Spontaneous BOLD waves – A novel hemodynamic activity in Sprague-Dawley rat brain detected by functional magnetic resonance imaging

Abstract

We report spontaneous hemodynamic activity termed “Spontaneous BOLD Waves” (SBWs) detected by BOLD fMRI in Sprague-Dawley rats under medetomidine anesthesia. These SBWs, which lasted several minutes, were observed in cortex, thalamus and hippocampus. The SBWs’ correlates were undetectable in electrophysiological recordings, suggesting an exclusive gliovascular phenomenon dissociated from neuronal activity. SBWs were insensitive to the NMDA receptors antagonist MK-801 but were inhibited by the α1-adrenoceptor blocker prazosin. Since medetomidine is a potent agonist of α2 adrenoceptors, we suggested that imbalance in α1/α2 receptor-mediated signalling pathways alter the vascular reactivity leading to SBWs. The frequency of SBWs increased with intensity of mechanical lung ventilation despite the stable pH levels. In summary, we present a novel type of propagating vascular brain activity without easily detectable underlying neuronal activity, which can be utilized to study the mechanisms of vascular reactivity in functional and pharmacological MRI and has practical implications for designing fMRI experiments in anesthetized animals.

Introduction

Functional MRI is an imaging tool that provides a real-time readout of brain activity in both humans and animals. It reflects functional hyperemia^1,2 represented by vasodilatation of small vessels within ∼200–250 µm from the foci of increased neuronal activity. This process is tightly controlled by both neuronal and astroglial regulatory mechanisms.^3–5 Neurons release vasoactive neuromediators such as acetylcholine, noradrenaline, serotonin dopamine or nitric oxide, whereas astrocytes use distinct signaling molecules (prostaglandins and arachidonic acid derivatives) depending on the brain region and physiological context.^3,6–9 Dynamic modulation of local blood flow is a cornerstone of blood oxygen level dependent (BOLD) functional MRI that is commonly used in research, in characterization of disease models and in the pre-clinical drug testing.¹⁰ Disruption of the components of neurovascular cascade is involved in pathogenesis of several major neurological disorders, such as migraine, neuropathic pain, Alzheimer disease and stroke.¹¹ Understanding the regulatory pathways that match local brain activity with the regional cerebral blood flow (CBF) is specifically important for dissection of neuronal and glial mechanisms in healthy or diseased nervous system.

The fMRI in preclinical and basic neurobiological research has some limitations and practical challenges.¹² In particular, deployment of fMRI requires careful selection of the subjects (species, strain, vendor) and experimental protocol. Despite recent advancements in awake fMRI, usually animals need to be anesthetized during the experiments, mainly to avoid motion artifacts and restrain-induced stress but also to allow concurrent invasive procedures, stimulation and sampling. Effects of anesthesia on neurovascular coupling and interpretation of experimental data have been a subject of debate since the dawn of fMRI.¹³ Anaesthetic agents such as isoflurane, halothane, α-chloralose, urethane, medetomidine (MED), etc. have distinct and largely different modes of action and influence fMRI readouts. Numerous studies have found differential effects of anesthesia on the resting state fMRI as well as on fMRI recordings of responses to various functional and pharmacological stimuli.^14,15

MED is a potent and selective α2 adrenoceptor agonist with extremely high α2/α1 selectivity ratio (>1500¹⁶). It is used in humans as well as in laboratory animals with hypnotic, sedative and analgesic effects.¹⁷ Initially, MED was employed as a supplementary component to provide analgesia and myorelaxation or to reduce the requirement for volatile anesthetics such as halothane.¹⁸ Nevertheless, in recent years, MED has joined the group of anesthetics for functional neuroimaging in small animals.^19,20

Regulation of local vascular tone involves a complex interplay of multiple factors, such as changes in pH, ion fluxes (K⁺, Ca²⁺ and Na⁺) and molecules (NO, arachidonic acid and its derivatives), which directly affect myocytes of arterioles.^5,21 Noradrenaline released from ubiquitous projections of locus coeruleus neurons,²² is particularly important for regulation of blood flow. In the brain, α2 adrenoceptors are localized in neurons (mainly presynaptically being a part of feedback system) and in neuroglia.²³ In astrocytes, noradrenaline is the main regulator of Ca²⁺ signaling in the adult brain,^24,25 and it is also involved in maintaining tonic regulation of blood flow independent from phasic, activity-related coupling mechanisms.²⁶ These mechanisms are involved in pre-adjusting the whole-brain blood flow to respond to salient stimuli, startle and attention shift.²⁷

