Sending light and recording electrical activity in deep brain structures using optrodes
When optogenetic tools for controlling neural activity came into play in the late 2000s, it became obvious that hybrid probes for delivering light into the brain and recording photoevoked activity simultaneously would become a basic component of the optogenetic toolkit. And this for at least two reasons. First, microelectrode recordings are still the golden standard for measuring neuronal firing, surpassing by far optical sensors at least at the single cell resolution and millisecond time scale. And second, electrical recordings are and will probably remain the ultimate readout of the efficiency of optogenetic activation and inactivation protocols. In other words, the only way to show that your optogenetic activator or silencer does the job is to directly record its effects on neuronal firing.
Various types of optical⁄electrical probes for deep brain photostimulation and recording. Top left: An optical⁄electrical probe having single optical fiber and single electrode. Bottom: Three types of single optical fiber–multiple electrode combination. Bottom left – wire-wound tetrodes are combined with an optical fiber. Bottom center – “Michigan probe” integrated with an optical fiber. Bottom right – “Utah” multi-electrode array combined with a tapered optical fiber. The optical fiber is Au-coated, and also works as an electrode. Top right: A combination of multiple optical fiber and multiple electrode. Multiple optical fiber cores are bundled with a spacing of 3.3 µm in a single optical fiber bundle. From Hayashi et al., 2012.
Light delivery and electrical recordings can easily be decoupled in any preparation that can be put under a microscope. But for deep-brain structures (> 1 mm of depth), light must be guided along the recording electrodes to reach the region of interest. A little armada of hybrid implantable devices combining one or several optical fibers with micro-electrodes, often called opto-electrodes (“optrodes”), emerged in the last 6 years. The trick usually consisted in simply adding single optical fibers to existing microelectrode systems. The simplest optrode is an optical fiber and a single microelectrode glued together. Other versions are based on multielectrode arrays such as silicon electrode arrays (“silicon probes”). In this case, an optical fiber is glued onto the shaft of the silicon probe. The fiber can be equipped with its own LED or laser diode at the opposite end (“pigtailed”), as shown by the Buzsaki lab. Another way to build silicon optrodes is to include a light guide (waveguide) during the fabrication step. This can be done by depositing a layer of some kind of light conducting resin (such as the SU-8 photoresist). Silicon optrodes are starting to be commercially available, for example from Neuronexus. Other designs for this type of optrode are under study, stemming for example from the work of the NeuroProbes consortium. Optical fibers can also be simply glued to existing tetrode systems like the standard 4 twisted wires, or can be integrated into 3D arrays of single electrodes such as the one produced by Innovative Neurotechnologies. Two other “exotic” solutions were developed in parallel: one uses gold-metalization of the fiber itself to make it behave as an electrode. The other uses a dual core fiber: one core is used for light transmission and the other one is hollow and filled with an electrolyte, and can be used for electrical recording. This dual-core optrode can be pulled like conventional glass micropipettes, give an equivalently good signal-to-noise ratio (usually superior to metal electrodes), and is by definition not exhibiting any light artifact in the recording.
The need for more channels in light delivery systems
One obvious path for improving devices that either send or detect signals is to increase their resolution, in other words to increase their number of independent channels. For optical devices you can think of this as increasing the number and decreasing the size of the points where light is sent or collected (such as pixels). For electrical devices used in neuroscience, the game is mostly to increase the number of sites where the extracellular potential is measured, which is the purpose of multi-electrodes. From the overview given in the above paragraph, it is pretty clear that existing optrodes have a better resolution in the electrical domain than in the optical one. Indeed, few researchers had envisaged the need for multi-point light delivery deep into the brain, which was only motivated by the problematics of deep-brain imaging. But this might change. Given the progress in optogenetic strategies for controlling neuronal activity, it is becoming relevant to refine the way we deliver light in deep brain regions. Using sub-millimeter light patterning techniques could indeed help us bridge the functioning of micro and macro-circuits (let’s call this the study of meso-circuits?). A first attempt toward this was the production of a multi-site light delivery probe by the lab of Ed Boyden at MIT. In this probe, several independent waveguides are built onto the same shank and can send light perpendicularly in the tissue at different levels of the shank using micro-mirrors.
