Jingting Liang Thesis Defense (Yun Zhang, Advisor)

Title: Brain-wide neural dynamics for aversive olfactory learning in C. elegans

Abstract:  Associative learning allows animals to modify behavior based on prior experience, yet how learned experience is encoded across brain-wide neural dynamics remains unclear. In C. elegans, naïve animals are attracted to the pathogenic bacteria Pseudomonas aeruginosa PA14, but reduce their preference for PA14 after 4-6 hours of exposure, providing a tractable model for systematically dissecting how experience reshapes neural activity across the entire nervous system. In my thesis, I combine brain-wide calcium imaging with behavioral analysis, neuronal manipulation, singular value decomposition and modeling, network modularity analysis, and genetic perturbation to investigate how aversive learning reorganizes neural activity in C. elegans. In Chapter 1, I describe the development of a brain-wide calcium imaging pipeline using multicellular promoter labeling, microfluidic odor delivery, and cell-type-resolved imaging to record neural responses to bacterial odor stimulation patterns in the discrimination task (OP50-PA14) and the detection task (PA14-Buffer) aligned with behavioral preference assays. To improve automated cell identification in multicellular images, I collaborated with Hyun Jee Lee to develop CRF_ID 2.0, an algorithm with neuron-identification accuracy similar to that of human annotators. In Chapter 2, I identify learning-modulated neurons whose odor-evoked activity changes significantly after PA14 training. I find that aversive training produces widespread, context-gated modulation of neural responses across sensory neurons, interneurons, and motor neurons, suggesting that learning and memory are encoded across distributed neuron types. Neuronal manipulation experiments further support functional roles for candidate learning-modulated neurons, including ASEL, URY, and OLL, in learned behavior. In Chapter 3, I examine training-dependent changes in population dynamics that are revealed by brain-wide recordings. Together with my collaborators Sahil Moza and Sihoon Moon, who performed temporal component analysis, manifold analysis, network modularity analysis, and fixed-point analysis, we show that learning reorganizes population dynamics through changes in temporal structure, functional network modularity, neural trajectory rotation and contraction, and shifts of PA14-evoked fixed points toward preexisting buffer-like states. In Chapter 4, I further investigate the mechanism underlying this learning-dependent reorganization and find that INX-7-mediated electrical synapses regulate aversive learning and brain-wide activity reorganization. Loss of inx-7 disrupts learned behavior and impairs training-dependent changes in network modularity and neural manifold geometry from sensory-inter group I neurons, while many sensory responses remain relatively preserved. Collectively, in the thesis I present a brain-wide imaging and analysis framework for studying learning in C. elegans and reveal that aversive olfactory learning is associated with INX-7-mediated, context-gated reorganization of distributed neural dynamics across the nervous system.

Committee:  Yun Zhang (Advisor), Bence Ölveczky, Mansi Srivastava (Chair), Venkatesh Murthy