www.elsevier.com/locate/ynimg

NeuroImage 32 (2006) 520 – 530

Spatial flow-volume dissociation of the cerebral

microcirculatory response to mild hypercapnia

Elizabeth B. Hutchinson, Bojana Stefanovic, Alan P. Koretsky, and Afonso C. Silva*

Laboratory of Functional and Molecular Imaging, National Institute of Neurological Disorders and Stroke, National Institutes of Health,

10 Center Drive, Building 10, Room B1D114, Bethesda, MD 20892-1065, USA

Received 19 September 2005; revised 7 March 2006; accepted 16 March 2006

Available online 19 May 2006

The spatial and temporal response of the cerebral microcirculation to

mild hypercapnia was investigated via two-photon laser-scanning

microscopy. Cortical vessels, traversing the top 200 Mm of somatosen-

sory cortex, were visualized in A-chloralose-anesthetized Sprague–

Dawley rats equipped with a cranial window. Intraluminal vessel

diameters, transit times of fluorescent dextrans and red blood cells

(RBC) velocities in individual capillaries were measured under

normocapnic (PaCO2 = 32.6 T 2.6 mm Hg) and slightly hypercapnic

(PaCO2 = 45 T 7 mm Hg) conditions. This gentle increase in PaCO2 was

sufficient to produce robust and significant increases in both arterial

and venous vessel diameters, concomitant to decreases in transit times

of a bolus of dye from artery to venule (14%, P < 0.05) and from artery

to vein (27%, P < 0.05). On the whole, capillaries exhibited a significant

increase in diameter (16 T 33%, P < 0.001, n = 393) and a substantial

increase in RBC velocities (75 T 114%, P < 0.001, n = 46) with

hypercapnia. However, the response of the cerebral microvasculature

to modest increases in PaCO2 was spatially heterogeneous. The

maximal relative dilatation (range: 5–77%; mean T SD: 25 T 34%,

P < 0.001, n = 271) occurred in the smallest capillaries (1.6 Mm–4.0 Mm

resting diameter), while medium and larger capillaries (4.4 Mm–6.8 Mm

resting diameter) showed no significant changes in diameter (P > 0.08,

n = 122). In contrast, on average, RBC velocities increased less in the

smaller capillaries (39 T 5%, P < 0.002, n = 22) than in the medium and

larger capillaries (107 T 142%, P < 0.003, n = 24). Thus, the changes in

capillary RBC velocities were spatially distinct from the observed

volumetric changes and occurred to homogenize cerebral blood flow

along capillaries of all diameters.

Published by Elsevier Inc.

Keywords: Cerebral blood flow; Brain; Red blood cell velocity; Hyper-

capnia; Two-photon microscopy

Introduction

The most prominent methodologies for non-invasive imaging

of human brain function – functional magnetic resonance, positron

1053-8119/$ - see front matter. Published by Elsevier Inc.

doi:10.1016/j.neuroimage.2006.03.033

* Corresponding author. Fax: +1 301 480 2558.

E-mail address: Silvaa@ninds.nih.gov (A.C. Silva).

Available online on ScienceDirect (www.sciencedirect.com).

emission tomography, and near-infrared spectroscopy – all rely on

tight coupling between focal cerebral hemodynamics and neuronal

activity (Raichle, 2003). A detailed understanding of the mecha-

nisms of local cerebral circulatory regulation is thus critical for the

establishment of accuracy and range of the applicability of these

techniques.

Like the neuronal network itself, the cerebral vascular tree

exhibits both hierarchy and spatial specialization. At rest, local

blood flow varies as much as 18-fold between the different regions

of a rat brain (Fenstermacher et al., 1991), likely resulting from

regional differences in capillary density (Patlak et al., 1984) as well

as from transit time variations (Rosen et al., 1991). Moreover, CBF

undergoes local (in addition to global) regulation so that, on a sub-

millimeter scale, maps of hemodynamic changes closely follow

those of neuronal activity (at least at the columnar level) under a

variety of conditions (Cox et al., 1993; Woolsey et al., 1996;

Malonek and Grinvald, 1996; Duong et al., 2001; Logothetis,

2002). A growing number of studies have suggested that both the

density of the capillary beds as well as the amplitude and the

temporal evolution of blood flow response to functional activation

follow the cortical neuronal architecture (Cox et al., 1993; Gerrits

et al., 2000; Harrison et al., 2002; Silva and Koretsky, 2002; Lu

et al., 2004). While the existence of capillary level structures for

very fine hemodynamic regulation has been demonstrated in

various species (Ehler et al., 1995; Rodriguez-Baeza et al., 1998;

Harrison et al., 2002), the spatial limit of hemodynamic adjust-

ments remains unclear and is a subject of current interest

(Lauritzen, 2001), as it dictates the theoretical limit on the

functional specificity of flow-weighted neuroimaging techniques.

Indeed, it is the microcirculatory CBF control that is of most

interest to brain function investigations due to the proximity of the

capillary network to the activated parenchyma, and thus it is crucial

to understand how capillary diameter and red blood cell (RBC)

velocities are regulated. While a heterogeneous profile of

microcirculatory CBF adjustments has long been suspected

(Rosenblum, 1965), data on the spatial pattern of microvascular

flow regulation in the brain have been scarce, likely due to the

intrinsic difficulty of achieving the required spatial resolution in

vivo. Conventionally, the pial microvessels have been directly

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530 521

observed via intravital microscopy in animals equipped with closed

cranial windows (Navari et al., 1978) and the vessel diameters

measured by a calibrated video microscaler (Morii et al., 1986;

Wagerle and Degiulio, 1994; Parfenova et al., 1995; Ishimura et al.,

1996; Hudetz, 1997; Takenaka et al., 2000, 2003; Iliff et al., 2003;

Xu et al., 2004). Although important information on the effects of a

range of vasoactive agents has been collected, these studies have

been constrained to the pial surface and, typically, to vessels with

diameters larger than 10 Am.

The advent of scanning confocal microscopy enabled imaging

of vessels below the cortical surface, nonetheless with limited

penetration capabilities. Microvascular CBF increases elicited by

hypercapnia (Villringer et al., 1994; Seylaz et al., 1999), as well as

decreases under ischemic conditions (Morris et al., 1999; Pinard et

al., 2000), have been investigated with scanning confocal

microscopy. Two-photon laser-scanning microscopy, on the other

hand, brought in an additional number of important advantages,

particularly for in vivo biological applications, that have afforded

widespread use (Denk et al., 1990) and made significant impact on

in vivo microscopy. Recently, the applicability of two-photon

microscopy to the investigation of cerebral microcirculation has

been demonstrated in rat somatosensory cortex following vibrissae

or hindlimb stimulation (Kleinfeld et al., 1998), rat dorsal olfactory

bulb during odor stimulation (Chaigneau et al., 2003), mouse

somatosensory cortex during bicuculline-induced focal epilepti-

form activity (Hirase et al., 2004) as well as freely moving awake

rats (Helmchen et al., 2001).

