The treatment of epilepsy has been transformed over the past century by the development of anti-epileptic drugs (AEDs). Despite the almost exponential increase in the number and variety of available AEDs, almost 25% of patients remain refractory to pharmacological treatment (Kwan et al, 2011). This equates to a significant burden of disease. Within the UK almost 100,000 patients require continuing hospital-based medical care. A third of these patients continue to have seizures at least monthly and a sixth have at least one major seizure per month (Neligan and Sander, 2009).

For patients with refractory focal epilepsy, resective surgery offers a potentially curative option. Those who have mesial temporal sclerosis can see seizure freedom rates of 80-90% within two years following a selective amygdalo-hippocampectomy (Sagher et al, 2012). However, this number declines to 50-60% over longer-term follow-up, and other forms of resective surgery have lower success rates (de Tisi et al, 2011). Resective surgery faces a number of complications. Any surgery has its attendant risks of morbidity and mortality. Success is dependent on the seizure focus being accurately localised, in non-eloquent cortex, and able to be resected in its entirety. There is a balance to be struck between the margins of the resection and likelihood of seizure freedom against the risk of neurological deficit. All this combines to mean that there is a low number of resective surgeries done in the UK relative to the number of potential beneficiaries (Lhatoo et al, 2003).

With the success of deep brain stimulation (DBS) in the treatment of Parkinson's disease, there has been a growing interest in the development of neuromodulatory devices. This includes DBS for epilepsy, vagus nerve stimulators (VNS) and the NeuroPace RNS system. There are several advantages to this type of approach over resective surgery. They are relatively non-invasive, with no required removal of brain tissue, and as a result these therapies are reversible, either by switching off the device or explantation. This minimises the risk of surgical complications or neurological deficits. It is not required to precisely know the location of the seizure focus as they act by modulating neural dynamics over a wide area of brain tissue. As such they can be used to control epilepsies with larger or multiple foci. They are also controllable, with the ability to provide either open-loop or closed-loop control.

Most neurostimulation systems used today act in an open-loop manner, with stimulation pre-programmed by a clinician and altered at intervals to try and improve seizure control. Open-loop control simply delivers the pre-programmed stimulation without reference to any information about the patient’s current physiology or state. Due to the limitations of such control, there has been a move towards the development of closed-loop control in these devices. In closed-loop control the device also records some aspect of the patient’s physiology and uses this information to control the stimulation parameters in real time. Theoretically, the device could detect an upcoming seizure or increased seizure risk and provide stimulation appropriately. Closed-loop therapies may therefore increase the efficacy, improve the clinical benefit and reduce the side-effects of stimulation, while also improving the power consumption of the device.

However, all current neuromodulatory devices use electrical stimulation which is imprecise, exciting all cell types in a non-selective manner. In order to overcome these weaknesses the CANDO project has turned to optogenetics.

Optogenetics is a form of gene therapy utilising naturally occurring light-sensitive proteins called opsins. At the heart of each opsin is a molecule called retinal, derived from vitamin A. The retinal molecule is bound to a partner opsin and together the opsin-retinal complex is referred to as rhodopsin. When a photon (a particle of light) strikes a retinal molecule it changes shape. This triggers a conformational change in its partner opsin.

This conformational change can drive a variety of different functions depending on the type of opsin. Bacteriorhodopsins and halorhodopsins act as ion pumps, where absorption of a photon provides the energy to move a number of ions across the cell membrane against their electrochemical gradient. Channelrhodopsins act as ion channels that open in response to light, permitting the flow of ions across the membrane along their electrochemical gradient. OptoXRs are coupled to an intracellular G-protein domain that can activate downstream G-protein coupled signalling pathways.

