Social stress induced neural network reconfiguration affects decision making and learning in zebrafish

In many social species, behavioral mechanisms of how social hierarchies formed and maintained have been studied extensively. However, the neural bases underlying behavioral decisions and dynamics of neural circuits that permit animals to adapt to changes in social rank are poorly understood. In this study we focused on two social stress induced behaviors in zebrafish [the Mauthner cell (M-cell) mediated startle escape response and swimming behavior] to investigate how social regulation affects intrinsic cellular and network properties that result in the behavioral differences between dominant (DOMs) and subordinate (SUBs) animals. We utilized a non-invasive technique that allowed us to monitor the activation pattern of the two neural circuits in freely behaving animals. High behavioral responsiveness and a low stimulus threshold for the initiation of escape in M-cell were observed in SUBs while DOMs showed the quicker habituation to repeated auditory stimulation compared to SUBs. We also observed that on average SUBs generated significantly less number of swim bursts compared to DOMs. These results suggest that social status induced stress can modify the startle plasticity as well as the local swimming circuit. The change in M-cell’s excitability due to the change in the presynaptic inhibitory drive may be responsible for the lowered threshold. On the other hand, the local neural circuits and their intrinsic modulatory components (motor neurons and interneurons) may be configured differently according to social status to produce status-dependent swim patterns. To test these ideas, we developed a biologically-based mathematical model whose network architecture is based on recent experimental data. The model is able to reproduce several hallmarks of social status induced behavioral differences that were experimentally observed between DOMs and SUBs, as well as some inherent activity patterns. Changing some intrinsic synaptic and network parameters was sufficient to obtain the transition between DOMs and SUBs activity patterns while maintaining the network architecture. Recent experiments show that the startle plasticity in M-cell can be modulated by endocannabinoids, 2-AG. We chose the availability of 2-AG in M-cell as one of main parameters for the simulation, whose dynamics is governed by the intracellular calcium level in M-cell. Model simulation shows that high behavioral responsiveness in SUBs results from the increased excitability in M-cell, which can be interpreted as the reduced inhibitory input to M-cell. To reproduce less swimming activity in SUBs, the hallmark of social status induced behavioral difference observed in our experiments, we chose another intrinsic parameter, the availability of 2-AG in inhibitory interneurons to represent 2-AG modulated local network property. Model simulation shows that less swimming activity in SUBs is produced by the increased inhibitory input to the swimming neural circuit via the 2-AG driven elevated interneuron activity.