The purpose of the present work was to investigate the spatial

and temporal evolution of microvascular CBF regulation using

two-photon laser-scanning microscopy. To elicit changes in vessel

diameter and in RBC velocities, we employed mild hypercapnia

while imaging the microcirculation of the rat somatosensory

cortex. Mild hypercapnia was chosen as an effective and simple

way to elicit robust increases in global resting blood flow, thereby

obviating the need for localization of the affected region and hence

greatly facilitating the use of two-photon microscopy. Using

fluorescently labeled dextrans, we quantified the changes in transit

times from arteries to venules and veins. The concomitant

vasodilatation was quantified by estimating the diameter of

arteries, veins and capillaries from stack maximum intensity

projections. Special attention was given to capillaries. The present

experiments yield new insight into the spatial distribution of CO2-

induced changes in hemodynamics, in terms of both RBC velocity

and intraluminal diameter, and provide detailed information on

microvascular CBF regulation.

Materials and methods

Animal preparation

Twenty male Sprague–Dawley rats, weighing 146–300 g, were

anesthetized with isoflurane (3% for initial induction and 1.5% for

maintenance) in O2-enriched medical air, orally intubated and

mechanically ventilated throughout the remainder of the experi-

ments (with 1.5% isoflurane concentration). Respiratory parame-

ters, including end-tidal CO2 levels, were carefully monitored by a

capnograph (BCI 300 Capnocheck, BCI, Inc, Waukesha, Wiscon-

sin). Rectal temperature was monitored and maintained at 37.0-Cby means of a feedback-controlled warm water bath. The right

femoral artery and vein were cannulated for blood gas analyses and

intravenous administration drugs. After surgery, all wounds were

treated with 2% lidocaine and closed.

Closed cranial window

The rats were secured to a custom-built stereotaxic equipped

with plastic ear pieces and a bite-bar. The dorsal area of the scalp

was shaved, and a rostral-to-caudal incision was made to expose

the skull. Connective tissue above the skull was removed, and local

bleeding was controlled either by cauterization or by coating the

skull with bone wax. Some portions of the temporal muscle on the

side of the cranial window were excised, and the location of a

cranial window over the primary somatosensory cortex was

determined based on stereotaxic coordinates (center at 4 mm

lateral to bregma, 0 mm caudal to bregma, diameter 5 mm). Using

a dremel with a burr bit, the perimeter around the window was

carefully thinned until pial vessels were visible. Ice-cold saline

solution was constantly applied during drilling to avoid over-

heating of the skull. The area of skull within the perimeter was then

removed with no. 7 forceps. In most cases (animal body weight

BW <250 g, n = 14), the dura matter below the cranial window

remained intact and was left in place. In a few cases, however (BW

�250 g, n = 6), it was necessary to remove the dura matter by

careful dissection, leaving the pial surface of the cortex exposed.

Once the craniotomy was completed, the ear bars were removed

and a metal frame was fixed around the craniotomy site with dental

cement and held in place by custom made metal bars attached with

pins on either side of the frame (Yoder and Kleinfeld, 2002). The

bars were secured to the stereotaxic stage by the ear bar holders

and adjusted until the cranial window was oriented horizontally.

The well over the cranial window was then filled with 1% wt./vol.

agarose and covered with a custom cut Corning glass #1 cover slip.

The animal and sample stage were then moved and fixed to the

microscope stage. Anesthesia was switched from isoflurane to an

intravenous solution of a-chloralose (80 mg/kg BW initial bolus,

followed by a constant infusion of 27 mg/kg BW/h) supplemented

with pancuronium bromide as needed (2 mg/kg BW/90 min) to

minimize motion artifacts. To visualize vessels, blood plasma was

fluorescently labeled by an intravenous injection of 150–300 Al of0.5 mg/ml rhodamine-labeled dextran (70,000 MW) in phosphate-

buffered saline.

Blood gases and hypercapnia

Blood gases were carefully monitored and adjusted throughout

experiments to ensure accurate normocapnic and hypercapnic

conditions. Ventilation parameters (rate, tidal-volume and oxygen

concentration in gas mixture) were adjusted to correct PaO2 and

PaCO2. Whenever necessary, sodium bicarbonate was injected

intravenously to correct for acidic blood pH. Once the blood gases

were in physiological range (normocapnic condition), the imaging

commenced. Blood gases were periodically sampled to ensure

physiological stability throughout imaging. Hypercapnia was induced

by adding 2–3% CO2 to the inhaled air mixture. A few minutes were

allowed before blood gases were sampled again to confirm

equilibration under hypercapnic conditions. After the last image

during the hypercapniawas acquired, the last blood samplewas drawn

to verify the preservation of the hypercapnic condition. Typical

experimental duration was 40 min (20 min for each of normo- and

hypercapnia). The trials in which either normocapnia or hypercapnia

could not be maintained were excluded from the analysis.

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530522

Optical microscopy and vessel diameter measurement

Imaging was performed with a Bio-Rad Radiance 2100 MP

two-photon laser-scanning microscope. The pump laser used was a

Tsunami Ti:S fs pulsed laser, which was tuned to 805 nm and

operated at 8.25 or 8.5 W. The lens used was a Nikon 10� water

immersion objective with a working distance of 2 mm and 0.3

numerical aperture. Emitted fluorescence was detected with a PMT

using a 620/100 nm bandpass filter.

High spatial resolution (matrix size of 512 � 512), Kalman

averaged, horizontal planar images were acquired at 0.8� 0.8 Am2 in

plane resolution and stepped in 1 Am increments from the pial surface

down to a cortical depth of 200 Am. The maximal pixel intensities in

these imageswere then projected onto a single plane to yield a 2Dmap

of the 3D vasculature. From this map, line profiles across vessels of

interest were created, from which the full-width at half-maximum

(FWHM) values were computed to yield the vessel diameter. This

method of measurement was employed due to the relatively large (¨3

Am in the z axis) focal volume of the objective, which exceeded the

step size and ensured imaging of the maximal diameter in the x –y

plane. This procedure was followed for each vessel under both normal

and hypercapnic conditions so that diameter estimates of each vessel

could be paired and changes quantified.

