Research Collaborations

Bioelectricity – a fundamental phenomenon underlying key functions of organisms at various levels – demonstrates extraordinary robustness to extensive variations in values of its many seemingly independent parameters. What makes this phenomenon particularly attractive for mechanistic analysis is that, unlike most other biological processes, the physics and chemistry of it are deeply understood, ab initio. Prof. Marom applies closed-loop methodological designs as means for both system identification and modelling of bioelectrical phenomena in proteins, cells and networks. At the network level, the labs of Dr. Eytan, Prof. Marom and Prof. Ziv, study large networks of bioelectrically active nerve cells (neurons) isolated from rodent brains. When many such neurons are placed together and given appropriate environmental conditions, they extend processes – axons and dendrites – form numerous synaptic connections, and develop complex patterns of activity. The groups of Eytan, Marom and Ziv apply and develop methodologies for simultaneous recording, stimulation and imaging of many individual neurons within such networks, which allow for continuous, long-term (weeks) concomitant monitoring of neuronal activity, neuronal structure, synaptic sizes and synaptic distributions. Using these approaches, these groups analyze the structure and dynamics of these networks, their sensitivities to developmental histories, environmental challenges, enforced constraints, chemical and physical modulators and structured input, the dynamics and variance of microscopic variables within the networks (e.g., synaptic sizes and single neuron activities), and how these give rise to stable, high level functional capacities such as input classification, adaptation and learning. Together with Profs. Meir, Brenner and Barak, attempts are made to develop principled understanding of such processes using concepts from dynamical systems, statistical mechanics, control theory and machine learning. Dr. Eytan’s group attempts to transfer these insights to the level of the intact brain using non-invasive electrical recordings in healthy and critically-ill subjects as well as searching for corollaries to these phenomena in other physiological systems.

In the lab of Prof. Braun, experiments focus on morphogenesis— the emergence of form and function in a developing organism, which is one of the most remarkable examples of pattern formation in nature. Morphogenesis provide a prime example for naturally occurring complex, far from equilibrium, processes showing the emergence of order from the underlying microscopic disorder. The experiments focus on the dynamic interplay of three type of processes underlying morphogenesis: Biochemical, mechanical and electrical which span all scales from the molecular to the entire organism. Hydra, a small multicellular fresh-water animal exhibiting remarkable regeneration capabilities, is utilized as a model system to: 1. Study mechanical forces and feedback that shape the body plan during morphogenesis (in collaboration with Prof. K. Keren, Physics Department). 2. Utilize an external electric field to stimulate electrical processes, attempting to modulate the course of morphogenesis in a controlled manner and study the interactions between electrical, mechanical and biochemical processes and their coordination in morphogenesis. Overall, the methodology and phenomenology exposed in these experiments offer a unique access for studying the physics underlying morphogenesis, one of the most fundamental processes in living systems. 

The characteristics of gene regulatory networks suggest a fundamental analogy between them and learning neural networks in their ability to create novel behavior in the face of change, while maintaining their functionality. Essential to both types of systems is a closed-loop relational dynamic with their environment. Professors Naama Brenner and Omri Barak study complex, high-dimensional dynamical systems with feedback from their environment to understand common themes and differences between the two classes of biological systems. For example, gene regulatory networks have heterogeneous structures including large hubs - master regulators - with large effect over the networks. They have shown that this structure is important for allowing a primitive form of learning under global feedback in such networks. In a collaboration between two of our theory groups – theoretical neuroscience and theoretical biophysics – we have found that these hubs can be mapped to feedback circuits and induce low-dimensional structures, similar to those that emerge in trained artificial neural networks. This work combines engineering approaches of systems theory with statistical mechanics of networks, shedding new light on the relation between structure and dynamics of complex networks. 

They strive to test the applicability of concepts from learning theory (studied in the group by Professors Soudry and Meir) to cell biology, and to uncover their potential biological functionality. In another joint project of the two theoretical groups, we aim to identify learning features in cancer cells adapting to novel environments as they colonize new tissue to create metastases. The statistical and dynamical nature of this process suggest that such epigenetic learning might be at play, and has been largely overlooked among the known cell-biological mechanisms supporting metastasis.

More generally, our theory group (Barak, Brenner, Meir and Soudry) aims to formulate a theory of learning that addresses both the commonalities and the differences between the various forms of learning in biology. In the Center, we study learning in single cells, in the brain and in machine learning settings. This has allowed us to detect common themes that arise in all cases. For instance, we showed that implicit feedback loops emerge through learning in all systems, and we are studying their implications for network function. On the other hand, we study the differences between these systems.