Long-Term, High-Spatiotemporal Resolution Recording From Cultured Organotypic Slices with High-Density Microelectrode Arrays

Abstract

A novel system to cultivate and record brain slices directly on high-density microelectrode arrays (HD-MEA) was developed. This system allows to continuously record electrical activity of selected individual neurons at high spatial resolution, while monitoring neuronal network activity at the same time. For the first time, properties of single neurons and the corresponding neuronal network in an organotypic hippocampal slice culture were studied over four consecutive weeks at daily intervals.

Introduction

In neuroscience, one of the major challenges includes to record the activity of multiple single neurons at high spatiotemporal resolution for extended time periods. Common approaches to record neuronal activity include: (1) intracellular recordings with patch clamp, (2) extracellular recordings with single electrodes or tetrodes, (3) extracellular recordings with micro- or multi-electrode arrays (MEAs), and (4) calcium imaging. The patch clamp method can detect sub-threshold signals at the single-cell level; however, it is difficult to record from multiple neurons at the same time. Extracellular single-electrode and tetrode recordings allow for access to multiple neurons at the same time; however, due to the usually large distance between the electrodes, it is difficult to record single-neuron activity within a network. Commercial MEAs with low spatial resolution can be used to simultaneously record from multiple neurons within a network, but because of the large electrode pitch, identification of single-neuron activities is difficult if not impossible. Calcium imaging is commonly used to record the activity of multiple single neurons within a network at high spatial resolution; calcium imaging, however, has a relatively poor temporal resolution that makes it challenging to discern individual action potentials.

Here, we report on a method that enables long-term cultivation of organotypic slices directly on a HD-MEA and the recording of neuronal network and single-neuron activities from the same culture at multiple time points. The HD-MEA provides high spatiotemporal resolution recordings from over 11,000 electrodes in an area of 2x2 mm². The high electrode density of the array allows to record cellular to sub-cellular electrical activity. The relatively large number of electrodes and the large electrode area of the array allow for selecting electrodes according to the physiological nature of the brain slice under investigation. In the current project, we investigated how the electrical activity of single neurons and the activity of local networks develop within organotypic slices under in vitro conditions.

Organotypic brain slice cultures are well-established in vitro models in neuroscience research. The two most common culturing methods are the “Stoppini interface method” [1] and the “Gähwiler roller tube method” [2]. Normally, preparations according to these two methods can be only used once for single-time-point recordings. Our approach, based on the Gähwiler method, allows for repeated recordings from the same neurons for a long time period.

Methods

CMOS-based HD-MEA

The HD-MEA was fabricated by using complementary-metal-oxide-semiconductor (CMOS) technology [3]. The array area is 2.0 × 1.75 mm², contains 11,011 planar platinum electrodes, with a center-to-center pitch of 17.8μm and features an electrode density of 3,150 electrodes / mm² [3]. There are 126 read-out or stimulation channels that can be simultaneously connected to a subset of electrodes. Depending on the goals of the experiment, different electrode configurations can be applied. The readout channels provide a programmable gain between 0 to 80dB, and signals are sampled at 20 kHz [3].

The HD-MEA surface was passivated with SiO₂ and Si₃N₄ layers, and the chip was bonded to a custom printed circuit board (PCB). A plastic ring of 8 mm height and 20 mm diameter was glued on the PCB around the array with epoxy (Epo-Tek 302-3 M) to contain the culture medium. The array area was pre-treated with oxygen plasma, then covered with a PDMS (Polydimethylsiloxane) stamp, and the device was heated in an oven to 100°C for 15 minutes. After the PDMS stamp was well attached on the array surface, epoxy was filled in the space between the array and plastic ring, to protect the bondwires. A layer of platinum black was deposited on the electrode surface by using a solution containing hexachloroplatinic acid and lead acetate (pH = 1, adjusted with hydrochloric acid). A current of 180 μA was simultaneously applied to all electrodes (current density of 0.5 nA/μm²). The platinum black layer is needed to reduce the electrode impedance and to improve the signal-to-noise characteristics of the recorded signals.

