Fig 1.
Coupled neuronal activity, gGM, PFI, and CSF fMRI signals during burst-suppression anesthesia.
(a) top: Experimental design: fMRI recording under deep sevoflurane anesthesia in a Philips Achieva 3.0T scanner. Bottom: Example scan positioning in a representative subject (#5). Schematic depiction of the fMRI volume superimposed on a sagittal T1-weighted image with representative slice positions for the gGM and PFI masks and slice 1, which contains the CSF mask. (b) Simultaneous recordings of the EEG with an MRI-compatible, 64-electrode cap with equidistantly arranged electrodes (Easycap) and of three fMRI masks, depicted in representative slices: gGM (white), PFI (blue), and CSF (red). (c) Simultaneously recorded signal time courses from one subject under burst-suppression anesthesia: EEG (green), gGM fMRI (black), PFI fMRI (blue), and CSF fMRI (red). Suppression and burst periods are indicated in light blue and light orange, respectively. (d) Subject- and event-averaged time courses of all transitions from suppression to burst (n = 22 events from 17 subjects). gGM fMRI (black), PFI fMRI (blue), and CSF fMRI (red). The shaded areas represent the standard error of the mean (SEM). Suppression and burst periods are indicated in light blue and light orange, respectively. The y-axis on the right contains y-values for the PFI and CSF fMRI signal. (e) Same as (d) for transitions from burst to suppression (n = 21). (f) Event-averaged time courses of d/dt gGM fMRI, -d/dt PFI fMRI, and -CSF fMRI, normalized to 100% for suppression-burst events. (g) Event-averaged time courses of -d/dt gGM fMRI, d/dt PFI fMRI, and CSF fMRI, normalized to 100% for burst-suppression events. (h) Scatterplot of the d/dt PFI and the CSF amplitude. Crosses represent transition events; the line depicts linear regression. (i) Same as (h) for all burst-suppression events. Source data for Fig 1h and 1i are available in the source data file (S1 Data).
Fig 2.
Bidirectional grey-matter CBF changes induced by a hypercapnic challenge.
(a) Schematic depiction of the experimental design: a gas mixture of CO2 and medical air was applied over a medical mask in awake subjects (top). Schematic time course of the inspiratory CO2 concentrations of the applied gas mixture, including two 180 s segments with elevated CO2 levels (5% vol/vol). Transition periods lasted ~ 30 s due to the ramp time of the gas mixer. (b) Example scan positioning for the pCASL imaging volume (red) and labeling plane (yellow) in a representative subject superimposed on a sagittal T1 image (left) and a sagittal reconstruction of a 3D-phase contrast angiogram (right). (c) Whole brain cerebral blood flow (CBF) maps (top) and extracted subject-average (n = 17) time course of gGM CBF (bottom). Mean (black) ± SEM (grey). CO2 application periods are color-coded in purple. (d) Subject-average maps of CBF changes for transitions from normocapnia to hypercapnia (N → H, left) or from hypercapnia to normocapnia (H → N, right). (e) Global grey matter ∆CBF values for N → H (n = 34, left) or H → N (n = 34, right) transitions (2 for each subject). One-sample t test (individual groups). ***p < 0.001. Source data for Fig 2e are available in the source data file (S1 Data).
Fig 3.
Coupled gGM and CSF fMRI signal changes induced by a transient hypercapnia challenge.
(a) Global grey matter (left) and CSF masks (right) for extracting fMRI signals, superimposed on axial T1-weighted images. (b) Time courses of the gGM (top, black) and CSF (bottom, red) fMRI signals. Mean (solid lines) ± SEM (shaded areas) of n = 17 subjects. CO2 application periods are color-coded in purple. (c) Averaged time courses of all normocapnia to hypercapnia (N → H) transitions (n = 34 transitions from 17 subjects). gGM (black) and CSF fMRI signal (red). The shaded areas represent SEM. CO2 application periods are color-coded in purple. (d) Same as (C) for transitions from hypercapnia to normocapnia (H → N, n = 34 transitions from 17 subjects). (e) Correlation coefficients between gGM and CSF fMRI signals during N → H (n = 17, left) and H → N (n = 17, right) transitions. *p < 0.05, n.s. not significant. One-sample t test (individual groups). Two-sample t test (between groups). Source data for Fig 3e are available in the source data file (S1 Data).In summary, these results demonstrate that H → N and N → H transitions, respectively, induce total CBV changes, which underpin gGM fMRI signal changes coupling inversely with CSF signal changes in the basal cisternae of the brainstem.
Fig 4.
Changes in global brain blood volume mediate CSF in- and outflux.
(a) Representative example of the positioning of the most caudal fMRI slices on a sagittal T1-weighted image. The CSF voxels of slice 1 are indicated in orange, slice 2 in pink, and slice 3 in green; the dashed lines are schematic depictions of the respective imaging slices. (b) Average time courses of all transitions from hypercapnia to normocapnia. gGM (black), CSF fMRI signal in slice 1 (orange), slice 2 (pink) and slice 3 (green). The shaded areas represent SEM. Hypercapnia and normocapnia periods are indicated in purple. *p < 0.05, repeated-measures ANOVA (between slices) per subject (n = 17). (c) Average time courses of all transitions from burst to suppression. Global GM (black), CSF fMRI signal in slice 1 (orange), slice 2 (pink) and slice 3 (green). The shaded areas represent SEM. Burst and suppression periods are indicated in light orange and light blue, respectively. *p < 0.05, repeated-measures ANOVA (between slices) per event (n = 21). (d) Mean fMRI-based CSF signal intensities (a.u.) per event of slices 1-3 for steady-state suppression or steady-state burst periods. ** p < 0.005, ***p < 0.001, n.s. not significant. Repeated measures ANOVA with Dunn-Sidak post-hoc comparison. (e) Same as (b) for all transitions from normocapnia to hypercapnia. *p < 0.05, repeated-measures ANOVA (between slices) per subject (n = 17). (f) Same as (C) for all transitions from suppression to burst. ** p < 0.005, ***p < 0.001, repeated-measures ANOVA (between slices) per event (n = 22). (g) Schematic model of the link between neuronal activity (blue), CBV (black), CSF flux (↓ - efflux from the acquisition volume, ↑ - influx into the acquisition volume), fMRI-based CSF signal in slice 1 (orange) and slice 2 (pink). The pie charts symbolize the saturation state of CSF, being either fully saturated (white) or unsaturated (black); the rectangles indicate the signal intensity in the respective slice, which results from the mixing of RF-saturated and fresh CSF. Source data for Fig 4b–4f are available in the source data file (S1 Data).