Here, for the first time, we report a spontaneous hemodynamic activity termed “Spontaneous BOLD Waves” (SBWs). Currently, there are few neuroimaging observations showing similarities to our experimental finding. These existing observations fall into three major categories: spreading depolarization (SD), epileptiform activity (EA) and calcium waves (CWs). Both SD and EA have distinct electrophysiological features and occur (being induced) only in specific experimental models. Spontaneous CWs have never been reported in fMRI studies before, since their detection requires more direct and sensitive methods such as multiphoton imaging. Therefore, we characterized and differentiated SBW from other analogous phenomena. We employed fMRI together with pharmacological interventions using glutamate antagonist MK-801 and α1 adrenoceptor inhibitor prazosin. The non-selective NMDA receptor channel blocker MK-801 is effective in suppressing SD,²⁸ whereas prazosin is inhibiting astroglial CWs in vivo.²⁵ We further performed electrophysiological measurements and studied the effects of experimental conditions on the SBW occurrence and parameters.

Materials and methods

Animals and procedures

Adult (10–14 weeks) male Sprague-Dawley (HSD, Harlan, Netherlands) rats (n = 24), 328 ± 22 g were used in this study. Group sizes were calculated using preliminary data with 100% SBW incidence rate in control group. Animals were randomly split into five groups shown in Table 1.

Table 1.

Blood gases, pH and oxygen saturation values in all groups obtained before and after the phMRI measurement (mean ± SD).

Group	n	pCO2	pO2	sO2	pH
Control, 7 ml/kg	5	42.4 ± 7.2	102 ± 16	96.8 ± 0.6	7.39 ± 0.11
Control, 10 ml/kg	7	31.6 ± 9.4	138 ± 11	99.3 ± 0.4	7.44 ± 0.1
MK-801	4	34.1 ± 4.8	124 ± 9	99.1 ± 0.5	7.41 ± 0.06
Prazosin	5	32.8 ± 6.1	120 ± 12	99.4 ± 0.4	7.4 ± 0.08
EEG	5	35.4 ± 5.7	134 ± 16	99.6 ± 0.2	7.42 ± 0.11

Animals were allowed to accommodate for at least two weeks after the delivery prior to experimentation. All animal procedures were approved by the Committee for the Welfare of Laboratory Animals of the University of Kuopio and the Provincial Government of Eastern Finland, and conducted in accordance with the guidelines set by the European Community Council Directives 2010/63/EEC. Rats were housed in groups of two to three per cage in rooms with controlled environment (12-h day/night cycle, temperature 22℃, humidity 50–60%) with ad libitum access to food and water.

Physiological monitoring

All parameters of the respiratory function obtained by arterial blood gas analysis during the experiments showed good stability of the subjects throughout the imaging and remained within commonly accepted physiological range (Table 1).

Partial pressure of carbon dioxide and pH was slightly higher in rats with ventilation tidal volume of 7 ml/kg compared to other groups, but the difference did not reach significance.

Animals were initially anesthetized with isoflurane (5% for induction and 2.0–2.3% for maintenance) in a N₂/O₂ 70/30%% gas mixture. Polyethylene cannulas (BD Intramedic™ PE-10, Franklin Lakes, NJ, USA) were surgically inserted into femoral artery for blood sampling and to femoral vein for constant infusion of MED and muscle relaxant (Pancuronium bromide). Tracheotomy was performed and custom-made tracheal tube (wedge-cut BD 14 G Angiocath catheter, Franklin Lakes, NJ, USA) was inserted for mechanical ventilation. The temperature of the animals was maintained at ∼37℃ during the surgery with a feedback-controlled heating pad (Harvard Apparatus Inc., Holliston, MA, USA).

After surgical procedures, rats were transferred to pre-heated rat MRI holder, fixed with a bite bar and earplugs. Then tracheal tube was connected to mechanical ventilator and pancuronium bromide 0.05 mg/kg was administered as an intravenous bolus. Approx. 10 min later, after physiology monitoring sensors were attached to the animal, infusion of MED 0.1 mg/kg/h and pancuronium bromide 0.05 mg/kg/h was initiated. Additionally, intraperitoneal line (BD Intramedic™ PE-20, Franklin Lakes, NJ, USA) with 26 G needle was inserted in the animals that were treated with MK-801 or Prazosin during the fMRI scan.