Bolus passage

As a measure of the changes in both blood volume and blood

flow under hypercapnia, the transit of a bolus of dye through the

vasculature was imaged. A reasonable imaging plane containing

both arterial and venous vessels was selected from within the stack

volume, and a 30-s time course recorded at a rate of 1 frame every

222 ms. A 150-Al bolus of the dextran-rhodamine dye was rapidly

injected (within less than 1 s, using a 1-cm3 syringe) via the femoral

vein catheter. The dextran-rhodamine dye sequentially filled the

appropriate vessels in the subsequently acquired frames. This bolus

method was repeated several times before saturation of the dye.

For the purposes of the analysis, the vessel that was both largest

in diameter and first to enhance was chosen as the arterial input

function to which the rest of the vascular responses were

referenced. The latest onset arteriole before the capillary bed, the

first venule after the capillary bed and the largest and latest onset

venous vessel were also identified. Time courses of the passage of

the bolus of dye were plotted for each of the selected vessels. The

time point at which the signal intensity first reached one-half of its

maximum value (t1/2) was determined and compared to the t1/2 of

the arterial input vessel to estimate the transit time from the arterial

input to the other three vessels types in both normocapnic and

hypercapnic states. A method similar to the one described above

was employed to measure the transit of a bolus of dye through the

capillary network. The temporal resolution was improved to 140

ms per frame by reducing the field of view. In these trials, the onset

times of capillaries of several different sizes, as well as venules,

were referenced to a feeding arteriole.

RBC velocity measurement

Very high temporal resolution measurements of RBC velocities

were obtained by 1.333 ms line scans along the axis of selected

capillaries of different diameters in the direction of flow as

described previously (Kleinfeld et al., 1998). The RBCs appeared

as dark segments in a line of brightly labeled plasma. With forward

motion over line scans taken through time, the RBCs created dark

tracks angled at a slope given by the length of the capillary divided

by the time to clear the capillary. From these scans, velocity

measurements were obtained for individual RBCs over a period of

30 s and averages over all RBCs in each capillary computed.

Data in text and figures are presented as mean T standard

deviation. Student’s t tests were performed to test for statistical

significance. Unless otherwise specified, a P value of <0.05 was

considered indicative of a statistically significant difference.

Results

Blood physiology

Rats were kept under stable conditions at all times throughout

the experiments. Rectal temperature and end-tidal CO2 were

monitored continuously, and arterial blood gases were sampled

periodically. Under normocapnic conditions, the across-subject

mean values for PaO2, PaCO2 and pH were 105.2 T 6.2 mm Hg,

33.0 T 3.2 mm Hg and 7.42 T 0.03, respectively, n = 20. During

hypercapnia, these values changed to 98.5 T 10.0 mm Hg (P <

0.002 compared to normocapnia), 45.4 T 6.8 mm Hg (P < 0.001)

and 7.31 T 0.04 (P < 0.001), respectively. Because the data were

paired (that is, values were sampled from each rat before and after

introducing CO2 to the breathing mixture), the three blood gases

parameters were significantly different after CO2 compared to

normocapnia (P < 0.001), showing that the addition of 2–3% CO2

to the breathing mixture was a gentle, yet effective way to raise

PaCO2 to cause moderate increases in CBF.

Measurement of large vessel diameters

Fig. 1 shows typical two-photon images of the rat cerebral

microcirculation. The experimental setup allowed visualization of

vessels located up to 200 Am below the pial surface (Fig. 1A).

However, most experiments described herein were performed at

depths ranging from 50 to 150 Ambelow the pial surface.Within this

volume, the typical structure of the rat cortical vasculature was

observed. The pial surface contained large arteries and veins running

along the cortical surface (Fig. 1B). Anastomoses were visible and

predominant on the arterial side. Penetrating arteries and arterioles

were also seen, together with venules and veins, running perpen-

dicular to the pial surface (Fig. 1B). Capillaries were scarce at the

pial surface, but much denser deeper in the cortex.

To measure vessel diameters independently of their orientation

and depth, a maximum intensity projection (MIP) stack of the 3D

vasculature onto a single horizontal plane was created, such as the

one displayed in Fig. 1C. This image shows a wide range of vessel

diameters, with capillary vessels as narrow as 2 Am and veins wider

than 90 Am in diameter. In general, capillaries were randomly

oriented below the pial surface and exhibited high interconnectivity,

with several anastomoses clearly visible in the MIP. Large arteries as

well as veins were oriented more regularly, usually running from

side to the middle of the brain. Penetrating arteries and arterioles can

also be seen in Fig. 1C, along with surface venules and veins. Based

on the diameter estimates (U) obtained from the stack MIPs, each

vessel was assigned to one of three classes (Nakagawa et al., 1995):

& the arterial network, made up of arteries (U > 50 Am) and

arterioles (U < 50 Am);

Fig. 1. Typical two-photon images of the rat cerebral microcirculation. (A) Single planar images at six different depths spaced at 25-Am increments throughout the

175-Am depth of the imaged space. (B) Reconstructed 3D views of the XZ (top) and YZ (bottom planes) show large vessels (arteries and veins) running along the

pial surface and smaller vessels below the pial surface. Penetrating arteries and arterioles, as well as draining venules and veins, run perpendicular to the pial

surface. (C) Stack projection of the 3D vasculature onto a single horizontal plane shows a wide range of vessel diameters. Scale bars = 100 Am.

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530 523

& the capillary network (U < 10 Am); or

& the venous network, consisting of venules (U < 60 Am) and

veins (U > 75 Am).

No veins or venules were found in the 60–75 Am diameter

range. In addition, the identity of arterial and venous structures

was verified using the bolus data, as described below in the

following subsection.

Large vessel diameters were computed for each condition and

their changes due to mild hypercapnia compared between the

arterial and the venous networks. Arterial and arteriole diameters

increased by 11.2 T 14.9% (4 rats, 28 arteries and arterioles, P <

0.004 for paired t test, P < 0.002 for Wilcoxon rank-sum test),

while venous and venule diameters increased by 7.3 T 13.6% (4

rats, 44 venules and veins, P < 0.002 for paired t test, P < 0.0002

for Wilcoxon rank-sum test). The large standard deviations

resulted from the heterogeneity in the sign of individual vessel

diameter changes: some vessels dilated; others did not react to the

presence of CO2; and others still were constricted by hypercapnia.

