Design, Surface Treatment, Cellular Plating, and Culturing of Modular Neuronal Networks Composed of Functionally Inter-connected Circuits

A SU8-2075 mold on a silicon wafer with a feature thickness of approximately 100 µm was used to shape the PDMS. The pattern was composed of squares of several dimensions, with a side length and distance varying between 200 and 700 µm (Figure 1B). The size of the square was chosen to fit the field of view of a 10X (for islands with a side length <800 µm) and of a 20X objective (for islands with a side length <400 µm). Three parameters, namely cell plating density, distance between circuits, circuits’ size are determinant for obtaining monolayers or clustered circuits as well as to impose and shape their connectivity. However, these parameters are not independent since, given a fixed distance between two circuits, the probability that they spontaneously establish a connection increase with the circuits’ size. In the example discussed in this work, for small neuronal modules of ~300 x 300 µm, connections were established when the distance was not larger than ~300 – 400 µm, while for large neuronal modules of ~700 x 700 µm, connections were established when the distance was smaller than ~700 µm. In general, by increasing the distance between the squares it was possible to change the probability to establish spontaneously generated inter-connections among the modules, passing from highly connected modules to isolated ones.

Neuronal modules of ~600 x 600 µm are shown at 4 days in vitro (DIV; Figure 3A). After few days in vitro neurons are mostly located within the coated areas while it is not possible to observe developed neuronal processes and connections. On the contrary, at 14 DIV (Figure 3B-C) neurons establish connections within and among modules. While larger neuronal circuits of ~600 x 600 µm could self-organize into monolayers (Figure 3A-C), circuits of smaller dimensions (e.g., ~300 x 300 µm in Figure 3D) tended to cluster. Therefore, in order to obtain the best calibration for two-dimensional circuits, cells were plated with different density varying from 0.25 x 10⁶ to 1 x 10⁶ cells/23 mm cover slip attached in a 35 mm Petri dish. For larger circuits of ~700 x 700 µm we found that 0.75 x 10⁶ cells/23 mm cover slip attached in a 35 mm Petri dish is the optimal density which induces the formation of monolayer circuits not preventing their spontaneous inter-connection (Figure 3C). The same result of monolayer circuit was obtained in smaller circuits of ~300 x 300 µm by plating 0.5 x 10⁶ cells/23 mm cover slip attached in a 35 mm Petri dish. In general, when considering cell plating density, it is important to highlight that, as previously discussed⁷, neuronal and glial cells have an innate tendency to cluster, so some degree of clustering is almost unavoidable after some time in culture. However, we observed that increasing the area destined to the supporting network would enhance the probability of circuits’ survival and their mono-layered organization, probably due a larger concentration of nutrients in the extracellular space. In any case, a trial and error approach is required to identify the optimal cell plating density to create mono-layered circuits for calcium imaging. Every 4 days the medium was diluted with fresh growth medium. From 6/7 DIV, after seeing connections between the islands, 10 µl/ml FUDR has been added to prevent glial overgrowth.

Figure 4 shows the dynamic of a modular network recorded using calcium imaging (performed as described in Bonifazi et al. 2013⁸). Calcium-fluorescence images were acquired using a 4X magnification objective, which enabled monitoring the activity of the entire network (Figure 4A). Analysis of the calcium imaging was performed as described in Bonifazi et al. 2013⁸. In Figure 4B raster plot of the calcium events onset displaying spontaneous activity is presented (colors correspond to modules marked in Figure 4A). Calcium imaging was performed at 30 Hz, and image size is 1,000 x 1,000 pixels.

The green and pink modules are highly synchronized as shown in the raster plot (Figure 4B) in agreement with their thick connecting bundles, possibly corresponding to a large number of connections (Figure 4A). The blue and red modules are weakly connected to the green and pink modules and therefore they occasionally synchronized with them (Figure 4B). A video of the recording is available online (movie M1).

