Mechanistic Connectomics in Drosophila

Mechanistic Connectomics uses synaptic-resolution connectomes in a causal manner to explain circuit function. Multiple connectomes now exist for the adult fruit fly, \textit{Drosophila melanogaster}. This thesis develops statistics and analyses for Mechanistic Connectomics using these datasets to analyse the circuit architectures that guide the interactions between innate and learned behaviours. Creating meaningful comparisons between connectomes, however, requires an identification of common elements within brains and across brains, enabling open questions on cell type stereotypy to be addressed. This thesis contains major cell typing efforts to identify cell types for ~3600 neurons in the Lateral Horn (LH) and extrinsic neurons of the Mushroom Body (MB). In \textit{D. melanogaster}, innate sensory processing and learning and memory occur in the LH and MB, respectively. This cell typing effort revealed strict stereotypy in cell type membership for the MB, but greater variation in cell type membership for the LH cell types. To leverage this cell typing effort for future connectomes, a machine learning method was developed to learn the structure of these neurons/cell types. The input structure into the MB was then analysed. Random connectivity into the MB is thought to optimise memory formation. However, a random connectivity model does not accurately describe how second-order olfactory projection neurons (PNs) connect onto Kenyon Cells (KCs), the intrinsic neurons of the MB: there are two subsets of PNs with random connectivity, but two PN populations with over- and under-represented connectivity. This finding is replicated within and between brains, and is also present between the sexes. PNs that encode food odours (food-PNs) are more likely to connect onto a given KC. The micro-circuit connectivity from PNs onto KC dendrites was then analysed through electrotonic modelling of KC dendrites. These analyses reveal that KC claw numbers vary significantly across KC sub-types, and that multiple PNs can target a single KC claw. Food-PNs connect to more single claws and multiple claws. These connectivity biases represent an innate constraint of the circuit architecture present in a learning and memory centre. How does odour information flow through the MB? Effective connectivity analyses demonstrate that food-odour encoding PNs have stronger effective connectivity onto approach-promoting MBONs. Aversive-odour encoding PNs have stronger effective connectivity onto avoidance-promoting MBONs. The output connectivity from the MB onto a novel class of neurons, called Lateral Horn Centrifugal Neurons (LHCENTs), reveals specific streams of experience-guided output. Neurotransmitter predictions reveal that the majority of LHCENTs are GABAergic, confirmed through split-GAL4 screens for lines targeting LHCENTs and immunohistochemistry analyses. Importantly, several of these neurons provide recurrent inhibitory feedback onto KC dendrites in the MB and LH neurons. LHCENTs co-target the same KCs that food-PNs connect to, revealing a compensatory circuit architecture for biased PN-KC connectivity. Causal modelling reveals that this inhibitory feedback further de-correlates predicted KC activity. Finally, one neuron called LHMB1 has the opposite polarity to LHCENTs: LHMB1 is dendritic in the MB and LH. This glutamatergic neuron is the shortest path from PNs to motor output, providing an excitatory shortcut for the promotion of approach-behaviours. In summary, this thesis advances Connectomic analyses, whilst specifically addressing cell type stereotypy and the circuit organisation of neurons that integrate innate and learned behaviours.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2023

Author dc:creator

• Pleijzier, Markus

Advisors dc:contributor.advisor

• Jefferis, Gregory
• Zlatic, Marta
• O'Leary, Timothy

• Connectomics
• Drosophila
• innate & learned behaviours
• mushroom body

Licence

• https://apollo8-f-pro.lib.cam.ac.uk/bitstreams/419d5a62-091d-4c3a-afac-d11260864b8c/download
• http://purl.org/NET/rdflicense/allrightsreserved

Language dc:language

eng

Author Identifier

0000-0002-7297-4547

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/379433