Skip to main content

A bioartificial brain with an artificial body: training a cultured neural tissue to support the purposive behavior of an artificial body


Embryonic day 18 rats were harvested by caesarean section from anaesthetised pregnant dam (animal care followed standard procedures that are in accordance with institutional guidelines). Cerebral cortices were isolated and chopped into small pieces and exposed to a trypsin (0.125%) digestion and mechanical dissociation through fine-tipped pipette. The resulting tissue was resuspended in Neurobasal medium (Invitrogen) supplemented with 2% B27 and 1% glutamax (both Invitrogen), no antimitotic agent was used to control glial cells proliferation. Dissociated cells were counted, diluted and plated, in a drop of serum-free medium from 7µl up to 150µl, both at high density and low density and, finally, placed in a humidified incubator at 37° C, 5% CO2 for about one hour. When cells adhered to the substrate, 1 ml of Neurobasal supplemented medium was added to the culture. Once a week 50% of old medium was replaced with fresh pre-warmed medium. Culturing was usually carried on from 5 up to 6 weeks, but spontaneous cortical network activity already arises at the end of the first week. Low-density culture (600cells/mm2) was used onto glass coverslips 20mm, pre-treated only with poly-lysine solution, to characterize neuronal system by immunofluorescence technique. We evaluate from DIV7 to DIV 28 a panel of neuronal markers as indicators of the synaptogenesis maturation( synaptophysin ,synapsin , V-Glut 1, V-GAT, or cytoskeleton neuron protein tubulin βIII and astrocytes fibrillary acidic protein GFAP. On the other hand, high-density culture was used onto microelectrode arrays, pre-coated with adhesion promoting molecules: Poly-lysine and Laminin, in order to monitoring the electrophysiological activity of neuron network over a long period. During these last two years we used two different types of micro-electrodes MEAS and IMT clustering chip. On the first commercial type we plate around 1000 - 1200cells/mm2, while on the IMT chip we plate on each of the five clusters 2000cells/mm2.
A neurophysiological mini-laboratory (NML) has been developed. It consists of a microtransducer array (MTA), temperature sensors (Pt-RTDs), heating elements, a glass reservoir and a mini-incubation chamber. The aim of this development was to provide a stand-alone, simple to use system allowing an easy mounting and dismantling of different parts. The MTA integrates an array of sixty Pt thin-film electrodes for electrical recording and stimulation of the network activity, EPON SU-8 clustering structures for structuring the network in interconnected sub-populations and two temperature sensors. The MTA is realized on Pyrex and mounted on a printed circuit board allowing optical observation by an inverted microscope. The overall dimensions of the chip are 14mm x 14mm. A glass reservoir (outer diameter of 24 mm, wall thickness of 2mm and a height of 8mm) which is glued on the MTA defines a culture media volume of 2.5ml. In order to limit the evaporation of the media, the reservoir is closed by a transparent semi-permeable FEP membrane (permeable to O2 and CO2). The PMMA incubation chamber has commercially available heating elements (thermal-clear heating) fixed on its internal side as well as on the top cover in order to avoid vapour condensation. The heating elements have been selected considering the minimal power needed for keeping the internal chamber air at 37°C and compensate an external environment temperature of 25°C. The cooling is realized by thermal dissipation and consequently, the system does not function with external temperatures higher than the regulated one. On the top-cover of the incubation chamber, two in-/outlets accesses for media and gas exchanges are provided. The realized NML is easy to use and allows to measure precisely the temperature of the media with the integrated Pt-RTDs and to stabilize the culture media temperature at 37°C. The NML is moreover compatible with commercially available temperature controllers, perfusion systems and MCS pre-amplifier.
A modular and programmable acquisition/stimulation system for standard Micro-Electrodes Arrays has been designed; specific boards plugged in a rack and PC controlled run analogue acquisitions and stimulation from and to MEAs. The rack receives from the MEA 4 to 60 analogue signals. The signals are processed (programmable amplification, filtering), digitised and transferred to the PC system via a PCI bus. The rack is also in charge of the generation of programmable stimulation signals, triggered by the PC. The rack functions are controlled by the user via an i2C bus from the computer. The signal channels are processed in parallel; the complete set-up of 60 channels requires 2 Neurobit racks, as each manages 32 channels for the acquisition and 16 channels for the stimulation. Specifications list: Acquisition (from MEA): Number of input channels: 4/Board Input Voltage: +/-1V Gain: 1 to 127000 Noise density at: 1kHz en =13nv/ Input impedance: >1010_ – 6pF Stimulation (to MEA): Number of output channels: 4/Board Input triggers: 4/Board Output voltage range: +/- 15V Waveforms: monophasic, biphasic, square pulses Operation mode: Digitally triggered Communication bus: i2c
The "NEURObit-TOOLS" library is a data analysis software tools developed to process and analyse signals recorded from “in vitro” networks of neurons, interfaced with the external environment through a matrix of microelectrodes (MEA).The "NEURObit-TOOLS" library offers a collection of classic algorithms and, in addition, some innovative methods based on the application of Wavelet Theory for noise reduction and the estimation of the Hurst Parameter for the determination of bursting activity. Some functionalities can be managed through a graphical interface structured in a way that simplifies the processing, while still allowing the user to have an immediate control over the parameters. The main characteristics of the tool are: - ease of use, even for inexperienced users; - versatility in the management of parameters, allowing optimisation of the processing strategy; - an innovative offering that facilitates evaluation of new analysis methodologies.