The image of a modular network grown on a MEA at 21 DIV is shown in Figure 5A. In order to obtain the best calibration for two-dimensional circuits, we plated cortical cells with different density varying from 250 x 10³ to 500 x 10³ cells/per MEA. 250 x 10³ cells/per MEA (30 mm ring) gave better results in terms of the formation of monolayer circuits, while not preventing their spontaneous inter-connection. Once patterned neuronal networks have been achieved on MEAs substrate (Figure 5A), electrophysiological activity has been monitored using a commercial system used for acquiring and recording electrophysiological signals. Spontaneous activity of cultures is detected using the Precise Timing Spike Detection (PTSD) algorithm²².

Figure 5B shows a raster plot corresponding to 5 min of spontaneous activity (only active electrodes are visible), colored coded according to the cluster number. By looking at these plots, it is possible to make qualitative assessments of the activity of single clusters and of the whole network at the same time. In particular, it is possible to observe a strong synchronization among the activities recorded within each cluster. A video of a representative recording from patterned cultures over MEA is available online (movie M2).

Figure 1. Silicon wafer. (A) The silicon wafer (~4 inch diameter) used to mold the PDMS stencils. Different SU8 structures represent different modular networks. (B) Representative feature design implemented on a silicon wafer for modular networks construction. Each green spot define an SU8 pillar structure and therefore the area for a single module. The designs within the dashed black rectangle correspond to three different stencils which can be used on standard square culture coverslips of 23 mm side length. In each stencil, an asterisk marks a macro-region for the supporting network. Depending on the distance between the spots, the finite size circuit or modules might be able to establish spontaneous neuronal connections among them. The design of the features was chosen to create different module size and inter-module distance to achieve isolated or interconnected neuronal circuits which could fit the field of view in different objective magnifications (4X, 10X and 20X; the field of view for each magnification is represented by the red squares). The features out of the dashed black rectangle are designed to fit the simultaneous use of MEA and calcium imaging. Please click here to view a larger version of this figure.

Figure 2. PDMS structures deposition on a MEA. (A) PDMS mold with a ring shape (marked with the red arrow) mounted on a MEA. The mold is used to coat the area destined to the supporting neuronal network and located at the periphery of the culturing area (this is limited by the ring mounted on the MEA and marked by the green arrow). (B) Alignment of a PDMS stencil on the MEA using a micro manipulator. (C) PDMS stencil deposited on the on MEA after alignment. (D) Image of a PDMS stencil on MEA using a 10x magnification (inter-electrode distance is 500 µm). It is possible to note the holes on the stencils in correspondence of the electrodes. The adhesive layer favoring cellular adhesion will be deposited only on the opened areas. Please click here to view a larger version of this figure.

Figure 3. Modular neuronal networks at different DIV. Bright field image of cortical modules of ~600 x 600µm (A) after 4 DIV and (B) after 14 DIV. (C) Note the spontaneous interconnections among the circuits after 14 DIV. Cells were plated at the optimal cell density for large circuits to organize into monolayers, i.e., 750 x 10³ cells/23 mm cover slip attached in a 35 mm Petri dish. (D) Using PDMS features of half size and same cell concentration, neuronal circuits organized into clustered structures. Please click here to view a larger version of this figure.

Figure 4. Calcium imaging of modular networks. (A) Cortical cells (18 DIV) loaded with the calcium indicator OGB. Different colours mark different modules. The field of view is 2 x 2 mm. (B) Raster plot of spontaneous activity. The different colors correspond to the cells within different modules as shown in panel A. Please click here to view a larger version of this figure.

Figure 5. MEA recording of modular networks. (A) Modular network (cortical neurons, 21 DIV) composed of 3 distinct circuits grown on a 4Q MEA. Neuronal circuits are distant ~600 µm interconnected to each other. (B) Raster plot showing 5 min of spontaneous electrophysiological activity recording. C1, C2 and C3 correspond, respectively, to the upper left, the lower right and the lower left cluster’s activities, highlighted in different colors. Please click here to view a larger version of this figure.