Izquierdo, Eduardo (2021) International Worm Meeting "Computational Neuroethology to Bridge the Gap between Connectome, Neural Dynamics, and Behavior"

Izquierdo, Eduardo

[International Worm Meeting, 2021]

One of the current scientific challenges is to understand how the brain works. To address this challenge, we must bridge the gap between the information available about neural circuitry, activity, and behavior. With the growing recognition of the central roles that embodiment and situatedness play, the true challenge is even more difficult: to understand how behavior is grounded in the dynamics of an entire brain-body-environment system. It is also increasingly recognized that the largest challenge we face is not in collecting the data, but in developing the computational modeling tools that will allow us to understand the system. C. elegans is a uniquely qualified target for such an integrated modeling of a complete animal. I will provide an overview of the current challenges and state of the art on the effort to develop such a model. Despite the substantial behavioral and anatomic knowledge available, we still do not know the strengths and signs of the synaptic connections or the response characteristics of the majority of the neurons. However, it would be a mistake to wait until more data is gathered before we begin to develop the computational models. I have developed a unique computational approach using optimization techniques to explore the space of unknown electrophysiological parameters of the nervous system necessary to generate organism-like behavior. Because the models are necessarily underconstrained, each successful parameter search produces an ensemble of models that are consistent with the known anatomy, physiology, and behavior of the organism. We then analyze the properties of this entire ensemble using techniques ranging from model neuron recordings, neural and behavioral manipulation and lesion studies to parameter clustering, dynamical systems theory, and information theory. The focus of the analysis is to identify different possible classes of solutions and to thoroughly understand the operation of the highest-performing exemplars of each. This insight can then be used to suggest specific experiments that could decide between the different possibilities. In addition to accelerating the discovery and understanding of the neural mechanisms underlying specific behaviors of interest, I will show how this methodology allows us to begin to address key theoretical challenges in a situated, embodied, and dynamical understanding of adaptive behavior more generally.

Izquierdo, Eduardo et al. (2009) International Worm Meeting "A minimal neural network model of klinotaxis behavior in C. elegans."

Lockery, Shawn, Izquierdo, Eduardo

[International Worm Meeting, 2009]

Previous experiments have shown that C. elegans chemotaxis is based in part on a biased random walk. Recently, however, Iino and colleagues have described a complementary strategy, called klinotaxis, in which the direction of locomotion is continuously aligned to the direction of the chemical gradient. Here we combined neural network modeling and mathematical analysis to identify simple circuit motifs for klinotaxis and to understand how they function. The model involves an idealized representation of C. elegans sensory neurons, motor neurons, and neck musculature. The direction of locomotion is determined by the angle of the head with respect to the body, which is set by the difference between the activation levels of dorsal and ventral neck muscles. The circuit has two sensory neurons, representing the chemosensory neurons ASEL and ASER, and two motor neurons collectively representing the dorsal and ventral neck muscle motor neurons. In keeping with functional differences between ASE neurons in the biological network, the sensory neurons in the model respond to the derivative of the attractant concentration such that the model''s ASEL neuron acts like an ON cell, whereas the model''s ASER neuron acts like an OFF cell. The motor neurons are modeled as simple nonlinear dynamical neurons. In addition to input from the sensory neurons, the motor neurons also receive out-of-phase sinusoidal inputs from a central pattern generator, which is not modeled explicitly; however, connections between motor neurons were not included. We used an evolutionary algorithm to optimize synaptic strengths and intrinsic neuronal properties to generate klinotaxis behavior in simulated radial gradients. Connections were constrained to be symmetric across the dorsal-ventral midline. We found that it was possible to generate realistic klinotaxis behavior within the constraints described above. Importantly, the effects of simulated ablations of sensory neurons in the model were consistent with the same ablations in real animals. In particular, ablation of ASER produced a strong chemotaxis deficit whereas ablation of ASEL produced little or no deficit, unless combined with ablation of ASER. Further analysis of the simulated ablations showed that ASEL contributes mainly to the time it takes the worm to reach the gradient peak (efficiency), whereas ASER contributes mainly to the likelihood of reaching the peak (reliability). Finally, a dynamical systems analysis of the motor neurons suggested a phase-dependent sensitivity to input. It is this mechanism that enables the state-dependence necessary for klinotaxis. These findings provide novel hypotheses that can now be tested on real worms.

Izquierdo, Eduardo J. et al. (2017) International Worm Meeting "Integrated neuromechanical model demonstrates stretch-reception can generate and propagate wave responsible for forward locomotion on agar."

Beer, Randall D., Izquierdo, Eduardo J.