Nevertheless, both the arterial as well as the venous vasodilatation

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530524

elicited by hypercapnia were significant (P < 0.05), but not

different from each other (P > 0.1).

Large vessel bolus transit time changes

Vessel-specific dynamics of cerebral blood flow were probed by

following the passage of a bolus of contrast agent through the

cerebral circulation. Fig. 2 shows typical time courses of a bolus

passage through an artery, arteriole, venule and vein. The dynamics

of the passage of a bolus through each of the four vessel types

show a clear distinction in onset times, times-to-peak and

dispersion (full-width at half-maximum). Recirculation of the

contrast was clearly visible in the form of an elevated baseline

after the passage of the bolus, as well as by secondary and tertiary

increases in signal intensity following the first passage. An

Fig. 2. The transit of a bolus of dye through the rat cerebral vasculature: (A)

passage of bolus from artery (top left) to arteriole (top right) to venule

(bottom left) to vein (bottom right), with regions of interest indicating each

of the four vessel types. (B) Time courses obtained from each ROI indicated

in panel A show a clear distinction in onsets, times-to-peak and full-widths

at half-maximum between the different blood vessels. Recirculation of the

bolus can also be seen in the form of an elevated baseline and secondary

and tertiary passages.

example of this phenomenon can be seen in Fig. 2B. The arterial

time course shows a second passage of bolus well before

conclusion of the venous washout of the first passage.

To avoid possible confounds introduced by recirculation, only

the onset to the first passage of bolus was used to probe vessel

reactivity to CO2. Fig. 3A shows averaged (n = 6) time courses

of the first passage of dextran-rhodamine in each of the four

vessel types, under normocapnic and hypercapnic conditions. The

onset (defined as the time point at which the signal intensity first

reached one-half of its maximum value) – t1/2 – was determined

for each vessel type, and compared to the onset time of the

arterial input vessel, to yield the bolus transit time for that vessel.

Under hypercapnic conditions, the average transit time of a bolus

from the artery to the vein was 1.77 T 0.62 s, 27% shorter (P <

0.05) than the corresponding transit time during normocapnia,

2.43 T 0.66 s. Venules showed a smaller yet significant (P <

0.05) decrease in transit time due to CO2. The transit time from

artery to venule was 1.66 T 0.50 s during normocapnia,

decreasing (P < 0.05) by 14% to 1.43 T 0.50 s during

hypercapnia. There was no detectable difference in the transit

times to arteriole during either condition (0.21 T 0.23 s during

normocapnia and 0.18 T 0.20 s during hypercapnia, P > 0.7).

However, these measurements were probably limited by the low

frame rate (222 ms) employed. Fig. 3B shows plots of the transit

times for arterioles, venules and veins during normocapnic and

hypercapnic conditions.

Capillary diameter changes

Once the larger vasculature reactivity to mild CO2 administra-

tion was characterized, as described above, the focus of the study

turned to the microcirculation. The size distribution of 393

capillaries sampled from 13 different animals was obtained from

stack MIPs of the microcirculation and plotted in Fig. 4A for

normocapnic and hypercapnic conditions. Capillaries with appar-

ent mean diameters as low as 1.6 Am were found, although their

incidence was rare, about 1% of the total number of capillaries

sampled. Hypercapnia induced vasodilatation of the cerebral

microvasculature, as noticed by a right shift of the histogram

obtained during hypercapnia, and manifested by a significant

decrease in the number of capillaries smaller than 3.2 Am which

was accompanied by a significant increase in the number of

capillaries larger than 5.2 Am. The mean capillary diameter during

hypercapnia was 4.1 T 1.1 Am, significantly larger than the average

capillary diameter during normocapnia, 3.7 T 1.1 Am (P < 0.005).

There was a pronounced heterogeneity in the hypercapnia-induced

changes in capillary diameter. As shown in Fig. 4B, the largest

changes in diameter (5–77%; mean T SD: 25 T 34%, P < 0.001)

induced by hypercapnia occurred in capillaries with small resting

diameters (�4.0 Am, n = 271, 69% of the total number of

capillaries), while no significant changes (P > 0.07) were observed

in capillaries of larger resting diameter (4.4 Am–6.8 Am, n = 122,

31% of the population), with the exception of capillaries that were

5.6 Am (4% of the capillary population) in diameter. These 5.6 Amdiameter capillaries showed a significant decrease (�14% T 19%,

n = 17, P < 0.004) in resting diameter with hypercapnia.

RBC velocity changes

Fig. 5 shows line scans of three different capillaries, circled in

Fig. 5A, used to measure the reactivity of RBC velocities to

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530 525

hypercapnia. The horizontal axis in Fig. 5C represents the length of

the capillary vessel in the direction of blood flow (indicated by the

green arrow in Fig. 5B), and the vertical axis represents time. Red

blood cells appear as dark streaks, with slopes that indicate their

velocities as they traverse the capillaries. Steeper slopes correspond

Fig. 4. Histograms of capillaries according to diameter (A) and percent

changes in diameter (B) induced by hypercapnia. Significant overall

vasodilatation is seen, with the largest changes occurring in the smallest

capillaries (*P < 0.03; error bars correspond to 1 standard deviation).

to slower RBC velocities. Hypercapnia elicited changes in RBC

velocities, measured by a change in the slope of the dark streaks in

the line scans of the different capillaries. In the example given in

Fig. 5, RBC velocity increased during hypercapnia in the medium

and large capillaries, but decreased in the small capillary.

Mild hypercapnia induced an increase in RBC velocities in

most capillaries. During normocapnia, the mean RBC velocity

across all capillaries was 1.12 T 0.89 mm/s, significantly lower

than the mean RBC velocity during mild hypercapnia, of 1.81 T1.60 mm/s (P < 0.0002, n = 46). The RBC velocity across the

capillary network thus increased by an average 75 T 114%. The

distribution of the capillary RBC velocity changes induced by mild

hypercapnia is shown in Fig. 6. As seen in Fig. 6A, a significant

increase in RBC velocities was observed in capillaries with

diameters from 2.8 to 6.0 Am. However, as shown in Fig. 6B,

relative changes in RBC velocities were more pronounced in the

medium and larger capillaries (resting diameters >4 Am, 107 T

Fig. 3. Transit of a bolus of dye through the vasculature during

normocapnic and hypercapnic conditions. (A) Normalized mean intensity

time courses from 6 animals. The onset (t1/2) of the bolus in each vessel was

computed and referenced to the leading artery. The vertical lines indicate a

clear reduction in transit time for venules and veins during hypercapnia. (B)

Plot of the mean (top) and individual (bottom) transit times for arterioles,

venules and veins at normocapnia and hypercapnia. There was a significant

reduction in transit times in venules and veins during hypercapnia (*P <

0.05, error bars correspond to 1 standard deviation).

