2.1.

Animal Preparation

A total of $N=11$ males C57/BL6 mice (postnatal 8 weeks old, 20- to 25-g weight) were used for this study, of which three died before data acquisitions due to procedures. Four mice were used for vascular ${\mathrm{PO}}_2$ measurements and four mice were used for tissue ${\mathrm{PO}}_2$ recordings. The Animal Research Ethics Committee of the Montreal Heart Institute approved all surgical procedures, which were performed according to the recommendations of the Canadian Council on Animal Care. All procedures were done according to the animal research: reporting of in vivo experiments guidelines (ARRIVE).

Animals were anesthetized with urethane (1 to $1.6\mathrm{g}/\mathrm{kg}$, intraperitoneal injection), tracheotomized, and maintained at constant body temperature (37°C) with a controlled physiological monitoring system (Labeotech, Lachine, Canada). A moderate flow of ambient air lightly supplemented with oxygen was supplied next to the tracheotomy (10% oxygen, 90% air, $1\mathrm{L}/\min$). The mice were then head-fixed with a stereotaxic device. After the injection of a local anesthetic (Xylocaine, subcutaneous, 0.2%) under the scalp, a $\sim 2\times 2\mathrm{mm}$ cranial window was opened over the right hemisphere to expose the somatosensory cortex and surrounding brain (anteroposterior: $-1.5\mathrm{mm}$ and dorsoventral: $+1.5\mathrm{mm}$). A small hole was drilled next to the cranial window for the injection of 4-AP. For tissue ${\mathrm{PO}}_2$ measurements, $\sim 1\mu \mathrm{L}$ of the oxygen-sensitive dye (PtP-C343, $200\mu \mathrm{M}$, synthetized based on published procedures)¹⁶ was injected into the tissue and the cranial window was sealed with 1% agarose in artificial cerebrospinal fluid (125 mM NaCl, 10 mM HEPES, 10 mM glucose, 5 mM KCl, 1,5 mM ${\mathrm{CaCl}}_2$, and 1 mM ${\mathrm{MgSO}}_4$) using a $150\text{-}\mu \mathrm{m}$-thick microscope coverslip. For vascular measurements, the same dye was injected through the tail vein at 10 to $15\mu \mathrm{M}$ initial concentration. A catheter in the femoral artery was used to monitor blood gases (${\mathrm{PCO}}_2$, 36 to 39 mm Hg and ${\mathrm{PO}}_2$, 110 to 160 mm Hg). The average blood pressure (80 to 110 mm Hg) was measured using volume–pressure recordings performed immediately prior to imaging. For each experiment, we labeled the blood plasma with fluorescein isothiocyanate conjugated with dextran (Sigma; 128 mM, $200\mu \mathrm{L}$) and used TPM to obtain a high-resolution structural image of the microvasculature. A larger image of the surface microvasculature ($\sim 2\times 2\mathrm{mm}$) was also obtained, which was later used for easier tracing of the pial arterioles and venules.

2.2.

Epileptogenesis and Electrophysiology

Epilepsy was induced by injecting the potassium channel blocker 4-AP (Sigma; 15 mM, $0.5\mu \mathrm{L}$)⁹ through a glass microelectrode using a syringe pump controller (UMP3, WPI) into a small hole with a depth of $\sim 300\mu \mathrm{m}$ next to the cranial window at a rate of $100\mu \mathrm{L}/\min$. Extracellular local field potentials (LFPs) were recorded with a tungsten electrode (impedance, $\sim 1$ to $2\mathrm{M}\mathrm{\Omega }$) lowered to a depth $\sim 300\mu \mathrm{m}$ into the neocortex. The signal from the electrode was filtered by a bandpass filter (1 to 5000 Hz), amplified with a microelectrode AC amplifier (model 1800, A-M system, Sequim, Washington), and digital filtered between 0.2 and 130 Hz.¹³

2.3.

