Ultrafast venous and sagittal sinus constrictions in the brain driven by abdominal pressure

Abstract

Nearly all the blood supplying the cortex exits via the bridging veins (BVs) that drain into the superior sagittal sinus (SSS), making these vessels key chokepoints for cerebral blood flow. Using optical imaging in head-fixed mice, we found that the SSS, BVs, and some other pial veins exhibit ultrafast constrictions (<0.1 s) at the onset of locomotion, following whisker stimulation, and upon awakening from sleep. Constrictions of the BVs and the SSS were strongly correlated with abdominal muscle EMG activity and were tightly correlated with respiration at rest. The rapid decrease in blood volume caused by venous constrictions resulted in spurious increases in fluorescence in mice expressing fluorescent reporter proteins, creating artifacts that could mimic functional signals. Venous constrictions with the same amplitude and dynamics could be generated in anesthetized mice by abdominal pressure application, showing that these constrictions were generated by mechanical coupling with the abdomen. Externally imposed abdominal pressures also drove a rapid but transient increase in blood flow. Unlike the pial and parenchymal microvasculature whose diameters are largely controlled by local signals, the diameters of SSS/BV are dynamically controlled during behavior in part by abdominal muscle regulation of intracranial pressure, establishing a pathway for regulation of cerebral hemodynamics via mechanical coupling between the central nervous system and the viscera.

Introduction

Under physiological conditions, the flow of blood in the brain is regulated by signals from neurons and glia on a time scale of seconds to tens of seconds (1, 2). Relaxation of smooth muscle around arterioles, penetrating arteries and pial arteries is the most rapid of these processes, occurring within a few hundred milliseconds after neural activity increases (3–6). In behaving mice, pharmacological infusions silencing local neural activity block arterial dilation in the awake animal (7), showing arterial dilation is driven by local neural activity. Arterial dilation is followed by a much slower dilation of ascending and pial veins (4, 8) and capillaries (3, 9). Because of their sluggish (minutes) response to vasoconstrictors (10), the control of the cerebral venous system has received less attention as veins have been assumed to have slow and passive dynamics.

Blood flow from the dorsal cortex drains into the superior sagittal sinus (SSS), a large dural sinus formed by bifurcation of the dural membrane (Figure 1A). The SSS drains the dorsal and frontal portions of cortex (11) as well as dural vessels (12). The SSS is fed by the large bridging veins (BV) that run nearly perpendicular to the SSS. The bridging veins have sphincter-like rings of collagen surrounding the entrance of the bridging vein to the SSS (13, 14) which have been hypothesized to play a role in venous pressure regulation (15). Any changes in caliber of the SSS will change the vascular resistance and could thus change the blood flow for a large portion of cortex. The intraluminal pressure decreases through the cerebral vascular network, so that in anesthetized animals (15–18) and in awake humans (19), SSS pressure is lower than intracranial pressure (ICP). This negative transmural pressure is maintained due to the relative stiffness of the venous wall which can resist collapse in the face of small negative pressure gradients, much like the hull of submarine allows the crew compartment to have a lower pressure than the water around it. Because nearly all the blood flowing through the cortex drains through the bridging veins and superior sagittal sinus, changes in the diameters of these vessels could have an influence on perfusion of all the cortex. In anesthetized animals, artificial elevation of ICP causes bridging veins to constrict (20). In behaving mice, ICP is dynamic and can rise from a baseline of 5–10 mmHg up to 25 mmHg or more during locomotion (21, 22). These ICP increases are caused by abdominal muscle contraction that drive increased intra-abdominal pressure which is transmitted to the central nervous system via a vascular network that spans the vertebral bones (23). Determining whether these behaviorally driven ICP increases have any impact on bridging veins and the SSS is important because diameter changes in these vessels could drive hemodynamic changes and affects fluid drainage (24, 25).

Figure 1. Venous constriction during voluntary locomotion in awake, head-fixed mice.

(A) Schematic showing the pial vasculature in the mouse brain. Left, dorsal view in vivo under 530 nm illumination with veins traced in blue and arteries in magenta; Top right, images showing the dural venous sinuses after transcardial perfusion with FITC and gelatin and fixation; Bottom right, schematic showing the coronal view of the mouse brain. (B) Left, schematic of the experimental setup for widefield IOS imaging of awake head fixed mice. Right, image of the cerebral vasculature under 530 nm illumination through a thinned-skull window spanning the parietal cortices of both hemispheres. Colored lines denote locations of vessel diameter measurements shown in subsequent figures. (C) Example data showing hemodynamic changes of veins at different locations (shown in B) during voluntary locomotion. FL/HL, forelimb/hindlimb representation of the somatosensory cortex; Wh, vibrissae cortex. ΔHbT, total hemoglobin; ΔHbO-HbR, differences of oxy- and deoxy-hemoglobin. The shaded area denotes the period of locomotion. (D) Averaged locomotion onset- and offset-triggered responses of ΔD/D₀ (n = 7 mice) in veins at different locations. (E) Average venous diameter change (ΔD/D₀) during the initial phase of locomotion (0–1 second after locomotion onset, left) and sustained locomotion (2–5 seconds after locomotion onset, right) in veins at different locations. There was a rapid constriction in superior sagittal sinus (SSS, −1.18 ± 0.42%, Wilcoxon rank sum test, p = 0.0006) and bridging veins (BV, −0.76 ± 0.84%, Wilcoxon rank sum test, p = 0.0169) in response to locomotion onset, while pial veins in the FL/HL (−0.63 ± 1.22%, Wilcoxon rank sum test, p = 0.1801) and Wh (0.13 ± 0.51%, Wilcoxon rank sum test, p = 0.1801) did not change significantly. In response to sustained locomotion, the initially constricted BV returned to baseline level (−0.34 ± 2.01%, Wilcoxon rank sum test, p = 0.6900), while the SSS stayed constricted (−2.13 ± 0.92%, Wilcoxon rank sum test, p = 0.0006) and returned to baseline level with a prolonged delay after the cessation of voluntary locomotion. (F) Significant negative relationship (slope = −0.0059, 95% confidence interval [−0.0093, −0.0025], goodness of fit R² = 0.333, p = 0.0013) between the average initial venous diameter change and baseline vessel diameter, showing the large collecting vessels constrict in response to locomotion. Dashed line indicates 95% confidence bounds. * p < 0.05 compared to zero.