BOLD measurements

All fMRI experiments were performed on Bruker 7 T PharmaScan imaging system. A volume coil (Bruker Biospin GmbH, Ettlingen, Germany) was used for transmission and a surface quadrature array coil was used for receiving (Rapid Biomedical GmbH, Rimpar, Germany). Anesthetized animals were fixed to a head holder and positioned in the magnet bore in a standard orientation relative to the gradient coils. For anatomical reference, coronal T2-weighted TurboRARE MR images covering the whole cerebrum were acquired with following parameters: TR 3.5 s, Effective TE 26 ms, FOV 2.5 cm × 2.5 cm, matrix size 256 × 256 (in-plane resolution of 97 µm × 97 µm). Single-shot spin-echo echo-planar imaging (SE-EPI) sequence was used to obtain BOLD fMRI data with following parameters: TR 2 s, TE 45 ms, 4500 scans (2.5 h scan time), 25 × 25 mm field of view and 64 × 64 acquisition matrix, yielding in-plane resolution of 391 × 391 µm. Slice pack of 11 × 1.5 mm axial slices was placed to cover the whole cerebrum with olfactory bulb edge used as a landmark for identical positioning.

Blood samples were obtained immediately before the start and right after the end of each fMRI scan. Values for pCO₂, pO₂, sO₂ and pH were obtained from the samples with an i-STAT blood gas analyzer (Model 300, Abbott Point of Care Inc., Princeton, NJ, USA) immediately after sampling. In-between physiological state was assessed by continuous monitoring using MR-compatible small animal monitoring system (Model 1025, Small Animal Instruments Inc), including rectal temperature probe, respiration pneumatic sensor, pulse oximeter and capnograph. The animals were maintained at a stable body temperature of ∼37℃ in the animal MRI holder with a warm water circulation system (Thermo Fischer Scientific, Loughborough, England).

Simultaneous LFP recordings

Scalp of the animal was removed and a small craniotomy (0.5 mm in diameter) was performed to access the right hemisphere (0.5 mm anterior from bregma, 3 mm lateral from midline, at the depth of 0.5 mm from the surface of the brain (Paxinos, Watson 1998). A 30 mm long Teflon-coated tungsten wire (50 µm diameter, 7.9 µm² tip size, California Fine Wire, Grover Beach, CA) electrode was implanted into the somatosensory cortex using a stereotaxic electrode holder. The electrode was subsequently bent towards the neck of the animal and glued (Permabond Engineering Adhesives Ltd, UK) to the skull to prevent detachment during MRI surface coil placement. Partially insulated chloridized silver wire (0.25 mm diameter) reference and ground electrodes were inserted subcutaneously in the neck.

Electrophysiological recordings were performed using a BrainAmp MR plus system with BrainVision Recorder software (Brain Products GmbH, Munich, Germany). Signal from the electrodes was amplified and digitized with sampling rate of 5 kHz.

EEG measurements

Four recording screw electrodes were located in sensorymotor and visual cortices on left and right hemispheres, 1 mm anterior and 5 mm posterior from bregma, 3 mm lateral from midline. In addition, two similar screw electrodes were implanted in cerebellum as ground and reference electrodes. The connector was attached to a head stage (HST/16-TR-GR-G1, Plexon, USA) and the signal was further amplified × 1000 with an AC amplifier, (16-Channel Extracellular Differential AC Amplifier Model 3500, A-M System, USA) and digitized at 2 kHz per channel (DT2821 series A/D board; Data Translation, Marlboro, MA, USA). The signal was bandpass-filtered between 1 and 3000 Hz. The data were acquired using Sciworks 5.0 program (DataWave Technologies, Loveland, CO, USA).

Pharmacological intervention

All compounds (Sigma Aldrich, Seelze, Germany) were fresh formulated at the experimental day. Prior to the injection, all solutions were slowly pre-warmed to 36℃ using the heated water bath. Dosing was performed during the imaging through the pre-implanted ∼70 cm long PE20 (intraperitoneal) or PE10 (intravenous) cannula and flushed with 0.2 ml of saline. Non-selective glutamate receptor antagonist MK-801 was prepared as 10 mg/ml saline solution and injected as 1 ml/kg bolus. The α1 receptor blocker prazosin was given at a dose of 0.1 mg/kg saline solution as a slow intravenous bolus injection.

