Large-scale recording and computational modelling of midbrain raphe microcircuitry during emotional learning

Emotional well-being is crucial for healthy living. Yet emotional disorders are common and disabling, and they contribute significantly to diminished quality of life in an ageing population. Serotonin (5-hydroxytryptamine; 5-HT) neurons in microcircuits of the midbrain raphe region are critical to the normal regulation of emotion. Environmental and genetic variation in 5-HT has striking effects on emotional processing in animals and humans, and this variation generates increased risk for developing emotional disorders such as anxiety and depression. Although 5-HT targeted drugs have important therapeutic effects, current agents lack the efficacy and tolerability required, and improved treatment options are sorely needed.

Despite the strong links between 5-HT and emotional control, after more than 40 years of research we still do not know what emotional information 5-HT neurons signal nor how such signals are encoded. Thus, theories on the role of 5-HT in emotional control in health and disease are being developed without knowledge of what midbrain raphe 5-HT neurons actually do. Progress is slow for two main reasons. Firstly, the lack of methods to record the activity of identified 5-HT neurons during behaviour; in the few previous such attempts the identity of the neurons is assumed and not known. Secondly, the newly discovered high complexity of the midbrain raphe microcircuitry; multiple 5-HT neuron subtypes interacting in unknown ways with neighbouring neurons to generate their signals. Capturing this complexity will require large-scale recordings of identified neurons in behaving animals, and the generation of large biological datasets. However, analysis and interpretation of such datasets using standard biological approaches will be a major challenge.

In the current proposal we aim to identify emotional information signalled by 5-HT neurons by taking, for the first time, a systems biology approach. This is an approach by which biological questions are addressed through integrating the collection of complex data with computational modelling to produce a better understanding of biological systems. To this end we will take advantage of rapid progress in large-scale electrophysiological recording methods to monitor the firing of identified neurons in the midbrain raphe microcircuitry, in animals engaged in the processing of emotional information. We will identify 5-HT and other important neurons in the raphe circuit using a method called 'optotagging' which is a form of optogenetics. Optotagging is a new technology in which a virus is used to target a light-sensitive protein to neuron types in a highly specific way such that when exposed to a short burst of light, the neurons are briefly activated and reveal themselves in the recordings. We will monitor 5-HT neurons alongside another neuron type, GABA neurons, that we predict encode the signals generated by 5-HT neurons. We will carry out these recordings whilst animals are engaged in different stages of an emotional learning task that we know is sensitive to 5-HT, and therefore likely involves 5-HT signalling. In separate experiments we will use optogenetics to manipulate 5-HT and GABA neurons to establish their causal role in emotional learning.

The final critical part of our systems biology approach is to use machine-learning techniques to capture and analyse the large amount of complex data generated in our electrophysiological experiments, and integrate it into a computational model of the functional connectivity of the raphe microcircuit. Through this we will develop a multi-scale computational model of 5-HT neural signalling during the processing of emotional information. We can then use the model to make further predictions that we can test and eventually get a much more complete explanation of how 5-HT signalling in the raphe microcircuit is changed by risk genes, drug therapies, and environmental factors such as when we get older.

Emotional well-being is crucial for healthy living, yet emotional disorders contribute much to diminished quality of life during ageing. Midbrain raphe 5-HT neurons are linked strongly to emotional control, but significant research has not revealed what emotional information 5-HT neurons signal or how such signals are encoded. This is because methods used to record raphe neurons during behaviour do not allow neuron identification, and fail to take into account the high complexity of the midbrain raphe microcircuitry; multiple 5-HT and other neuron subtypes with diverse firing (and thus functional) properties. Capturing this complexity requires large-scale recordings of identified neurons in behaving animals, and the generation of large biological datasets that are difficult to analyse and interpret using standard biological approaches.

Here we aim to identify emotional information signalling by 5-HT neurons by taking, for the first time, a systems biology approach. We will obtain large-scale multiunit recordings of raphe neurons in mice engaged in an aversive learning task that is highly 5-HT sensitive, and thus a likely generator of 5-HT signals linked to emotional processing. We will monitor 5-HT neurons as well as GABA neurons, which our data suggest are a likely 5-HT encoder. We will identify these neurons by 'optotagging' - activation of a light-sensitive opsin (ChR2) targeted to 5-HT or GABA neurons using viral vector gene delivery in a 5-HT- and GABA-specific Cre-driver lines of transgenic mice (SERT-Cre and VGAT-Cre, respectively) - and use optogenetics to bidirectionally manipulate 5-HT and GABA neurons to establish their causal role in behaviour. Finally, we will use machine-learning techniques and causal connectivity analysis to capture the highly complex biological dataset, and integrate it into a multi-scale computational model of raphe microcircuit function during emotional processing, for further prediction and testing.

Who will benefit?
In addition to the academic community (see Academic Beneficiaries), the other potential beneficiaries of our work will be clinicians, health policy makers, the pharmaceutical industry, and ultimately patient populations and the general public.

How will they benefit?
Emotional disorders such as anxiety and depression are common and disabling, and are the source of a major socio-economic burden and poor quality of life for many individuals. Their prevalence increases with age and so as people live longer, their contribution to disability increases.

The neurotransmitter 5-HT is heavily implicated in genetic and environmental vulnerability to emotional disorders, and their successful drug treatment. Recent thinking is that 5-HT's main influence is on emotional learning, rather than on mood directly. Understanding the mechanisms through which 5-HT supports or influences emotional (aversive) learning is therefore of paramount importance for understanding what puts individuals at risk of emotional disorder, and for improving their treatment. SSRIs are currently the pharmacological treatment of choice for anxiety and depression but suffer from poor tolerability and not all patients receive adequate benefit.

Before we can understand the contribution of 5-HT to emotional learning/mood in patient populations, we need to understand what emotional information is signalled by 5-HT, and how these signals are generated and contribute to aversive memory. However, the 5-HT system is of such high complexity that capturing and understanding it will require the generation of large biological datasets, which will be challenging to analyse and interpret using standard biological approaches.

By taking a systems biology approach in which we will use machine-learning techniques to capture and analyse the large amount of complex data generated in our neurophysiological experiments, we will develop a multi-scale computational model of 5-HT neural signalling during the processing of emotional information. We can then use the model to make further predictions that we can test. Ultimately we believe that we will get a much more complete explanation of what 5-HT signaling does, and also a computational model that realistically predicts how 5-HT signalling is changed by risk genes, drug therapies, and environmental vulnerability factors such as age.

Such an understanding and increased capacity to predict the future could foreseeably be of importance to clinicians who are treating patients, developing prevention strategies, and trying to detect patients at risk, and even to Governments and other bodies who are developing health policy. It will also be of value to industry that is investing in new drug therapies that are currently without good predictors. And it will also be of great value to patients and their carers themselves, who are trying to understand and cope with the disorders.