Organotypic culture

Hippocampal tissue of newborn C57Bl/6J mice (age between postnatal day 5 - 7) was dissected and sliced (300 μm slice thickness) in oxygenized ice-cold dissection solution (Hank’s balanced salt solution, 45% D-Glucose, kynurenic acid, penicillin-streptomycin). The HD-MEA surface was coated with polyethyleneimine (PEI). The hippocampus slices were placed directly on top of the HD-MEA and attached to the array surface by using a mixture of chicken plasma and thrombin from bovine plasma (Fig1. A). Slices were then cultured in custom culturing chambers that allow for sufficient nutrient supply and air exchange (Fig1. B). The HD-MEA with the slices on top were continuously rotated (1.5 min per rotation cycle) and alternately immersed in culturing solution and exposed to air [2]. Physiological culturing conditions were provided by an incubator with controlled temperature (36 °C), humidity (90%), and CO₂ level (5%). Culture medium containing basal eagle medium without L-glutamine, Hanks’ balanced salt solution, inactivated horse Serum, 45% D-Glucose, GlutaMAX, penicillin-streptomycin, and B27 Supplement [2] has been changed at the third day in vitro (DIV3), and once per week during the subsequent cultivation period.

Figure 1. Organotypic slice culturing device.

(A) Hippocampus slice placed on top of the HD-MEA. (B) The rotation system used to maintain nutrition and gas exchange during slice culturing.

Electrophysiology recordings

To record extracellular spontaneous neuronal activities from the slice culture, the HD-MEA PCB was connected to a custom-designed chip support board, which provides interfacing circuitry to communicate with the HD-MEA chip for receiving data and sending configurations. Data is streamed to a Xilinx Virtex II pro FPGA board (Digilent Inc., Pullman, USA) [3.4], which provided an Ethernet interface to a desktop PC. Custom made software (“MEAbench” [5]) was running on the PC to visualize experiments and perform recordings.

During a recording session, the cultivation chamber including an HD-MEA chip with a slice culture on top, was removed from the cultivation rack and plugged into the recording device located inside of the incubator (temperature 36 °C, humidity 65%, and CO₂ 5%). Each recording session lasted between 30 minutes to 3 hours, depending on the experimental design. The slice location on the HD-MEA was determined by taking a photo of the slice culture during slice plating by means of an eyepiece camera inserted in a stereomicroscope with 5x magnification. On the day before the recording sessions, the slice culture was checked for activity by recording from a series of randomized electrode configurations (126 channels were randomly connected to electrodes for each configuration, and each electrode was only recorded during one configuration). To record from all 11,011 electrodes, 97 configurations were generated, and each configuration was recorded for 60 seconds. On the first day of recording, a series of configurations with high electrode density (3,150 electrodes / mm²) were generated in the array areas that were covered with the slice and showed neuronal activity. Each high-density configuration block (17 x 6 electrodes) was designed to have electrodes overlapping with adjacent configuration blocks (Fig 2.).

Figure 2. A series of high-density configuration blocks was used to record spontaneous activity of the organotypic slice culture on the HD-MEA.

Results

Organotypic hippocampal slice activities seemed to be stable during the cultivation period over 30 days. Neural network activity and single neuron activity from the same slice culture were compared using data recorded every 24 hours from DIV 4 to 30.

Network activity

The network activity of the slice culture showed that the sub-network architectures were preserved over extended time periods (Fig. 3). The spiking activities on each electrode were analyzed with respect to spike amplitudes (μV) and firing frequencies (Hz). The averaged spike amplitudes on each electrode were plotted, and by putting the activity patterns of the different recording configurations together, a slice network activity map on the HD-MEA was generated (Fig.3).

Figure 3. Electrical activity maps of a hippocampal slice culture at 4 different time points (6 -34 days in vitro, DIV).

Single-neuron activity

The spike footprints of single neurons changed and slightly moved in their location over time (Fig.4). Spike-triggered averages of extracellular signals (footprints) of single neurons were analyzed from recordings at different days (Fig.4). Single-neuron activities were extracted by applying principal-component-analysis (PCA)-based spike sorting on spike waveforms recorded from multiple channels [6].

Figure 4.

A. The electrical “footprint” of one single neuron within the indicated area is shown. B. The electrical footprint of this neuron was traced over days and is shown at 3 different time points.

Conclusion

The presented method, based on the Gähwiler roller tube method, allows for cultivating and recording from hippocampal slices that have been placed directly on top of the HD-MEAs. The same slice culture can be used for experiments at different time points over many weeks. During the cultivation period, a good interface between the slice culture and the electrodes was formed, which yielded good signal-to-noise characteristics of the recorded neuronal signals. The neuronal network structure was preserved, as visualized by the electrical activity map constructed from the multi-neuron spike activities. High-spatial resolution single neuron activity and single-neuron electrical footprints were extracted from the same data set. The single-neuron electrical footprints were traced at different days during the cultivation period and showed that the neuronal activities were generally stable but exhibited minute spatial movements. The method presented here provides a tool to observe single neuron activity and neural network activity at the same time. It entails minimal user-interference with the slice cultivation process, and, therefore, the same slice culture can be continuously observed over a long time period.