A systems approach to understanding sensory-motor control of aimed limb movements

Our own limb movements - and even those of insects - far exceed in dexterity and robustness those of any robot. Accidents or medical disorders such as large fibre sensory neuropathy that impair or prevent controlled limb movements have profound effects on the quality of life of those affected. We seek to understand how the brain controls aimed limb movements so that it is possible to better understand what goes wrong in disease processes, and to develop better medical interventions. Brain function in humans is exceptionally complex, and it is difficult or impossible to carry out many of the sorts of experiments that are required to investigate it. We therefore work with a much simpler animal - a locust - in which we can record, analyse and manipulate the activity of individual nerve cells while it makes aimed movements. The movements are analysed using video-based movement tracking and such data are used to test our computer models. A great advantage of our approach is that we have a much more complete understanding of the roles of particular nerve cells in a locust than we do in humans or any other mammals. The problems that a locust must solve in making an aimed movement are the same as those faced by humans, so we seek out the general principles of organisation that underpin all such movements. In this proposal we set out to analyse how groups of nerve cells operate together to generate aimed movements, and to determine how signals are transferred between different parts of the nervous system. A second strand of our research develops software that enables both us and clinicians with whom we collaborate to analyse nerve cell signals recorded from the brain. We have developed powerful methods to permit detection of the activity of single nerve cells in recordings made from the brains of awake patients, and have used these to reveal important aspects of how these cells respond to complex stimuli. We now wish to develop these methods to permit the detection of many nerve cells using recordings from multiple electrodes, and to work much more quickly. Such advances in signal processing will be very important for improving our understanding of human brain function and will be crucial in the development of prosthetic limbs that are controlled interactively by the activity of a patient's brain. Developing such methods requires the processing of large amounts of data from real recordings, which is very difficult and costly to obtain from human patients. We will instead carry out our development work using signals recorded from locusts. Our ability to recognise individual identified cells in a locust provides us with extremely powerful ways of validating our approach. To achieve our aims we have the following main objectives: 1. Develop, validate, use and make available to other users a powerful improved version of our software for processing neural data. What aspects of nerve cell signals permit us to best identify their activity in a complex recording? How can we classify the firing of these cells most accurately and most rapidly? 2. Characterise the separate roles of motor nerve cells that drive leg flexion during an aimed limb movement. How do the signals of each cell differ, and what are the relationships between the patterns seen in the different cells? We will develop new techniques for recording many single cells simultaneously. 3. Analyse the patterns of activity in nerve cells that carry information between different parts of the nervous system. What are their inputs and outputs? Can our software automatically distinguish between different types of cells in complex multiple-cell recordings? 4. Characterise the responses of sensory nerve cells that signal leg position. How do these influence an aimed movement, and how do they change after damage to the sense organ?

How does neuronal activity drive aimed limb movements, and how should these complex signals be recorded and analysed? Our interdisciplinary collaboration will use an established invertebrate preparation to address these questions. In locusts, the kinematics of aimed limb movements can be related to the activity of identifiable cells in several small populations, and we have the ability to experimentally manipulate the system in ways that are not possible in vertebrates. We will take advantage of this to provide data that can be used to answer questions about sensory-motor control of movements, and to develop and validate analysis tools that will be distributed for widespread use. A strength of our project is that we will quantify the performance of our spike sorting algorithm using real data. Our main objectives are to: 1. Develop, optimise, use and disseminate an unsupervised spike sorting algorithm. We will address the major issues of (a) dealing with multi-electrode recordings, (b) automatically validating clustering outcomes, (c) performing spike sorting online, and (d) dealing with overlapping spike waveforms. 2. Characterise the separate roles of the 9 members of a population of motor neurones that drive tibial flexion during an aimed limb movement. 3. Analyse the transmission of information between body segments during an aimed limb movement. 4. Determine the population response of sensory neurones from the femoro-tibial chordotonal organ, and investigate the origins of behavioural plasticity following surgical perturbations to this proprioceptive feedback. We will use electrophysiological techniques that are all well established in our labs, and couple these with the development of multi-electrode recordings to provide an important step forward in our understanding of aimed limb movements. The development of analysis tools will likewise build on our extensive background in spike sorting to provide a general solution to key issues in this area.

Our project will contribute to a detailed knowledge of how nerve cells represent information about directed limb movements, and will develop computational methods with widespread applicability to the analysis of neuronal signals. These questions are of great interest to neuroscientists seeking to understand brain function, and are also intrinsically related to major clinical applications such as the development of prosthetic devices driven by brain signals. Beneficiaries of our project will therefore include the scientists carrying out the work, our academic collaborators around the world, the wider neuroscience research community, members of the public, clinicians and commercial enterprises. Our work will contribute to the body of knowledge underpinning society's understanding of the world around us and our place and function within it. Scientists working on the project will benefit from genuine multidisciplinary training in specialised research techniques, and in transferable skills that improve their ability to perform their jobs and enhance their career prospects. These benefits will be realised within the 3 year duration of the project, and will be long lasting. Our project will therefore help to provide the skilled researchers needed for both academic research and industrial R&D, which is a BBSRC Strategic goal. Research collaborators will benefit from our development of knowledge that will inform their related research, and from scientific exchanges of staff and techniques. We have a proven track record of hosting and training researchers from our collaborators' labs, and our staff have benefitted from reciprocal arrangements. Members of the public will benefit in the longer term from improvements in medical interventions that improve their quality of life, such as the eventual goal of developing neuroprosthetic devices. Our unique approach, using a well-studied insect model of limb movement, will help to establish the UK in this area of research which has been dominated by US research groups. In the long term, clinicians may also benefit from the availability of improved diagnostic methods and medical devices. Commercial exploitation will enhance UK competitiveness in biomedical signal processing. Benefits to clinical researchers will begin to arise during the course of the project, whereas transfer of benefit to clinicians and patients will take many years, but have long-lasting consequences. Our results will be published in prestigious scientific journals, presented at international conferences and in talks at academic institutions, discussed during a 1-day public workshop, and summarised in public open days and media communications. Information will be available from dedicated public web pages and a specialised scientific database. We have already begun to explore the commercial exploitation of our software with a major biomedical equipment company (Neuralynx), and will be further assisted in this by the University's Technology Transfer Service. We already collaborate with clinicians, thus providing a direct and established route for communication and engagement beyond academia. These ongoing collaborations - along with our other academic collaborations - will continue to engage these beneficiaries throughout the project and beyond it. The main concepts underlying our work are attractive and easily comprehensible to a general audience. This will be used to our advantage in reaching the public through non-technical articles, press releases, open meetings and public lectures. Dissemination and public engagement will be facilitated by the University Press Office. In the last 5 years our work has led to 9 press releases by the University, which have been widely disseminated by the media, including several radio interviews and articles in The New York Times, The Washington Post, The Independent, Daily Mail, Scientific American, Discover Magazine, and New Scientist.