Fig. 5. Typical line scan data for determining RBC velocities and their

reactivity to hypercapnia. (A) MIP showing the location of the capillary

vessels. (B) High-resolution image of a single capillary, indicating the

direction of flow (bottom). (C) Line scans of a small (left, U = 2.4 Am),

medium (middle, U = 5.6 Am) and large (U = 8.0 Am) capillary under

normocapnic and hypercapnic conditions. The horizontal axis represents the

length of the capillary vessel in the direction of blood flow, while time is

shown along the vertical axis. RBCs appear as diagonal dark streaks, with

slopes that indicate their velocities as they traverse the capillaries. Steeper

slopes correspond to slower velocities.

Fig. 6. (A) Mean RBC velocity distribution during normocapnia and

hypercapnia as a function of capillary diameter shows increases in RBC

velocity in all capillaries. (B) Mean RBC velocity percent changes as a

function of the resting capillary diameter (*P < 0.05).

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530526

142%, P < 0.003, n = 24) than in smaller capillaries (resting

diameters = 4 Am, 39 T 51%, P < 0.004, n = 22). This result is in

contrast to the changes in capillary diameters of Fig. 4B, indicating

a spatial dissociation between increases in microvascular RBC

velocities and blood volume.

Discussion

In the present study, spatial and temporal patterns of cortical

microcirculatory response to mild hypercapnia were investigated

via two-photon laser-scanning microscopy of rodent somatosen-

sory cortex. Carbon dioxide was chosen to produce robust global,

yet gentle vasodilatation, manifested as an increase in the diameter

of arteries, arterioles, venules and veins, as well as a decrease in

transit time of the fluorescent dextrans from the arterial to venous

side of the vasculature. Complimentary to most previous studies of

the reactivity of the cerebral vasculature to CO2, special attention

was given to the capillary vessels (resting diameter <10 Am)

located at least 50 Am and as deep as 200 Am below the pial

surface. Histograms of capillary diameters were obtained during

normocapnia and following CO2-induced vasodilatation. RBC

velocity profiles were also obtained during both conditions. Most

notably, the data illustrate a pronounced spatial dissociation

between volume and velocity changes in the cortical microcircu-

latory response to hypercapnia. In the following, the significance

of the present results is discussed in detail.

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530 527

Hypercapnic modulation

Carbon dioxide, a very widely studied vasodilator, was used to

produce a robust and rapid rise in global CBF (Poulin et al., 1996).

While acetazolamide – an erythrocyte carbonic anhydrase

inhibitor – has cerebrovascular effects that are very similar to

those of CO2 (Ehrenreich et al., 1961; Vorstrup et al., 1984), CO2

administration has distinct advantages of easy dosage adjustments,

prompt agent removal and low cost. Addition of 2–3% CO2 to the

inspired air produced an average increase of 12.4 T 7.5 mm Hg in

PaCO2. This was a gentle effect, just enough to produce mild yet

robustly detectable changes in CBF. Carbon dioxide readily reacts

with water to form carbonic acid, which dissociates into carbonate,

bicarbonate and hydrogen ions. The slight drop in pH by 0.11 T0.05 mm Hg following the gentle increase in PaCO2 was expected

and is consistent with the mild hypercapnic conditions, under

which the vasodilatory effects of CO2 cannot be completely

disentangled from H+-induced hyperemia. It is important to point

out, however, that the purpose of this study was not to take the

cerebral vasculature to a point of maximal vasodilatation, but only

to instigate robust and detectable changes in CBF, without running

into physiological complications of acute hypercapnia, such as

acidosis, vascular adaptation, etc.

Bolus tracking

In this work, blood plasma was labeled by intravenous

injections of dextran-rhodamine, a fluorescent intravascular mark-

er. This approach enabled visualization of vessels of all sizes (Fig.

1). To distinguish the arterial from the venous size of the

vasculature, images were acquired during the administration of a

bolus of dye (Fig. 2) and transit times were computed (Fig. 3). The

use of bolus tracking techniques is well established for imaging

blood flow both by MRI and PET (Ostergaard et al., 1998;

Calamante et al., 1999). To the best of our knowledge, this is the

first time such a strategy has been employed in conjunction with

two-photon microscopy. The volumetric increases of arteries,

arterioles, venules and veins along with the decrease in the

arterial–venous transit times were clear indication that mild

hypercapnia produced a robust vascular response. However, no

attempts were made at quantification of regional blood flow or

volume as is typically done in other neuroimaging techniques

(Ostergaard et al., 1998; Calamante et al., 1999) since this would

require careful quantification of the length of each vessel. In future

studies, the length and diameters of specific vessels will be

measured and used to quantify blood flow and volume from bolus

tracking data. That approach will be complementary to the RBC

velocity measurements presented here and could be used as an

alternative method to quantify flow in vessels that are too large to

allow velocity measurements of individual RBCs.

Capillary dynamics

In the present study, capillary diameters ranged from 2 to 6.8

Am, in agreement with a recent two-photon microscopy study of

the rat olfactory bulb glomeruli which considered vessels of

diameters below 6 Am as capillaries (Chaigneau et al., 2003). The

observed distribution of capillary sizes also agrees well with

another study (Hudetz et al., 1993). However, the presently

observed microvessels are of considerably smaller caliber than

those studied in a number of earlier experiments (Morii et al.,

1986). In particular, all capillary diameters in an earlier two-photon

microscopy experiment exceeded 5 Am (Kleinfeld et al., 1998).

This discrepancy may well be a result of an anesthesia-induced

arterial PaCO2 elevation in the latter study (no information about

resting arterial pCO2 was given therein). In the present work,

careful steps were undertaken to maintain normal arterial PaCO2

(of about 35 mm Hg) throughout the experiments. Indeed, the high

sensitivity of the resting vessel diameter to PaCO2 was demon-

strated by detection of capillaries with diameters below 2 Amduring hypocapnic conditions (data not shown). Nonetheless, the

use of maximum intensity projections (MIP) for vessel diameter

measurements may have caused a systematic underestimation of

vessel sizes, depending on the signal to noise ratio of the

measurements. In all cases, however, the underestimation of vessel

diameter could not have been any bigger than one pixel width (i.e.,

0.8 Am). To investigate whether any such bias was present, we

performed line scans at higher sensitivity in a subset of subjects but

found no significant differences with respect to the data presented.