[International Worm Meeting, 2017]

Behavior is grounded in the interaction between an organism's brain, its body, and its environment. With only 302 neurons and a fully-reconstructed neural and muscle anatomy at the cellular level, C. elegans is an ideal candidate organism to study behavior with the help of computational models. Of particular interest is understanding the neuromechanical basis of locomotion, since nearly its entire behavioral repertoire is expressed through movement. How the rhythmic pattern is generated and propagated along the body is not yet well understood. We report on the development and analysis of a model of forward locomotion that integrates known neuroanatomy, neurophysiology and body mechanics of the worm. Our model is the first to consider recent experimental analysis of the structure of the ventral cord circuitry and the effect of local body curvature on nearby motorneurons. We develop a neuroanatomically-grounded neural model of the ventral nerve cord subcircuit, using a neural model capable of reproducing the full range of electrophysiological properties observed in C. elegans neurons. We integrated the neural model with a reconstruction of a biomechanical model of the worm's body from published descriptions, with updated musculature and stretch receptors. Unknown parameters were evolved using a genetic algorithm to match the speed of the worm on agar. We performed 100 evolutionary runs and consistently found electrophysiological configurations that reproduced realistic control of forward movement. The ensemble of successful solutions reproduced key experimental observations that they were not designed to fit, including the curvature profile of the body's movement, and the wavelength and frequency of the propagating wave. Analysis of the ensemble revealed forward locomotion is possible without intrinsic oscillations in either the head or the rest of the ventral nerve cord. Circuits were capable of initiating oscillations in the head using only stretch reception, providing a novel hypothesis. Similarly, circuits relied on stretch reception to propagate the dorsoventral oscillation, without the need for bistability in the motorneurons, as had been previously proposed, and with gap junctions across neural units playing only a minor role. Altogether, we provide an existence proof for forward locomotion through stretch-reception in an up-to-date neuromechanical model of the worm, as well as a series of testable hypotheses about its operation.

Ikeda, Muneki et al. (2021) International Worm Meeting "Klinotactic versus klinokinetic steering strategies implemented in neuroanatomical models for C. elegans thermotaxis"

Ikeda, Muneki, Matsumoto, Hirotaka, Izquierdo, Eduardo

[International Worm Meeting, 2021]

Steering behavior is a prevalent strategy of navigation in which animals make gradual adjustments of their moving direction. The current view holds that steering behavior is a direct orientation to a destination based on temporal comparisons of sequential stimulus samples. This orientation strategy is categorized as klinotaxis. However, for a small-size animal, C. elegans, the difference of signal intensity sampled through its sinusoidal locomotion is subtle and potentially overwhelmed by noise, variability, and non-uniformity in environmental signal. We herein constructed neuroanatomically-grounded models that reproduce C. elegans thermotactic steering behavior and found that model worms implemented klinokinetic steering strategy; persistent thermal input sensed not through sinusoidal but through forward locomotion of model worms controlled their curving rates, thereby steering indirectly to a destination temperature. An evolutionary algorithm was employed to find configurations of the model that reproduce empirical positive thermotaxis in C. elegans population. In all the evolved models, steering curvature were modulated by temporally persistent thermal signals sensed beyond the time scale of sinusoidal locomotion of C. elegans. Persistent rise in temperature decreased steering curvature resulting in straight movement of model worms, whereas fall in temperature increased curvature resulting in crooked movement. This relation between temperature change and steering curvature reproduced the empirical thermotactic migration up thermal gradients and steering bias toward higher temperature. Further, spectrum decomposition of neural activities in model worms showed that thermal signals were transmitted from a sensory neuron to motor neurons on the longer time scale than sinusoidal locomotion of C. elegans. These results indicate a previously unrecognized mechanism of C. elegans steering behavior which does not rely on its sinusoidal locomotion. The klinokinetic steering strategy proposed in this work is potentially employed in real worms and enabling reliable orientation behavior overcoming environmental noise and variability.

Bettinger JC et al. (2000) West Coast Worm Meeting "State-dependent learning in C. elegans."

McIntire SL, Bettinger JC

[West Coast Worm Meeting, 2000]

The execution of learned behaviors may be triggered by contextual information, consisting of environmental cues or the internal state of the organism. State-dependent learning refers to the ability of an organism to more effectively execute learned behaviors if the organism experiences internal contextual influences similar to those experienced when the learning occurred. Such contexts can be pharmacologically manipulated by treating with cholinergic compounds, opiates, cocaine and amphetamines, and with neurodepressants such as ethanol and barbituates1. Worms become intoxicated by ethanol in a manner similar to that of most other organisms tested (see abstracts by A. Davies and H. Kim, this meeting). We have sought to study the mechanisms of ethanol-induced state-dependent learning in worms. Olfactory adaptation in C. elegans is a decrease in the chemotaxis response to an odorant as a result of prior exposure to the odorant2. We demonstrate a form of state-dependent learning in worms by pairing olfactory adaptation and ethanol administration. Ethanol does not interfere with olfactory adaptation, however, worms exposed to an odorant while being treated with ethanol will only show subsequent adaptation to the odorant if ethanol is again administered during chemotaxis testing. If the odorant is presented without ethanol during testing, the animals behave as nave animals and therefore fail to alter their behavior based on their previous experience or prior exposure to the odorant. Further, we demonstrate that the state-dependent effects of ethanol require normal dopaminergic function. The dopamine-defective cat-14 and cat-25 mutants are able to adapt to volatile odorants, however, they do not show state-dependency when they are adapted to volatile odorants while intoxicated by ethanol. These results suggest that C. elegans is capable of a form of learning and indicate a conserved role of dopamine in the modulation of behavioral responses to ethanol. 1 Izquierdo. in Neurobiology of Learning and Memory (eds Lynch, McGaugh, and Wienberger) 333 (Guilford, New York, 1984). 2 Shulz, Sosnik, Ego, Haidarliu and Ahissar (2000) Nature 403, 549, and references therein. 3 Colbert and Bargmann (1995) Neuron 14, 803. 4 Duerr et al. (1999) J. Neuroscience 19, 72. 5 Lints and Emmons (1999) Development 126, 5819.

