Cortical silencing results in paradoxical fMRI overconnectivity

Carola Canella, Federico Rocchi, Shahryar Noei, Daniel Gutierrez‐Barragan, Ludovico Coletta, Alberto Galbusera, Stefano Vassanelli, Massimo Pasqualetti, Giuliano Iurilli, Stefano Panzeri, Alessandro GozzibioRxiv 2020 · DOI

TL;DR

This study challenges the common assumption that functional connectivity (as measured by fMRI) directly reflects anatomical connections. By silencing specific brain regions (the prefrontal cortex) in mice, researchers found that instead of decreasing, functional connectivity paradoxically increased. This suggests that low-frequency brain rhythms, rather than direct connections, play a crucial role in shaping large-scale brain network dynamics.

What Problem Does This Paper Address?

For years, researchers have used resting-state functional Magnetic Resonance Imaging (rsfMRI) to map brain networks. The prevailing interpretation is that areas showing correlated activity are directly connected anatomically. This makes intuitive sense: if two brain regions are talking to each other, they likely have a physical pathway. However, this interpretation has been challenged by observations of increased functional connectivity in conditions where brain damage is present, which seems counterintuitive. The authors aimed to causally test the relationship between direct neural connections and rsfMRI-based functional connectivity. They wanted to understand if silencing a brain region reduces connectivity (as predicted by structural models) or if other mechanisms are at play.

Methods

The researchers employed a multi-pronged approach combining advanced techniques:

  • Chronic Neural Silencing (Kir2.1): They used a viral vector to express the potassium channel Kir2.1 in the mouse prefrontal cortex (PFC). Kir2.1 is known to reduce neuronal excitability, effectively "silencing" the targeted neurons for an extended period.
  • Acute Neural Silencing (Chemogenetics/DREADDs): They used Designer Receptors Exclusively Activated by Designer Drugs (DREADDs). This involves introducing a modified receptor into PFC neurons that can be activated by a specific drug (clozapine-N-oxide, CNO). When CNO is administered, it inhibits neuronal activity in a time-controlled manner.
  • Resting-State fMRI (rsfMRI): This is the standard technique used to measure brain activity when a subject is at rest. By analyzing the spontaneous fluctuations in the BOLD signal, researchers infer functional connectivity between different brain regions.
  • In Vivo Electrophysiology: This involved implanting electrodes in the brains of mice to directly record neural activity (firing rates and local field potentials - LFPs) at a much finer temporal and spatial scale than fMRI. This allowed them to investigate the underlying neural mechanisms driving the fMRI signals.

Datasets

Not discussed in the paper. The study utilized data collected from experiments performed on mice.

Key Insights and Ideas

  • Paradoxical Overconnectivity: The most striking finding is that silencing a brain region (PFC) led to increased functional connectivity, not decreased connectivity as predicted by structural models. This directly challenges the simple "more connections = more connectivity" assumption.
  • Delta Band Dominance: Electrophysiological recordings revealed that while high-frequency activity (gamma) decreased with silencing, low-frequency delta band activity significantly increased and became more synchronized between brain regions. This suggests that slow oscillations, rather than fast ones, are key drivers of the observed rsfMRI overconnectivity.
  • Thalamus as a Hub: The study points to the polymodal thalamus as a critical relay station for these slow oscillations, propagating the synchronized delta activity to widespread cortical areas and contributing to the observed overconnectivity.
  • Rethinking Functional Connectivity: The findings necessitate a revision of how we interpret rsfMRI. It's not just about direct anatomical links but also about the influence of large-scale, low-frequency network dynamics.

Key Results

  • Chronic silencing of the PFC using Kir2.1 resulted in significantly increased rsfMRI connectivity in regions like the posterior cingulate/retrosplenial cortex and centromedial thalamus.
  • Acute chemogenetic silencing of the PFC with DREADDs replicated these findings, showing increased rsfMRI connectivity in similar DMN regions.
  • Electrophysiological recordings showed that PFC silencing led to reduced gamma activity and increased delta power in the silenced area.
  • Crucially, there was a robust increase in delta band coherence between the PFC and overconnected regions, which strongly correlated with the observed rsfMRI overconnectivity.
  • The study found that the observed overconnectivity was not due to vascular or hemodynamic confounds.

Limitations and Future Work

  • Animal Model: The study was conducted in mice, and while it provides crucial mechanistic insights, direct translation to human brain function requires further investigation.
  • Sedation: The use of light sedation in the animals might influence brain states and arousal levels, potentially affecting rsfMRI signals. The authors acknowledge this and suggest future studies could explore this further.
  • Generalizability: While the study focused on the PFC, future work could investigate whether similar paradoxical overconnectivity occurs when silencing other brain regions.
  • Clinical Relevance: The authors suggest their findings offer a new network-level mechanism for understanding rsfMRI overconnectivity observed in neurological disorders like stroke and Alzheimer's disease, opening avenues for future research in these areas. They propose that this overconnectivity might reflect a shift in the balance between direct communication and global low-frequency synchronization, particularly in early stages of neurodegeneration.

Summary updated

Abstract

ABSTRACT fMRI-based measurements of functional connectivity are commonly interpreted as an index of anatomical coupling and direct interareal communication. However, causal testing of this hypothesis has been lacking. Here we combine neural silencing, resting-state fMRI and in vivo electrophysiology to causally probe how inactivation of a cortical region affects brain-wide functional coupling. We find that chronic silencing of the prefrontal cortex (PFC) via overexpression of a potassium channel paradoxically increases rsfMRI connectivity between the silenced area and its thalamo-cortical terminals. Acute chemogenetic silencing of the PFC reproduces analogous patterns of overconnectivity, an effect associated with over-synchronous fMRI coupling between polymodal thalamic regions and widespread cortical districts. Notably, multielectrode recordings revealed that chemogenetic inactivation of the PFC attenuates gamma activity and increases delta power in the silenced area, resulting in robustly increased delta band coherence between functionally overconnected regions. The observation of enhanced rsfMRI coupling between chemogenetically silenced areas challenges prevailing interpretations of functional connectivity as a monotonic index of direct axonal communication, and points at a critical contribution of slow rhythm generators to the establishment of brain-wide functional coupling.
BibTeX
@misc{canella2020cortical,
  author = {Carola Canella and Federico Rocchi and Shahryar Noei and Daniel Gutierrez‐Barragan and Ludovico Coletta and Alberto Galbusera and Stefano Vassanelli and Massimo Pasqualetti and Giuliano Iurilli and Stefano Panzeri and Alessandro Gozzi},
  title = {{Cortical silencing results in paradoxical fMRI overconnectivity}},
  year = {2020},
  doi = {10.1101/2020.08.05.237958},
  howpublished = {bioRxiv},
}
Figure