Fluctuating and sensory-induced vasodynamics in
rodent cortex extend arteriole capacity
Patrick J. Drewa,b,c, Andy Y. Shiha, and David Kleinfelda,d,1
aDepartments of Physics and Neurobiology anddCenter for Neural Circuits and Behavior, University of California, San Diego, CA 92093; andbCenter for Neural
Engineering andcDepartments of Engineering Science and Mechanics and Neurosurgery, Pennsylvania State University, University Park, PA 16802
Edited* by Nikos K. Logothetis, Max Planck Institute for Biological Cybernetics, Tübingen, Germany, and approved March 28, 2011 (received for review
January 10, 2011)
Neural activity in the brain is followed by localized changes in blood
flow and volume. We address the relative change in volume for
arteriole vs. venous blood within primary vibrissa cortex of awake,
measure spontaneous and sensory evoked changes in flow and
volume at the level of single vessels. We find that arterioles exhibit
slow (<1 Hz) spontaneousincreases in their diameter, aswell aspro-
nounced dilation in response to both punctate and prolonged stim-
ulation of the contralateral vibrissae. In contrast, venules dilate only
in response to prolonged stimulation. We conclude that stimulation
that occursonthetime scale ofnaturalstimulileads toanetincrease
in the reservoir of arteriole blood. Thus, a “bagpipe” model that
highlights arteriole dilation should augment the current “balloon”
model of venous distension in the interpretation of fMRI images.
blood oxygen level-dependent functional MRI|neurovascular coupling|
two-photon imaging|vasomotion|vibrissa somatosensation
neural activity. For two important imaging modalities, blood ox-
ygen level-dependent functional magnetic resonance imaging
(BOLD fMRI) (1) and intrinsic optical signal imaging (IOS) (2),
the signals are generated by a complex interplay of the rate
of oxidative metabolism, the flux of blood in the underlying
angioarchitecture, and changes in vascular volume (3). The locus
for the increase in vascular volume that follows sensory stimula-
tion (i.e., arterioles or venules) is an enduring controversy that
bears directly on interpreting and quantifying fMRI signals (4–6).
To resolve this question, we used in vivo two-photon laser-
scanning microscopy to image spontaneous and sensory evoked
vascular dynamics in the vibrissa area of parietal cortex of awake,
head-fixed mice. All data were collected through a reinforced
thin-skull window (7) (Fig. 1A); this method obviates potential
complications from inflammation or changes in cranial pressure
that may occur with a craniotomy (8).
ocalized changes in the flow and volume of oxygenated blood
in the brain are commonly used as a correlate of heightened
Images of the pial surface and measurements of the diameter of
the lumen of surface arterioles and venules were performed while
baseline diameter occurred with frequencies of 0.07 ± 0.05 Hz
(mean ± SD, n = 118 arteries in six mice). The diameters of short
segments of arterioles (red in Fig. 1B) exhibited relatively large
spontaneous increases in the spectral range between 0.1 Hz and 1
Hz(Fig.1 C and D). Thepeak amplitude ofthesefluctuations was
23% ± 10% ofthe initial vessel diameter across all arterioles, with
instances of a 50% increases in diameter. Further, these low-
frequency oscillations were strongly coherent and synchronous
overa distanceofseveral hundredmicrometers acrosscortex (198
pairs of vessels, in six mice, that were separated by 20–315 μm;
slope of coherence = 0.001 μm−1[nonsignificant (NS)] and slope
of phase shift = 0.0009 rad/μm (NS)) (red in Fig. 1E). In contrast,
the coherence in fluctuations in diameter across midrange fre-
quencies, defined as 1–3 Hz, was substantially reduced (red in
Fig. 1E). Last, the slow spontaneous oscillations in arterial
diameter seen in awake animals (Fig. 1C and Fig. S1) are greatly
attenuated by urethane anesthesia (Fig. S2).
In contrast to the case for arteries, spontaneous changes in the
diameter of surface venules are relatively small (Fig. 1 B and C
andFig.S2).Thespectralcontent ofthesesignalsis similartothat
seen for arterioles, although the amplitudes at low frequencies
are substantially reduced (Fig. 1 C and D). Further, there was
significant yet greatly reduced spatial coherence between pairs of
venules (33 pairs of vessels), or even pairs of venules and arte-
rioles (124 pairs of vessels), compared with pairs of arterioles in
the 0.1- to 1-Hz range (Fig. 1D). The coherence at the midrange
of frequencies remained low (Fig. 1E). We conclude that spon-
taneous fluctuations are relatively weak within the venous return
from cortex, consistent with the dearth of contractile tissue in
To examine the responses of cerebral vasculature to sensory
stimuli, contralateral vibrissae were deflected with puffs of air;
puffs aimed at the tail served as a control for general arousal (Fig.
