Using expression profiles of Caenorhabditis elegans neurons to identify genes that mediate synaptic connectivity

Abstract

Synaptic wiring of neurons in Caenorhabditis elegans is largely invariable between animals. It has been suggested that this feature stems from genetically encoded molecular markers that guide the neurons in the final stage of synaptic formation. Identifying these markers and unraveling the logic by which they direct synapse formation is a key challenge. Here, we address this task by constructing a probabilistic model that attempts to explain the neuronal connectivity diagram of C. elegans as a function of the expression patterns of its neurons. By only considering neuron pairs that are known to be connected by chemical or electrical synapses, we focus on the final stage of synapse formation, in which neurons identify their designated partners. Our results show that for many neurons the neuronal expression map of C. elegans can be used to accurately predict the subset of adjacent neurons that will be chosen as its postsynaptic partners. Notably, these predictions can be achieved using the expression patterns of only a small number of specific genes that interact in a combinatorial fashion.

Figure 1. Context Specific Independencies Reduce the Complexity of the Model and Make It Easier To Interpret.

(A) A complete decision tree for the simplified example of a lock-and-key molecular identifiers mechanism: only when the presynaptic neuron expresses a lock molecule and the postsynaptic neuron expresses a key molecule, a synapse is formed between them. (B) A simpler decision tree that captures the same logic but exhibits context specific independence. In the context in which the lock is not expressed in the presynaptic neuron, the formation of a synapse between adjacent neurons is independent of the expression of the key in the postsynaptic neuron.

Figure 2. The Neural Network of the C. elegans Provides Examples for Learning the Patterns of Synaptic Wiring.

(A) A standard schematic of the worm's head (taken from Wormatlas [43]) with a network depiction of a part of C. elegans's neural network on the right side of the nerve ring. Neurons are in their real relative location (data taken from the authors of [44]). (B) An example of a neighborhood of one neuron. The neuron AIBL introduces all types of combinations of synaptic relations with other neurons. For each such combination one neuron has been chosen to demonstrate it. For example the neuron RIVL is the representative of the group of neurons that forms only electrical synapses with AIBL. Each cross on a synapse represents one more additional identical synapse that was observed. The neighborhood of a neuron is defined as the group of neurons that forms a synapse with it (chemical or electrical synapse in either direction). Neurons that are in the same neighborhood must be in spatial proximity in the worm's body. A positive example is created when a neuron “chooses” to be presynaptic to another neuron in its neighborhood and a negative example is created when a neuron “chooses” not to be presynaptic to another neuron in its neighborhood.

Figure 3. Summary of the Prediction Performance as a Function of the Maximal Depth of the Tree-CPD after 30 AdaBoost Iterations.

Standard deviation of the real data was calculated on 50 iterations of 5-fold cross validation, each time for a different division of the data to train and test sets. Standard deviation of the random models was calculated on 50 iterations of 5-fold cross validation, each time for a different shuffling of the data.

Figure 4. The Highest Confidence Rules That Were Learned in Bootstrap of Tree-CPDs.

The highest confidence rules that were learned in bootstrap of tree-CPDs of maximal depth of one (A), two (B), three (C), four (D), and five (E). The confidence of each rule is written in parentheses. (F) The final, most confident, tree-CPD for the chemical synapse. This tree was constructed by combining the rules from (A–E).