Having identified these light sensitive opsins, we need a means of getting our target cell, which we want to make light-sensitive, to manufacture and express these opsins in its cell membrane. To do so we turn to gene therapy. As early as 1966 it was recognised that viruses possessed properties that could be very useful in delivering genes into cells of interest (Tatum, 1966). A virus, on infecting a cell, inserts DNA into the host cell’s genome and uses the host cell’s own machinery to replicate itself. If it were possible to remove the pathological genes in a virus and replace the inserted DNA with a gene of our choice, then we would have a vector capable of delivering a gene into the DNA of target cells. It wasn’t until the 1980s that we had acquired the appropriate tools for this kind of genetic manipulation and the first human trials of gene therapy were done (Rosenberg et al, 1990).

Since then, there has been a proliferation in the types of viruses that can be used and the applications of gene therapy, but the fundamental principles are similar. A virus is reengineered such that it remains capable of entering a cell and inserting its genetic code into the DNA of the host cell. This DNA contains the code for our desired gene and a promoter. The promoter is a little bit of genetic code which lets the host cell know there is a gene which can be translated into a protein, and starts this process. Different types of cell – nerve, muscle, skin, etc – will translate genes following specific promoters. As a result, by selecting the right promoter sequence, even if you infect lots of different cells, only a specific cell type will actually manufacture and express the inserted gene.

The first application of these opsins to neuroscience came in 2004, when researchers incorporated channelrhodopsins into mammalian neurons (Boyden et al, 2005). This was shortly followed up by a demonstration of the use of halorhodopsin (Zhang et al, 2007). This opsin pumps one chloride ion into the cell for each photon it absorbs. The result is a hyperpolarising current which acts to inhibit the neuron and prevent spiking activity. This is important for epilepsy as it is characterised by the over-activity of various neural circuits. The ability to inhibit neurons optically, along with the ability to target this inhibition to specific cell types, has therefore led to a range of investigations into the control of epilepsy. Early experiments incorporated these inhibitory opsins into glutamatergic principle neurons in the hippocampus. These cells could then be hyperpolarised in response to light, suppressing epileptiform activity (Tønnesen et al, 2009).

Following this pioneering work, there has been an explosion in the types and properties of opsins as research groups genetically manipulate opsins to improve upon or alter various characteristics. This includes tuning opsins to be responsive to specific frequencies of light. There has also been work done altering opsin kinetics, so that they can respond incredibly quickly to the onset or offset of stimulation, allowing millisecond-level control of neural activity (Lin et al, 2009). In the other direction, the kinetics can be slowed to the point that a brief light stimulus can lead to long-lasting effects on neural activity; even up to 30 minutes (Zhang et al, 2007). More recently, channelrhodopsins have been engineered to allow the passage of anions instead of cations, acting as a much more efficient inhibitory current (Berndt et al, 2014).

Research into epilepsy has taken advantage of this ever-expanding toolset to allow much more focused interrogation and control of neural circuits. This led to research by Ivan Soltesz’s laboratory demonstrating closed-loop control of a mouse model of temporal lobe epilepsy using either inhibition of excitatory principle cells or excitation of inhibitory interneurons (Krook-Magnuson et al, 2013). Seizures were detected via depth electrodes in the induced seizure focus, and optical stimulation delivered via fibre optic cable. Optical stimulation was triggered by seizure detection and all animals responded with silencing of electrographic seizures. Significantly, activation of a subset of inhibitory interneurons also resulted in seizure abortion, despite this population of cells only making up 5% of hippocampal neurons. This suggests that even targeting only a small proportion of the network can be effective in controlling wider network dynamics.

The result of all this work is a hugely powerful and versatile toolset for very specific manipulation of neural activity. With earlier studies providing proof of concept, our aim is to translate this research into clinical practice, working towards a first-in-man trial of optogenetic control of focal epilepsy in 2021. Optogenetics helps overcome a number of weaknesses of current invasive methods for treatment of focal epilepsy. It is far more specific; stimulation is delivered to a specific type of cell, rather than all neural tissue, and this stimulation can be either excitatory or inhibitory. In combination with electrical recording, optogenetic neuromodulation has the potential to deliver appropriate stimulation to suppress developing seizure activity.