2A). Repetitive stimulation for 30 s, similar to that used in fMRI
experiments with anesthetized rodents (4) but much longer than
the 0.5–2 s required for decision making in a vibrissa-based dis-
crimination task with mice (9), caused a rapid and transient di-
lation of arterioles that was followed, on average, by a steadily
increasing dilation (Fig. 2 A–C). The high variance of the latter
dilation results in part from intermingled transient events (Fig.
S3). The same stimulus caused only a slow distension in venules
that was also substantially weaker than the distension in arterioles
(Fig. 2 B and C); control puffs did not lead to sustained distension
(Figs. S3 and S4). These changes do not result from a change in
blood gases, which remained unchanged across the period of
stimulation, with pCO2= 31.1 ± 1.7 mm Hg; pO2= 108 ± 8 mm
Hg, and pH = 7.37 ± 0.01 (four animals). Shorter periods of
stimulation accentuated the difference between arteriole and
venuole responses. In particular, a single air puff ledto a transient
dilation of the arterioles, with a peak amplitude of 7.3% ± 6.5%
of the initial vessel diameter, but produced no discernable venous
dilation (Fig. 2C and Table 1).
dilation of individual arteriole diameters was 8 ± 8 s, whereas that
for the change in the venous time constants was 44 ± 35 s (Fig. 2).
After the termination of prolonged vibrissae stimulation, arteries
constricted below their initial diameter (−2.2% ± 4.6% averaged
20–60 s after stimulus offset; P < 0.005). Urethane anesthesia
greatly reduced the arterial response and completely blocked
venous distension to sensory stimulation (Fig. S2 C–F). None-
Author contributions: P.J.D. and D.K. designed research; P.J.D. and A.Y.S. performed re-
search; D.K. contributed new reagents/analytic tools; P.J.D. analyzed data; and P.J.D. and
D.K. wrote the paper.
The authors declare no conflict of interest.
*This Direct Submission article had a prearranged editor.
1To whom correspondence should be addressed. E-mail: email@example.com.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
| May 17, 2011
| vol. 108
| no. 20
theless, the biphasic behavior seen in arterioles matched that
observed after strong stimulation to anesthetized rodents (10).
compartments associated with different stimuli showed an ap-
proximately fourfold greater change in the arterial volume than in
veins, as well as a slight poststimulus decrease in arterial blood
volume (right-hand scale in Fig. 2C). Because the venous com-
ponentoftotalbloodvolume isthought tobeapproximately three
to four times that of the arterial component (6), the arterial
volume changes dominate for brief stimuli and are likely to ex-
ceed the venous changes even for long stimuli. Finally, the di-
ameter of subsurface capillaries (diameter = 2.9 ± 0.5 μm) was
unchanged for all conditions of stimulation (Fig. S4), although
capillaries with twice these diameters are reported to constrict in
response to neuromodulation by central neurons in in vitro cer-
ebellar and retina preparations (11).
Small arterioles dilated proportionally more than larger arte-
rioles for both spontaneous dilation events and the rapid dilation
that followed vibrissa simulation and vibrissae-evoked dilations
(Fig. 3 A and B). In contrast, the delayed, second peak in the di-
lation does not show a significant dependence on vessel size (Fig.
3B), nor does the response to control stimuli (Fig. 3C). Under all
conditions, venous distensions had no dependence on initial di-
ameter. The dependence of spontaneous and evoked diameter
changes onvesselsize andtypearesimilartothoseobservedunder
arterial responses were similar in magnitude to spontaneous ar-
terial dilations, with typical dilations of 30% and maximum dila-
tions near 50% for both cases (Fig. 3 A and B).
We now ask whether the sensory evoked dilations of surface
arterioles are reflected in the velocity of RBCs in the underlying
capillaries (Fig. 4A). We find that a single puff of air to the vi-
brissae causes a transient increase in the speed of RBCs (Fig. 4 B–
Dand Table 2); systemic changes in blood pressure are unlikely to
account for this effect because the increases in speed were not
locked to changes in the heart rate triggered by the stimulus (Fig.