Ventilation parameters

All animals were paralyzed and mechanically ventilated using the volume-controlled mechanical ventilator (Inspira ASV, Harvard Apparatus Inc.) connected to 4 m-long tubing (Tygon®, Saint-Gobain, France) routed into MRI room and animal holder. That enabled a full control over respiration and complete immobilization. The respiration rate was fixed to 65 bpm, and tidal volume was typically 10 ml/kg. In one group, the tidal volume was set to 7 ml/kg, leading to minor hypoventilation but still within the physiological range of values for this parameter.

Electrical forepaw stimulation

In a subset of animals (n = 2) that received prazosin treatment (n = 5), electrical forepaw stimulation was used as control to show that fMRI response to standard stimulus could be detected under present experimental conditions as described in Huttunen et al.²⁹ Briefly, bipolar 1 mA, 10 Hz electrical pulses were applied for 30 s every 10 min throughout the measurement using a constant current stimulator (A-M Systems, Sequim, WA) through stainless steel needle electrodes inserted into the right forepaw.

Data analysis and reporting

All fMRI data analysis was performed using in-house MATLAB scripts (The Mathworks Inc., Natick, MA, USA) as described previously.³⁰ Briefly, all volumes were first corrected for slice-timing differences, followed by motion correction and spatial smoothing with a 0.8 × 0.8 × 0.1 mm FWHM Gaussian kernel. Individual datasets were coregistered to the reference brain with manually delineated regions of interest (ROI) atlas. Each BOLD time series taken from specified ROIs was de-trended prior to statistical analysis. Statistical parametric mapping was done with SPM8 (http://www.fil.ion.ucl.ac.uk/spm) using generalized linear model, where blocks of individual SBWs were compared to preceding baseline of 10 min (300 data points). Resulting statistical activation maps (paired t-test, FDR-corrected, thresholded to p < 0.05) were overlaid on high-resolution T2-weighted image.

Due to heterogeneity of the signals, temporal SBWs characteristics were analyzed manually, from mean BOLD timeseries of all the voxels in selected ROIs, with onset, time to peak and duration values recorded for each subject.

Quantified SBWs features, such as incidence and duration in all test groups were compared against saline control group using Mann–Whitney two-tailed t-tests with a significance level set at p < 0.05. Values from multiple SBWs were first averaged within individuals and then within each group.

Exclusion criteria included persistent artifacts in the recordings and/or physiological monitoring data outside of the accepted range. Data analysis was performed blinded to the treatment group and results reported according to ARRIVE guidelines (https://www.nc3rs.org.uk/arrive-guidelines).

Results

Anesthesia specificity

The SBWs described below appeared only in MED-anesthetized animals and never occurred when either urethane (1.3 g/kg i.p. bolus) or isoflurane (1.2-1.3% imaging dose) protocols was used (data not shown).

Three types of SBWs

The SBWs were recorded in various brain areas in all untreated and unstimulated animals under MED anesthesia (n = 12). Based on the individual features, SBWs were arbitrarily classified into three main types (Figures 1 and 2):

• Type I: a subcortical SBW was presented in nine animals by unilateral hippocampal activation, often with simultaneous adjacent thalamic engagement (Figure 1(a), Suppl. Video 1) but without substantial cortical involvement.

• Type II: characterized by prominent symmetrical activation of midline cortical structures along the superior sagittal sinus, involving medial prefrontal, cingulate and retrosplenial cortices (n = 12, Figure 1(b), Suppl. Video 1) without subcortical contribution.

• Type III: progressive engagement of subcortical structures such as hippocampus and thalamus, followed by cortical surface activation (n = 10, Figure 1(c), Suppl. Video 2) resembling BOLD readout of the cortical spreading depression waves³⁰ (CSD) but without characteristic electrophysiological signature (Figure 4(c)).

Figure 1.

BOLD activation maps of three main types of SBW in different brain regions. (a) SBW Type I, unilateral activation in hippocampus with involvement of adjacent middle thalamic areas. (b) SBW Type II, bilateral “parting” of cortical activation, spanning along sagittal sinus from retrosplenial cortex in the back to cingulate cortex in the front. (c) SBW Type III, example of highly variable isolated cortical activation engaging majority of the cortical surface. Each panel shows a representative SPM map of a whole single SBW event from different individual animals overlaid on T2-weighted anatomical image of rat brain from back (top-left slice) to front (bottom-right slice). Color bar shows the t-value for all the voxels with p < 0.05 significance (FDR corrected).