A series of stimulation protocols have been designed and implemented in order to induce activity-dependent and pathway-specific modifications in cortical networks cultured over MEAs. The developed plasticity protocols are based on two stimulation paradigms presented in the literature, the Spike Timing Dependent Plasticity - STDP and the tetanic stimulation. The application of the two protocols clearly demonstrated the possibility to modulate the evoked response of a cortical network through a combination of specific stimulation patterns and by exploiting the features of the MEAs, which allow the delivery of stimulating signals from more than one site. The developed tetanic protocol consists in the simultaneous application of a tetanus and a low frequency stimulation pattern (i.e. tetanic co-activation) from two different sites. The general effect consists in the increase of the evoked response of the network and seems to be long lasting, since a test stimulus executed one hour after the delivery of the tetanus continued to show high responses in most of the recording channels. A different result was obtained by using a single tetanus (i.e. without a co-activating site), one of the few proposed in literature as plasticity protocol for the neural preparations employed for this project. Only the first stimulation after the tetanus was able to increase the network response: we can assume that this effect is short-term. Our innovative protocol does not affect only the entity of the evoked response, but also the timing (after the tetanus the evoked activity appears anticipated) and the recruitment process. By analysing the functional connectivity of the network and studying the changes in the pathways activated by each stimulating site, we found that a higher number of connections are involved after the tetanus and their strength results, in general, improved.
The result is a Matlab library of analytical tools for the analysis of the dependence of multi-site neural recordings on a (multidimensional) time-varying sensory signal. In particular, the toolbox implements one- and multi-dimensional reverse correlation techniques (Wiener filters), post-stimulus time histograms, information entropy, and mutual information coding efficiency. It also includes modules for analysis of the temporal structure of spike trains: for instance, to distinguish between bursts and isolated spikes.
A real-time hardware/software system was developed for closed-loop interactions between cultured neural preparations and an artificial body. The set-up is an organized around two closed-loop paths: one conveys the data bus (analogue and digital, for acquisition, processing and stimulation) and the other conveys the associated control bus. The set-up elements are hosted in 3 different functional entities: the MEA and its connectivity, the acquisition/stimulation HW/SW interface (described i a specific result) and the PC-hosted processing HW/SW. The dedicated hardware and software elements were specifically designed to optimise the real-time functionality in the closed-loop applications. The closed-loop system presents the following characteristics: - Due to the modular architecture of the Neurobit rack, the set-up can be customized (number acquisition and stimulation channels) depending on the user needs. - The whole analogue data path is implemented in a single hardware element (Neurobit rack), including A/D conversion, to minimize external noise coupling (no analogue signals in the PC). - The closed-loop data path delay is less than 50us on each of the 60 channels; biological real-time processing is possible including important data processing and/or analysis. - The control parameters can be dynamically programmed during the running of the experiment (real-time acquisition, processing, stimulation);
A device (microtransducer array MTA), based on a microelectrode array (MEA) integrated with a EPON SU-8 clustering structure comprising five clustering chambers interconnected with microchannels, for studying in-vitro neuronal network dynamics in interconnected sub-populations of cortical neurons has been developed. The MEA consists of 60 thin-film Pt microelectrodes of 30 micrometer in diameter microfabricated on a Pyrex substrate. The MEA-chip has the overall dimensions of 14 x 14mm2 and comprises also two resistive (Pt-RTD) temperature sensors. The microelectrodes distribution is as follows: 11 microelectrodes per lateral chamber, 1 microelectrode per channel and 12 microelectrodes in the central chamber. The five clustering chambers interconnected via microchannels are 350 micrometer in height and have a diameter of 3mm. The microchannels are 800 micrometer long and have a width of 300 micrometers. The device is mounted on a printed circuit board (PCB) providing a compatible interface with commercial pre-amplifier (MCS). The PCB has dimensions of 49mm x 49mm x 1.6mm and provides all the required electrical contacts as well as on opening (12mm in diameter) for optical observation by an inverted microscope. A glass ring is glued on the PCB in order to define a reservoir (volume of 2.5ml) for the physiological solution. The outer diameter of the reservoir is 24mm with a wall thickness of 2mm and a height of 8mm. The reservoir is closed by a semi-permeable membrane for preventing water evaporation and allowing CO2 and O2 permeability. The MTA can be integrated in an incubator chamber, which has been also developed in this project. It is fabricated in PMMA and has commercially available flexible heating elements (thermal-clear heating) fixed on the internal side of the chamber and on the top-cover. The technological developments performed allowed to achieve highly stable stimulation/recording capabilities, adequate biocompatibility and lifetime of the MTA devices. The cluster functionality was demonstrated by comparing the recordings of both spontaneous and evoked network activities on clustered and conventional MEAs.
