A Closed-Loop Optogenetic Platform

Abstract

Neuromodulation is an established treatment for numerous neurological conditions, but to expand the therapeutic scope there is a need to improve the spatial, temporal and cell-type specificity of stimulation. Optogenetics is a promising area of current research, enabling optical stimulation of genetically-defined cell types without interfering with concurrent electrical recording for closed-loop control of neural activity. We are developing an open-source system to provide a platform for closed-loop optogenetic neuromodulation, incorporating custom integrated circuitry for recording and stimulation, real-time closed-loop algorithms running on a microcontroller and experimental control via a PC interface. We include commercial components to validate performance, with the ultimate aim of translating this approach to humans. In the meantime our system is flexible and expandable for use in a variety of preclinical neuroscientific applications. The platform consists of a Controlling Abnormal Network Dynamics using Optogenetics (CANDO) Control System (CS) that interfaces with up to four CANDO headstages responsible for electrical recording and optical stimulation through custom CANDO LED optrodes. Control of the hardware, inbuilt algorithms and data acquisition is enabled via the CANDO GUI (Graphical User Interface). Here we describe the design and implementation of this system, and demonstrate how it can be used to modulate neuronal oscillations in vitro and in vivo.

Keywords: closed-loop; electrophysiology; neuromodulation; open-source; optogenetics.

Figure 1

Schematic of a closed-loop optogenetic platform for neuromodulation research. Brain signals are recorded via electrodes and optical stimulation is delivered via micro-LEDs located on the shaft of an optrode. An Application Specific Integrated Circuit (ASIC) headstage implements recording and LED driving, and communicates with a microcontroller-based control system via an Serial Peripheral Interface (SPI). The control system receives the recorded data and processes it in real-time to deliver appropriate stimulation commands. The closed-loop algorithm is configured and monitored by the user via a Graphical User Interface (GUI). The GUI also allows for the acquired data to be stored for post-processing.

Figure 2

Overview of the full system. We show two configurations to illustrate how the system can be used in combination with custom components (top) as well as only commercial components (middle). Custom components include (A) CANDO Control System, (B) CANDO headstage, (C) CANDO GUI, and (D) CANDO Optrodes. Commercial components include (E) Commercial Electrode Array & Light Fibre, (F) Intan Recording Controller, (G) Intan headstage, (H) Intan GUI, (I) 1401-4 Data Acquisition System, (J) Spike2 GUI, and (K) T-Cube LED Driver.

Figure 3

Block diagram (A) and physical implementations of the CANDO Control System (B) and CANDO Headstage (C). The headstage is based on a CANDO3 Application Specific Integrated Circuit (ASIC) comprising a Finite State Machine (FSM) digital block that controls the LED Drive circuits and Low Noise Amplifiers (LNAs). Communication with the control system is implemented by a Serial Peripheral Interface (SPI). The CANDO-CS is based on an MK22FN512VLH12 Microcontroller Unit (MCU) which contains Analog-to-Digital Converter (ADC), Digital-to-Analog Converter (DAC), SPI, Universal Asynchronous Receiver Transmitter (UART) and General-Purpose Inputs/Outputs (GPIOs) peripherals. There are four power management circuits, a level-shifting circuit for signals being introduced via the Intan Recording Controller, and a second order Anti-Aliasing Filter (AAF) with a cut-off frequency of 250 Hz. There is also a USB-to-UART converter for communication between the MCU and a PC. The Joint Test Action Group (JTAG) Connector allows for programming and debugging the MCU firmware.

Figure 4

Schematic of firmware implementation of closed-loop algorithms; (A) block diagram representation of the firmware architecture; the algorithm Interrupt Service Routine (ISR) determines the program flow and utilises functions from the Input/Output (IO) layer. The ASIC commands block allows for different drivers to be introduced according to the desired ASIC; different layers represent alternative commercial/custom systems. The main external/internal communication protocols are addressed in the MCU peripherals block; (B) flowchart of the main method and ISR; the main method is responsible for performing the code initialisation and control of the external DAC. The ISR is triggered every 1 ms and performs the recording, algorithm processing, stimulation and data transmission over UART. The program then returns to the main method. (C) block diagram representation of one inbuilt algorithm for phase-shifting neural oscillations via an FIR filter. (D) ISR execution time (no more than 547 μs) for a single ASIC being controlled (recording and stimulation) by the MCU and for the data to be processed and transmitted to the PC via UART.

Figure 5

Screenshots showing the main functionality of the GUI for configuring and controlling closed- or open-loop stimulation experiments.

Figure 6

Results from an in-vitro experiment on a mouse brain slice; (A) experimental set-up, in which both the Intan headstage and CANDO headstage are connected to different recording sites of a single 64-channel NeuroNexus probe; (B) dimensions of the Intan headstage, CANDO headstage and NeuroNexus probe; (C) experimental results show simultaneous recording from the Intan headstage (top) and CANDO headstage (bottom) of a seizure-like event elicited by 4-AP.

Figure 7

Closed-loop optogenetic stimulation in-vitro (mouse brain slice) and in-vivo (non-human primate). (A) Schematic of the in-vitro experimental set-up, in which optogenetic stimulation delivered via an optic fibre is controlled in real-time from LFP recordings; (B) LFP traces during no stimulation and stimulation during 0° and 180° phase shift; (C) Power spectra for each of the phase conditions; (D) Modulation of power relative to no stimulation epochs at each frequency for the different phase conditions revealing a phase-dependent enhancement (red) and suppression (blue) of oscillations around 10 Hz. Note that phase has been unwrapped and plotted over two cycles to reveal the pattern more clearly. (E) Schematic of the in-vivo experimental set-up in which closed-loop optogenetic stimulation is delivered to the brain of an anesthetised animal via an implanted optrode; (F) LFP traces during no stimulation and stimulation during 270° and 90° phase shift; (G) Power spectra for each of the phase conditions; (H) Power at each frequency for the different phase conditions revealing a phase-dependent enhancement (red) and suppression (blue) of oscillations around 10 Hz.