please click here to download this File.
This work describes an accurate functional in vitro model to evaluate the interaction between human hiPS-derived cortical glutamatergic neurons and brain tumoral cells in microfluidic devices. One of the crucial steps in the present protocol was the hiPS differentiation in glutamatergic neurons, which was confirmed by the decrease of Nestin and Sox2 immunofluorescent staining and simultaneous appearance of mGluR2 and vGLUT1 staining. Nevertheless, few neural progenitors remained as only half of the glutamatergic cells expressed mGluR2, which means a heterogeneous neuron cell population. Altogether, these results emphasize that particular care must be devoted to glutamatergic cell growth in such devices. Furthermore, the next step studying the electrical behavior of neurons in the presence of pHGG cells was relatively labor-intensive to set up the data processing to optimize the spike detection and interpretation. Nevertheless, as expected and described in other recently published works4,5,6,9,10, the presence of pHGGs and glutamatergic neurons enhances electrical activity confirming the excitatory feature of those neurons.
The major limitation of this modeling might be the heterogeneous distribution of neurons on the MEA itself that can impact electrophysiological recordings. It would require the maintenance of high-density culture in the microfluidic device. For seeding a secondary cell type onto the device, it is critical to maintain a rapid proliferation and migration initially for 48 h and obtain as for neurons a homogeneous distribution in the 48 h technique for recording. Another limitation is differentiating the different types of cell populations in such devices visually, but new approaches using fluorescent nanoparticles might help following both cell types26. Finally, the performance of this microfluidic approach was only done in two types of phGG lines and should be extended to more patient-derived cell lines bearing different molecular drivers. Complementary assessments might be done to understand this excitatory effect when co-culturing those cells. Nevertheless, it is feasible with pHGG cells bearing an H3.3 K27M driver mutation, such as in the BT35 line3.
The overall method developed in this work is one of the new approaches to explore the electrical impact. It has shown the capacity of neural network analysis to transduce the interaction between hiPS-derived glutamatergic neurons and pHGG tumoral cell lines in microfluidic culture conditions. This method is helpful for many applications, particularly for functional and mechanistic studies and for analyzing the effects of pharmacological agents that can block pHGG cell migration and interaction with neurons. It emphasizes the use of microfluidic devices, but specifically when experiments might record the electrophysiological activity and can be added to an MEA.
The authors have nothing to disclose.
This work was supported by grants from Satt Conectus program, Fondation de l'Université de Strasbourg, «J'ai demandé la lune», «Une roulade pour Charline», «LifePink», «Franck, Rayon de Soleil» and «Semeurs d'Etoile» associations. We thank the children and families affected by HGGs for their contributions to this research and their support.
256MEA100/30iR-ITO-w/o | MCS | 256MEA100/30iR-ITO-w/o | |
40 µm probe for Scepter counter | Dutscher | 53750 | |
60 µm probe for Scepter counter | Dutscher | 51999 | |
Accutase | Sigma | A6964 | |
Ala -Gln (GlutaMAX) | Sigma | G8541 | |
Axel Observer 7 Microscope | Zeiss | 431007-9904-000 | |
Cell culture flask with cap with filter membrane 70 mL Falcon® | Dutscher | 353109 | |
Class II Biological Safety Cabinet | Thermo Scientific | HERASafe type KS12 | |
Colibri 7 LED | Zeiss | 4230529710-000 | |
Cortical Glutamatergic Neurons
|
BrainXell | BX-0300 | |
DMEM/F-12 (1:1) GlutaMAX | Gibco | 31331-028 | |
DMEM/F12 Medium | Sigma | D8437 | |
DPBS 1X | Dutscher | L0615-500 | |
EasYFlaskTM cell culture flasks 75cm3 | Nunc | 156499 | |
Foetal Bovine Serum (FBS) | Dutscher | 500105 | |
GDNF | Peprotech | 450-10 | |
Geltrex | Life Technologies | A1413201 | |
Human BDNF | Peprotech | 450-02 | |
Incubator | Memmert | IC0150med | |
MCS InterFace Boarder | MCS | 181205-MEA2100-11240 | |
MEA2100 | MCS | 181205-MEA2100-11240 | |
Micropipette P10 | Sartorius | LH-729020 | |
Micropipette P100 | Sartorius | LH-729050 | |
Micropipette P1000 | Sartorius | LH-729070 | |
Micropipette P200 | Sartorius | LH-729060 | |
Microtube Eppendorf 1,5 ml Safe-Lock | Dutscher | 33290 | |
MultiChannel Experimenter | MCS | – | |
N2 Supplement-A | StemCell | 7152 | |
Neurobasal Medium | Life Technologies | 21103049 | |
Neurocult SM1 neuronal supplement | StemCell | 5711 | |
Non filter tip 0.1 – 10 µl ClearLine® sterile in removable-lid rack | Dutscher | 030570ACL | |
Non filter tip 1 – 200 µl ClearLine® sterile in removable-lid rack | Dutscher | 032260CL | |
Non filter tip 50 – 1250 µl ClearLine® sterile in removable-lid rack | Dutscher | 134760CL | |
Non-essential amino acids (NEAA) without L-glutamine | Dutscher | X0557-100 | |
Pipeteur Pipet-Aid XP Gravity | Drummond | 4000202/4038202 | |
Pipette for cell culture 10 mL Falcon® | Dutscher | 357551 | |
Pipette for cell culture 5 mL Falcon® | Dutscher | 357543 | |
Plaque chauffante (CultureTemp) | Belart | 370151000 | |
Poly-D-Lysine | Sigma | P6407 | |
Primovert microscope | Zeiss | 415510-1100-000 | |
Scepter (Handheld Automated Cell Counter) | Millipore | PHCC00000 | |
TGF-β1 | Peprotech | 100-21C | |
Tube with conical bottom 15 mL (bulk) Falcon® | Dutscher | 352096 | |
Tube with conical bottom 50 mL (bulk) Falcon® | Dutscher | 352070 |