Sean M O'Rourke et al. (2003) International Worm Meeting "Exploring the centrosome with proteomics and functional genomics."

Sean M O'Rourke, Bruce Bowerman

[International Worm Meeting, 2003]

The centrosome is a large eukaryotic organelle that organizes microtubules for cellular functions including setting up the mitotic spindle. Despite the important role of the centrosome, relatively little is known about its composition and regulatory mechanisms in metazoans. SPD-5 is a component of C. elegans centrosomes and contains multiple coiled-coil domains that may interact with other proteins thereby governing centrosomal structure (1). Because SPD-5 is thought to be a central component of centrosomes, it represents an excellent handle to pursue other genes and proteins that are also involved in centrosome function. We are taking two experimental routes to uncover additional centrosomal components and regulators.First, SPD-5 antibodies are being used to isolate protein complexes from embryonic extracts with the goal of purifying SPD-5-interacting proteins. Such proteins will be identified by mass spectrometry. To date, we have prepared embryonic extracts from N2 animals and we have successfully immunoprecipitated SPD-5 protein. Interestingly, two SPD-5 isoforms are apparent in our extracts, the significance of which remains to be determined. We are currently optimizing the purification procedure to obtain larger quantities of SPD-5 and interacting proteins. The interacting proteins we identify will be initially characterized by localization and RNAi inhibition studies.In the second route to understanding centrosome function, we are employing genome interaction screens to identify genes which, when knocked-down by feeding RNAi, alter the phenotypes of temperature-sensitive mutants. Building on the work of Jose Eduardo Gomes (see Gomes and Bowerman abstract), we have begun to use multiple-well plates for culturing worms and a robot to easily and accurately dispense the dsRNA-producing E. coli. We have performed a pilot semi-automated screen using the chromosome I library (2). Current results will be presented. In addition to examining spd-5, we are planning to perform genome-wide interaction RNAi feeding screens with 20 different temperature-sensitive mutants defective in several different processes. We envision generating a genetic interaction map that will be useful for determining how many different processes require any given gene. We anticipate that our use of many different sensitized genetic backgrounds will enable us to identify gene requirements missed in previous classical and genomic screens.(1) Hamill, D. R. et al. Developmental Cell 3, 673-684 (2002).(2) Fraser, A. G. et al. Nature 408, 325-330 (2000).

da Veiga Beltrame, Eduardo et al. (2021) International Worm Meeting "Single Cell Tools for WormBase"

Arnaboldi, Valerio, da Veiga Beltrame, Eduardo, Sternberg, Paul

[International Worm Meeting, 2021]

The number of single cell RNA sequencing (scRNAseq) publications has exploded in recent years, with over 1200 studies currently available and over 350 new studies in 2020 alone. This wealth of data presents new challenges and opportunities on how to integrate, query, and display results in ways that are useful and easy to use for scientists. Over 85% of scRNAseq studies use human or mouse samples, and the volume of data generated by these studies is so high that their integration and unified management represents a formidable engineering challenge in itself. However for other model organisms such as C. elegans, for which there are only on the order of a dozen scRNAseq studies in the literature, data integration and maintenance of tools covering most of the published data is manageable by a small team with simpler tools. This offers a fantastic opportunity for WormBase and other model organism databases that are part of the Alliance of Genome Resources (alliancegenome.org). We are yet to be engulfed by the tidal wave of high throughput single cell data that is swamping human and mouse communities, and can learn from their efforts to create tools that will be effective and useful for our communities. We are presenting an overview of two WormBase single cell tools that are currently in development: 1) A differential expression tool for selecting cell groups to compare and visualizing dynamically generated results. In the backend the scvi-tools framework (scvi-tools.org) is used for Bayesian decision theory to detect differentially expressed genes. 2) A framework for visualizing static data that takes in precomputed results in h5ad format, slices the data according to user selection, and returns a visualization. At the moment we have the following visualizations in development: - Heatmaps & dot plots to visualize mean gene expression across select genes and cell type within an experiment. - Ridgeline histograms to visualize individual gene abundances stratified by cell type and experiment. - Swarm plots to visualize expression of multiple genes across all cell types relative to one cell type. We want to hear your thoughts on these tools! Please come to chat with us during the poster session or email eduardo@wormbase with your thoughts. You can learn more about each tool the webpage and check their current development status at https://wormbase.github.io/single-cell/