This is an experimental apparatus and real-time software application for bi-directional interaction of cultured neurons and a mobile robot. More in general, it can be used for closed-loop electrophysiology experiments at population or organism level. It is based on off-the-shelf hardware (PCs and acquisition boards). Design tools for rapid prototyping of real-time control and hardware-in-the-loop applications have been used to implement the real-time software application. The control architecture is made of two PCs. PC1 runs a real-time operating system (QNX) and is responsible for (i) acquisition and sampling of electrophysiological signals, (ii) on-line spike detection and artifact blanking, (iii) decoding of the spike trains, (iv) robot control, (v) coding of robot proximity sensors signals and (vi) production of the pattern of stimuli that trigger the electrical stimulator. A second computer, PC2, connected to PC1 through an Ethernet link, is the experiment front-end. We used Simulink/ Real-Time Workshop (The Mathworks) and the RT-Lab package (Opal-RT) as hardware-in-the-loop development environments. The control loop runs at 10kHz, and robot control is updated at 10Hz. This system allows simultaneous acquisition of neural signals from up to 32 recording sites. A video-capture system, running on a third PC, is capable of monitoring in real time the robot trajectories. Such information can be used in learning experiments, which require on-line assessment of robot performance. The software part includes a library of general coding and decoding modules for bi-directional neural interfaces. These modules are the basis for the implementation of the neural interface. As for coding, we provide schemes for proportional (i.e., intensity) as well as event coding. Blocks for stimulus generation include Poisson and non-leaky integrate-and-fire. As regards decoding, there are command generation blocks based on estimates of the instantaneous firing rate. In particular, the library involves (i) spike detection, (ii) artifact suppression, (iii) firing rate estimation, and (iv) motor command generation modules. The final version of such library makes available a general tool for design and implementation of prototype neural interfaces.
The protocol for the preparation of dissociated cortical cell cultures has been standardized among the partners. From the onset of plating rat cortical nerve cells in a culture chamber the neurons start growing out dendritic and axonal arborisations and form synaptic connections. After several days in vitro the developing neuronal network starts to exhibit electrical activity from the spontaneous action potential firing of the individual neurons. This firing activity displays two typical phases, one of uncorrelated firing among the neurons at low firing rate, and one of highly synchronized firing and strong network interactions (called network bursts). In the project the spontaneous firing dynamics has been characterized for its firing rate and spatio-temporal pattern during network development. Although network bursts were highly variable to each other, the underlying probabilistic patterns of firing in space-time appeared to be highly stable during development. As during this period outgrowth still continues and synapses are being formed, this is a surprising and important outcome. Electrical stimulation has been shown to have lasting effects on the spontaneous firing rates of individual neurons, ranging from increased or decreased firing rates up to the complete activation or silencing of neurons. Overall spatio-temporal patterns of firing within network bursts, however, remain highly stable. Conditions for maximal responses of electrical stimulation need to be optimised for each individual culture. Different electrical stimulation protocols have been explored in order to induce changes in the response patterns of the cultured networks. Tetanic (co-) stimulation was able to significantly potentiate responses, measured by the area of the post-stimulus time histogram. With the burst-phasing protocol the responsiveness of the network could also be altered. The one-site paired pulse stimulation protocol was found to generally enhance the responsiveness of the network. The two-site paired-pulse stimulation protocol was able to depress the response at one site of a pair, elicited by a stimulus at the other site. These experiments have demonstrated the sensitivity of the network to alter its responsiveness differentially after carefully designed stimulation protocols.
The experiments conducted in the Neurobit project produce very large recording files to analyse. This large amount of data requires (i) a fast, reliable and partially automatic signal processing and (ii) facilities to visualize and extract classical (ISIH, IBIH and PSTH) and customized analysis (correlation maps and coefficients, JPSTH and Fourier analysis). For this purpose, we developed the ANALYSER application optimised for the signal processing of 60 channels microelectrode arrays (MEA) recordings. It markedly allows self-adaptive threshold estimation as well as all the aforementioned analysis. It also provides fast raw-data visualization tools allowing quick experiment survey and feedback for an optimal data management.