このコンテンツを視聴するには、JoVE 購読が必要です。 サインイン又は無料トライアルを申し込む。

概要

Here, we present an easy-to-use co-culture assay to analyze glioblastoma (GBM) migration on patterned neurons. We developed a macro in FiJi software for easy quantification of GBM cell migration on neurons, and observed that neurons modify GBM cell invasive capacity.

Abstract

Glioblastomas (GBMs), grade IV malignant gliomas, are one of the deadliest types of human cancer because of their aggressive characteristics. Despite significant advances in the genetics of these tumors, how GBM cells invade the healthy brain parenchyma is not well understood. Notably, it has been shown that GBM cells invade the peritumoral space via different routes; the main interest of this paper is the route along white matter tracts (WMTs). The interactions of tumor cells with the peritumoral nervous cell components are not well characterized. Herein, a method has been described that evaluates the impact of neurons on GBM cell invasion. This paper presents an advanced co-culture in vitro assay that mimics WMT invasion by analyzing the migration of GBM stem-like cells on neurons. The behavior of GBM cells in the presence of neurons is monitored by using an automated tracking procedure with open-source and free-access software. This method is useful for many applications, in particular, for functional and mechanistic studies as well as for analyzing the effects of pharmacological agents that can block GBM cell migration on neurons.

Introduction

Primary malignant gliomas, including GBMs, are devastating tumors, with a medium survival rate of 12 to 15 months reported for GBM patients. Current therapy relies on large tumor mass resection and chemotherapy coupled with radiotherapy, which only extends the survival rate by few months. Therapeutic failures are intimately related to poor drug delivery across the blood-brain barrier (BBB) and to invasive growth in perivascular spaces, meninges, and along WMTs¹. Perivascular invasion, also called vascular co-option, is a well-studied process, and the molecular mechanisms are beginning to be elucidated; however, the process of GBM cell invasion along WMTs is not well understood. Tumor cells migrate into the healthy brain along Scherer's secondary structures². Indeed, almost one century ago, Hans-Joachim Scherer described the invasive routes of GBM, which are now referred to as perineuronal satellitosis, perivascular satellitosis, subpial spread, and invasion along the WMT (Figure 1A).

Some chemokines and their receptors, such as stromal cell-derived factor-1α (SDF1α) and C-X-C motif chemokine receptor 4 (CXCR4), but not vascular endothelial growth factor (VEGF), seem to be implicated in WMT invasion³. More recently, a transcellular NOTCH1-SOX2 axis has been shown to be an important pathway in WMT invasion of GBM cells⁴. The authors described how GBM stem-like cells invade the brain parenchyma on partially unmyelinated neurons, suggesting the destruction of myelin sheaths by GBM cells. A milestone was reached in 2019 when three articles wereconsecutively published in Nature journal, underlining the role of electrical activity in glioma development^5,6. Seminal work by Monje and collaborators shed light on the central role of electric activity in the secretion of neuroligin-3, which promotes glioma development.

Winkler and collaborators described connections between GBM cells (microtubes) being crucial in invasive steps, and lately, interactions between GBM cells and neurons via newly described neuroglioma synapses. Those structures favor glutamatergic stimulation of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors located at the GBM cell membrane, which promotes tumor development and invasion. Tumor cell invasion is a central process in the dissemination of metastases or distant secondary foci, as observed in GBM patients. Several factors have been identified to be important in GBM invasion such as thrombospondin-1, a transforming growth factor beta (TGFβ-regulated matricellular protein, or the chemokine receptor CXCR3)^7,8.

Here, a simplified biomimetic model has been described for studying GBM invasion, in which neurons are patterned on tracks of laminin, and GBM cells are seeded onto it, as single-cells or as spheroids (Figure 1B). The two experimental settings are aimed at recapitulating invasion on neurons, which is observed in GBM^9,10,11. Such models have been developed in the past as aligned nanofiber biomaterials (core-shell electrospinning) that allow studying cell migration by modulating mechanical or chemical properties¹². The co-culture model described in this article allows a better understanding of how GBM cells escape on neurons by defining new molecular pathways involved in this process.