Here, we visualized the dynamics of bridging veins and the SSS using optical imaging in head fixed mice and found a constriction of the SSS and bridging veins (as well as some pial veins) with ultra-rapid onset (~100ms) during bouts of voluntary locomotion. Sensory stimulation to the whiskers also drove abdominal muscle contraction and corresponding SSS/BV constriction. In mice expressing fluorophores, this constriction decreased light absorbance by hemoglobin which could drive artifactual functional signals. The SSS/BV constriction was tightly correlated with abdominal muscle contraction (as measured with electromyography) and could be recapitulated in anesthetized mice by squeezing the abdomen. Our results suggest that the diameters of the SSS/BV, as well as some larger pial veins, may be at least partially impacted during behavior by abdominal muscle regulation of intracranial pressure. Cerebral hemodynamics may not just reflect local neural activity but also could be influenced by abdominal muscle contractions via mechanical coupling between the central nervous system and the viscera.

Results

We measured brain hemodynamics using widefield intrinsic optical signal (IOS) imaging, two-photon laser scanning microscopy (2PLSM) and laser Doppler flowmetry in 7 head fixed C57BL/6 mice, as well as using mini-scope in one free-behaving C57BL/6 mouse. We also measured green fluorescent protein (GFP) responses in 6 head fixed CAG-EGFP mice. Intracranial pressure measurements in 10 C57BL/6 mice were from our previous publications (21, 22) and re-analyzed here.

Voluntary locomotion drives rapid constriction of SSS, BV and pial veins in FL/HL, and slower dilation of pial veins.

We first assessed the spatial patterns of venous responses and their relationship to voluntary locomotion using widefield IOS imaging. We performed imaging through thinned skull in awake C57BL/6 mice (n = 7 mice, 2 female) head fixed on a spherical treadmill (26, 27). Because the SSS is underneath the midline suture, the midline and coronal sutures were thinned to provide optical clarity. We illuminated the brain with alternating 530 nm and 470 nm light, providing information about blood volume, vessel diameter and brain oxygenation (26–28). We calculated the vessel diameter changes using the full-width at half maximum (FWHM) (27, 28) from the IOS images under 530 nm illumination. We recorded electromyography (EMG) signals from the abdominal musculature, which are engaged during locomotion and drive brain motion (23).

We first looked at the vascular responses to voluntary bouts of locomotion, averaging vascular changes evoked by movement into locomotion triggered averages. Consistent with our previous results (8, 21, 29), voluntary locomotion evokes a large increase in arterial diameter within a few hundred milliseconds (Supplementary Figure 1A and C). Both the abdominal muscles (23) and nuchal muscle (26) showed a strong increase in activity during locomotion (Figure 1C and Supplementary Figure 4B). In contrast, there were branch-specific changes in venous diameter (ΔD/D₀) in response to locomotion (Figure 1C–E). Specifically, the SSS, bridging veins, and pial veins in FL/HL rapidly constricted within hundreds of milliseconds of the locomotion initiation (Figure 1D, Supplementary Figure 2C), consistent with previous reports (30). The constriction of pial veins in FL/HL is rapidly followed by a dilation. The smaller more lateral veins in the whisker (Wh) representations dilated with a delay during locomotion. At the cessation of locomotion, pial veins in the vibrissae cortex and bridging veins returned to their baseline diameters within 3 seconds of locomotion offset. In contrast, pial veins in FL/HL remained dilated, while the SSS stayed constricted for a longer period (Figure 1D, Supplementary Figure 2C). The venous constriction occurred prior to any significant vasodilation in the somatosensory cortex (Figure 1C and D, Supplementary Figure 2). Similar constrictions were observed using two-photon microscopy (Supplementary Figure 1A–C). Moreover, we also observed bridging vein constriction in a freely moving mouse during locomotion using a miniscope (Supplementary Figure 1F), showing the observed responses are not due to body posture difference caused by head fixation. As an alternate metric of vasodilation, we quantified the change in blood volume using changes of total hemoglobin (ΔHbT, Supplementary Figure 2) around these veins, and saw similar dynamics, with the decrease in blood volume in SSS and BV, and increase in the smaller, more lateral veins (Supplementary Figure 2B and D). Interestingly, though the blood volume decreased in both the SSS and BV, the blood oxygenation (ΔHbO-HbR) increased (Supplementary Figure 2E) (27, 31). The contractions of these veins were ultrarapid (~100 ms), far too quickly to be due to mural cell contractions (3, 9), so we sought other mechanisms. We observed a significant negative correlation between the basal vein diameter and its diameter change during locomotion (Figure 1F). Because the smaller veins feed into larger veins, the blood pressure inside the vein decreases with size, while the pressure outside all the veins (the intracranial pressure) will be roughly constant. If the ICP rises enough, then the balance of pressures favors a decrease in the diameter in the larger veins. Movements in both humans and mice are often preceded by the activation of abdominal muscles (32), whose contraction stiffens the core in anticipation of movement. Previous studies have shown that intraabdominal pressure (IAP) is coupled to ICP (33) and abdominal muscle contraction drives increases in IAP (32), leading to forces on the brain that drive brain motion (23) and increase ICP (34). To understand how these muscle contractions were related to the rapid changes in venous diameter, we recorded oblique abdominal muscle electromyograph (EMG_Abd) signals (n = 4 mice, all male) during locomotion. We observed a robust increase in EMG_Abd power that were strongly correlated with locomotion (Figure 1C and Supplementary Figure 4B). When we compare the cross-correlation curve between venous constriction and abdominal EMG activity (Supplementary Figure 4D) or locomotion speed (Supplementary Figure 4E), we observed that the constriction of SSS invariably followed EMG activity, but often preceded locomotion, suggesting that abdominal muscle contraction drives the SSS constriction. We also quantified cortical parenchymal hemodynamic signals adjacent to the imaged veins during locomotion (Supplementary Figure 2B, H–K). These measurements revealed similar parenchymal hemodynamics to those observed in the adjacent veins. Abdominal muscle contraction applies force (35) to the abdomen which increases abdominal pressure. The abdominal pressure is transmitted to the central nervous system by the vertebral venous plexus, a network of veins connecting the abdominal cavity to the spinal column, which functions as a hydraulic system that transfers force between the two compartments (23). Our results suggest that this abdominally driven force also drives constriction of the large veins in the cortex.

