Learning via mechanosensitivity and activity in cytoskeletal networks

Area of Science:

• Biophysics
• Systems Biology
• Computational Biology

Background:

• The actomyosin cytoskeleton is crucial for cellular mechanics and function.
• Biological systems exhibit remarkable adaptability and homeostasis in response to environmental changes.
• Understanding learning mechanisms in biological systems is a key challenge.

Purpose of the Study:

• To investigate a minimal, biologically plausible learning mechanism based on actomyosin networks.
• To demonstrate how mechanosensitive proteins and molecular motors can enable learning from environmental perturbations.
• To explore the implications of this mechanism for cellular adaptation and homeostasis.

Main Methods:

• Development of a coarse-grained network model inspired by the actomyosin cytoskeleton.
• Implementation of a contrastive learning framework to train the network.
• Simulation of environmental perturbations to assess the network's learning capabilities.
• Analysis of the network's response in terms of adaptation and homeostasis.

Main Results:

• The actomyosin-inspired network successfully learned from environmental perturbations within the contrastive learning framework.
• The presence of mechanosensitive proteins and molecular motors was essential for the network's learning ability.
• The model reproduced key phenomenological aspects of biological adaptation and homeostasis.

Conclusions:

• Force-sensitive proteins and molecular motors provide a general strategy for learning in biological systems.
• A minimal, biologically plausible learning mechanism can be based on actomyosin dynamics.
• This framework offers insights into cellular adaptation and homeostasis.