Extracellular Recordings Unveil Hidden Patterns of Neuronal Connectivity

Sensations and perceptions arise due to a complex interplay of neuronal connections, each with varying strength. These communication patterns form the “connectome”, which has been examined in numerous studies to further understand and quantify neuronal computations. While many of these networks have been well documented in the literature, scientists from the University Hospital Bonn (UKB), the University Medical Center Mainz and the Ludwig-Maximilians-University Munich (LMU), and a research team from the Max Planck Institute for Brain Research in Frankfurt uncovered meaning behind seemingly random neuronal connections.

“Connectomics” describes researchers’ active effort to generate a map detailing approximately 86 billion neurons within the brain. The connections’ strength between individual neurons has been shown to be essential for learning and cognitive performance. Since there are numerous neuronal networks within the brain, each connected by thousands of synapses, it has been challenging for researchers to differentiate meaningful signal from noise. 

“…each synapse is unique and its strength can vary over time. Even experiments that measured the same type of synapse in the same brain region yielded different values for synaptic strength. However, this experimentally observed variability makes it difficult to find general principles underlying the robust function of neuronal networks,” says senior author Prof. Tatjana Tchumatchenko, research group leader at the Institute of Experimental Epileptology and Cognitive Research of the UKB and at the Institute of Physiological Chemistry of the University Medical Center Mainz.

To begin their work, the team recorded visual stimuli transmitted by the eye via the thalamus, a switching point for sensory inputs in the diencephalon, and examined the activity of active neurons. To do this, the researchers measured the joint response of two neuronal classes to different visual stimuli in a mouse model. They also used mathematical models and the “stabilized supralinear network” (SSN) to predict the strength of these connections.

“[The SSN] is one of the few nonlinear mathematical models that offers the unique possibility to compare theoretically simulated activity with actually observed activity,” says Prof. Laura Busse, research group leader at LMU Neurobiology. “We were able to show that combining SSN with experimental recordings of visual responses in the mouse thalamus and cortex allows us to determine different sets of connection strengths that lead to the recorded visual responses in the visual cortex.”  

While the measured synaptic strength was variable, the team uncovered hidden order beneath the noise. For example, the connections from excitatory to inhibitory neurons were always observed as the strongest, while the reverse connections in the visual cortex were weaker. While the absolute values of synaptic strengths varied in their models, they also maintained a specific order, suggesting that the relative ratios are crucial for the course and stability of measured activity. “It is remarkable that analysis of earlier direct measurements of synaptic connections revealed the same order of synaptic strengths as our model prediction based on measured neuronal responses alone,” says Simon Renner, Ph.D., of LMU Neurobiology.

“Our results show that neuronal activity contains much information about the underlying structure of neuronal networks that is not immediately apparent from direct measurements of synapse strengths. Thus, our method opens a promising perspective for the study of network structures that are difficult to access experimentally,” states Nataliya Kraynyukova, Ph.D., from the Institute of Experimental Epileptology and Cognitive Research of the UKB and Max Planck Institute for Brain Research in Frankfurt. 

The interdisciplinary team recently published their findings in PNAS.