Skip to main content

Tracing memory formation in a behaving animal: analysis of learning-induced morpho-functional plasticity along the bee’s olfactory system

Periodic Reporting for period 1 - MEM-ENTO (Tracing memory formation in a behaving animal: analysis of learning-induced morpho-functional plasticity along the bee’s olfactory system)

Reporting period: 2020-04-01 to 2022-03-31

With a relatively small brain, honeybees are capable of advanced cognitive abilities, providing a useful model to investigate the neural correlates of sensory coding and memory formation. Moreover, anatomical and functional knowledge of its olfactory circuit allows investigating how experience impacts on the coding of an olfactory cue along different stages of the olfactory pathway and different intervals from the conditioning procedure.
The goal of this proposal was to perform high spatial and temporal resolution calcium imaging analysis within the same individuals to identify the neural correlates of short and long-term memory across the three main neuropils of the bee's olfactory system: the antennal lobe, the mushroom body, and the lateral horn. With this approach, I aimed to observe how the neural activity elicited by rewarded and unrewarded odorants is modified by associative learning in terms of response intensity, latency, and neural connectivity based on partial correlation analysis.
The project pipeline was seriously impacted by the Covid-19 pandemic in multiple aspects - including the access to laboratory facilities, and the interaction with the research group - as well as by the early termination of the fellowship.

Work completed included:

1) Implementation of an associative conditioning protocol to be performed on restrained bees at the imaging setup. To facilitate the olfactory learning procedure, I have tested the possibility to induce memory formation without the classical sugar reward (i.e. where the animal receives a drop of sugar to ingest upon odorant delivery), but with the sole stimulation of the antennae with a sugar-coated toothpick. Briefly, I performed a differential conditioning protocol, where five stimulations with 1-hexanol (or nonanal) were rewarded by a stimulation of the antennae with a 50% sucrose solution (CS+) and five stimulation with nonanal (or 1-hexanol) were not rewarded. After five trials, all tested animals learned to discriminate the two stimuli (Figure 1A-B) and after 1 or 24 hours they show specific memory for the conditioned stimulus even in absence of sugar feeding and comparable to the classical conditioning protocol (Figure 1C-D).

2) Selective labelling of antennal lobe projection neurons (Figure 1E) and of mushroom body Kenyon cells was optimized (Figure 1F).

3) Calcium imaging of antennal lobe projection neurons. One of the first goals of the project was to observe how the odorants' representation in the first processing centre, the antennal lobe, is altered by associative learning. I was able to complete a setup phase and the analysis pipeline. Briefly, bees were stimulated with three odorants up to 20 times and acquisition was conducted at high temporal resolution (130 fps) to observe fine changes in response dynamics, among which response latency, duration, response shape, frequency components (Figure 2). Preliminary experiments allowed me to set up all analytical tools, to assess that preparations were viable for up to 6h, with a stable neuronal activity and limited brain movements.

4) I developed the analytical tools to extract causality relationships from the activity of individual glomeruli. This is based on Granger causality, an information theory algorithm investigating the influence of the past timepoints of one node on the present points of the other nodes of the network. This allows determining if the activity of one node/glomerulus is influenced by the activity of other glomeruli (in presence and absence of an olfactory stimulation), thus providing a functional connectivity map across antennal lobe glomeruli.
This project was expected to investigate how memory influences sensory coding across different neuropils of the brain. The observation of neural activity before and after learning in the same animal would have allowed to pinpoint learning-related changes in neural activity. In addition, the adoption of high-temporal resolution multiphoton microscopy allows the quantification of stimulus-induced responses in the time frequency domain and to study the development of functional connectivity during a learning phase. These last two approaches are well-established for electorophysiology and EEG methods, but still under-exploited in the domain of calcium imaging, and would provide a new layer of information in the analysis of sensory coding. Due to the limitations mentioned in the previous paragraph, only the setup phase of the different work packages was accomplished.
Figure 2. Calcium imaging analysis pipeline.
Figure 1. Olfactory conditioning and neuronal labelling.