The role of different midbrain dopamine neuron populations in signalling reward and cost

The brain must learn to use environmental cues to predict whether an action will have positive or negative consequences and the associated cost (e.g. effort) of performing the action. This type of learning involves the chemical messenger dopamine and there are good theoretical models that explain how the nerve impulses generated by dopamine-releasing cells might signal reward. However, despite the elegant simplicity of these models, at the cellular level there is greater complexity. For example, dopamine cells seem to consist of several populations which convey different aspects of the reward signal. Therefore, to better understand reward processes and when these go awry (e.g. in addiction), we must first understand how different dopamine-cell populations signal different parts of the reward signal and to which parts of the brain these signals are sent.

To do this, we will divide populations of dopamine cells according to the brain regions that they innervate and use the combination of different molecules present in each population as a kind of barcode to identify them. We will then use newly-developed, powerful techniques to record nerve impulses from individual dopamine cells in mice during a reward-task and label each recorded neuron. We will investigate how different populations of dopamine cells signal 1) positive and negative consequences and 2) the amount of effort required to obtain reward. We will use each labelled dopamine cell's 'barcode' to tell us where in the brain the signals were sent. To further test the role of different dopamine-cell populations, we will use cutting-edge technologies to measure dopamine released in a particular brain-region and then switch-off one of the dopamine-cell populations. We will also use computer simulations to help us interpret how the dopamine nerve-impulses translate into dopamine release in different regions of the brain.

This research will help us to better understand some of the complexity of reward-related signalling and enhance our theoretical models of learning. We will define populations of dopamine cells and reveal which brain regions receive different components of the reward signal. The 'barcodes', data, and computer models we generate will enable us and other researchers to build a better picture of reward learning and understand how it goes wrong in brain disorders.

In order to choose optimum actions, the brain must associate environmental stimuli with an outcome, distinguish whether the outcome is positive, and determine the cost (e.g. effort) associated with obtaining it. Dopamine is thought to provide a uniform teaching signal which guides such learning. However, the neurons that generate the signal are heterogeneous, with different neurons signalling different aspects of reward. How then can such diverse neurons transmit a coherent signal to guide learning?
Recent evidence suggests that subpopulations of neurons innervating different regions encode different aspects of reward; it is therefore essential to define how a neuron encodes reward in the context of which brain region it innervates. To achieve this, we will identify combinations of molecular markers which define populations of midbrain dopamine neurons projecting to particular regions of the nucleus accumbens and striatum. We will then record and label single dopamine neurons in head-fixed, behaving mice, and use the molecular signatures to determine how different populations of neurons signal reward and the cost of obtaining it.
We will first examine how neurons projecting to different regions differentially encode positive and negative outcomes. Then, to investigate how neuronal activity translates into dopamine release, we will optogenetically silence one of the target-defined populations and measure dopamine release using fast-scan cyclic voltammetry during a positive/negative outcome task. To examine encoding of cost we will use an instrumental task where the effort required to obtain reward is varied. We will record the activity of different, target-defined dopamine neurons during high- and low-effort trials. These cutting-edge experiments will elucidate how different aspects of reward are encoded by discrete populations of dopamine neurons and transmitted to different forebrain regions.

The main deliverables from this work are: 1) Identification of molecular markers that define populations of dopamine neurons projecting to different brain regions. 2) Mechanistic insight into how different populations signal positive and negative outcomes and the cost associated with obtaining them. These findings will increase our understanding of how discrete populations of dopamine neurons encode different aspects of reward and how these signals are transmitted to different brain regions. Defining molecular signatures of dopaminergic populations will enable researchers in the field to better understand the development of different groups of dopaminergic neurons and to generate new tools to study the function of discrete populations. The results will also enable us and other researchers to better understand reward learning and how it goes wrong. For example, while this research is not focussed on mechanisms of addiction, further understanding the fundamental processes underlying reinforcement learning will improve our knowledge of how differences in these processes may result in overeating, drug abuse or other forms of addiction (which are estimated to have social and economic costs for the UK exceeding £60bn per year).

The findings from this research will primarily benefit academic labs researching reward, the basal ganglia, development, addiction, and disorders involving the dopaminergic system (e.g. Parkinson's, Schizophrenia). Further understanding signalling by dopaminergic neurons may also provide insight to druggable targets for new or improved therapies. To realise these benefits we will disseminate our findings through talks at scientific meetings, by rapid publication in open-access journals and will make available data, computational models and equipment designs arising from this work.

The other major beneficiary is the post-doctoral research assistant on this programme. They will receive training in cutting-edge in vivo techniques only available in a few laboratories worldwide. This will be of significant value both to them and UK science, given the shortage of researchers with in vivo expertise in both academia and industry. In addition to practical techniques, they will also gain skills in writing, presentation and project management, which would be of benefit in all employment sectors.