S5). Further, stimulation for 30 s caused a transient increase in
velocity, followed by a progressive increase that parallels the di-
lation in the surface arteries (Fig. 4D). These increases were not
limited to the vibrissa area of parietal cortex, because vessels in
the immediately adjacent medial region also respond to sensory
stimuli (Fig. 4C). There was no significant difference in the rel-
ative increase in speed for vessels inside vs. outside the vibrissa
area (P > 0.32, Kolmogorov–Smirnov test); the fraction increase
in speed was 1.2 ± 0.1 (n = 28 vessels) within the vibrissa area,
compared with an increase of 1.3 ± 0.2 (n = 3 vessels) outside the
induced increase in blood volume, especially during the early
portion of a sensory stimulation (Fig. 5). This is consistent with
the overshoot of net-oxygenated blood, inferred from fMRI (6)
and IOS (2) studies, that occurs after sensory stimulation. The
expanded volume of arteriole blood leads to an increase in the
velocity, but apparently not volume, of capillary blood. Constant
capillary volume is consistent with a lack of capillary dilation even
during epochs of seizure activity (12). For prolonged stimulation
of the form typically used in fMRI experiments (4), but more than
cortex of awake mouse. (A) Schematic of the experimental
setup. The awake mouse is head-fixed by means of a bolt and
sits passively in an acrylic cylinder beneath the two-photon
microscope. Air puffers for sensory stimulation are aimed at
the vibrissa and as a control at the tail. (B) Example image of
surface vessels. Arteries are highlighted in red, veins in blue.
Colored boxes show regions where vessel diameter is mea-
sured. The intensity has been logarithmically compressed. (C)
Left: Plot of vessel diameters of a quietly resting mouse from
a representative 5-min span. Right: Power spectrum of the
fluctuations in diameter of same vessels. Colors indicate the
location of the segment of the vessel shown in B. (D) Plot of
the maximal magnitude of the spectral coherence, |C(f)|, be-
tween pairs of vessels for 0.1- to 1-Hz band; bandwidth typi-
cally 0.1 Hz. Arteriole-to-arteriole |C| = 0.84 ± 0.15 (mean ± SD),
198 pairs (red). Venue-to-venule |C| = 0.58 ± 0.20, 33 pairs
(blue). Arteriole-to-venule |C| = 0.55 ± 0.16, 124 pairs (black).
Filled circles are for significant (P < 0.01; inverse of twice the
number of degrees of freedom) values, hollow circles for
nonsignificant values. Arrows on right indicate mean values
across types of pairs. (E) Plot of the maximal magnitude of the
spectral coherence between pairs of vessels for the 1- to 3-Hz
band; 0.1 Hz bandwidth. Filled circles are for significant
coherences, hollow circles for nonsignificant. Arteriole-to-ar-
teriole |C| = 0.54 ± 0.10 (red). Venue-to-venule |C| = 0.51 ± 0.07
(blue). Arteriole-to-venule |C| = 0.54 ± 0.08 (black).
Basic setup and spontaneous vascular dynamics in the
| www.pnas.org/cgi/doi/10.1073/pnas.1100428108Drew et al.
an order of magnitude longer than behavior time-scales, the
continuous increase in flux eventually leads to an observable dis-
tension of the venules. In this case, the greater total volume of
venous vs. arteriole vasculature (6) suggests that the increase in
venous volume may be on parity with the increase in arteriole
volume. The delayed increase in the volume of venous blood upon
prolonged stimulation (Fig. 3), and the subsequent slow return of
this volume to baseline levels, has been ascribed as evidence for
“windkessel” (5) or “balloon” (4) models, in which the veins form
a sink for excess blood. Rather, evolution has developed an arte-
image of vessels averaged 3–5 s before start of stimulation and
the same vessels averaged 25–27 s after stimulus onset. In-
tensity has been logarithmically compressed. Illustration shows
the location of veins and arteries in images. (B) Time course of
the diameters of arterioles and venules in response to stimu-
lation. Colored lines are in response to vibrissa stimulation,
with the color indicating location of the diameter measure-
ment in A; gray lines are control stimuli. Error bars indicate SD.
(C) Normalized population volume changes (mean ± SD) for
arterioles (Upper) and venules (Lower) for single puffs (28
arterioles and 7 veins), 10 s of puffs (17 arterioles and 8 ven-
ules), and 30 s of puffs (39 arterioles and 45 venules).