Protocol

Informed written consent was obtained from all patients (from the Haukeland Hospital, Bergen, Norway, according to local ethics committee regulations). This protocol follows the guidelines of Bordeaux University human and animal research ethics committees. Pregnant rats were housed and treated in the animal facility of Bordeaux University. Euthanasia of an E18-timed pregnant rat was performed by using CO₂. All animal procedures have been done according to the institutional guidelines and approved by the local ethics committee. All commercial products are referenced in the Table of Materials.

1. Preparation of the patterned slides

1. Substrate preparation for micropatterning
   1. Treat 18 mm circular glass coverslips by air/plasma activation for 5 min. Place the coverslips in a closed chamber with 100 µL of (3-aminopropyl) triethoxysilane in a desiccator for 1 h.
   2. Incubate with 100 mg/mL of poly (ethylene glycol)-succinimidyl valerate (molecular weight 5,000 (Peg-SVA)) in 10 mM carbonate buffer, pH > 8, for 1 h. Rinse extensively with ultrapure water, and dry under a chemical hood.
      ​NOTE: At this stage, the sample can be stored at 4 °C in the dark for further use.
   3. Add the photoinitiator, 4-benzoylbenzyl-trimethylammonium chloride (PLPP), at 14.7 mg/mL in phosphate-buffered saline (PBS).
      NOTE: A concentrated form of PLPP, a PLPP gel, can also be used. It results in a shorter ultraviolet (UV) illumination time required to degrade the PEG brush (100 mJ/mm²).
2. Photoinitiator gel deposition
   1. Prepare a mixture of 3 µL of PLPP gel and 50 µL of absolute ethanol to deposit in the center of the slide. Place the sample under a chemical hood until complete evaporation of the absolute ethanol.
      NOTE: At this stage, the sample can be stored at 4 °C in the dark for further use.
3. Glass slide micropatterning
   1. Mount the coverslip in a Ludin chamber, and place it on the motorized stage of a microscope equipped with an auto-focus system.
   2. Load images corresponding to the envisioned micropatterns into the software. Apply these parameters: replication 4 x 4 times, spacing of 200 µm, UV dose of 1,000 mJ/mm². After the automatic UV-illumination sequence, rinse the PLPP away with multiple PBS washes.
      NOTE: If PLPP gel was used, remove it by extensive washes with deionized water, dry in a stream of N₂, and store at 4 °C.
   3. Incubate with laminin (50 µg/mL in PBS) for 30 min. Wash extensively with PBS.
      ​NOTE: A fluorescent solution of purified green fluorescent protein (GFP, 10 µg/mL in PBS) can be mixed with laminin to visualize the micropatterns by fluorescence microscopy.