Figure 2.

BOLD signal time series from individual ROIs in the control group of naïve Sprague-Dawley rats. Each set of signal traces represents an individual animal, whereas each line in a set is the BOLD signal time course from single ROI (ROI’s: CinCtx: cingulate cortex; SSCtx: somatosensory cortex; Thal: thalamus, Hip: hippocampal ROI; Ventr: ventral periventricular areas). SBW is marked as follows: I-predominantly cortical; II-subcortical; III – large scale combined waves.

Figure 4.

A representative example of SBW shown in (a) BOLD time series from 3 × 3 voxels (∼1.5 mm³) ROI located around the LFP electrode tip; (b) DC recording from the same electrode (black trace); (c) example of BOLD and (d) LFP signal recordings of cortical spreading depression (adapted from Shatillo et al.³⁰).

No hemisphere preference or systematic succession of activated regions was observed regardless of SBWs’ type.

Region-specific activation features

Next, we compared the temporal profile of SBWs observed in various brain regions (Figure 2).

In SBWs Type I, hippocampal involvement regardless of laterality and initial onset had average duration of 286 ± 62 s. Majority of SBWs Type I events involving thalamus had a sharp onset and relatively short duration (288.1 ± 58.7 s). However, a small group of thalamic events had a sustained activation lasting up to 20 min (850.7 ± 269 s, n = 6 in four out of seven animals).

Similar temporal heterogeneity was observed in SBWs Type II and III. While in majority of cases (n = 22 in all animals), these events lasted for 262.9 ± 39.8 s, in a smaller subset (n = 6 observed in three rats), there was ∼ 4-fold longer duration (1000 ± 71 s). Notably, in SBWs Type I and III that were originating in subcortical structures, a negative BOLD signal (120.3 ± 27.8 s) was detected in posterior periventricular regions. In contrast, when only cortical areas were activated (SBW Type II), this negative component was absent.

Intensity of ventilation affects SBW

The BOLD signal recorded under normal conditions are tightly coupled with CBF and cerebral blood volume (CBV) changes,³¹ which are, in turn, linked to pCO₂ and pO₂.^32,33 Since animals were anesthetized and paralyzed, we were able to adjust pCO₂ and pO₂ by controlling mechanical ventilation, which allowed modulating the CBF to test how mechanical ventilation affects the incidence of SBWs. During the 2-h imaging session in five rats with stable ventilation (respiration rate 65 breaths/min), the tidal volume (Vt) was stepwise increased from 7 ml/kg to 10 ml/kg. Surprisingly, this resulted in a significant rise in the rate of SBWs (medians of 0.5 vs. 1.6 events/h, p < 0.01, Figure 3(a)). Notably, this inhibition was not associated with considerable changes in pH, which remained stable in these conditions.

Figure 3.

Modulation of SBWs incidence in rat’s cortex. (a) Modulation of SBW incidence by tidal volume of mechanical ventilation with stepwise increase of tidal volume from 7 ml/kg to 10 ml/kg within 1 h imaging time intervals (n = 5, *** p < 0.01). All data shown as Mean ± SD. (b) Pharmacological treatment with NMDA receptors blocker MK-801 (n = 4), and α₁-adrenoceptors selective antagonist prazosin (n = 5) vs. saline control (n = 7). ND: not determined.

SBWs are not accompanied with robust changes in electrical activity

Commonly recorded waves of CSD are manifested by both vascular BOLD signals and electrical local field potentials (LFPs), reflecting neuronal activity.³⁰ In order to test if SBW represents a novel type of the electrical brain activity, similar to CSD, we performed LFP recordings in direct current (DC) mode during the BOLD measurements (Figure 4). Unlike CSD, electrical brain activity corresponding to SBW was not detectable in LFP recordings. Power spectrum of low frequency (0.01–0.5 Hz) fluctuations demonstrated only minimal graduate increase in overall power over time, likely reflecting ongoing electrochemical processes.