The average hypercapnia-elicited change in diameter, taken

across all capillaries, was small but significantly above zero,

consistent with perfused brain studies (Duelli and Kuschinsky,

1993) and in vivo confocal microscopy study of surface vessels

(Villringer et al., 1994). Notably, most – if not all – of the observed

capillary bed dilatation was concentrated in the smallest capillaries

imaged (with vessel diameters �4 Am). There was no significant

change in diameter of medium or larger capillaries (4.4 Am–6.8 Amdiameters). Some of the larger capillaries exhibited vasoconstriction

(Fig. 4B). The width of the distribution of capillary vessel diameters

decreased during hypercapnia as the smallest vessels dilated and the

larger ones did not respond, thus producing amore uniform capillary

bed flow in the hypercapnic condition relative to the normocapnic

condition. This is consistent with previous results derived from

confocal microscopy (Villringer et al., 1994) and also from

microscopical analysis of in-vivo-marked plasma perfusion

(Bereczki et al., 1993; Abounader et al., 1995).

RBC velocities

To determine whether the pattern of capillary dilatation led to a

corresponding pattern in capillary RBC velocity changes, two-

photon microscopy was used in line scan mode to measure RBC

velocities (Fig. 5). The range of RBC velocities measured agrees

quite well with previously reported values as does the large

variation in RBC velocities across individual capillaries (Villringer

et al., 1994; Hudetz, 1997; Kleinfeld et al., 1998; Seylaz et al.,

1999; Pinard et al., 2000; Helmchen et al., 2001; Chaigneau et al.,

2003). The variation in the mean capillary bed RBC velocity is

likely influenced by the number and size of the capillaries

considered in addition to the choice and level of anesthesia (Wei

et al., 1993; Hirase et al., 2004). Alternatively, such variations

could come from spatial and temporal heterogeneities in capillary

flow derived from the network architecture (branching order) or

from random and periodic fluctuations caused by pressure differ-

ences, by arteriolar vasomotion or by plasma skimming (for a

review, see Hudetz, 1997).

Consistent with previous confocal scanning microscopy studies

(Villringer et al., 1994), all vessels were filled with fluorescent

dextrans under normocapnic conditions. There was thus no

evidence for recruitment (i.e., ‘‘all or none’’ opening) of capillaries,

in agreement with earlier studies (Kuschinsky and Paulson, 1992;

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530528

Villringer et al., 1994; Seylaz et al., 1999; Pinard et al., 2000).

However, some capillaries were very poorly perfused with RBCs

and some showed periods when RBCs stopped moving (data not

shown). In accordance with earlier work (Villringer et al., 1994;

Hudetz, 1997), these findings support the existence of functional

reserve capillaries and hence allow for functional recruitment of

capillaries under increased flow conditions (Kuschinsky and

Paulson, 1992; Akgoren and Lauritzen, 1999). Moreover, the

variability in flow pattern illustrates the highly dynamic nature of

the microcirculatory system and points out the challenges in

characterizing the microcirculatory network via measurements of

individual capillaries.

Spatial dissociation between microvascular blood volume and

RBC velocities

The changes in capillary RBC velocity induced by hypercapnia

were not coupled to the observed capillary diameter changes. The

overall increase in RBC velocity across all capillaries and a

broadening of the RBC velocity distribution with hypercapnia

found here are consistent with findings of previous studies

(Villringer et al., 1994; Hudetz, 1997). The average RBC velocity

increase of 75% across the capillary bed agrees well with earlier

hypercapnia studies in rats (Villringer et al., 1994; Barfod et al.,

1997). On the other hand, the fact that most of the changes in RBC

velocity occurred in the larger capillaries suggests a spatial

dissociation between capillary volume and RBC velocity changes.

The increase in RBC velocity in the larger capillaries without any

change in volume indicates that the major resistance to flow is

upstream of the capillary, at the level of arteriole or at the pre-

capillary sphincter (Nakai et al., 1981; Jokelainen et al., 1982). The

change in diameter of the smaller capillaries that is accompanied

by a change in RBC velocity suggests that there may be some

resistance to flow at the level of these capillaries. While such

distention may be passive (due to a resistance drop upstream), there

is anatomical evidence – namely the existence of vascular

pericytes on the abluminal surface of the capillary wall or close

to the capillary branching points as well as the presence of actin-

and myosin-like filaments in the cytoplasm of endothelial cells and

pericytes (Owman et al., 1977; Le Beux and Willemot, 1978; Ehler

et al., 1995; Rodriguez-Baeza et al., 1998; Harrison et al., 2002;

Hirase et al., 2004) – that supports active control of the capillary

diameter. Further studies are needed to address this issue.

It is unclear whether the observed dissociation between

increases in RBC velocities and capillary dilatation in response

to mild hypercapnia would also be present during increased

hyperemia induced by neuronal activation. A number of recent

studies have used two-photon microscopy to observe vasodilata-

tion and changes in RBC velocities during somatosensory

stimulation (Kleinfeld et al., 1998) or odor stimulation (Chaigneau

et al., 2003) in rats and during bicuculline-induced focal

epileptiform activity in mice (Hirase et al., 2004). It will be

interesting to investigate the changes in capillary volume and RBC

velocities elicited by somatosensory stimulation, and this is the

subject of ongoing experiments in our laboratory.

Implications for mechanism of CBF regulation

It is tempting to speculate on the relative blood flow increase that

occurs in different capillaries in response to mild hypercapnia. Using

the Central Volume Principle (Stewart, 1894), which relates the flow

F to the volume V and the transit time t, one can write the flow along

a vessel of diameterD and length L as F ¼ Vt¼ AL

t¼ p

4D2v, where

v is the mean velocity over the vessel. Thus, if an increase in flow

resulted from an increase in velocity alone, one would expect a linear

relation between velocity and flow. On the other hand, if flow

increases were solely due to a rise in cross-sectional area of the

vessel, then the flow increases should be a quadratic function of the

vessel diameter changes. Macroscopically, both vasodilatation and

RBC velocity increase take place, so that the exponent of the

empirically attained power law, describing the dependence of flow

changes on volume changes, was reported to be on the range of

2.63–3.45 for hypercapnic perturbations in rodents, rhesus

monkeys and humans (Grubb et al., 1974; Ito et al., 2003; Rostrup

et al., 2005; Jones et al., 2005). In the current study, we observed an

¨107% increase in velocity (and hence in flow) in capillaries larger

than 4 Am without any dilatation (Fig. 6B). On the other hand, the

¨39% increase in velocity observed in capillaries smaller than 4 Am(Fig. 6B) was accompanied by¨25% increase in their diameter (Fig.

