Respiration-driven brain network: neural underpinning of breathing correlates with resting-state fMRI signal

Read the full article See related articles


Respiration can induce motion and CO 2 fluctuation during resting-state fMRI (rsfMRI) scans, which will lead to non-neural artifacts in the rsfMRI signal. In the meantime, as a crucial physiologic process, respiration that can directly drive neural activity change in the brain, and may thereby modulate the rsfMRI signal. Nonetheless, this potential neural component in the respiration-fMRI relationship is largely unexplored. To elucidate this issue, here we simultaneously recorded the electrophysiology, rsfMRI and respiration signals in rats. Our data show that respiration is indeed associated with neural activity changes, evidenced by a phase-locking relationship between slow respiration variations and the gamma-band power of the electrophysiologic signal recorded in the anterior cingulate cortex. Intriguingly, slow respiration variations are also linked to a characteristic rsfMRI network, which is mediated by gamma-band neural activity. In addition, this respiration-related brain network disappears when brain-wide neural activity is silenced at an iso-electrical state, while the respiration is maintained, further confirming the necessary role of neural activity. Taken together, this study identifies a respiration-related brain network underpinned by neural activity, which represents a novel component in the respiration-rsfMRI relationship that is distinct from respiration-related rsfMRI artifacts. It opens a new avenue for investigating the interactions between respiration, neural activity and resting-state brain networks in both healthy and diseased conditions.

Article activity feed

  1. Evaluation Summary:

    This paper will be of interest to researchers studying control of respiration and also those developing functional magnetic resonance imaging methodology. The work provides insight into the relationship between brain activity (measured directly) and non-invasive functional magnetic resonance imaging measures. The authors find that the respiration signal is associated with the gamma band in the cingulate cortex, and both the gamma signal and respiration signal correlate with distributed neuronal networks across the brain. This contributes to our knowledge of the contribution of respiration on neuro and neuro-vascular signals during resting conditions.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1 agreed to share their name with the authors.)

  2. Reviewer #1 (Public Review):

    The study by Tu and Zhang is very strong, from its technical implementation, the interesting question being addressed, and a clear presentation of the results. Indeed, the visual guides in the figures allow for easy navigation of the results and help the readers make his/her own inferences seamlessly. The quality of the MRI combined with electrophysiological recordings is excellent, as far as I can tell without looking at the data made available by the authors. The experiments and analysis follow a logical progression that makes sense. If any weakness is to be found, perhaps the authors overstep their inferences of respiration -> neuronal signal causality in the discussion.

  3. Reviewer #2 (Public Review):

    Functional magnetic resonance imaging is confounded by physiological artefacts such as the heart beat and respiration. Usually researchers remove such physiological signals in the analysis phase via various statistical methods. However, there is a body of research interest in defining higher cortical centres that control respiration - e.g. to understand how emotion influences breathing (and vice-versa) and to understand perceptions of breathlessness.

    This study was performed in 7 rats that had probes implanted in their anterior cingulate cortices, and underwent resting-state functional magnetic resonance imaging at various levels of sedation and anaesthesia. The main finding of the study was that respiration is associated with neural activity changes (in the anterior cingulate cortex). Additional findings include that respiration is linked to a characteristic resting state FMRI network, and that respiration-related activity disappears when an iso-electric state is induced with high dose barbiturates.

    The main strength of the paper is that it helps us better understand the resting-state functional magnetic resonance imaging signal (i.e. that there are specific respiratory related contributions to this).

    The main conclusion of the paper (that respiration is associated with cortical neural activity) is supported by work in awake humans with deep-brain stimulator electrodes, which does limit the novelty somewhat. Anaesthesia has profound effects upon brain physiology (i.e. neurovascular coupling) and on respiration - which may confound the findings. Finally the sample size of 7 rats is rather low. In conclusion, although the results support the authors' conclusions, there needs to be caution in interpreting this study as conclusive.

  4. Reviewer #3 (Public Review):

    This study investigates the neuronal correlates of low-frequency changes in respiration volume per unit time (RVT). The authors report distributed patterns of correlations between RVT and fMRI that may represent a respiration-driven brain network. The ability to demonstrate that this pattern has neuronal origins would make an important contribution to the fMRI field, especially as physiological signals are typically treated as artifacts in fMRI analysis.

    A major strength of this paper is the use of concurrent fMRI, physiological monitoring, and invasive electrophysiology (electrode in the anterior cingulate cortex; ACC) in the anesthetized rat, which allows for directly measuring local neuronal activity associated with changes in respiration. A second strength is that the authors demonstrate coherence between respiration (the raw signal as well as RVT) and gamma-band power in the ACC, and furthermore replicate prior findings of a close link between gamma-band power and the BOLD fMRI signal. The authors also take care to ensure that the pattern of correlation between RVT and fMRI is distinct from artifacts resulting from breathing-induced static field changes as well as from CO2-related effects of breathing on the BOLD signal. The findings are clearly presented throughout the paper.

    I believe that additional information would help to more strongly support the main claim, i.e., that the reported RVT-fMRI correlation pattern is of neuronal origin. One analysis that supports this claim is that in the lightly anesthetized state, regressing out the gamma-band power signal considerably reduced correlations between RVT and fMRI. However, the more direct test of this possibility involves the experiment in which neural activity across the brain is silenced (isoelectric state) while respiration is artificially maintained. The resulting disappearance of correlation between RVT and fMRI data points to the neuronal nature of RVT-fMRI correlation. Yet, since the amount of temporal variation in RVT during the iso-electric state was not reported, it was not clear whether RVT itself also exhibited less temporal variation in the isoelectric state. Since respiration was maintained by a ventilator in the isoelectric state, I wondered if the respiration depth and volume was more constant compared to in the lightly anesthetized state, in which it is mentioned that spontaneous respiration occurred. Importantly, the authors do mention that the respiration patterns were visually similar between these conditions (Fig. 5C and line 219), which is very promising, but quantification of RVT properties would be important to provide as well.