In order to confirm with different approach, the lack of obvious electrical activity during SBWs, we performed EEG recordings (n = 5) with experimental procedures and conditions identical to fMRI experiments. A thorough inspection of individual EEG traces was performed along with FFT and histogram analyses for all the available recording times in all four channels. Interestingly, no abnormal electrical activity was detected in any of the subjects during 3-h recording time (Supplementary Figure 1).

Pharmacological modulation of SBW

What neurochemical mechanism underlies the SBW? In order to address this issue, we first tested the action of the general NMDA blocker MK-801 which completely prevented the BOLD waves of CSD in rats in our previous study.²⁸ However, in contrast to CSD, even a high concentration of MK-801 (10 mg/kg) did not reduce the occurrence of SBWs (n = 4, Figures 3(b) and 5). This result implied that glutamatergic neurotransmission mediated by NMDA receptors is not required for SBW generation.

Figure 5.

Average duration of BOLD activation in selected brain regions in the control group (n = 6, white bars) vs. animals treated with MK-801 (n = 4, black bars) and prazosin (n = 5, grey bars).

In contrast, pharmacological blockade of noradrenergic receptors was highly efficient in modulation of SBW. Thus, the potent α₁-adrenoreceptor antagonist prazosin significantly diminished the occurrence of SBW, eliminating them from cortical regions and reducing their duration and amplitude in subcortical areas (Figures 3(b) and 5). Thus, noradrenergic system was apparently involved in control of generation of SBWs in the brain.

Strong vasodilatation caused by prazosin could potentially alter fMRI readout by diminishing normal hemodynamic response below the BOLD detection limits, potentially rendering SBW invisible to our measurement.

To explore this possibility, we measured BOLD signals during forepaw stimulation as described in Huttunen et al.²⁹ in two prazosin-treated animals. Figure 6 shows that although prazosin reduced the activated area in somatosensory cortex, even mild (intensity 1 mA) forepaw stimulation produced clearly detectable activation pattern in forepaw projection areas of contralateral primary somatosensory cortex (S1).

Figure 6.

Group activation maps of repeated forepaw stimulation before (a) and after (b) prazosin treatment (n = 2, 5 ON/OFF blocks/animal). Overall magnitude of the response is diminished but the activation was still present in the primary somatosensory cortex (S1).

Discussion

In this study, we present a novel type of spontaneous BOLD activity observed in the rat brain. This phenomenon was detected in Sprague-Dawley rats in fMRI experiments under MED anesthesia; importantly, it was absent when using another anesthetic agents. Main characteristics of the phenomenon, which we called SBW, include: (i) spontaneous onset; (ii) strain-specificity; (iii) facilitation by MED; (iv) dependence on ventilation parameters; and most interestingly, (v) large and propagating brain activation without corresponding robust electrical activity.

Over the course of five years of fMRI research in our lab, we have repeatedly observed large amplitude spontaneous SBWs.³⁴ Initially rendered as an artefact, these events were nonetheless consistent, always appearing in specific experimental setting. This consistency instigated further investigation. SBWs always appeared in Sprague-Dawley rats (HSD, Harlan, Netherlands) anesthetized with MED, paralyzed and artificially ventilated, as described in our previous papers.^35,36 These SBWs were undetectable in simultaneously performed LFP recordings and occurred spontaneously in random fashion during the imaging session. Well-defined BOLD signal signature included high amplitude, long duration and consecutive involvement of several brain areas.

SBWs coverage of large areas of the brain (Suppl. Video 2) resembled the waves of CSD.^28,37,38 Well characterized, CSD is a slowly propagating cortical wave of activity involving neurons, glial cells and vessels^39,40 implicated in migraine with aura and stroke.⁴¹ Of note, CSD similarly to SBWs can propagate to the thalamus⁴² potentially suggesting equivalent underlying mechanism(s). However, further analysis revealed distinct properties of SBWs. First, we could not detect any substantial electrophysiological correlates of SBWs neither in EEG nor in LFP recordings. Second, the broad-spectrum NMDA receptor blocker MK-801, which completely prevented CSD²⁸ had no effect on SBWs. These results ruled out the involvement of neurons and glutamatergic mechanisms in SBW. Third, the time-course of SBW was longer than in CSD. Notably, BOLD signals measured in this study also originate from the vascular events reflecting changes in CBF and CBV. Taken together, these data indicate that SBWs are different from CSDs and SBWs are primarily associated with propagating vascular events (Suppl. Videos 1 and 2). Spontaneous occurrence, highly variable temporal profiles of SBWs, ranging from ∼5 to 20 min, absence of consistent electrical activity and no correlation with the blood gases highlight the impairment in the regulatory mechanisms of the vascular tone. Relatedly, Winder et al.⁴³ recently reported only weak correlation between hemodynamic signals and neural activity during the resting state, however, in the absence of clearly detectable waves. Together, these studies indicate that hemodynamic readouts are not always coupled with neuronal activity during resting state, which have implications in interpretation of rs–fMRI data. Especially, if MED is used in Sprague-Dawley rats, such slow events as SBWs, if present, may profoundly impact rs-fMRI analysis and consequently study results.

