Biophysical genetics of collective feeding in C. elegans

The goal of behavioural genetics is to understand what aspects of behaviour are inherited and which DNA sequence differences are responsible. Geneticists have long dismissed the notion that there are genes 'for' particular behaviours and recognise that reality is more complicated. DNA sequence differences change how proteins work inside of cells and these changes at the molecular level influence the cells that make up organs and ultimately a whole animal. A given animal will typically contain multiple sequence differences that can interact with each other, further complicating prediction. That is why it is often useful to study the simplest organism that displays a behaviour of interest.

In this proposal, we are trying to understand the genetics of collective behaviour--that is, the behaviour of groups of animals. Think starling flocks, ant colonies, and traffic jams. This is a daunting task, but fortunately, the nematode worm C. elegans, a small animal with only 302 neurons, is also capable of a simple kind of collective behaviour: some C. elegans strains feed in large groups while others feed alone. Over the last fifteen years, geneticists have identified several genes that disrupt worm collective feeding when they are mutated, but to understand which aspects of the behaviour these genes affect and how they work together in an intact animal, we need a model of collective feeding.

Studies of other kinds of collective behaviour have shown that animals following relatively simple rules, which depend only on what their neighbours are doing, can give rise to complex group behaviours. In this approach, we write down a set of rules that capture what we think we know about the system and then simulate the results on a computer. If the simulated animals behave like the real animals, then we gain confidence that we have identified the right rules. One of the keys to good modelling is to have good data and so the first step is to record movies of worms as they form groups and to quantify exactly how they move once the groups have formed. To see how worms move in tight groups we will use a technique called fluorescence microscopy that allows us to see particular parts inside of worms, which makes them easier to identify. We will then use our knowledge of how worms move to design a tracking algorithm that can automatically identify individual worms in the group. Tracking real animals will give us the same kind of information that we will derive from the simulated results enabling us to precisely compare the simulated and real animals. If we need to, we will update the rules in the simulation to achieve better agreement between the simulations and experiments and learn how worms interact in the process. By repeating the cycle of experiment and simulation with mutant worms, we will also begin to understand which aspects of the behaviour are under genetic control.

Despite years of study, we actually still do not know why some worm strains form groups. Is there some advantage to feeding collectively instead of alone? It could be that these animals with very limited computational capacity are able to take advantage of the wisdom of the crowds to find high quality food patches more efficiently. We will use our computational model to predict food patch arrangements that favour aggregating or solitary animals and then test those predictions experimentally. If we find environments where collective feeding is advantageous this might tell us why collective behaviour has evolved in worms.

This work will advance our understanding of how sets of genetic mutations can give rise to changes in complex behaviours. It will also give some insight into how nematode worms forage as groups which may have implications in agriculture because some plant parasitic nematodes, which cause more than $100 billion in crop damages annually, can form swarms.

We will use quantitative fluorescence imaging, automated tracking, and computational modelling to identify the behavioural rules that govern collective feeding in the nematode worm C. elegans. We will determine how these rules are affected by mutations as well as how the resulting collective behaviour may provide a foraging advantage for aggregating strains to find high-quality food in patchy environments. Specifically, we will image aggregating (CB4856) and non-aggregating (N2) strains that express GFP in sparse patterns at high spatio-temporal resolution at increasing worm densities to study the transition from solitary to group behaviour. A key advance will be the application of multi-hypothesis particle filter tracking algorithms to follow individual worms even in dense groups. Group-size distributions and position/velocity correlations will be used to quantify the behaviour. We will develop a self-propelled particle model that incorporates putative behavioural mechanisms including roaming and dwelling, oxygen avoidance, and touch sensation. A systematic comparison between the experimental and simulated systems will identify the most important mechanisms and which aspects of the behaviour they control. Determining the interaction parameters that are changed by mutations that affect known neural circuits (oxygen sensation with gcy-35 mutants, neuropeptide signalling with npr-1 mutants, and touch with mec-4 and glr-1 mutants) will show how these pathways exert their effects on behaviour quantitatively. Finally, we will use the validated model to search for arrangements of different quality food patches that favour aggregation and solitary feeding, and test these predictions experimentally. The proposed research will give insight into how complex behaviour is regulated genetically, and how an animal with a compact nervous system is able to forage collectively in heterogenous environments.

In addition to the Academic Beneficiaries described above, the proposed research will have concrete benefits in two sectors of UK industry and on the general public in the UK and abroad.

Agriculture
Plant parasitic nematodes cause more than $100 billion in damage to crops annually, but the pesticides that are used to control them are under increasing regulatory pressure due to their environmental and health risks. There are therefore clear social and economic reasons for finding safer pesticides or alternative approaches to controlling parasitic nematodes. The quantification of nematode behaviour plays an important role in these efforts both in UK industry and in BBSRC-supported projects focussing on plant parasitic nematodes. The greatly improved tracking algorithms we will develop in the proposed research will benefit these efforts directly. In the longer term, our integrative approach that combines experiments with computational modelling will open new ways of working in this area that are data-driven and quantitative.
Lameness and mastitis in dairy cows costs UK industry £100 million annually and lameness in sows is also a substantial problem, with a prevalence of approximately 4% with at least one lame sow being found in 50% of surveyed breeding units. Efforts to identify lameness before obvious symptoms arise will save farmers money and improve animal welfare. There are UK companies that offer radio-frequency tagging systems for tracking cows and pigs as well as BBSRC-supported projects to advance these systems. Position and acceleration data can be informative, but the richer data available from video monitoring would likely improve the accuracy of animal health monitoring systems and enable earlier diagnosis. Again, our work on improved animal tracking algorithms and quantitative analysis methods will benefit these projects.

Manufacturing
A 2013 government report on The Future of Manufacturing identified the need to support small and medium enterprises (SMEs) in the UK "specialising in the production of high value sophisticated components for equipment manufacturers." The PI has worked closely with SMEs that work in the technology sector to develop an advanced small animal tracking system, which included both hardware and software development. The new technology we develop during the course of this grant will feed into future collaborations with these SMEs and enhance the kind of innovative equipment they are able to produce. This will improve the UK's international competitiveness in behaviour monitoring and analysis technology.
The report also emphasised the increasing reliance of UK manufacturing on highly skilled workers and the role that 'Big Data' will play in the future for these companies. The RAs will both receive extensive training in analysing large datasets and will finish as highly skilled workers who will contribute to the UK's creative output.

General public
The PI will continue to work with Zooniverse on WormWatchLab, a citizen science project that engages the public to help classify videos of worm egg-laying behaviour. This generates valuable data but also provides the general public with an opportunity to be directly involved in a scientific project and to learn about how model organisms contribute to biological research. This enhances the reach of research results by also providing a platform for discussing new results with a wider audience. People have a natural interest in the complex patterns produced by animals behaving in groups, and the fluorescence videos of aggregating worms produced in the proposed project will visually striking. This will provide a hook for public engagement through social media to ensure our work maximally benefits the public understanding of science.