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

Respiration can induce motion and CO2 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.


54
Resting state fMRI (rsfMRI), which measures spontaneous blood-oxygen-level 55 dependent (BOLD) signal, is a powerful tool for non-invasively investigating brain-wide 56 functional connectivity [1][2][3] . Due to its hemodynamic nature, the rsfMRI signal is 57 susceptible to systemic physiological changes such as respiration and cardiac 58 pulsations 4-7 , and these effects are usually treated as non-neuronal artifacts in rsfMRI 59 data. 60 Respiration is a major physiological process that drives fluctuations in cerebral 61 blood flow and oxygenation 5 . Respiration can affect the BOLD signal 8-10 with two types 62 of effects resulting from slow respiration variations and fast cyclic changes, respectively. hippocampus, and this effect has been consistently found across species [26][27][28][29] . In 77 addition, respiration changes can be associated with arousal and/or emotion-related 78 brain state changes, which covary with cortical activity [30][31][32][33] . Therefore, in addition to the 79 artifactual effects aforementioned, respiration may affect the rsfMRI signal by directly 80 modulating the neural activity. However, this potential neural component in the 81 respiration-fMRI relationship is largely unexplored.

82
To gain a comprehensive understanding of the relationships between respiration, 83 neuronal activity and rsfMRI signal, here we simultaneously acquired rsfMRI, 84 electrophysiology and respiration data in anesthetized rats. Anesthesia was used to 85 ensure our results are not confounded by the animal's motion, which affects all three 86 signals. Based on these measures, an RVT-correlated rsfMRI network was identified. 87 Importantly, regressing out gamma activity or silencing neural activity across the brain 88 disrupted this respiration-related network, suggesting that this respiration-rsfMRI 89 relationship is mediated by neural activity.

92
To determine the potential role of neural activity in the respiration-rsfMRI 93 relationship, we simultaneously recorded the electrophysiology and respiration signals 94 along with rsfMRI data in rats (Fig. 1A). The respiration signal was recorded by a 95 respiration sensor placed under the animal's chest (Fig. 1A). Representative raw 96 respiration signal, the respiration rate distribution across all scans, and the averaged    Given that the gamma band power is respectively associated with the RVT and 178 rsfMRI signals, we specifically asked how RVT is related to the rsfMRI signal in lightly 179 sedated animals by calculating voxel-wise correlations between the RVT and rsfMRI 180 signals (Fig. 4A). This analysis generates a respiration-related rsfMRI network (Fig. 4B), 181 involving key brain regions controlling respiration such as the piriform cortex. It is known 182 that the piriform cortex receives inputs from the olfactory bulb, and can be directly The respiration-related rsfMRI network is absent at the isoelectric state 203 To further confirm the necessary role of neural activity in the respiration-related 204 rsfMRI network, we experimentally silenced the neural activity in the whole brain by 205 inducing an isoelectric brain state using high-dose sodium pentobarbital, while

248
Respiration-related brain network is distinct from respiration-related rsfMRI artifacts 249 To confirm the respiration-driven neural network we observed is different from Importantly, this breathing-related brain network might be altered in brain disorders, and 294 thus our finding might potentially provide important clinical value.

296
The respiration network is very likely linked to respiration-entrained brain-wide neural generally believed to relate to the BOLD signal 37,42 , which is also demonstrated in our 306 study (Fig. 3). These results suggest that respiration can drive gamma oscillations, 307 which further lead to BOLD signal changes in the respiration network.

308
In addition to the mPFC, previous work has shown that respiration-generated         A. an example of raw electrophysiology signals before the denoising of MRI artifacts; B. an example of the MRI artifact template. C. LFP after MRI artifact denoising.