Developing a device which can be used clinically throws up a number of unique challenges. While optical stimulation has many advantages, the brain is not transparent; it absorbs and scatters light. This makes delivering a sufficient intensity of light to activate opsins over a significant volume of brain tissue, difficult. This has two knock-on effects. Firstly, it requires a lot of electrical power, placing demands on battery capacity and performance, and recharging must only be necessary at reasonable intervals. Secondly, this power demand is ultimately dissipated as heat, and care must be taken to ensure that tissue around the implant is not overheated. The brain is also not a friendly milieu for foreign objects and any implantable device must be protected from degradation within the body, as well as vice versa. These challenges must be met within a device which is small enough to be implanted within the brain, without causing any significant damage or scarring by its presence.

To this end, the CANDO project is truly multidisciplinary. A large team of engineers is working to meet the technical challenges of developing such a device. In parallel with them is a team of researchers, working to enable this device to detect abnormal network dynamics and prevent this activity progressing to a seizure. The challenges are formidable but surmountable, and the potential benefit for patients in whom all other options have been exhausted is immense.

1. Kwan PMDP, Schachter SCMD, Brodie MJMD. Current Concepts: Drug-Resistant Epilepsy. The New England Journal of Medicine. 2011;365(10):919-26.

2. Neligan A, Sander JW. The incidence and prevalence of epilepsy. London: UCL Institute of Neurology. 2009.

3. Sagher O, Thawani JP, Etame AB, Gomez-Hassan DM. Seizure outcomes and mesial resection volumes following selective amygdalohippocampectomy and temporal lobectomy. Neurosurgical Focus. 2012;32(3):E8.

4. de Tisi J, Bell GS, Peacock JL, McEvoy AW, Harkness WFJ, Sander JW, et al. The long-term outcome of adult epilepsy surgery, patterns of seizure remission, and relapse: a cohort study. The Lancet. 2011;378(9800):1388-95.

5. Lhatoo Samden D, Solomon Juliet K, McEvoy Andrew W, Kitchen Neil D, Shorvon Simon D, Sander Josemir W. A Prospective Study of the Requirement for and the Provision of Epilepsy Surgery in the United Kingdom. Epilepsia. 2003;44(5):673-6.

6. Tatum EL. Molecular Biology, Nucleic Acids, and the Future of Medicine. Perspectives in Biology and Medicine. 1966;10(1):19-32.

7. Rosenberg SA, Aebersold P, Cornetta K, Kasid A, Morgan RA, Moen R, et al. Gene Transfer into Humans — Immunotherapy of Patients with Advanced Melanoma, Using Tumor-Infiltrating Lymphocytes Modified by Retroviral Gene Transduction. New England Journal of Medicine. 1990;323(9):570-8.

8. Boyden ES, Zhang F, Bamberg E, Nagel G, Deisseroth K. Millisecond-timescale, genetically targeted optical control of neural activity. Nat Neurosci. 2005;8(9):1263-8.

9. Zhang F, Wang L-P, Brauner M, Liewald JF, Kay K, Watzke N, et al. Multimodal fast optical interrogation of neural circuitry. Nature. 2007;446(7136):633-9.

10. Tønnesen J, Sørensen AT, Deisseroth K, Lundberg C, Kokaia M. Optogenetic control of epileptiform activity. Proceedings of the National Academy of Sciences. 2009;106(29):12162-7.

11. Lin JY, Lin MZ, Steinbach P, Tsien RY. Characterization of Engineered Channelrhodopsin Variants with Improved Properties and Kinetics. Biophysical Journal. 2009;96(5):1803-14.

12. Berndt A, Lee SY, Ramakrishnan C, Deisseroth K. Structure-Guided Transformation of Channelrhodopsin into a Light-Activated Chloride Channel. Science. 2014;344(6182):420-4.

13. Krook-Magnuson E, Armstrong C, Oijala M, Soltesz I. On-demand optogenetic control of spontaneous seizures in temporal lobe epilepsy. Nat Commun. 2013;4:1376.