What triggers the SBWs? The occurrence of SBWs under anesthesia with the selective α2-adrenoceptor agonist MED and elimination of waves by the α1-adrenoceptor antagonist prazosin indicate the key role of the noradrenergic system in this process. The α1 - and α2-adrenoreceptors are abundant in the CNS.^44,45 These receptors are in particular localized in smooth muscle cells of blood vessels regulating the local blood flow. Noradrenaline is one of the main regulators of the global and local vasculature tone in the brain, mediating functional hyperemia. The release of this neurotransmitter is coordinated by locus coeruleus (LC) neurons, which have widely distributed projections throughout the brain. Noradrenaline release from these projections orchestrates astroglial Ca²⁺-signaling in rodents.²⁷ Prazosin, which disrupts astroglial Ca2 + signals in mice,²⁵ was also efficient in reducing SBWs suggesting possible role of astrocytes. Combining with our data, these results further support gliovascular mechanisms of the SBWs, detached from neuronal activity.

Under normal conditions, ventilation parameters such as Vt and respiration rate influence cerebral vascular tone through pCO2- and pH-mediated mechanisms. However, dependence of the SBWs occurrence on ventilation parameters cannot be explained by changes in pCO2 and pH, as they were not significant and remained in physiological range. Alternatively, increased SBWs incidence can be explained by altered reactivity of the vascular system under this combination of experimental parameters. Changes in respiration parameters may also modulate brain stem and locus coeruleus activity through autonomic nervous system feedback circuits. Yet, both of these hypotheses remain to be experimentally investigated in future studies.

Although highly consistent and robust in all rats from different batches of Harlan Sprague-Dawley strain, SBWs were never observed (in identical settings) in a large cohort of Sprague-Dawley rats from different vendor (Crl:SD, Charles River, Germany), neither SBWs were detected in Wistar rats. This can be explained by the large heterogeneity and even vendor-induced variance in laboratory animal lines. There are known inter- and intra-strain differences in cerebral anatomy,⁴⁶ effects of the pharmacological agents,⁴⁷ reactivity in different pathological conditions,⁴⁸ and behavior⁴⁹ in rats. In particular, Sprague-Dawley rat strain has internal vendor-related (e.g. Harlan vs. Sasco) differences in α1-adrenergic projections in the spinal cord and in the anti-nociceptive effect produced by administration of dexmedetomidine.⁵⁰ If identified, genetic underpinnings of SBWs triggered by specific set of conditions could deepen our understanding of basic mechanisms of BOLD, neurovascular coupling and hemodynamic regulation in health and disease. To address the electrophysiological origin of SBWs and its relation to CSD, induction of CSD should be performed within the same measurements as a conditional control.

In conclusion, the novel type of spontaneous activity, characterized in this study, using BOLD fMRI in naïve Sprague-Dawley rats, represents a rare case of likely non-neuronal cerebrovascular activity found by non-invasive tools. Strong and sustained BOLD activation of large areas of the brain has no robust signature in electrophysiological recording, which points towards its non-neuronal but likely vascular or vascular-glial origin. Highly reproducible in Sprague-Dawley rats over the years, the SBWs activity warrants future investigation of underlying mechanisms and possible applications as a pathophysiological non-invasive model of neurovascular dysregulation in neurological diseases.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Authors’ contributions

The study was devised by AS, RG, OG and HT. Experiments were performed by AS and AL. Data were analysed by AS, RS and AL. Results were discussed by all the co-authors, reviewed, interpreted and compiled into final manuscript by AS, HT, AV, RG and OG.

Supplementary material

Supplementary material for this paper can be found at the journal website: http://journals.sagepub.com/home/jcb