Sensory evoked dilation of surface vessels. (A) Baseline
Table 1. Stimulus-evoked vasodilation
Punctate10-s stimulus30-s stimulus
(0.5–1.5 s after onset) Peak
(0.5–10.5 s after onset)Peak
(0.5–30.5 s after onset)Peak
Vibrissa (%)4.4 ± 4.5
(P < 10−4)
1.9 ± 3.0
(P < 3 × 10−3)
8.9 ± 6.1
(P < 10−5)
5.0 ± 3.8
(P < 10−5)
13.7 ± 7.0
(P < 10−4)
3.2 ± 3.3
(P < 10−3)
22.0 ± 8.4
(P < 10−5)
9.6 ± 7.1
(P < 10−5)
17.0 ± 9.6
(P < 10−4)
2.9 ± 4
(P < 10−4)
25.8 ± 12.5; early* (P < 10−5)
27.8 ± 10.7; delayed†(P < 10−5)
12.1 ± 8.6; early* (P < 10−5)
9.7 ± 8.4; delayed†(P < 10−5)
13.1 ± 7.5
27.9 ± 13.716.5 ± 10.4
−0.7 ± 0.8 (NS) 1.2 ± 1.2 (NS)1.8 ± 3.5 (NS)5.6 ± 3.3
(P < 10−5)
5.3 ± 4.1
(P < 3 × 10−3)
4.1 ± 4.2 (P < 10−4) 10.2 ± 6.2 (P < 10−5)
−0.1 ± 0.1 (NS)0.8 ± 0.9 (NS)0.8 ± 1.4 (NS)
0.5 ± 3.3 (NS)6.8 ± 5.7 (P < 10−5)
50.7 ± 23.7 17.0 ± 16.413.9 ± 8.3
−0.1 ± 3.0 (NS)
−0.2 ± 1.7 (NS)
3.3 ± 0.6
30–50 μm below pia
2.0 ± 5.1 (NS)
0.3 ± 2.5 (NS)
2.9 ± 0.5
20–50 μm below pia
All numbers are mean ± SD.
*Early peak, which occurs in the 1- to 10-s period after stimulation.
†Delayed peak, which occurs 10–31 s after stimulation.
Drew et al.PNAS
| May 17, 2011
| vol. 108
| no. 20
rial “bagpipe” that serves as a reservoir of fresh blood to support
the ongoing and anticipated increase in brain metabolic activity.
The BOLD fMRI signal is determined by an interplay of me-
tabolism, blood flow, and vascular volume (13). Thus, the in-
terpretation of the BOLD signal relies on understanding the
changes in each of these components during functional hyperemia.
The present optical-based results (Fig. 2) may be combined with
MRI-based results (5, 14–17) to synthesize a view in which the rel-
ative contributions of changes on arterial and venous volumes
depends upon the temporal characteristics of the stimulus. The
rapid increase in blood flow to cortex that follows sensory stimula-
tion, which occurs over a period of seconds, is accompanied by
an increase in arterial cerebral blood volume. This causes an in-
crease in oxygenation of the venous blood on the time scale of the
arterial–venous transit time (16). If the stimulation is prolonged,
≈20 s or more in duration, a slow increase in venous blood volume
gradually becomes a significant component of the blood volume
change (14, 15). Under some conditions, capillary volume may
also increase (18). In summary, transient sensory stimuli evoke
substantial changes in the arteriole volume, whereas prolonged
Our realization of two-photon microscopy (Fig. 1A) allows vi-
sualization of blood flow dynamics at the level of single vessels with
transcranial illumination and detection in awake animals. However,
it is currently limited to imaging the superficial aspects of the cor-
tex, and even with extensive averaging we could not resolve changes
in diameter less than 0.2 μm. Thus, changes in venous diameter of
less than 2% and changes in capillary diameter of less than ≈7%
would not be detected in our experiments but would contribute to
total blood volume changes measured with MRI. Future work,
using longer wavelengths (19) and adaptive optics (20–23) to cor-
rect for scattering by the thinned skull and brain tissue, should
enhance the resolution and maximum depth of our images.
A final point concerns the low-frequency, coherent fluctuations
in arterial volume that are seen in both the absence (Fig. 1C) and
a capillary that lay 40 μm below the pial surface. (B) Velocity measurements
from capillary in A. Upper: Raw line-scan data from a single trial showing
increase in speed subsequent to vibrissa stimulation. Lower: Averaged ve-
locity response to vibrissa (purple) and control (gray) stimuli. (C) Evoked
increases in speed to single puffs, measured in and around the vibrissa-re-
lated area. Center shows histological reconstruction, green shows location of
cytochrome-oxidase–rich regions that correspond to cortical columns, or
“barrels,” and large surface arterioles and venules are highlighted in red and
blue, respectively. The symbols M and A refer to medial and anterior
directions, respectively. (D) Upper: Plot of normalized population velocity
response to a single air puff, effectively the “impulse response” (63 vessels).