4B). Thus, in these capillaries,

FCO2¼ p

4D2

CO2vCO2

¼ p4

1:25D0ð Þ21:39v0 ¼ 2:17p4D2

0v0

��

¼ 2:17F0:

This flow increase is thus similar to the estimated flow change

in the larger capillaries. Such uniformity in the flow change across

the capillaries – despite the heterogeneity in the microvascular

architecture – is indicative of a heterogeneous microcirculatory

CBF regulation and, in particular, heterogeneity in the CO2

reactivity of the microvessels that act to raise the blood-flow-

carrying capacity of the entire capillary network as a whole. The

fact that both diameter and RBC velocity increase in small

capillaries makes them a much more efficient target for microcir-

culatory CBF changes.

Acknowledgments

The authors would like to acknowledge Ruperto Villadiego for

excellent machinery services. This research was supported by the

Intramural Research Program of the NIH, NINDS (Eugene Major

and Henry McFarland, Acting Scientific Directors).

References

Abounader, R., et al., 1995. Patterns of capillary plasma perfusion in brains

in conscious rats during normocapnia and hypercapnia. Circ. Res. 76,

120–126.

Akgoren, N., Lauritzen, M., 1999. Functional recruitment of red blood cells

to rat brain microcirculation accompanying increased neuronal activity

in cerebellar cortex. NeuroReport 10, 3257–3263.

Barfod, C., et al., 1997. Laser-Doppler measurements of concentration and

velocity of moving blood cells in rat cerebral circulation. Acta Physiol.

Scand. 160, 123–132.

Bereczki, D., et al., 1993. Hypercapnia slightly raises blood volume and

sizably elevates flow velocity in brain microvessels. Am. J. Physiol.

264, H1360–H1369.

Calamante, F., et al., 1999. Correction for eddy current induced Bo shifts

in diffusion-weighted echo-planar imaging. Magn. Reson. Med. 41,

95–102.

Chaigneau, E., et al., 2003. Two-photon imaging of capillary blood flow

in olfactory bulb glomeruli. Proc. Natl. Acad. Sci. U. S. A. 100,

13081–13086.

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530 529

Cox, S.B., et al., 1993. Localized dynamic changes in cortical blood flow

with whisker stimulation corresponds to matched vascular and neuronal

architecture of rat barrels. J. Cereb. Blood Flow Metab. 13, 899–913.

Denk, W., et al., 1990. Two-photon laser scanning fluorescence microscopy.

Science 248, 73–76.

Duelli, R., Kuschinsky, W., 1993. Changes in brain capillary diameter

during hypocapnia and hypercapnia. J. Cereb. Blood Flow Metab. 13,

1025–1028.

Duong, T.Q., et al., 2001. Extracellular apparent diffusion in rat brain.

Magn. Reson. Med. 45, 801–810.

Ehler, E., et al., 1995. Heterogeneity of smooth muscle-associated proteins

in mammalian brain microvasculature. Cell Tissue Res. 279, 393–403.

Ehrenreich, D.L., et al., 1961. Influence of acetazolamide on cerebral blood

flow. Arch. Neurol. 5, 227–232.

Fenstermacher, J.D., et al., 1991. Functional variations in parenchymal

microvascular systems within the brain. Magn. Reson. Med. 19,

217–220.

Gerrits, R.J., et al., 2000. Regional cerebral blood flow responses to

variable frequency whisker stimulation: an autoradiographic analysis.

Brain Res. 864, 205–212.

Grubb Jr., R.L., et al., 1974. Effects of changes in PaCO2 on cerebral blood

volume, blood flow, and vascular mean transit time. Stroke 5, 630–639.

Harrison, R.V., et al., 2002. Blood capillary distribution correlates with

hemodynamic-based functional imaging in cerebral cortex. Cereb.

Cortex 12, 225–233.

Helmchen, F., et al., 2001. A miniature head-mounted two-photon

microscope. high-resolution brain imaging in freely moving animals.

Neuron 31, 903–912.

Hirase, H., et al., 2004. Capillary level imaging of local cerebral blood flow

in bicuculline-induced epileptic foci. Neuroscience 128, 209–216.

Hudetz, A.G., 1997. Blood flow in the cerebral capillary network: a review

emphasizing observations with intravital microscopy. Microcirculation

4, 233–252.

Hudetz, A.G., et al., 1993. Imaging system for three-dimensional mapping of

cerebrocortical capillary networks in vivo. Microvasc.Res. 46, 293–309.

Iliff, J.J., et al., 2003. Adenosine receptors mediate glutamate-evoked

arteriolar dilation in the rat cerebral cortex. Am. J. Physiol.: Heart Circ.

Physiol. 284, H1631–H1637.

Ishimura, N., et al., 1996. Nitric oxide involvement in hypoxic dilation of

pial arteries in the cat. Anesthesiology 85, 1350–1356.

Ito, H., et al., 2003. Changes in human cerebral blood flow and cerebral

blood volume during hypercapnia and hypocapnia measured by positron

emission tomography. J. Cereb. Blood Flow Metab. 23, 665–670.

Jokelainen, P.R., et al., 1982. Non-random distribution of rat pial arterial

sphincters. In: Heisted, D.D., Marcus, M.L. (Eds.), Cerebral Blood

Flow: Effects of Nerves and Neurotransmitters. Elsevier/North-Holland,

New York, pp. 107–116.

Jones, M., et al., 2005. The effect of hypercapnia on the neural and

hemodynamic responses to somatosensory stimulation. NeuroImage 27,

609–623.

Kleinfeld, D., et al., 1998. Fluctuations and stimulus-induced changes in

blood flow observed in individual capillaries in layers 2 through 4 of rat

neocortex. Proc. Natl. Acad. Sci. U. S. A. 95, 15741–15746.

Kuschinsky, W., Paulson, O.B., 1992. Capillary circulation in the brain.

Cerebrovasc. Brain Metab. Rev. 4, 261–286.