Whisker stimulation drives constriction of BV and SSS, and transient constriction of pial veins in FL/HL.

Sensory evoked responses can drive both covert and overt muscle contractions (36, 37). Since many imaging experiments make use of sensory-evoked responses, such as whisker stimulation (38), we then asked if constriction of the SSS and BV could be driven by sensory stimulation of the whiskers, potentially mediated by abdominal muscle contractions. Brief (0.1s at 10 psi) unilateral stimulation of the whiskers drove vasodilation of arteries in the contralateral sensory cortex with a ~1 second delay (Supplementary Figure 3B), consistent with previous reports (4, 6, 39). However, we also observed a rapid (< 0.2 seconds from stimulation onset) constriction in the SSS, BV and pial veins in FL/HL, but not of the smaller veins in the whisker representation of the somatosensory cortex (Figure 2B–D, Supplementary Figure 3). Quantification of ΔHbT also showed a decrease in response around stimulation onset (Supplementary Figure 3B and C), and in contrast to locomotion, there was a brief dip in net oxygenation of the blood (Supplementary Figure 3D), likely because the stimulus was unexpected and was not accompanied by an increase in respiration (Supplementary Figure 1E) that raises systemic blood oxygenation (27). Similar rapid hemodynamic changes were also observed in cortical parenchyma adjacent to these veins (Supplementary Figure 3B, G–J). The dip in oxygenation is reminiscent of the “initial dip” seen in some imaging experiments (40, 41). Electromyography recordings showed a large increase in abdominal EMG (EMG_Abd) activity time locked to the whisker stimulation (Figure 2B and Supplementary Figure 4C). As a control we looked at the responses to auditory stimulation, which drives whisking and similar (though somewhat smaller) functional hyperemia in the whisker representation of the cortex due to whisking (4, 42). Auditory stimulation did not drive significant abdominal EMG activation (Figure 2B and Supplementary Figure 4C), and there was no significant constriction of the SSS and BV (Figure 2B–D), consistent with the hypothesis that abdominal muscle contraction drove the constriction of the largest veins.

Figure 2. Superior sagittal sinus and bridging vein constriction in response to whisker stimulation in awake, head-fixed mice.

(A) Top, schematic of the experimental setup for widefield IOS imaging of awake head fixed mice during whisker stimulation. Bottom, image of the cerebral vasculature under 530 nm illumination through a thin-skull window spanning the parietal cortices of both hemispheres. Colored lines denote locations of vessel diameter measurements shown in subsequent figures. (B) Example single trial responses showing hemodynamic changes of veins at different locations (shown in A) in response to whisker stimulation. FL/HL, forelimb/hindlimb representation of the somatosensory cortex; Wh, vibrissae cortex. ΔHbT, total hemoglobin; ΔHbO-HbR, differences between oxy- and deoxy-hemoglobin. The shaded area denotes the period of whisker stimulation. (C) Left, group average of a short (100 ms) stimulation triggered response of ΔD/D₀ (n = 4 mice) in veins at different locations in response to contralateral and ipsilateral whisker stimulation, as well as auditory stimulation control. Right, a zoom-in view of ΔD/D₀ responses immediately before (330 ms) and after (660 ms) stimulation. (D) Average venous diameter change (ΔD/D₀) during the initial phase of stimulation (0–0.5 second after stimulation onset, left) and after stimulation (0.5–3 seconds after stimulation onset, right) in veins at different locations. In response to contralateral whisker stimulation, there was a rapid constriction in superior sagittal sinus (SSS, −1.24 ± 0.77%, Wilcoxon rank sum test, p = 0.0286), bridging veins (BV, −0.25 ± 0.30%, Wilcoxon rank sum test, p = 0.0286) and pial veins in the FL/HL (−0.56 ± 0.56%, Wilcoxon rank sum test, p = 0.0286), while pial veins in the Wh (−0.005 ± 0.09%, Wilcoxon rank sum test, p = 0.3143) did not change significantly. After the initial constriction, SSS (0.18 ± 0.57%, Wilcoxon rank sum test, p =0.3143) and BV (0.17 ± 0.67%, Wilcoxon rank sum test, p = 0.3143) returned back to baseline levels, while the pial veins in FL/HL (1.27 ± 1.11%, Wilcoxon rank sum test, p = 0.0286) dilated. In contrast, the pial veins in the Wh did not change significantly (0.88 ± 1.54%, Wilcoxon rank sum test, p = 0.3143). In response to ipsilateral whisker stimulation, there was a rapid constriction in SSS (−0.85 ± 0.28%, Wilcoxon rank sum test, p = 0.0286) and BV (−0.51 ± 0.27%, Wilcoxon rank sum test, p = 0.0286), while pial veins in the FL/HL (−0.36 ± 0.75%, Wilcoxon rank sum test, p = 0.3143) and Wh (−0.09 ± 0.13%, Wilcoxon rank sum test, p = 0.3143) did not change significantly. After the initial constriction, SSS (−0.08 ± 0.66%, Wilcoxon rank sum test, p = 0.3143) and BV (0.69 ± 0.92%, Wilcoxon rank sum test, p = 0.3143) returned back to baseline levels, while the pial veins in FL/HL (1.63 ± 0.90%, Wilcoxon rank sum test, p = 0.0286) and Wh (0.39 ± 0.61%, Wilcoxon rank sum test, p = 0.0286) dilated. In response to auditory stimulation, no significant diameter changes were observed during the initial phase following the stimulation (SSS: −0.23 ± 0.36%; BV: −0.11 ± 0.13%; Wh: 0.06 ± 0.07%), with the exception of pial veins in the FL/HL, which exhibited a small but significant constriction (0.18 ± 0.19%, Wilcoxon rank sum test, p = 0.0286). Likewise, no significant diameter changes were detected during the later phase after stimulation (SSS: 0.11 ± 0.29%; BV: 0.16 ± 0.32%; FL/HL: 0.32 ± 0.46%; Wh: 0.16 ± 0.31%). * p < 0.05 compared to zero.