2. Preparation of neurons and GBM cells for co-culture

1. Culture of embryonic rat hippocampal neurons
   1. Dissect the hippocampus of embryonic (E18) rats, and transfer the tissue into a Hank's balanced salt solution (HBSS)/1 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES)/penicillin-streptomycin solution in a 15 mL tube. Remove excess solution without drying the hippocampus.
   2. Add 5 mL of trypsin-ethylenediamine tetraacetic acid (EDTA) supplemented with penicillin (10,000 units/mL)/streptomycin (10,000 µg/mL) and 1 mM HEPES, and incubate for 15 min at 37 °C. Wash 2x with the HBSS/HEPES/penicillin-streptomycin solution, and let the tissue remain in this solution for 2-3 min.
   3. Dissociate the tissue using two flame-polished Pasteur pipettes, by pipetting up and down 10x with each tissue, taking care to minimize foaming. Count the cells, and evaluate the viability of the cell suspension. Plate the neurons on micropatterned coverslips as indicated below.
      NOTE: Cell viability rate is 85-90% after extraction.
2. Cell culture for neurons on micropatterned coverslips
   1. Rehydrate micropatterned glass slides with PBS, and incubate complete neuronal cell culture medium.
   2. Seed the hippocampal neurons obtained from E18 Sprague-Dawley rats directly over the micropatterned glass coverslip at a density of 50,000 cells per cm² in neurobasal medium (NBM) enriched with 3% horse serum. Place the micropatterned neurons in the incubator (37 °C, 5% CO₂) for 48 h.
      NOTE: After ~6 h, primary hippocampal neurons can be seen adhering to the laminin micropatterns.
3. Co-culture of human GBM stem-like cells on neurons
   ​NOTE: For this study, membrane GFP-positive and nuclear-tomato patient-derived GBM cells were grown according to previous published protocols¹⁰.
   1. As the spheroid-shaped cells grow in suspension, centrifuge the suspension for 5 min at 200 × g. Wash the spheroids with 5 mL of PBS, and incubate the cells with 0.5 mL of the cell dissociation reagent (see the Table of Materials) for 5 min at 37 °C.
   2. Add 4.5 mL of complete NBM (complemented with B27 supplement, heparin, fibroblast growth factor 2, penicillin, and streptomycin, as described previously⁸), and count the cells using an automatic counting technique.
   3. Seed 1,000 GBM cells over the micropatterned neuronal culture in NBM enriched with 3% horse serum. Incubate the plate at 37 °C, 5% CO₂, and 95% humidity.

3. Live cell imaging

1. Immediately after GBM cell seeding, place the sample on the stage of an inverted microscope equipped with a thermostat chamber. Perform live-cell imaging on the microscope equipped with a motorized stage for recording multiple positions by using a multidimensional acquisitions toolbox in the software. Acquire brightfield and epifluorescence GFP/Tomato images every 5 minutes over 12 h with a 20x objective in a temperature (37 °C) and gas-controlled (5% CO₂) environment.

4. Image analysis

NOTE: Using Fiji, two-dimensional (2D) image stack were semi-automatically preprocessed or processed by using a homemade and user-friendly tool (available at this address: https://github.com/Guyon-J/Coculture_Gliomas-Neurons/blob/main/README.md), which is written in IJ1 macro language (Figure 2A). The automated workflow and procedures are summarized in Figure 2B.

