Circuitry of inhibition and selectivity in a Drosophila learning centre

A major role of the brain is to coordinate appropriate behavior and for this it has to process an enormous palette of information across a range of senses, and within each sense, recognize 'objects' that influence output behaviors. These behaviors are not limited to simple reflexes, like withdrawal of the leg from a needle, but can be shaped by experience. One example of this is our sense of smell; humans and animals can discriminate a vast range of odors, and a characteristic smell can evoke a highly specific flood of memories. The ability of our brain to recognize a particular smell as a perceptual 'object', implies that it combines a process of odor discrimination with formation of specific associations (memories). Understanding how this happens is a major challenge for neurobiology. This has implications not only for basic brain science but also for understanding human behavior, and potentially even conditions including schizophrenia, in which memories of objects including smells may be inappropriately retrieved as hallucinations. The fruitfly Drosophila offers many advantages to understand the neuronal circuits that discriminate and use specific odor information. It can discriminate many odors; it can learn and remember experiences associated with specific odors; remarkably the structure of its olfactory system shares many common features with humans; it is easy to breed in the laboratory; it has powerful genetic tools that can monitor and manipulate activity in specific neurons. We therefore have a toolkit to dissect neuronal circuits including those for learning and memory and to test their roles in behavior, a toolkit that is probably more powerful than in any other organism. Our experimental system is the Drosophila larva. It contains a fully functional olfactory system, since it is capable of olfactory learning, but has a numerically and anatomically simple olfactory system, compared to either adult flies or vertebrates. It has only 21 olfactory sensory neurons, each sensing a different 'odor quality'. However, just as only three kinds of photoreceptor can define our entire range of color vision, inputs from these 21 neurons can be combined to define potentially thousands of smells. This occurs in a brain region called the mushroom body, which has many similarities to human sensory cortex, with which it may share a common evolutionary origin. The mushroom body is required for olfactory learning, and contains several hundred neurons called Kenyon cells (KCs). Individual KCs can combine inputs that originate from around 6 of the 21 olfactory neurons, but their responses to odors are much more selective, sometimes so selective that many KCs do not respond to any odor tested. Therefore KCs only respond when many of their ~6 inputs are activated simultaneously. Therefore a smell is defined by the combination of sensory neurons that it activates, and by the small number of KCs that integrate this combination. The high selectivity of KCs results from their inhibition by other neurons, which keeps them from firing when they receive only a small number of olfactory inputs. This balance between activation and inhibition is critical to their function - too little inhibition, and KCs will become less selective and unable to discriminate among odors; too much inhibition, and KCs will never respond to odors. Our goal is to understand which neuronal circuits cause this inhibition, and the mechanisms by which it affects the selectivity of KCs, and ultimately its consequences for both learning and retrieval of memories that are associated with specific odors. This work will reveal how integration of olfactory information in the mushroom bodies is regulated, in a simple and highly accessible system. The basic principles that we uncover should also be relevant to similar processes in higher cortical areas of the human brain, which is much harder to study.

Encoding of sensory objects by the higher brain depends on sparse and highly selective patterns of neuronal activity. This in turn critically depends on inhibitory innervation, to prevent inappropriate neuronal activity that would degrade the specificity of sensory representations. Many features of this process are shared between mammals and insects, and our goal is to understand its circuit mechanisms. To this end we want to understand the role of inhibition in olfactory representations used for learning in the higher brain of the Drosophila larva. This system offers an anatomically simple brain, full capability for odor discrimination learning, and powerful circuit analysis tools. Higher level olfactory discrimination and formation of associative odor memories in Drosophila requires a brain region known as the mushroom body. Its input region, the calyx, receives heavy GABAergic innervation. This is of critical importance in regulating the timing, sparseness and specificity of odor-evoked responses of mushroom body neurons. However, the circuit mechanisms of this are poorly understood. To address this problem, we first aim for a comprehensive neuroanatomical understanding of the logic of GABAergic calyx innervation, focusing on those neurons that are strongest candidates to regulate the selectivity of mushroom body responses. In the process we will generate tools for targeted expression in subsets of calyx GABAergic neurons. We will use these tools to monitor activity in specific GABAergic neurons during olfactory discrimination and learning. We will also use them to manipulate activity of these cells, and examine the effects both on activity of other cells in the circuit, and on larval olfactory and learning abilities. This work will give an integrated view of how inhibition regulates selectivity of responses in the higher brain, in a system that is analogous and possibly homologous to mammalian sensory cortex, but much easier to monitor and manipulate.

The work proposed here is basic underpinning research, designed to reveal the logic of how the brain recognizes, discriminates and remembers complex sensory objects. The principal beneficiaries will be academic, but the insights are likely to interest a wider field. First, the mechanisms that we dissect here are to at least some degree a model for sensory discrimination in human higher brain. Discrimination is intimately linked to formation and retrieval of memories, and a knowledge of the underlying neural mechanisms is a prerequisite to understand disease processes that involve either loss of these representations (e.g. dementia) or inappropriate recall (e.g. schizophrenia). Second, there are important applications of the science of odor discrimination in insects. Chemical attractants (e.g. pheromones) and repellents (e.g. DEET) have long been in use against insect pests and disease vectors. The former in particular have a less deleterious environmental impact than insecticides, since they can be adapted to specific insects. More recently there has been tentative but increasing interest (and investment) in use of insects as chemodetectors for substances such as explosives, e.g. http://www.inscentinel.com/ (based at BBSRC Rothamsted), or the Los Alamos Stealthy Insect Sensor Project (http://www.youtube.com/watch?v=_T7d0bze4kM). While peripheral olfaction mechanisms are most directly applicable to these problems, understanding higher brain mechanisms can provide valuable background understanding such as the potential discriminatory power of the system. Third, understanding the nature and function of brains is of wide public interest, largely because of questions such as what makes us human, and our understanding of animals and their welfare requirements. Indeed, the more sophisticated we are at interrogating insect behavior, the more sophisticated we realize it is. The wide public interest that this evokes is illustrated by the interest of national and international media in Drosophila courtship (http://news.bbc.co.uk/1/hi/sci/tech/7350403.stm), smarter flies that learn better (http://www.nytimes.com/2008/05/06/health/06iht-06dumb.12604485.html), and by the fascinated response of school groups to regular outreach talks given by CJO'K. A finding that any manipulations to the GABAergic system could increase learning ability would be likely to arouse media interest, and would certainly arouse great interest in outreach activities. Fourth, insect behavior, particularly locomotion and vision, continues to provide blueprints for robotics (e.g. Halloy et al., Social integration of robots into groups of cockroaches to control self-organized choices, Science 318: 1155-8, 2007; 135 other papers in PubMed on insects and robots/robotics). So far there is less robotic work based on insect olfactory discrimination and associative learning, but we believe there is potential - but this depends entirely on a better understanding of the circuitry, which is the major goal of our work.