Rapid venous constrictions can generate artifactual fluorescence increases with kinetics similar to functional indicators.

Hemoglobin absorbs light, and changes in hemoglobin levels will change the absorbance of excitation and emission light (43), which may confound interpretation of neural activity using fluorescent calcium indicators (44–47). This is particularly salient for the SSS/BV constrictions we observe, which have onset and offset kinetics similar to genetically-encoded calcium indicators (48), rather than the slower dynamics of blood volume changes (29). With mesoscopic imaging, the entire dorsal cortical surface can be imaged during sensory stimulation and behavior (49). Because sensory input and behavior drive activity changes not just in a few brain regions, but across the entire brain (50, 51), it becomes important to understand how hemodynamic signals everywhere in the cortex could confound fluorescent activity indicators. Rapid venous constrictions could drive artefactual signals in any paradigm in which the abdominal muscles of mice are engaged. To investigate the role of venous constriction in generating fluorescence artifacts, we quantified the effects of hemodynamics on fluorescence measured from CAG-EGFP mice (n = 6, all male) head fixed in a tube using widefield fluorescence imaging in response to whisker stimulation (Figure 3A). Because green fluorescent protein (GFP) is not responsive to neuronal activity, changes in fluorescence will be entirely due to changes in the local concentration of light-absorbing hemoglobin, and the signals we see here would recapitulate what would be reported by the fluorescence of calcium indicators if there was no change in neural activity. We observed large amplitude fluctuations that were restricted mainly to the midline vasculature and typically included fast transient spikes in fluorescence (Figure 3B). These fluctuations resulted from changes in venous blood volume along the SSS that relate to movements or postural changes (Figure 3B–D, Supplementary Figure 5C). We also examined brain hemodynamics (Supplementary Figure 5D) and fluorescence signals (Supplementary Figure 5E) in the surrounding brain parenchyma adjacent to these veins. These areas exhibited rapid transients with temporal dynamics similar to those observed over those veins, indicating that these hemodynamic artifacts are not restricted to the large surface veins but can also influence fluorescence measurements from the adjacent cortex. We then tested three commonly used methods for correcting hemodynamic contamination, including hemodynamic correction using only green reflectance (i.e., single-wavelength regression method) (52), hemodynamic correction using estimated excitation and emission attenuation (i.e., Ex-Em method) (46, 47), and hemodynamic correction using a spatially detailed regression-based method to estimate hemodynamics contamination (i.e., spatial model) (44). For the artifacts caused by initial rapid constrictions and slow dilation of veins, both the spatial model and regression model can successfully remove the artifact, but the Ex-Em method can cause overcorrection (Figure 3B–F). We also compared the responses in the brain parenchyma and found that the three hemodynamic correction methods showed similar performance to that observed directly over the veins, with the spatial model and regression approach providing effective correction while the Ex-Em method overcorrected the rapid transients (Supplementary Figure 5E). These results show that the constriction of veins can cause changes in fluorescence independent of neural activity in many locations of the dorsal cortex.

Figure 3. Ultrafast venous constriction induces contamination in widefield fluorescence imaging.