1. Neuronal network analysis (Figure 2Bi)
   1. Select one image of the stack. Right-click on the Network tool to open the corresponding Options dialog box and adjust the settings (e.g., Threshold = Triangle, Li, Huang…, Gaussian Blur, and Median filters = 1, 2, 3…) to produce a precise segmentation of images. Then, click on OK.
   2. Left-click on the Network tool to automatically activate the following procedure.
      1. Duplicate the selected image, and split it into three color channels (Red-Grey-Green).
      2. Select the grey channel (brightfield), and perform contrast stretch enhancement (CSE) to enhance the separation between different areas. Use the Sobel edge detector (SED) to perform the 2D signal processing convolution operation already grouped under the Find Edge command.
      3. For double-filtering (F), apply a Gaussian blur and median filter to reduce noise and smooth the object signal. Convert to Mask (CM) by executing adapted threshold algorithms to obtain a binary picture (BIN-grey) with black pixels (cell area) and white pixels (background). Skeletonize (Sk) the cell area into a simple network (NET), and filter particles (EP) in a NET image by removing small, non-networked particles.
      4. For red and green channels (nucleus and membrane), perform double-filtering, convert to Mask using the adapted thresholding method, and allow BIN-green to determine cell morphology with the Analyze Particles command.
      5. Merge all channels using their region of interest (ROI) with the OR (combine) operator, and readjust their initial color into a simple RGB image.
2. Single-cell motility analysis (Figure 2Bii)
   1. Right-click on the Single Cell Tracking tool to open the corresponding Options dialog box, and adjust the settings (e.g., Trail type = Nucleus or Membrane, Threshold = Triangle, Li, Huang…, Z projection = Max intensity, Sum Slices…, Gaussian Blur and Median filters = 1, 2, 3…) to produce a precise segmentation of images. Then, click on OK.
   2. Left-click on the Single Cell Tracking tool to automatically activate the following procedure.
      1. Remove the grey channel. Apply a Z projection on the stack, which will generate an image corresponding to an image stack according to the time (T-stack). Double-filter and convert to Mask the trails left by cells. Remove small particles to the BIN-red/green image.
      2. By using ROI, select each contour of the cell trace, and check the box Skip edge detection in the Options dialog box to skip this preprocessing step for subsequent steps.
      3. Isolate the red channel (Trail type = Nucleus) on the original stack. Select one ROI and remove the outer area. Double-filter all images and convert to Mask (BIN-red). Determine the centroid X/Y position of each binarized nucleus.
   3. Using an already published macro for spreadsheet software¹³, calculate the mean square displacement, directionality ratio, and average speed for this cell.
3. Multiple-cell tracking analysis (Figure 2Biii)
   1. Right-click on the Tracking tool to open the corresponding Options dialog box and adjust the settings (e.g., Threshold = Triangle, Li, Huang…, Gaussian Blur and Median filters = 1, 2, 3…) to produce a precise segmentation of images. Then, click on OK.
   2. Left-click on the Tracking tool to automatically activate the following procedure.
      1. Remove the grey channel.
      2. Split the red and green channels, double-filter, and convert to Mask.
      3. Merge the channels using Image Calculator… command with the AND operator, leaving only the nucleus signal found in the membranes.
         NOTE: Several plugins in Fiji can be used to determine the X/Y position of several cells in this binary preprocessed image at the same time (see the Table of Materials).
   3. Using a previously described macro¹³, calculate the trajectory plot, mean square displacement, directionality ratio, and average speed for these cells.
4. Spheroid migration on the neural mat (Figure 2Biv)
   1. Right-click on the Migration tool to open the corresponding Options dialog box and adjust the settings (e.g., Threshold = Triangle, Li, Huang…, Gaussian Blur and Median filters = 1, 2, 3…) to produce a precise segmentation of images. Then, click on OK.
   2. Left-click on the Migration tool to automatically activate the following procedure.
      1. Remove the red channel.
      2. Split the green and grey channels.
      3. For the grey channel, draw manually the contour of the neuronal mat, and measure its area.
      4. For the green channel, double-filter the stack and convert to Mask. Remove the area outside the pattern (BIN), and determine the binarized cell area for each image.
         ​NOTE: Parameters described above are calibrated by changing their values by left-clicking on the icon of interest. This processing can be done manually. However, for a large number of images (approximately a hundred per acquisitions), channels (generally 3 channels), and processing steps, an automated or semi-automated tool would be preferable.

Representative Results

Patterned neurons co-cultured with fluorescent GBM cells were prepared as described in the protocol section, and tracking experiments were performed. GBM cells quickly modified their shape while migrating on the neurons (Figure 1B: panel 6 and Video 1). Cells migrated along the neuronal extensions, in a random motion (Video 1). Fluorescent GBM cells and non-fluorescent neurons can be easily distinguished, and this allowed the tracking of cell movements by using the Fiji macro, as described in the protocol section (Figure 2). Fiji is an open license software that facilitates image processing and analysis. Manual procedures that are relatively time-consuming for the analysis of numerous images can be automatized in a macro. Images are imported by drag-and-drop procedure, processed, and quantified, generating data which are available using an ROI manager.