Multiwaveguide Implantable Probe by the Boyden lab. (a) Schematic of an example waveguide probe, 360 μm wide, and containing 12 waveguides, with inputs, bends, shanks, and corner mirrors labeled. (b) Photomicrograph of a waveguide probe fabricated according to the design in (a), with light coupled into 3 of the 12 waveguides and immersed in a scattering medium, showing emission of 473 nm and 632 nm light out three separate ports. (c) Cross section of a single waveguide, taken through the shank section of the probe. (d) SEM of a single waveguide, for the cross section shown in (c). From Zorzos et al., 2012.
Endomicroscopy: the way to go for deep brain light patterning?
Let’s come back to the pioneers in deep-brain imaging. Their goal is to perform microscopy in deep brain structures with minimal tissue damage, usually pulling their hairs out to find ways of sending light to and collecting light from many points in space through a thin conduct called an endoscope. The endoscope is often composed of a fiber bundle which transmits incoming and outgoing light through as many channels (pixels) as there are single cores in the bundle. In other words, the spatial mapping across the hundreds or thousands of individual fiber cores at one end is preserved at the other end. Therefore fiber bundles can be suited for laser scanning microscopy: the scanner moves a laser beam successively into each core of the fiber bundle and fluorescence light collected from individual cores is focused onto a light detector. The fiber bundle can terminate with no additional optics, allowing only soft-contact surface imaging of the tissue (in this case resolution is provided by the density of fibers in the bundle), or with a gradient refractive index microlens (GRIN lens) that provides higher resolution. A company called Mauna Kea has pushed this technology close to its limits by building confocal endomicroscopes that can provide X-Y resolution down to about 2.5 microns.
From Benjamin Abrat (Mauna Kea) and Andrew Masters (Coherent). Endoscopic Confocal Microscopy Moves into the Clinic, Biophotonics International, Nov 2006.
Sending light to different points in space is precisely what many people want to do to refine optogenetic stimulation experiments. So why not learn directly from what endomicroscopists have achieved? And if we are to use their technology, what not also use it for its original purpose, which is imaging the brain and optically monitoring its activity? So here we are: now is the time to image and control deep neuronal tissue in vivo with high spatio-temporal resolution using electro-microendoscopes.
Fiber bundles + microelectrodes = the “electro-endomicroscope”
In a paper published in the European Journal of Neuroscience, Hayashi et al. investigate the use of fibered endomicroscopy coupled to microelectrodes for combined patterned light delivery, optical imaging and electrical recording. The rationale is simple: single cores of optical fiber bundles can be used both as individual pixels for constructing an image of brain tissue and as independent waveguides for light delivery; in parallel, several microelectrodes are associated to the fiber bundles in order to provide a simultaneous multi-site readout of neuronal activity. Hayashi’s probe consists in 3 optical fiber bundles (80 or 125 microns diameter each, made of 1.9-um-diameter single-mode optical fibers spaced every 3.3 um) and 10 tungsten microwires inserted into a stainless steel tube. The spacing between microelectrodes is at least 40 um. The probe has a 45 deg beveled or a conical shape to minimize brain damage during insertion. A confocal scanner is used to raster a laser beam over the input facet of the bundles through an objective lens.
Map of spike-generation points in the endoscopic field of view. Large colored dots represent electrode position. Small dots represent spike-generation points. Small dots are color-coded according to the electrodes by which spikes were detected. Scan speed was 8 ms ⁄ line and light intensity was 2.0 mW at the probe tip. Superimposed spike waveforms detected with each electrode are shown on both sides. Vertical bars indicate 100 lV, horizontal bars indicate 1 ms. From Hayashi et al., 2012.
Hayahi et al started by making fluorescence pictures of cortical layer 2/3 in an anesthetized mouse. Expression of ChR2-YFP was obtained by in utero electroporation. The density of fibers in the fiber bundles allowed discriminating individual neurons in the tissue. For photostimulation, the laser beam was raster-scanned in rectangular areas of the endoscopic field of view. Depending on the region illuminated and the amount of light used, the photoevoked activity could be picked up by the nearest microelectrode only. Using their probe in a Thy1-ChR2 mouse, the authors show that they can evoke single-whisker deflections by stimulating small areas through a single core of the bundle, arguing that the spatial selectivity of their photostimulation method is at least as good as that of electrical microstimulation.
One thing the authors did not do is to assess whether they could also image neuronal activity with their probe, using for example genetically-encoded calcium indicators. An earlier study by Vincent et al. showed that functional optical imaging in vivo is indeed feasible with fiber bundles. Let’s stay tuned