(A) Top, schematic of the experimental setup for widefield IOS and fluorescence imaging of awake head fixed mice during whisker stimulation. Bottom, image of the cerebral vasculature under 530 nm illumination (bottom left) and image of the fluorescent signals under 470 nm illumination (bottom right) through a thin-skull window spanning the cortices of both hemispheres. Colored lines denote locations of vessel diameter measurements shown in subsequent figures. (B) Averaged spatial distribution of ΔHbT, raw fluorescence signal (ΔF/F_{0, uncorrected}) and corrected fluorescence signal using different methods (ΔF/F_{0, SpatialModel}, ΔF/F_{0, Ex-Em} and ΔF/F_{0, Regression}) in response to contralateral whisker stimulation (n = 47 events in 1 mouse). (C) An example showing averaged trace of contralateral whisker stimulation triggered response of venous diameter (ΔD/D₀), raw fluorescence signal and corrected fluorescence signal using different methods (n = 47 events in 1 mouse) in veins at different locations in the same animal shown in (B). (D) Group (n = 6 mice) average of whisker stimulation triggered response of ΔD/D₀, raw fluorescence signal and corrected fluorescence signal using different methods in veins at different locations. (E) Average fluorescence change (ΔF/F₀) during the initial phase of contralateral whisker stimulation (0–0.8 second after whisker stimulation onset) in veins at different locations. During the initial phase after the contralateral whisker stimulation, we observed a fast increase of fluorescence in superior sagittal sinus (SSS, 0.85 ± 0.47%, Wilcoxon rank sum test, p = 0.0022), bridging veins (BV, 1.86 ± 0.81%, Wilcoxon rank sum test, p = 0.0043) and in pial veins in the FL/HL (0.75 ± 1.19%, Wilcoxon rank sum test, p = 0.0022), while fluorescence of and pial veins in the Wh (0.20 ± 0.43%, Wilcoxon rank sum test, p = 1) did not change significantly. These artifactual fluorescence signals were effectively corrected using the spatial model method (SSS, 0.05 ± 0.05%, paired t test, p = 0.0085; BV, 0.15 ± 0.20%, paired t test, p = 0.0133; FL/HL, 0.03 ± 0.17%, paired t test, p = 0.1465; Wh: 0.05 ± 0.06%, paired t test, p = 0.1891) and regression method (SSS, 0.04 ± 0.05%, paired t test, p = 0.0080; BV, 0.11 ± 0.08%, paired t test, p = 0.0068). In contrast, the Ex-Em method overcorrected these signals (SSS, −0.84 ± 0.41%, paired t test, p = 0.0045; BV, −1.35 ± 0.71%, paired t test, p = 0.0076; FL/HL, −0.49 ± 0.67%, paired t test, p = 0.1635; Wh, −0.21 ± 0.24%, paired t test, p = 0.1928). (F) Average fluorescence change (ΔF/F₀) during 1–3 seconds after whisker stimulation in veins at different locations. During later phase after the whisker stimulation, we observed a decrease of fluorescence signals in SSS (−1.65 ± 1.24%, Wilcoxon rank sum test, p = 0.0022), BV (−1.28 ± 0.69%, Wilcoxon rank sum test, p = 0.0044), as well as pial veins in the FL/HL (−2.26 ± 0.57%, Wilcoxon rank sum test, p = 0.0022) and the Wh (−2.23 ± 0.58%, Wilcoxon rank sum test, p = 0.0022). These artifactual fluorescence signals can be corrected using spatial model method (SSS, −0.02 ± 0.08%, paired t test, p = 0.0232; BV, −0.15 ± 0.22%, paired t test, p = 0.0092; FL/HL, −0.49 ± 0.34%, paired t test, p = 0.0001; Wh, −0.36 ± 0.25%, paired t test, p = 0.0002), regression method (SSS, −0.22 ± 0.12%, paired t test, p = 0.0380; BV, −0.20 ± 0.04%, paired t test, p = 0.0209; FL/HL, −0.37 ± 0.15%, paired t test, p = 0.0001; Wh, −0.30 ± 0.10%, paired t test, p = 0.0002), and the Ex-Em method (BV, −0.18 ± 0.29%, paired t test, p = 0.0374; FL/HL, −0.22 ± 0.22%, paired t test, p = 0.0001; Wh, 0.05 ± 0.25%, paired t test, p = 0.0003). However, Ex-Em method did not perform well in SSS (0.10 ± 0.73%, paired t test, p = 0.0750). * p < 0.05 compared to zero. # p < 0.05 compared to raw fluorescence signal.

Intracranial pressure dynamics during behavior.

The rapid constriction of veins could be mediated by increases in intracranial pressure (ICP), as artificial elevations of ICP are known to compress bridging veins (20) and ICP is elevated during locomotion (21, 22, 26). To address this, we reanalyzed a previously published dataset (21, 22) of intracranial pressure recordings made from mice head fixed on a spherical treadmill. In the reanalysis, we used broadband signals with minimal temporal filtering in order to capture the fast temporal dynamics. An example ICP trace, along with locomotion dynamics is shown in Figure 4A. It is apparent that there are large (>10 mmHg) increases in ICP during locomotion that dominate all other fluctuations, consistent with ICP increases being the driver of the constrictions of the largest veins. The locomotion-triggered averages show an increase in ICP that precedes the onset of locomotion, rises ~8 mmHg above the resting pressure, and then returns to baseline shortly after the cessation of locomotion (Figure 4C), consistent with the dynamics of the venous constriction (Figure 1, Supplementary Figure 2) and EMG activation (Supplementary Figure 4) during locomotion. Inspection of periods of quiescence and of the power spectrum reveal ~3 Hz fluctuations in ICP that are less than 1 mmHg (0.41 ± 0.11 mmHg, n = 10 mice) in amplitude (Figure 4A and B). There are also ~10 Hz fluctuations (at the cardiac frequency) visible in the power spectrum (Figure 4B) that are at least an order of magnitude smaller than the respiration fluctuations, too small to have an appreciable effect on venous diameter. While both cardiac and respiratory-induced pressure fluctuations are at least an order of magnitude smaller than those that accompany locomotion, the ~1mmHg fluctuations due to respiration could potentially have an impact on venous volume (53, 54). To address this, we measured respiration with a thermocouple (26, 27) during widefield IOS imaging (n = 6 mice, two mice were excluded from analyses due to insufficient resting data). Increases and decreases in the thermocouple temperature correspond to the expiratory and inspiratory phases of the respiratory cycle, respectively (Figure 4D). During resting period, there was substantial correlation between the diameter of the SSS, BV and pial veins in FL/HL, and the respiration signal (Figure 4D and E). Intra-abdominal pressure (and therefore ICP) rises during inspiration (55), which tends to compress the SSS, though these pressure fluctuations are much smaller than those caused by locomotion. We observed that the diameter of the cerebral veins (especially SSS and BV) depends on respiratory cycles (it decreases during inspiration, and increases during expiration) (Figure 4D and E). These results indicate that ICP fluctuations driven by locomotion and respiration are associated with corresponding changes in BV and SSS diameter.