When deposited in the Fiji.app | macros | toolsets folder, the Gliomas-Neurons tool was available in the More Tools menu and displayed several icons (Figure 2A). A workflow of the image processing, obtained by clicking on the icons, is illustrated in Figure 2B (i-iv). Cell shape can be analyzed on the neural network (Figure 2B, i). Several parameters from cell tracking can be obtained for one (Figure 2B, ii) or several cells (Figure 2B,iii) by using two distinct image processes. The speed of recovery by spheroid GBM cells on the neurons can be also analyzed (Figure 2B, iv). Cells seeded onto neurons displayed an elongated shape with multiple protrusions following neuron tracts (Figure 3A, i), but had a round shape when cultured directly on laminin (Figure 3A, ii). Cells cultured on neurons efficiently modified their shape, although they were not elongated when cultured on laminin (Figure 3A, iii-v). Thin protrusions, sometimes linking two cells, were seen in cells co-cultured with neurons at later stages (Video 1).

The migratory capacity of GBM cells seeded on neurons was compared to cells directly seeded on laminin. Cells seeded on neurons had greater migratory capacities than on laminin alone (Figure 3B, i,ii). Random movement of P3 cells was detected in both conditions, with greater distance for P3 cells on the neurons, as shown in the trajectory plot (Figure 3B, i,ii). Cell motility was quantitatively estimated by the mean square displacement (MSD), and its log representation was fitted with a linear function¹⁴ (Figure 3B, iii). Directionality and average speed were also calculated for both conditions (Figure 3B, iv,v). Cell migration of P3 spheroids was also followed by detecting the fluorescent area over time on the neurons and compared with migration on laminin alone (Figure 3C, i,ii and Video 2). Half of the pattern with neurons was covered with GBM cells after 500 min in a linear profile; however, spheroids did not adhere to the laminin pattern (Figure 3D, iii and Video 2).

Figure 1: Experimental setup of glioblastoma cells migrating on patterned neurons. (A) Representation of glioblastoma cells invading the contralateral hemisphere through the corpus callosum. (B) Experimental setup. In step 1, the plate surface is coated with an antifouling PEG layer. The photoinitiator is added in step 2, covering the entire coating. In step 3, a UV widefield image is projected through the objective of the microscope, which locally activates the photoinitiator molecules. The activated photoinitiator locally cleaves PEG molecules and allows the subsequent adsorption of laminin. In step 5, neurons are seeded and adhere on the laminin arrays. P3 (GFP/Tomato_nuclear) glioblastoma cells are then deposited on the neuronal pattern, and images are acquired (step 6). Abbreviations: PEG = polyethylene glycol; UV = ultraviolet; GFP = green fluorescent protein. Please click here to view a larger version of this figure.

Figure 2: Fiji tool presentation and analysis workflow. (A) The tool called Gliomas-Neurons is available in the More Tools menu when the macro is added to Fiji.app folder | macros | toolsets (left panel). It is composed of several action tools described below (right panel). (B) i. Network tool: image processing, which is used to draw the neural network and glioma cells in a simplified representation. ii. Single-cell tracking tool: image processing, which is used to draw and select the cell movement area for analyzing the displacement of single cells. iii. Tracking tool: image preprocessing steps for the use of pre-installed Fiji tracking plugins. iv. Relative migration tool: image processing, which is used to determine the relative cell migration on a manually selected pattern. Some parameters can be calibrated with a left click on the tool button. Abbreviations: CSE = Contrast Stretch Enhancement; SED = Sobel Edge Detector; F = double Filtering; CM = Convert to Mask; Sk = Skeletonize; EP = Elimination of particles; OR (combine) = Union operator on selected ROIs; AND = Conjunction operator on selected ROIs. Please click here to view a larger version of this figure.