Lauritzen, M., 2001. Relationship of spikes, synaptic activity, and local

changes of cerebral blood flow. J. Cereb. Blood Flow Metab. 21,

1367–1383.

Le Beux, Y.J., Willemot, J., 1978. Actin- and myosin-like filaments in rat

brain pericytes. Anat. Rec. 190, 811–826.

Logothetis, N.K., 2002. The neural basis of the blood-oxygen-level-

dependent functional magnetic resonance imaging signal. Philos. Trans.

R. Soc. Lond., B Biol. Sci. 357, 1003–1037.

Lu, H., et al., 2004. Spatial correlations of laminar BOLD and CBV

responses to rat whisker stimulation with neuronal activity localized by

Fos expression. Magn. Reson. Med. 52, 1060–1068.

Malonek, D., Grinvald, A., 1996. Interactions between electrical activity

and cortical microcirculation revealed by imaging spectros-

copy: implications for functional brain mapping. Science 272,

551–554.

Morii, S., et al., 1986. Reactivity of rat pial arterioles and venules to

adenosine and carbon dioxide: with detailed description of the closed

cranial window technique in rats. J. Cereb. Blood FlowMetab. 6, 34–41.

Morris, D.C., et al., 1999. High resolution quantitation of microvascular

plasma perfusion in non-ischemic and ischemic rat brain by laser-scan-

ning confocal microscopy. Brain Res. Brain Res. Protoc. 4, 185–191.

Nakagawa, H., et al., 1995. Blood volumes, hematocrits, and transit-

times in parenchymal microvascular systems of the brain. In:

LeBihan, D. (Eds.), Diffusion and Perfusion Magnetic Resonance

Imaging – Applications to Functional MRI. Raven Press, Ltd, New

York, pp. 193–200.

Nakai, K., et al., 1981. Microangioarchitecture of rat parietal cortex with

special reference to vascular ‘‘sphincters’’. Scanning electron micro-

scopic and dark field microscopic study. Stroke 12, 653–659.

Navari, R.M., et al., 1978. Comparison of the open skull and cranial

window preparations in the study of the cerebral microcirculation.

Microvasc. Res. 16, 304–315.

Ostergaard, L., et al., 1998. Cerebral blood flow measurements by magnetic

resonance imaging bolus tracking: comparison with [(15)O]H2O

positron emission tomography in humans. J. Cereb. Blood Flow Metab.

18, 935–940.

Owman, C., et al., 1977. Immunohistochemical demonstration of actin and

myosin in brain capillaries. Acta Physiol. Scand., Suppl. 452, 69–71.

Parfenova, H., et al., 1995. Inhibitory effect of indomethacin on

prostacyclin receptor-mediated cerebral vascular responses. Am. J.

Physiol. 268, H1884–H1890.

Patlak, C.S., et al., 1984. An evaluation of errors in the determination of

blood flow by the indicator fractionation and tissue equilibration (Kety)

methods. J. Cereb. Blood Flow Metab. 4, 47–60.

Pinard, E., et al., 2000. Dynamic cerebral microcirculatory changes in

transient forebrain ischemia in rats: involvement of type I nitric oxide

synthase. J. Cereb. Blood Flow Metab. 20, 1648–1658.

Poulin, M.J., et al., 1996. Dynamics of the cerebral blood flow response to

step changes in end-tidal PCO2 and PO2 in humans. J. Appl. Physiol.

81, 1084–1095.

Raichle, M.E., 2003. Functional brain imaging and human brain function.

J. Neurosci. 23, 3959–3962.

Rodriguez-Baeza, A., et al., 1998. Perivascular structures in corrosion casts

of the human central nervous system: a confocal laser and scanning

electron microscope study. Anat. Rec. 252, 176–184.

Rosen, B.R., et al., 1991. Contrast agents and cerebral hemodynamics.

Magn. Reson. Med. 19, 285–292.

Rosenblum, W.I., 1965. Cerebral microcirculation: a review emphasizing

the interrelationship of local blood flow and neuronal function.

Angiology 16, 485–507.

Rostrup, E., et al., 2005. The relationship between cerebral blood flow and

volume in humans. NeuroImage 24, 1–11.

Seylaz, J., et al., 1999. Dynamic in vivo measurement of erythrocyte

velocity and flow in capillaries and of microvessel diameter in the rat

brain by confocal laser microscopy. J. Cereb. Blood Flow Metab. 19,

863–870.

Silva, A.C., Koretsky, A.P., 2002. Laminar specificity of functional MRI

onset times during somatosensory stimulation in rat. Proc. Natl. Acad.

Sci. U. S. A. 99, 15182–15187.

Stewart, G.N., 1894. Researches on the circulation time in organs and on

the influences which affect it. J. Physiol. 15, 1–89.

Takenaka, M., et al., 2000. Intrathecal dexmedetomidine attenuates

hypercapnic but not hypoxic cerebral vasodilation in anesthetized

rabbits. Anesthesiology 92, 1376–1384.

Takenaka, M., 2003. The comparative effects of prostaglandin E1 and

nicardipine on cerebral microcirculation in rabbits. Anesth. Analg. 96,

1139–1144 (table).

Villringer, A., et al., 1994. Capillary perfusion of the rat brain cortex. An in

vivo confocal microscopy study. Circ.Res. 75, 55–62.

E.B. Hutchinson et al. / NeuroImage 32 (2006) 520–530530

Vorstrup, S., et al., 1984. Effect of acetazolamide on cerebral blood

flow and cerebral metabolic rate for oxygen. J. Clin Invest. 74,

1634–1639.

Wagerle, L.C., Degiulio, P.A., 1994. Indomethacin-sensitive CO2 reactivity

of cerebral arterioles is restored by vasodilator prostaglandin. Am. J.

Physiol. 266, H1332–H1338.

Wei, L., et al., 1993. The velocities of red cell and plasma flows through

parenchymal microvessels of rat brain are decreased by pentobarbital.

J. Cereb. Blood Flow Metab. 13, 487–497.

Woolsey, T.A., et al., 1996. Neuronal units linked to microvascular modules

in cerebral cortex: response elements for imaging the brain. Cereb.

Cortex 6, 647–660.

Xu, H.L., et al., 2004. Influence of the glia limitans on pial arteriolar

relaxation in the rat. Am. J. Physiol.: Heart Circ. Physiol. 287,

H331–H339.

Yoder, E.J., Kleinfeld, D., 2002. Cortical imaging through the intact mouse

skull using two-photon excitation laser scanning microscopy. Microsc.

Res. Tech. 56, 304–305.