Figure 4. Intracranial pressure increases during locomotion and fluctuates with respiration.

(A) Example intracranial pressure (ICP) dynamics during voluntary locomotion. Top, black trace shows ICP, and black tick marks show locomotion events. Inset showing a zoom-in of respiration-driven ICP oscillations. Bottom, spectrogram of ICP during rest and locomotion. (B) Representative power spectrum of the ICP as well as its power law fit (dashed line) using the data from the resting period. The residual power (i.e., the difference between the power spectrum of the observed ICP signal and the power law fit) is shown in the bottom. (C) Locomotion onset and offset triggered ICP changes. Note that ICP rises before locomotion onset (onset time: −0.56 ± 0.29 s, Wilcoxon rank sum test, p = 0.0001). (D) Example trace showing the diameter change of veins at different locations is correlated with respiration. Inset, a zoom in view of the temporal relationship between respiration and venous diameters. The respiration signal here is from the temperature change measured outside the nostril, i.e., an increasing signal is the expiration phase, and decreases in the temperature happen during the inspiration phase. (E) Left, cross-correlation between respiration signal and venous diameter change during rest. Right, cross-correlation between respiration signal and venous blood volume (ΔHbT). A positive peak with positive time lag suggests that increasing respiration signal (i.e., expiration) causes venous dilation, and decreasing respiration signal (i.e., inspiration) causes venous constriction.

SSS and bridging veins dilate during sleep and constrict upon awakening.

Next, we looked at the dynamics of the large veins during sleep. Sleep is accompanied by large dilations of arteries (56, 57) and global increases in blood volume (58) that are much larger than those that are seen in the awake state. Sleep, particularly rapid eye movement (REM) sleep, is accompanied by large decreases in muscle tone (59), and drops in blood pressure (60). We hypothesized that the net effect of these changes during sleep would drive dilation of the SSS and BV. We performed widefield IOS imaging on head-fixed mice (n = 4, all male) in a tube and tracked whisker position, body movement, and abdominal muscle EMG (Figure 5A–B), which were used to determine the arousal state of the animal (42, 56). We performed sleep scoring as previously described (56, 57, 61) and categorized the arousal state into three stages: awake, REM, and non-rapid eye movement (NREM) sleep. In contrast to voluntary locomotion and whisker stimulation evoked venous constrictions, we found venous dilation associated with sleep, and saw large constrictions upon transitions from REM or NREM to the awake state (Figure 5B–D). We also saw large decreases in total hemoglobin and, in contrast to the constrictions during locomotion, decreases in blood oxygenation around these vessels in transitions to the awake state (Supplementary Figure 6). As with the venous dynamics in the awake state, venous constrictions were associated with higher abdominal EMG power (Figure 5B and C), again suggesting that there is abdominal modulation of venous dynamics across many arousal states.

Figure 5. Awakening from NREM and REM sleep drives venous constriction.

(A) Left, schematic of the experimental setup for widefield IOS imaging of head fixed mice. Right, image of the cerebral vasculature under 530 nm illumination through a thin-skull window spanning the parietal cortices of both hemispheres. Colored lines denote locations of vessel diameter measurements shown in subsequent figures. (B) Example data showing hemodynamic changes of veins at different locations (shown in A) during transitions between arousal states. Top, arousal state scored from EMG/ECoG/whisker and body motion. White break denotes break in data collection between trials. FL/HL, forelimb/hindlimb representation of the somatosensory cortex; Wh, vibrissae cortex. ΔHbT, total hemoglobin; ΔHbO-HbR, differences of oxy- and deoxy-hemoglobin. The shaded area denotes the periods of different arousal states. (C) Changes of abdominal muscle EMG (ΔP_EMG/P₀) and venous diameter (ΔD/D₀) during the transition of NREM into awake (n = 4 mice) and REM into awake (n = 2 mice). Note that scales are different across conditions. (D) Bar plot showing the mean changes in diameter in response to each arousal state transition. Awakening from NREM induced constrictions in superior sagittal sinus (SSS, −5.55 ± 1.19%, Wilcoxon rank sum test, p = 0.0286) and in pial veins in the FL/HL (−1.60 ± 0.89%, Wilcoxon rank sum test, p = 0.0286), dilations in bridging veins (BV, 1.54 ± 0.96%, Wilcoxon rank sum test, p = 0.0286), while pial veins in the Wh (−0.32 ± 0.27%, Wilcoxon rank sum test, p = 0.3143) did not change significantly. * p < 0.05 compared to zero.

Passively applied abdominal pressure drives SSS, BV, and FL/HL constriction and transient increases in blood flow.

Externally applied abdominal pressure drives brain movement like that seen during behavior, mediated through a network of valveless veins (vertebral venous plexus) that link the spinal cavity with the abdominal cavity (23). We asked if this pressure application could drive constriction of the SSS and BV. We anesthetized mice (1% isoflurane) and placed them in a computer controlled pneumatic pressure cuff (23) and applied one-second long squeezes of the abdomen while monitoring both cortical vessel diameter and cerebral blood flow (CBF) using laser Doppler flowmetry (62, 63). During cuff inflation, the pressure applied to the abdomen increases and reaches half-max level within ~0.17s. We found that externally applied abdominal pressure drove reliable constrictions of the SSS, BV, initial constriction in the FL/HL, but not in the Wh veins (Figure 6B and C), recapitulating the patterns seen during locomotion (Figure 1 and Supplementary Figure 2) and during sensory stimulation (Figure 2 and Supplementary Figure 3). Interestingly, the constriction also drove a flow increase in the parenchyma (3.02 ± 1.47%, Figure 6D) and oxygenation increases across the whole cortex (Supplementary Figure 7), indicating a coordinated brain-wide hemodynamic response accompanying venous constriction. It seems that the abdominal pressure can drive a very rapid blood flow increase, and this could be a mechanism by which flow to the brain is increased peremptorily by abdominal muscle contractions, such as those that accompany the startle response (64).

