Sleep cycle-dependent vascular dynamics in male mice and the predicted effects on perivascular cerebrospinal fluid flow and solute transport

Abstract

Perivascular spaces are important highways for fluid and solute transport in the brain enabling efficient waste clearance during sleep. However, the underlying mechanisms augmenting perivascular flow in sleep are unknown. Using two-photon imaging of naturally sleeping male mice we demonstrate sleep cycle-dependent vascular dynamics of pial arteries and penetrating arterioles: slow, large-amplitude oscillations in NREM sleep, a vasodilation in REM sleep, and a vasoconstriction upon awakening at the end of a sleep cycle and microarousals in NREM and intermediate sleep. These vascular dynamics are mirrored by changes in the size of the perivascular spaces of the penetrating arterioles: slow fluctuations in NREM sleep, reduction in REM sleep and an enlargement upon awakening after REM sleep and during microarousals in NREM and intermediate sleep. By biomechanical modeling we demonstrate that these sleep cycle-dependent perivascular dynamics likely enhance fluid flow and solute transport in perivascular spaces to levels comparable to cardiac pulsation-driven oscillations.

Introduction

Perivascular spaces (PVS) lined by the astrocytic endfeet, are key passageways for movement and exchange of fluids and solutes, and play important roles for drug delivery into the brain and waste clearance^1,2. Removal of extracellular waste products like β-amyloid (Aβ) is crucial for brain health and the prevention of neurodegenerative diseases such as Alzheimer’s disease (AD)³. Already in the 1970s pioneering work of Helen Cserr and colleagues identified the perivascular compartments of the brain as pathways enabling bulk flow and efficient transport of solutes out of the brain⁴. A model for brain waste clearance—the glymphatic system as proposed in a seminal study ten years ago¹—states that cerebrospinal fluid (CSF) flows along pial arteries, enters the brain via PVS of penetrating arterioles, then flows through the parenchyma collecting extracellular waste, before it exits in PVS along veins⁵. Other studies also found evidence for clearance along arteries⁶. From there, waste may exit the brain along cranial and spinal nerves, via arachnoid granulations (although recently debated)⁷ and meningeal lymphatic vessels^8,9, which all drain into the cervical lymphatic vasculature¹⁰. In addition, there might exist a bidirectional CSF and solute flow along the arteries⁶. Some of the mechanistic underpinnings of the glymphatic system, such as driving forces for perivascular and parenchymal flow and exit pathways, are still debated¹¹. Even so, there appears to be a consensus that the PVS serve as highways for efficient transport of extracellular fluid and solutes in the brain.

Flow along the PVS is thought to be facilitated by mechanical forces created by the vasculature^12,13. Specifically, heartbeat-driven pial artery pulsations have been demonstrated to propel fluorescent microspheres along vessels at the brain surface¹³. Moreover, pharmacologically manipulating blood pressure and heart frequency seems to affect clearance of waste products from the brain¹³. Recent studies have also shown that vasomotion of longer time scales observed in wakefulness may play a role in propelling CSF^14,15.

Brain waste clearance has been shown to be considerably more active in sleep^3,16. This was first demonstrated in 2013 in naturally sleeping head-fixed mice¹⁶. This link between sleep and brain waste clearance has later been reinforced by studies in humans showing that sleep deprivation is associated with reduced clearance and increased build-up of harmful proteins such as Aβ in the brain, and increased risk of neurodegenerative diseases³. The mechanisms underlying the enhancement of waste clearance during sleep are not well understood but have been proposed to depend on an increased extracellular space¹⁶ and coupled blood-CSF flow patterns in non-rapid eye movement (NREM) sleep¹⁷. Recently, sleep state coupled changes in brain perfusion and blood flow dynamics of large surface vessels have been demonstrated in mice^18,19. Yet the importance of a complete sleep cycle, including NREM sleep, intermediate state (IS)²⁰, REM sleep, microarousals and awakening after each sleep cycle²¹ on vascular dynamics and its consequences for PVS dynamics, has not been demonstrated.

Given that CSF fluxes in the brain are dependent on vascular dynamics and that there are prominent blood flow changes in sleep, we hypothesized that sleep specific vasomotion could govern the size of the PVS and facilitate flow of CSF. Hence, we set out to measure the changes in the PVS across the different sleep-wake states by two-photon microscopy. We show that blood vessels exhibit sleep state dependent dynamics. Specifically, we observed slow, large amplitude oscillations of both pial and penetrating arterioles in slow-wave sleep, coupled to reciprocal changes to the PVS of penetrating arterioles, followed by large arterial and arteriolar dilations that started during IS sleep and reached maximum dilation during REM sleep and lasted for the entire duration of REM episode, coupled to a near obliteration of the PVS. During the brief arousal at the end of a sleep cycle after REM sleep and microarousals in NREM and IS sleep, arteries and arterioles constrict and PVS enlarges. Using biomechanical modeling we demonstrate that these sleep cycle-dependent PVS dynamics, and in particular slow vasomotion in NREM sleep, likely play a salient role in driving fluid flow and solute transport in the PVS.

Results

Two-photon imaging of vascular dynamics in natural sleep

To assess the vascular dynamics throughout the sleep cycle, we performed two-photon microscopy linescans across blood vessels in the somatosensory cortex of naturally sleeping GLT1-eGFP transgenic mice expressing enhanced green fluorescent protein (eGFP) in astrocytes with the vasculature outlined by Texas Red-labeled dextran (Figs. 1a, b and 2a, b)²². Pial arteries and veins with corresponding penetrating vessels were identified by their morphology and direction of blood flow. We classified sleep-wake states using an infrared sensitive camera, electrocorticography (ECoG) and electromyography (EMG) (Supplementary Fig. 1). We identified three different behavioral states of wakefulness by analyzing mouse movements on the IR video footage (Supplementary Fig. 1a): locomotion, spontaneous whisking, and quiet wakefulness. In mice, locomotion is tightly associated with whisking²³, and for this reason our locomotion behavioral state comprises both movement and whisking. Using standard criteria on ECoG and EMG^20,24 signals, we identified three sleep states: NREM sleep, IS sleep, and REM sleep (Supplementary Fig. 1b). NREM sleep and IS sleep are sub-states of SWS, where IS sleep is a transitional state from NREM to REM sleep, characterized by an increase in sigma (10–16 Hz) and theta (5–9 Hz) power, and a decrease in delta (0.5–4 Hz) ECoG power^20,22. The mice were trained to fall asleep without any use of anesthesia or sedatives, as established in Bojarskaite et al. 2020, enabling us to monitor a natural progression of sleep states²²….