Two-Photon Microscope Setup, Acquisition, and Processing

Measurements were collected on a custom-built time-domain two-photon phosphorescence lifetime microscope setup with 80 MHz, 150 fs pulses from a MaiTai-BB laser oscillator (840 nm, Newport Corporation, Irvine, California) with output going through an acousto-optic modulator (ConOptics) to adjust the gain for depth-dependent two-photon excitation intensity and gate the laser during lifetime imaging. Reflected light was collected by a water-immersion objective ($20\times$, 1.0 NA; Olympus) and then separated into two beams (phosphorescent photons and fluorescent photons) by dichroic mirrors. Phosphorescent light was passed through a filter centered at 680 nm and detected by a first photomultiplier tube (H7422, Hamamatsu, Photonics, Japan) and fluorescent light was filtered around center wavelength of 520 nm and detected by the second photomultiplier tube (R3896, Hamamatsu Photonics, Japan). Scanning and data recordings were controlled by custom-designed software written in MATLAB^® (MathWorks, Natick, Massachusetts). Excitation pulses of $25\mu \mathrm{s}$ followed by $275\mu \mathrm{s}$ lifetime recording were averaged 1000 times. Phosphorescence lifetime was fitted with a single-exponential function using a least-squares method. The lifetime was converted to ${\mathrm{PO}}_2$ using a calibration plot obtained in independent oxygen titration experiments of the same dye batch.¹⁵

For each animal, a picture of the craniotomy along with the real-time display of the surface vessels scanned was used to localize arterioles and venules. These and real-time imaging of two-dimensional vascular planes were used to guide manual positioning of ${\mathrm{PO}}_2$ recordings on and along targeted vessels. Tissue ${\mathrm{PO}}_2$ measurements were at a depth $\sim 200$ to $300\mu \mathrm{m}$ and were chosen close to arterioles. Vascular ${\mathrm{PO}}_2$ measurements were recorded from the surface until a depth $\sim 250\mu \mathrm{m}$.

2.4.

Simulation of Oxygen Diffusion in Tissue

The Krogh model¹⁷ was used to simulate the initial dip in tissue oxygen tension at different distances from the arterioles and compare with experimental data

To roughly model the undershoot-minima time versus distance from the vessel, the diffusion equation was solved for the same geometry with the initial conditions of baseline ${\mathrm{CMRO}}_2$ and $T_0$ assuming only ${\mathrm{CMRO}}_2$ changes during epilepsy. The time to reach the steady state was used as an estimate for the undershoot-minima time.

3.1.

Oxygen Partial Pressure Changes in Tissue at the Epileptic Focus during Seizure-Like Activity

Seizure-like activity was characterized by first rhythmic spiking of increasing amplitude and decreasing frequency, evolving into rhythmic spikes and slow wave activity prior to gradual offset [see e.g., Figs. 1(a) middle and 1(b) middle].

Fig. 1

Representative changes in ${\mathrm{PO}}_2$ in (a) tissue and (b) arteriole, in response to epileptic events in a local area. Grayscale angiogram of cortical pial tissue with points of interest [with arterioles shown by the red arrows, a venule shown by the blue arrow, and a typical ${\mathrm{PO}}_2$ time course shown for the blue points, top of (a) and (b) panels]. The time course of LFPs indicates seizure initiation. (a) Epileptic activity induced a transient dip in tissue ${\mathrm{PO}}_2$ followed by an increase in ${\mathrm{PO}}_2$ at the focus. (b) In the arteriole, the ${\mathrm{PO}}_2$ profile was also biphasic with an early dip followed by an increase.

A typical change in tissue ${\mathrm{PO}}_2$ is shown in Fig 1(a). The ${\mathrm{PO}}_2$ time-profile was biphasic with an early dip after ictal onset, followed by a longer duration increase in ${\mathrm{PO}}_2$. These results were in agreement with previous works where similar responses were observed using confocal microscopy or oxygen electrodes.^6,13

We then investigated the spatial distribution of this dip along small arterioles. Four mice were used to measure the ${\mathrm{PO}}_2$ in tissue at multiple locations near an arteriole ($\sim 16$ arterioles per mouse) that was located in the epileptic focus, i.e., $<1.5\mathrm{mm}$ from the 4-AP injection site. Figure 2(a) shows that the percent ${\mathrm{PO}}_2$ changes during the dip were significantly lower in tissue closer to arterioles than that located father away.