Figure 6. Abdominal compression causes venous constrictions.

(A) Top, schematic of the experimental setup for abdominal compression experiments. Bottom, image of the cerebral vasculature under 530 nm illumination through a thin-skull window spanning the parietal cortices of both hemispheres. Colored lines denote locations of vessel diameter measurements shown in subsequent figures. (B) Example trace showing changes of brain hemodynamics at different locations during abdominal compression. (C) Population average of abdominal compression evoked responses of venous diameter (ΔD/D₀, n = 4 mice). Brief, gentle compression of abdomen reduced venous diameter in superior sagittal sinus (SSS, −3.18 ± 1.89%, Wilcoxon rank sum test, p = 0.0286) and bridging veins (BV, −0.53 ± 0.51%, Wilcoxon rank sum test, p = 0.0286), and no change of veins in FL/HL (−0.10 ± 0.79%, Wilcoxon rank sum test, p = 0.3143) and Wh (0.19 ± 0.04%, Wilcoxon rank sum test, p = 0.2857). (D) As in (C) but for changes of cerebral blood flow (ΔCBF). Significant increases of ΔCBF were observed in the parenchyma in response to 1 s squeezing (0.2–1.2 seconds after squeezing onset, 3.08 ± 1.47%, Wilcoxon rank sum test, p = 0.0286).

Discussion

We observed rapid constrictions of bridging veins and the superior sagittal sinus during voluntary locomotion, whisker stimulation, and upon awakening from sleep, all conditions were accompanied by abdominal muscle contraction. We also saw a transient constriction, followed by a dilation, in the large pial veins in the FL/HL to voluntary locomotion and sensory stimulation. Externally imposed abdominal pressure generated patterns of venous constriction that were very similar in amplitude compared to those seen during locomotion and sensory stimulation. These results are consistent with a model where abdominal pressures are transmitted to the central nervous system. These forces can cause brain movement (23), and here we show they can cause compression of the largest veins on a sub-second time scale.

Previous studies looking at more lateral and smaller veins have observed dilation (4, 62), something we have seen as well. The cause of the size dependent difference in venous responses is likely due to differences in intraluminal pressure throughout the network. Intraluminal pressure decreases monotonically as blood transits through the vascular network (65), so the larger veins will have a lower pressure than the smaller veins that feed them. This means that if ICP is elevated, then the first vessels in the brain to be compressed will be the large veins, as was observed here. Transiently induced ICP increases have been shown to be closely mirrored by cortical venous pressure, while not affecting SSS pressure (66). This observation is consistent with our finding that some large pial veins initially constrict before subsequently dilating, suggesting that rising cortical venous pressure progressively offsets the external compressive force generated by elevated ICP (67). The transient change could be caused by any combination of the venous pressure increasing due to upstream arterial dilation, elevation of central blood pressure, or a decrease in the ICP after an initial onset tansient. ICP changes will depend on the coupling from the abdominal cavity to the spinal cord (Figure 7A). These results, coupled with the observed ICP increases during locomotion, point to a model where the central nervous system is mechanistically linked to the abdominal cavity, and pressures generated in the abdomen are transmitted to the brain, leading to increase in ICP and brain motion (23). If the ICP rises, it will be above the intraluminal pressures of the bridging veins and SSS (but not above the intraluminal pressures of the smaller veins) (Figure 7B–D), causing a spatial pattern of venous responses like the one we see during locomotion (Figure 1F). We note that locomotion is not the only behavior that engages the abdominal muscles that increase intraabdominal pressure. In humans, many movements are preceded by abdominal muscle activation (68), and it is likely the same muscles will be engaged in many behaviors in mice. There are several caveats to our work. The superior sagittal sinus does not have a circular cross section, it is roughly triangular in histological sections (69), so our estimates of diameter changes may over- or under-estimate the actual changes. The changes in the diameter of the SSS (which reflect transmural pressure) during sleep could have contributions from other causes besides the mechanical coupling with the abdomen. ICP will be impacted not only by abdominal contractions, but also by the dilation of cerebral arteries during sleep (56), which will raise ICP (70). The blood pressure in the SSS will likewise be affected by changes in systemic blood pressure, peripheral resistance, and the dilation of the arteries of the cortex that feed the SSS. Furthermore, contractile cells around the SSS may be engaged during sleep and/or awakening (69, 71). There are other large veins in the brain that we did not image (e.g., transverse sinuses) that may undergo similar ICP-driven compression. Finally, ICP will not be completely spatially homogenous, and the venous pressure could vary dynamically, which will affect the sensitivity of veins to elevations in ICP.

Figure 7. Elevated intracranial pressure can drive differential responses in venous diameter.

(A) Schematic of the hypothesis that increases in intrabdominal pressure (due to abdominal muscle contraction) forces blood from the caudal vena cava to the vertebral venous plexus within the vertebral column. The increased blood volume in an enclosed space applies pressure to the dural sac, forcing the cranially-directed cerebral spinal fluid flow that increases intracranial pressure (ICP). (B) Intraluminal pressure of veins decreases as a function of location in the venous tree, with the largest veins having the lowest pressures. ICP on all these veins is approximately the same. (C) Respiration induces small variations of intracranial pressure, which will surpass the intraluminal pressure of superior sagittal sinus, leading to its compression. (D) During locomotion, sensory stimulation or other times abdominal muscles contract, ICP is elevated above the intraluminal pressures of the bridging veins and superior sagittal sinus, leading to their compression.

