Identifying reproducible resting state networks and functional connectivity alterations following chronic restraint stress in anaesthetized rats

Background: Resting-state functional MRI (rs-fMRI) in rodent models have
the potential to bridge invasive experiments and observational human studies,
increasing our understanding of functional alterations in the brains of patients
with depression. A major limitation in current rodent rs-fMRI studies is that there
has been no consensus on healthy baseline resting-state networks (RSNs) that
are reproducible in rodents. Therefore, the present study aimed to construct
reproducible RSNs in a large dataset of healthy rats and then evaluate functional
connectivity changes within and between these RSNs following a chronic restraint
stress (CRS) model within the same animals.
Methods: A combined MRI dataset of 109 Sprague Dawley rats at baseline and
after two weeks of CRS, collected during four separate experiments conducted
by our lab in 2019 and 2020, was re-analysed. The mICA and gRAICAR toolbox
were first applied to detect optimal and reproducible ICA components and then a
hierarchical clustering algorithm (FSLNets) was applied to construct reproducible
RSNs. Ridge-regularized partial correlation (FSLNets) was used to evaluate the
changes in the direct connection between and within identified networks in the
same animals following CRS.
Results: Four large-scale networks in anesthetised rats were identified: the DMNlike, spatial attention-limbic, corpus striatum, and autonomic network, which are
homologous across species. CRS decreased the anticorrelation between DMNlike and autonomic network. CRS decreased the correlation between amygdala
and a functional complex (nucleus accumbens and ventral pallidum) in the right
hemisphere within the corpus striatum network. However, a high individual variability
in the functional connectivity before and after CRS within RSNs was observed.
Conclusion: The functional connectivity changes detected in rodents following
CRS differ from reported functional connectivity alterations in patients with
depression. A simple interpretation of this difference is that the rodent response
to CRS does not reflect the complexity of depression as it is experienced by
humans. Nonetheless, the high inter-subject variability of functional connectivity
within networks suggests that rats demonstrate different neural phenotypes, like
humans. Therefore, future efforts in classifying neural phenotypes in rodents
might improve the sensitivity and translational impact of models used to address
aetiology and treatment of psychiatric conditions including depression.