Fig. 2

(a) Changes in the amplitude of the percent initial dip around small and large arterioles, as a function of the perpendicular distance. The solid line and dashed line show simulated results with small and large arterioles separately. Error bars represent standard error of the mean (SEM). (b) Timing of the dip extrema as a function of arteriolar size and perpendicular distance with simulated results. (c) Percent change of the initial dip in arterioles and venules in response to seizure-like activity, grouped by vessel diameter, with corresponding standard error reported. (d) The undershoot-minima time in arterioles and venules, grouped by vessel diameter, with corresponding SEM.

This change was vessel size dependent as seen by separating data according to diameter in two groups, $12.60\pm 0.83\mu \mathrm{m}$ ($n=38$, $\text{mean}\pm \mathrm{SD}$) and $14.65\pm 1.05\mu \mathrm{m}$ ($n=25$) with significantly larger changes in tissue surrounding smaller arterioles ($t$-test, $p<0.001$).

The undershoot-minima time [Fig. 2(b)] was also seen to be lower when the point was located closer to the arteriole or next to a larger vessel ($t$-test, $p=0.005$). Both trends of initial dip and undershoot-minima time could be predicted from simulations with the Krogh–Erlang model and diffusion equation [lines in Figs. 2(a) and 2(b)].

3.2.

Epileptic Seizures-Induced Oxygen Partial Pressure Response in Vasculature at the Epileptic Focus

Since smaller arterioles are likely to be located further downstream than larger ones, we investigated ${\mathrm{PO}}_2$ changes in upstream and downstream vascular segments during similar ictal events. Four mice were used to gather recordings at multiple locations in arterioles and venules at the epileptic focus ($<1.5\mathrm{mm}$ from the injection site). Figures 2(c) and 2(d) show the percent changes associated with the initial dip and undershoot-minima time in arterioles and venules categorized by diameter. Figure 2(c) shows a size-dependent initial dip with smaller vessels seeing larger changes. Moreover, the observed variations in ${\mathrm{PO}}_2$ were larger in venules than arterioles ($t$-test, $p=0.03$). Figure 2(d) further indicates that the undershoot-minima time of large arterioles and venules occurred earlier than small arterioles and venules. There was also a significant difference between the undershoot-minima time values of the arterioles and venules ($t$-test, $p=0.01$).

4.1.

Relationship between the Initial Dip and Distance from Arterioles

Exploiting spatial measures around small arterioles, we investigated how tissue oxygen pressure changed near arterioles located in the epileptic focus region during epileptiform activity. Our data indicate a decreased ${\mathrm{PO}}_2$ percent change surrounding small arterioles during early phases of epileptic activity (initial dip), which can be attributed to increased tissue metabolic consumption using a Krogh–Erlang model. While capillary delivery may increase its contribution, in these events they do not fully compensate for consumption needs with small and large arterioles remaining largely responsible for oxygen distribution in the cortex according to modeling.¹⁹ This is consistent with previous findings of oxygen exchange between large arterioles and tissue.¹³ Our results further suggest that it is tissue areas located far from larger arterioles, in the capillary bed, which are more susceptible to hypoxia rather than tissue next to arterioles. Temporally, our data show that points far away from the arteriolar wall take more time to recover basal ${\mathrm{PO}}_2$ than locations near small arterioles with a size-dependent arteriolar influence. The observed delay times are larger than typical blood transit time ($\sim 2\mathrm{s}$) in the neurovascular unit. Overall these results suggest that the increased CBF and cerebral blood volume in the focus during epileptic events will supply more oxygen to the tissue near arterioles but may not always meet the demands of oxygen metabolism in capillary areas. An overshoot in CBF might be required to supply these areas adequately to avoid hypoxia.

4.2.

Relationship between the Initial Dip and Diameter of Arterioles and Venules

The largest fractional decreases in vascular ${\mathrm{PO}}_2$ were measured in small venules and small arterioles. As upstream vessels deliver more oxygen during intense neural activity, downstream and smaller vessels see decrease both due to upstream consumption and local tissue need. The lower dip observed in large venules also suggest the presence of short vascular paths between arterioles and venules or shunts.