Previous work has observed rapid constriction of bridging veins (27, 30) and dural venous sinus (44, 45), though their origin was not understood. There have also been reports of stimulation-induced small, rapid blood volume decreases whose origin is mysterious (72), and of artifactually increased fluorescence just prior to locomotion onset in mouse two-photon imaging (43) that likely have contributions from the venous constriction we describe. Our work extends on previous work showing the brain is mechanically coupled to the abdominal cavity (23, 73, 74), and shows that the mechanical coupling to the abdomen can control blood flow in the brain with rapid dynamics. Although our findings are consistent with passive compression by transient ICP elevations, recent work demonstrating contractile smooth muscle-like cells surrounding the superior sagittal sinus, but not bridging veins (69, 71), raises the possibility that active regulation may also contribute on slower time scales. Bridging veins contain few, if any, contractile smooth muscle-like cells and are therefore likely to behave predominantly as passive vessels whose diameter is governed by changes in transmural pressure. In contrast, although SSS diameter is also influenced by transmural pressure, the presence of contractile smooth muscle-like cells (69, 71) together with sympathetic and peptidergic innervation (69, 75, 76) suggests that active vasomotor mechanisms could contribute to its dynamics.

Besides the increase in blood flow elicited by the constriction of these veins, there may be other possible roles for this constriction. Firstly, the constriction of these veins could help compensate for ICP increases by providing compliance in the cranial cavity (70). Secondly, the SSS is a dural sinus, constriction of the SSS would apply tension to the mechanosensitive dura (77), which may account for some of the movement or deformation in the dura seen during locomotion in mice (78). Thirdly, the SSS is innervated by trigeminal and cervical spinal nerves (79) and is pain sensitive (80), so changes in the mechanical dynamics of the SSS could also play a role in migraine and headache (77). Fourthly, the constriction of these veins could also help move cerebrospinal fluid (24, 25), which is known to be moved by dilation or constriction in arteries (81–83). The venous constrictions may also help with the pumping action of the lymphatic vessels seen near the superior sagittal sinus (84–87). Finally, they may serve a mechanical signaling role. Mechanical coupling between the abdomen and brain is especially interesting considering the functional mechanosensitive channels in central nervous system neurons (88) and glia (89), and the rapid constriction of these veins could also activate mechanosensitive channels in the brain, acting as a direct interoceptive signaling pathway to the brain. These pathways may contribute to cardiac and respiratory modulation of neural activity (90, 91).

While compression (stenosis) of veins by elevated intracranial pressure has been noted under pathological conditions (92–94), our work shows that this compression may occur naturally, albeit transiently, in the behaving brain. However, our results could have implications for brain pathologies as well. Pathologies like obesity elevate intra-abdominal pressure (95) and ICP (96), which could disrupt the normal flow of blood back and forth between the abdominal cavity and spinal canal. Elevated blood pressure may attenuate or block these venous constrictions, potentially playing a role in the adverse effects of hypertension on brain health (97). From the clinical perspective, changes in the normal dynamics of SSS/BV constrictions could be used as non-invasive indicators of intracranial pressure (98–100).

In summary, this work supports a model in which transient increases in intra-abdominal pressure are mechanically transmitted to the cranial cavity, producing rapid ICP elevations, brain motion, and dynamic, size-dependent compression of cerebral veins. This pressure-mediated pathway likely works in concert with contractile cells in the superior sagittal sinus to regulate venous diameter. This work establishes a previously unrecognized pathway through which the brain and cerebral vasculature is mechanistically coupled with the viscera, providing a dynamic signaling pathway that can rapidly communicate information about the body state to the brain.

Materials and Methods

All experimental procedures were performed in accordance with the National Institute of Health guidelines and the Institution of Animal Care and Use Committee of the Pennsylvania State University (protocol #201042827). A total of 24 mice, including 18 (2 female) C57BL/6 mice (Jackson Laboratory, #000664), and 6 (all male) C57BL/6-Tg(CAG-EGFP)131Osb/LeySop mice (CAG-EGFP, Jackson Laboratory, #006567), were used. Mice were implanted with polished and reinforced thinned-skull windows for brain hemodynamic measurements using widefield intrinsic optical signal imaging, two-photon laser scanning microscopy, laser Doppler flowmetry or miniscope. For electromyography recordings, electrodes were implanted to nuchal muscle and/or abdominal muscle. For intracranial pressure measurements, a pressure measuring catheter was inserted into the cortex. After recovering from the surgery, mice were gradually acclimated to head-fixation on either a spherical treadmill or a plastic tube. For respiration measurements, a thermocouple was placed near the mouse’s nose. Responses of brain hemodynamics, fluorescent signals, electromyography, intra-cranial pressure, and respiration were recorded during voluntary locomotion, brief whisker stimulation, sleep state transitions or experimentally applied abdominal compression. Cross-correlation analysis was performed between simultaneously recorded neural, respiratory and hemodynamics signals to quantify the relationship between fluctuations. Data are presented as mean ± standard deviation (SD) unless stated otherwise. Additional experimental procedures and analysis methods are provided in the SI Materials and Methods.

Significance Statement.

Zhang et al. identified a novel, peripheral mechanism regulating cerebral blood flow: ultrafast constriction of bridging veins and the superior sagittal sinus driven by abdominal muscle activity. These constrictions, triggered during behavioral transitions and mimicked by increasing abdominal pressure, reflect mechanical coupling between the viscera and brain vasculature. This process alters venous blood volume and can generate fluorescence signals that resemble neural activity, posing a confound for functional imaging. Unlike microvascular responses driven by local neural signals, this pathway operates via body-generated mechanical forces. Their findings reveal an unexpected brain-body interaction that challenges current models of neurovascular regulation and highlights the need to account for systemic physiological signals and behavior in interpreting brain imaging data.