Figure 3: Comparison of P3 cells or spheroids on patterned neurons vs. laminin coating. (A) Shape descriptors of dissociated P3 GFP/Tomato_nuclear cells on neurons or on laminin. Example of processed networks of P3 cells on i. patterned neurons or on ii. laminin coating. Scale bar = 50 µm. iii. Average cell area on patterned neurons or on laminin. iv. Formfactor is the ratio of the circumference to the area normalized to a circle, providing parameters on cell elongation and cell branching. v. Aspect ratio is the ratio of the major axis to the minor axis of the cell. In iii, iv, and v, 15 cells were analyzed per field; 4 independent patterns; data are represented as mean ± S.E.M. (B) Tracking analysis of dissociated P3 cells. One representative chart plot of cells on (i) patterned neurons and (ii) on laminin coating. iii. MSD of P3 cells migrating on neurons or on laminin coating. X/Y values are in logarithmic scales. iv. Directionality ratio. v. Average cell speed migration. In iii, iv, and v, 15 cells were analyzed per field; 4 independent patterns; data are represented as mean ± S.E.M. (C) Migration analysis of P3 GFP spheroids. Representative images at different time points of P3 spheroids on (i) patterned neurons or on (ii) laminin coating. Scale bar = 100 µm. iii. Spheroid migration represented by the pattern confluency; 4 independent patterns; data are represented as mean ± S.E.M. Abbreviations: GFP = green fluorescent protein; S.E.M. = standard error of the mean; MSD = Mean Square Displacement. Please click here to view a larger version of this figure.

Video 1: P3 cells on neurons or laminin recorded over 8 h (imaged every 5 min). The video shows P3 single-cell migration on neurons (left) and on laminin coating (right). Cells expressed green fluorescent protein (GFP, green color) and nuclear Tomato (red color). Bar = 50 µm. Please click here to download this video.

Video 2: P3 spheroids on neurons or laminin recorded over 8 h (imaged every 5 min). The video shows P3 spheroid migration on neurons (left) and on laminin coating (right). Cells expressed green fluorescent protein (GFP, green color). Bar = 100 µm. Please click here to download this video.

Discussion

Glioblastomas extensively invade the parenchyma by using different modes: co-option of surrounding blood vessels, interstitial invasion, or invasion on WMTs¹⁸. This latter mode is not well characterized in the literature because of the difficulty in finding suitable in vitro or in vivo models related to WMT invasion. Here, a simplified model has been proposed in which cultured rodent neurons were patterned on laminin-coated surfaces, and fluorescent GBM stem-like cells were seeded on top of the neurons. A grid-shape pattern was used in this study to improve the analysis of tumor cell attachment, invasion, and proliferation. GBM cells migrated more efficiently on top of the neurons than directly on the matrix, which was laminin in these experiments. The cell shape changed throughout the recording process, and the cell surface area increased at the same time. GBM stem-like cells are likely to be attracted by neuron tracts via the activation of specific signaling pathways (i.e., NOTCH2/SOX2)⁴ or secreted factors and signals from the neurons themselves. This system is well-suited to analyze the molecular exchanges between GBM cells and neurons, which may include metabolites, neurotransmitters, or cytokines.

Recently, Venkatesh and collaborators described the formation of a neuroglioma synapse in which glutamate activates its receptor, AMPAR⁶. Are similar processes involved during WMT invasion of GBM cells? This can be investigated with this experimental system using pharmacological inhibition or genetic approaches.

The following critical points should be noted. First, during the PEGylation steps, the substrate should not be allowed to dry out to avoid affecting the integrity of the anti-adhesive coating. Of note, it is possible to extensively wash the PEG-SVA solution with ultrapure water and to dry it out under a stream of nitrogen before storing it at 4 °C in the dark. Second, the macro developed for analyzing GBM cell migration on neurons is in an open-access mode and is compatible with FiJi software. Although this macro and its updates are available on GitHub, it requires an appropriate calibration for detecting cells. Hence, it may be useful to check the samples manually as a quality control while starting the analysis. The flexibility of the system used here allows different shapes of the pattern, with parallel lines separating the neurons to different extents. With this approach, the shape of several cerebral structures can be mimicked, as observed in the corpus callosum-the largest white matter structure in human brain-where WMT invasion is mainly observed. Alternatively, the same UV-light projection apparatus has been shown to structure UV-sensitive non-adhesive hydrogels in a z-controlled manner¹⁵, allowing the standardization of spheroid formation.

