*2.3. 3D Cell Printing and Post-Processing of Printed Samples*

The bioink was loaded in a reservoir consisting of a micro-tube coil of known internal volume, as schematized in Figure 1A. The reservoir was placed in a poly(methyl methacrylate) cylindrical tank (3 cm radius, 15 cm height) covered with a thermal isolating tape (Armaflex L414, Armacell, Munster, DE) and filled with ice in order to prevent the gelation of Matrigel in the reservoir. The total dimension of the system, shown in Figure S1, was rationalized in order to limit the encumbrance of the extruder while ensuring the maintenance of a temperature around 0 ◦C for the total duration of the printing step (~1 h). The reservoir was connected with the internal needle of a coaxial wet-spinning extruder, while the outer needle was fed with a calcium chloride solution (225 mM CaCl2 in HBS). Two independent microfluidic pumps (Cetoni, Korbussen, DE) controlled the flow of the bioink (5 μL/min) and the crosslinking solution (3.5 μL/min) through the coaxial extruder. The extruder and the ice-bath tank were mounted on a three-axis motorized system (PI-miCos, Eschbach, DE) with a computer-numerical-control (CNC) interface (Twintec, Auburn, WA, USA). Printing instructions were expressed in g-code language, and printing codes were generated using a custom MATLAB algorithm. The geometry described in these codes consisted of two alternating perpendicular layers of microfibers with theoretical diameter of 100 μm, separated by gaps of 200 μm, a layer thickness of 100 μm, deposited at a speed of 240 mm/min, forming a squared fiber mesh of 5 mm × 5 mm × 200 μm. Typically, printing time for generated codes was around 40–50 s, and each sample was constituted of 3 to 5 μL of solution depending on the desired dimension of the construct. Printed samples were collected, washed with sterile saline solution and placed in cell culture incubator for 10 min to trigger the gelation process of Matrigel. Samples were then transferred in 12-well cell culture plates, soaked with cell culture media and incubated for additional 2 h to terminate Matrigel gelation. Afterwards, samples were exposed to alginate-lyase enzyme (Sigma Aldrich, 0.2 μg/mL in cell culture media) overnight, washed with fresh media and maintained in culture for characterization and maturation.

#### *2.4. RNA Analysis*

Total RNA was extracted with the Quick RNA MiniPrep (Zymo Research, Freiburg, DE) and retrotranscribed with iScript Reverse Transcription Supermix for RT-qPCR (Bio-Rad, Hercules, CA, USA). Targets were analyzed by PCR with the enzyme MyTaq DNA Polymerase (Bioline, Boston, MA, USA). Thirty cycles of amplification were used for *PAX6* and *GAPDH*, while *FOXG1*, *TBR1*, *TBR2* and *GFAP* were amplified for 34 reaction cycles. The internal control used was the housekeeping gene *GAPDH*. Primer sequences are reported in Table S1.

#### *2.5. Live*/*Dead Cell Analysis*

Cell viability was assessed with the LIVE/DEAD Viability/Cytotoxicity Kit (Thermo Fisher Scientific), which uses green-fluorescent calcein AM and red-fluorescent ethidium homodimer-1 to identify live and dead cells, respectively, according to the manufacturer's instructions. Briefly, bioprinted constructs were incubated with calcein-AM and ethidium homodimer-1 at 37 ◦C for 30 min, followed by a PBS wash, and a further media change before acquisition. Image acquisition was performed with a custom fluorescent integrated system (Crisel Instruments, Rome, IT) based on an IX73 Olympus inverted microscope equipped with the x-light spinning disk module (Crestoptics, Rome, IT) for confocal acquisition, Lumencor Spectra X LED illumination and a CoolSNAP MYO CCD camera (Photometrics, Tucson, AZ, USA). The widefield images were acquired using Metamorph software version 7.10.2 (Molecular Device, San Jose, CA, USA) with 10×, 20× and 40× air objectives. The construct was sectioned in z with a step size of 5 μm to obtain at least five optical planes per construct to capture the whole structure. For the live/dead cells quantification, the entire image stacks were analyzed in 3D and cells were counted using Spots in Imaris 8.1.2 (Bitplane, Belfast, UK); nine fields were analyzed for each time point. Percentage of viability is reported as the mean value ± standard deviation of the mean, from three independent bioprinted constructs at DPP1 and DPP7 and one bioprinted construct at DPP50.

#### *2.6. Immunostaining*

Cells were fixed in 4% paraformaldehyde for 15 min at room temperature and washed twice with PBS. Fixed cells were then permeabilized with PBS containing 0.2% Triton X-100 for 10 min at room temperature and incubated overnight with primary antibodies at 4 ◦C. The primary antibodies used were anti-PAX6 (1:50, sc-81649, Santa Cruz Biotechnology, Dallas, TX, USA), anti-NCAD (1:100, ab18203, Abcam, Cambridge, UK), anti-TUJ1 (1:1000, T2200, Sigma-Aldrich), anti-MAP2 (1:2000, ab5392, Abcam), anti-NeuN (1:50, MAB377, Merck Millipore), anti-GFAP (1:500, MAB360, Merck Millipore) and anti-TBR1 (1:150, 20932-1-AP, Proteintech, Rosemont, IL, USA). The secondary antibodies used were goat anti-mouse Alexa Fluor 488 (1:250, Immunological Sciences, Rome, IT), goat anti-chicken Alexa Fluor 488 (1:500, Thermo Fisher Scientific), goat anti-rabbit Alexa Fluor 488 (1:250, Immunological Sciences), goat anti-mouse Alexa Fluor 594 (1:500, Thermo Fisher Scientific), goat anti-rabbit Alexa Fluor 594 (1:500, Thermo Fisher Scientific) and goat anti-rabbit Alexa Fluor 647 (1:500, Thermo Fisher Scientific). DAPI (Sigma-Aldrich) was used to label nuclei.

