Engineered Biological Neural Networks on High Density CMOS Microelectrode Arrays

Abstract

In bottom-up neuroscience, questions on neural information processing are addressed by engineering small but reproducible biological neural networks of defined network topology in vitro. The network topology can be controlled by culturing neurons within polydimethylsiloxane (PDMS) microstructures that are combined with microelectrode arrays (MEAs) for electric access to the network. However, currently used glass MEAs are limited to 256 electrodes and pose a limitation to the spatial resolution as well as the design of more complex microstructures. The use of high density complementary metal-oxide-semiconductor (CMOS) MEAs greatly increases the spatial resolution, enabling sub-cellular readout and stimulation of neurons in defined neural networks. Unfortunately, the non-planar surface of CMOS MEAs complicates the attachment of PDMS microstructures. To overcome the problem of axons escaping the microstructures through the ridges of the CMOS MEA, we stamp-transferred a thin film of hexane-diluted PDMS onto the array such that the PDMS filled the ridges at the contact surface of the microstructures without clogging the axon guidance channels. This method resulted in 23 % of structurally fully connected but sealed networks on the CMOS MEA of which about 45 % showed spiking activity in all channels. Moreover, we provide an impedance-based method to visualize the exact location of the microstructures on the MEA and show that our method can confine axonal growth within the PDMS microstructures. Finally, the high spatial resolution of the CMOS MEA enabled us to show that action potentials follow the unidirectional topology of our circular multi-node microstructure.

Keywords: CMOS; PDMS microstructures; bottom-up neuroscience; engineered networks; in vitro; microelectrode arrays.

Figure 1

Comparison of recording electrode densities on 60-channel glass MEAs and CMOS MEAs. (A) Commonly used 60-channel glass MEAs allow for simultaneous recording from an engineered biological neural network at 4 electrodes (blue dots) located at the microstructure channels. (B) The same microstructure on a CMOS MEA is covered with more than 700 electrodes, allowing for both axonal and somatic signal acquisition. (C) Electron micrograph of the chip surface. While the glass MEAs are flat, the surface of the CMOS MEA is non-planar due to recessed electrodes and trenches between columns of electrodes. Such a surface topology impairs the adhesion of the PDMS microstructure to the surface compromising its cell and axon guidance characteristics.

Figure 2

Transferring PDMS microstructures to a CMOS MEA using an intermediate gluing step with uncured PDMS. (A) Uncured PDMS is diluted with hexane in a 1:9 volume ratio and spin coated on a wafer at 2000 rpm for 60 s. (B) The PDMS microstructure is placed on the spin coated PDMS thin film without releasing it and lifted after contact was established, leaving a thin layer of uncured PDMS on the bottom side of the PDMS microstructure. (C) The PDMS microstructure with the intermediate layer of uncured PDMS is placed on the CMOS MEA and subsequently moved to an oven at 80 °C for 2 h to allow the intermediate PDMS layer to fully cure and create a tight seal between the surface of the MEA and the PDMS microstructure.

Figure 3

Detection of the PDMS microstructure on the CMOS MEA. (A) Applying a sinusoidal signal on the reference electrode with an on-chip signal generator yields an undamped signal on exposed electrodes and a highly attenuated signal on covered electrodes. (B) Repeating the procedure on 25 routing patches across the MEA reveals the location and openings of the PDMS microstructure.

Figure 4

Neuron viability and growth assessment (top row) using live-dead staining (green: alive, red: dead) and corresponding PDMS microstructure brightfield images after mask removal (bottom row). (A) The adhesion of the microstructure is low when heptane is used instead of hexane as a dilutant, since the same spincoating parameters resulted in a thinner PDMS film on the wafer. Consequently, less PDMS is transferred by the microstructure onto the CMOS array resulting in incomplete sealing of the trenches on the chip surface. Therefore, axons can escape the PDMS microstructure more easily and instead follow the topology of the MEA surface rather than the channels of the microstructure. After removing the microstructure from the MEA, a phase contrast image of the microstructure reveals that the MEA surface left almost no imprint on the PDMS microstructure, confirming that the intermediate PDMS layer did not fill the trenches of the MEA surface. (B) Using a spincoated 1:9 PDMS:Hexane film yields the best microstructure adhesion and axonal guidance. After removing the microstructure, the imprint of the surface of the CMOS MEA is visible. While some clogging is observable in the thin channels, the large channels interconnecting the nodes remain open. (C) When pressing the PDMS microstructure onto the CMOS MEA or using less diluted PDMS solutions, clogging is observed in all channels and only little axonal outgrowth is possible. The phase contrast image of the microstructure confirms the clogging, as the MEA surface imprint is also visible in the channel regions.

Figure 5

Electrical recordings of cortical neurons growing within a PDMS microstructure at DIV 14. (A) Fluorescently stained neurons imaged at DIV 31 reveal high viability, axonal growth within the microstructure channels and a full intra-node connectivity. (B) Recording of a single electrode shows a high signal-to-noise ratio. The 5σ threshold is indicated as well as spikes that were detected using the spike detection method described in the methods section. (C) Spike average and frequency map over 60 s on all routed electrodes within the network with color coded spiking frequency. It is evident that the spiking frequency is highest in the channels of the microstructure. The arrow indicates the electrode location at which the signal in (B) stems from. (D) A temporal analysis of spontaneous activity within a 10 ms time window indicates clockwise signal propagation.