In this context, 3D neurospheres could be generated to test GBM invasion. This technique can be also applied for patterning other cerebral cells, such as brain endothelial cells, to reproduce vessel-like shape or mimic microglial or other immune cells. Thus, synergistic or inhibitory effects of cerebral cells can be observed when co-cultured with GBM cells. One limitation of this study is the use of embryonic rat neurons in co-culture with human GBM cells, which may not mimic true physiological conditions. One way to overcome this drawback would be to use human induced pluripotent stem cell-derived neurons to avoid species cross-reactivity¹⁶. However, GBM cells quickly adhere and efficiently migrate on rat neurons, as shown in these experiments. It has also been demonstrated than rat cerebral cells (Schwann cells) could be efficiently co-cultured with human neurons¹⁷. Other methods include the use of 3D nanofibers, which offer a good model to study glioma cell migration¹², but limit cell-cell contact as nanofibers are considered non-living structures.

Furthermore, 2D cultures are reductionistic, simplify the observation of cellular processes, and may limit their validity for the in vivo context¹⁰. Thus, 3D co-cultures of GBM cells and neurons are better representatives of the in vivo situation. Complex brain organoids, such as mini-brains¹⁸, have been used in confrontation-culture invasion assays¹⁹. The main advantage of the strategy described herein is the reproducibility of the co-culture approach, i.e., the primary neurons are geometrically constrained on size-controlled micropatterns, and the interaction with the injected GBM cells cannot occur elsewhere. Furthermore, the spatial organization of neurons can be tuned because of the versatility of the UV-projection system, allowing for further optimization. Ultimately, the development and validation of such biomimetic approaches could also help in reducing the number of animal models used in biomedical research.

Materials

(3-aminopropyl) triethoxysilane	Sigma	440140-100ML	The amino group is useful for the bioconjugation of mPEG-SVA
96-well round-bottom plate	Sarstedt	2582624	Used to prepare spheroids
Accutase	Gibco	A11105-01	Stored at -20 °C (long-term) or 4 °C (short-term), sphere dissociation enzyme
B27	Gibco	12587	Stored at -20 °C, defrost before use
Basic Fibroblast Growth Factor	Peprotech	100-18B	Stored at -20 °C, defrost before use
Countess Cell Counting ChamberSlides	Invitrogen	C10283	Used to cell counting
Coverslips	Marienfeld	111580	Cell culture substrate
Dessicator cartridges	Sigma	Z363456-6EA	Used to reduce mosture during (3-aminopropyl) triethoxysilane treatment
DPBS 10x	Pan Biotech	P04-53-500	Stored at 4 °C
Fiji software, MTrack2 macro	ImageJ		Used to analyze pictures
Flask 75 cm²	Falcon	10497302
HBSS	Sigma	H8264-500ML
Heparin sodium	Sigma	H3149-100KU	Stored at 4 °C
Laminin		114956-81-9	Promotes neuronal adhesion
Leonardo software			loading of envisioned micropatterns
MetaMorph Software	Molecular Devices LLC	NA	Microscopy automation software
Methylcellulose	Sigma	M0512	Diluted in NBM for a 2% final concentration
Neurobasal medium	Gibco	21103-049	Stored at 4 °C
Nikon TiE (S Fluor, 20x/0.75 NA)			inverted microscope equipped with a motorized stage
Penicillin – Streptomycin	Gibco	15140-122	Stored at 4 °C
PLPP	Alveole	PLPPclassic_1ml	Photoinitiator used to degrade the PEG brush
Poly(ethylene glycol)-Succinimidyl Valerate (mPEG-SVA)	Laysan Bio	VA-PEG-VA-5000-5g	Used as an anti-fouling coating
PRIMO	Alveole	PRIMO1	Digital micromirror device (DMD)-based UV projection apparatus
Trypan blue 0.4%	ThermoFisher	T10282	Used for cell counting
Trypsin-EDTA	Sigma	T4049-100ML	Used to detach adherent cells