#### *2.7. Microscopy Imaging*

Confocal images of panels 1D, 2D and 2D' were acquired at the Olympus iX83 FluoView1200 laser scanning confocal microscope using an air 10× NA0.4 or a silicon oil 30× NA1.05 objective (Olympus, Shinjuku, JP) and 405, 473, 559 and 635 nm lasers. Filter setting for DAPI, Alexa Fluor 488, Alexa Fluor 594 and Alexa Fluor 647 were used when needed. Each stack consisted of individual images with a z-step of 0.5, 5 and 1 μm respectively. Stack images of 1024 × 1024 pixels were stitched together in a mosaic view with the Multi Area Viewer tool of Fluoview 4.2 image software (Olympus). The 3D rendering shown was performed using the Imaris image analysis software v.8.1.2 (Bitplane). Widefield images of panels 2C were acquired with the same system described above with a 20× air objective.

## *2.8. Patch Clamp Recordings*

Whole-cell patch-clamp recordings were used for the functional characterization of 3D bioprinted constructs. Cells in the 3D bioprinted structures were visualized with a BX51WI microscope (OLYMPUS), in a recording chamber continuously perfused with an external solution containing 140 mM NaCl, 2.8 mM KCl, 2 mM CaCl2, 2 mM MgCl2, 10 mM HEPES and 10 mM D-glucose (pH 7.3 with NaOH; 290 mOsm) at room temperature. Borosilicate pipettes were filled with a solution containing 140 mM K-gluconate, 5 mM BAPTA, 2 mM MgCl2, 10 mM HEPES, 2 mM Mg-ATP and 0.3 mM Na-GTP (pH 7.3 with KOH; 280 mOsm). Voltage- and current-clamp recordings were performed using Axon DigiData 1550 (MOLECULAR DEVICES). Signals were filtered at 10 KHz, digitized (25 kHz) and collected using Clampex 10 (MOLECULAR DEVICES). Whole-cell capacitance (Cm), cell membrane resistance (Rm) and resting membrane potential (RMP) were measured on-line by Clampex. Cells were

clamped at –70 and 0 mV to measure spontaneous activity. An on-line P4 leak subtraction protocol was used for all recordings of voltage-activated currents. Voltage steps (50 ms duration) from −80 to +40 mV (10 mV increment; holding potential −70 mV) were applied to study voltage-activated sodium currents. Voltage-activated potassium currents were evoked by voltage steps (50 ms duration) from −80 to +40 mV (10 mV increment; holding potential −70 mV). Firing properties were investigated in current-clamp mode, injecting current pulses (1 s duration) of increasing amplitude (from 10 to 80 pA; 10 pA increment), after imposing a membrane potential of −70 mV to each cell (injection of −79 ± 20 pA). Data were analyzed off-line with Clampfit 10 and Origin 7 software.

#### *2.9. Calcium Imaging Recordings*

Fluorescence images were acquired at room temperature using a customized digital imaging microscope. Between 5–8 field of views (FOVs) per bioprinted construct were recorded in each experiment session. Excitation of calcium dye was achieved using a 1-nm-bandwidth monocromator (Cairn Optoscan, Faversham, UK) equipped with a 150 W xenon lamp. Fluorescence was visualized using the upright microscope Olympus BX51WI equipped with a 40× water immersion objective and a CoolSnap Myo camera. Image acquisition and processing were obtained using MetaFluor software (Molecular Devices). Changes in the intracellular Ca2<sup>+</sup> level were monitored using the high-affinity Ca2<sup>+</sup>-sensitive indicator Fluo4-AM (Invitrogen). 3D bioprinted constructs were loaded by incubating for 30 min at 37 ◦C in external solution containing the following: 140 mM NaCl, 2.8 mM KCl, 2 mM CaCl2, 2 mM MgCl2, 10 mM HEPES, 10 mM D-glucose (pH 7.3 with NaOH; 290 mOsm) plus 5 μM Fluo4-AM. A custom-made MATLAB guided user interface (GUI) was used to perform calcium imaging data analysis. Fluorescence data collected as a series of images were converted to three-dimensional MATLAB files and the neurons were manually selected from the time-averaged fluorescence recording before running the trace extraction and analysis. Tens of neurons were identified for each field scanned, depending on the confluence and seeding density of the cultures. The calcium traces were acquired with a sampling frequency of 2 Hz and signals were normalized as a function of ΔF/F0 = (F − F0)/F0, where F is the current fluorescence intensity at any time point and F0 is the basal fluorescence intensity. Single calcium events were detected on the basis of a previously published method [20]. Threshold for peak amplitude, initially set to 3% of the baseline value, was manually adjustable to improve the detection depending on the recording conditions (signal-to-noise ratio, acquisition rate, etc.). Results were visualized, and eventual false or missing detections were manually corrected. Raster plots were created to visualize asynchronous (appearing as sparse vertical lines) and synchronous (appearing as a series of vertically aligned lines) results and the linear dependence between each pair of neurons was calculated by means of Pearson correlation coefficient from binary traces. Neuron firing rate, amplitude of the events and synchrony of the network (evaluated as the relative number of simultaneous events) were exported in Microsoft Excel to perform statistics. Statistical analysis was performed in GraphPad Prism 7 or OriginPro 6.0.

#### **3. Results**
