Mapping the Dynamic Synaptome
Synapse-level imaging and analysis
In 1894, Santiago Ramón y Cajal first suggested that memories are formed by changes in synaptic connections, a view that is widely held by neuroscientists. Over the last 50 years, studies have examined the behavior of individual synapses to better understand synaptic plasticity of a single synapse. However, understanding how memories are encoded in the brain requires means to visualize collections of synapses in an intact, living organism at different points in time to allow us to generate a map of synaptic connections.
To this end, we have developed and applied a set of tools that enables direct observation and mapping of synapses. Specifically, we can map and measure synaptic changes in the brain of a living vertebrate, using high-contrast light sheet imaging of an endogenous postsynaptic density protein labeled with a GFP-coupled reporter. We use this approach, in conjunction with a new classical conditioning paradigm that we developed, to study changes in the microstructure of the pallium that occur during learning in the larval zebrafish, a species whose neuroanatomy has relevant simularities to that of a human.
Using these tools, we have conducted a study that shows that contrary to the common wisdom, memory formation due to classical conditioning is associated with generation of new synapses rather than changes in existing synapses. Counter to what one would expect, with learning, the total number of synapses actually decreases, and synapses are destroyed in one area of the brain and new ones are created in a different area of the brain.
Understanding how synapses change during learning could have broad implications. For example, new deep learning methods that simulate synapse change might be developed or me may gain a better understanding of aberrant memory formation that can lead to diseases such as post-traumatic stress disorder and addiction diseases.
A New Approach to Discovery
Our study is an exemplar of a highly-collaborative, multidisiplanary scientific exploration that required a team of biologists, computer scientists, and microscope developers from different laboratories working closely together on a daily basis. Traditional methods for defining experiments, collecting data and sharing results quickly broke down. We were only able to develop a new measurement technique, create a new behavioral training paradigm and collect and analyize data yielding new neuroscience results by taking a new approch to scientific discovery that placed an emphisis on data driven collaboration and reproducability.
Our entire investigation was conducted such that all data was Findable, Accessible, Interoperable, and Reusable (FAIR data principles) from the first experiment to these published results. As a consequence, preparing our data for publication took place over a period of days, not months as is often the case. The complete data behind every figure, back to the raw data obtained from the microscopes, analysis programs, and human interpretation are available in a structured, queriable, and downloadable format.
The tools and data on this site are the same ones we use on a daily basis during the lifetime of our investigation. Using them you can search, browse, interact and download all of the structured data, code and methods that were used to obtain our results.
- Complete Study Data
- All Public Data
- Main Results. Clicking on these links will take you to a figure and all data and code supporting the figure:
- Tail flick conditioning (TFC), a robust classical conditioning paradigm for larval zebrafish
- Spines sampled within a 14-16 dpf zebrafish brain tend to move less than 4 ¬µm in five hours
- Cells in the anterolateral pallium project dendrites within the region of synapse gain
- Total synapse number and distribution of PSD-95.FingR labeling intensity does not change with memory formation
- Phosphorylated ERK (pERK) staining shows increased anterolateral activity in learners exposed to the CS following training and in na√Øve fish exposed to the US
- PSD-95.FingR-GFP localizes to synapses in living zebrafish larvae
- PSD-95.FingR expression does not alter the learning ability of 14-16 dpf zebrafish larvae
- An algorithm for semi-automated identification of synapses from raw SPIM data
- Changes in synapse number for zebrafish exposed to control behavioral paradigms after TFC are not significant
- Distributions of lost and gained synapses in the pallium following TFC
- Intensity of synaptic labeling does not change with TFC
- Anterolateral cell population within the 14-16 dpf zebrafish brain projects out of the dorsal pallium compartment ventrally and posteriorly
- Rapid synapse turnover in larval zebrafish
- Synapse formation occurs preferentially in the ventrolateral half of the dorsal pallium in learner, but not nonlearner, fish
- Experimental setup and analysis of tail-flick conditioning (TFC), a behavioral paradigm that stimulates associative memory formation in zebrafish larvae
- Neurons within the anterolateral pallium respond to the CS in learner fish and the US in naïve fish
Overview of Synapse Data Model
The figure below shows the structure of the all of the data used to derive our results and the relationships between those data. Click on the text in the ovals to go to the different types of data. You can navigate to the connected elements from the location you select