Lower: Plot of normalized population velocity responses during 30 s of
stimulation (15 vessels).
Velocity increases caused by sensory stimulation. (A) Image of
vasculature. Brief stimulation leads only to arteriole distension, whereas
a prolonged period of stimulation leads to an initial increase in arterial
distension followed by a rise in capillary flow and finally venous distention.
Schematic summary of stimulus-induced changes in the cerebral
(A) Plot of peak spontaneous dilations for arteries, in red, and veins, in blue.
Gray area shows the 0.2-μm resolution limit of detectable changes. The line
shows the linear regression for arterioles, with a slope of −0.004 μm−1(r2=
0.13, P < 0.001), whereas the slope for veins (not shown) is not significantly
different from zero. (B) Plot of peak averaged dilation responses to 30-s vi-
brissae stimulation. Early arterial peaks, in the 0- to 10-s interval after stim-
ulation, are denoted by red circles; regression slope = 0.007 μm−1(r2= 0.15,
P < 0.02). Late arterial peaks, more than 10 s after onset, are denoted by red
triangles; the linear regression (not shown) is not significant. Venuoles are
denoted by blue dots; the linear regression is not significant. (C) Response to
control stimulation. Note that spontaneous events had negligible contribu-
tion to the estimate of stimulus-induced events in B and C.
Relationship between peak value of the dilation and vessel diameter.
| www.pnas.org/cgi/doi/10.1073/pnas.1100428108Drew et al.
presence of stimulation (Fig. 2B). The typical amplitude of the
fluctuations can be larger than the average amplitude of the stim-
ulus-initiated response. Thus, from the perspective of homeostatic
regulation, changes in vascular diameter that occur in response to
neuronal activation by external stimuli are comparable to sponta-
neous changes. Further, the frequency dependence and spatial
coherence of oscillations seen here for surface arterioles is similar
tothatobservedfor slowspontaneousoscillationsinthe velocityof
RBCs across capillaries (7). We conjecture that these result from
vasomotion (24). The nature of these fluctuations seems similar to
fluctuations in the BOLD fMRI signal in nonhuman (25) and hu-
man (26) primates and in the IOS and Doppler signals from
rodents (24). Thus, the transcrianal imaging technique we used
may be of use to explore the mechanistic basis of these ubiquitous
oscillations (27) in behaving mice (28, 29).
Surgery and Experimental Procedures. Allsurgicalandexperimentalprocedures
were performed in accordance with local Institutional Animal Care and Use
Committee guidelines. Our subjects were 21 male c57/bl6 mice with masses
greater than 20 g (Charles River). They were implanted with polished and
reinforced thin-skull windows (7) over the vibrissa region of primary somato-
sensory cortex. A head-bolt was glued to the skull, and two or three self-tap-
head-bolt and skull via dental cement. Animals were allowed to recover for 2
d and then were acclimated to head fixation and vibrissa stimulation over 2–7
d. Imaging took place up to 4 mo after the initial surgery. Urethane (Sigma)
chlorprothixene (Sigma) i.m. (30). Additional animals for blood gas measure-
ments (n = 3) were acclimated and had catheters inserted in their left femoral
artery, as previously described (31), along with a head bolt.
Imaging and Data Analysis. Animals were imaged (32) using a custom two-
photon microscope (33) that employed photon counting detection (34) and
was controlled by MPScan software (35). We further used a 10× 0.3 N.A.
(Zeiss), 20× 0.5 N.A. (Olympus), 20× 0.95 N.A. (Olympus), or 40× 0.8 N.A.
(Olympus) dipping objective. Capillaries were imaged in the upper 200 μm of
cortex. Power exiting the objective was typically 15–20 mW for imaging
surface vessels and varied from 10 to 200 mW for capillaries. Baseline values
of velocity and diameters exhibited no noticeable changes during 3–20 min
of prolonged scanning.
Before each imaging session, animals were briefly anesthetized with
isoflurane (2% in air) and infraorbitally injected with ≈0.05 mL 5% (wt/vol) 70
kDa fluorescein-conjugated dextran (Sigma) (36). Imaging sessions typically
lasted 1 to 2 h. Sensory stimulation to the macrovibrissae of awake animals
consisted of single 20-ms puffs of air (20 psi), delivered a distance of 3 cm
from the vibrissae, either singly or in 8-Hz trains. This is known to cause
a robust electrical response in the primary vibrissa area of the contralateral
parietal cortex (37). Control stimuli were presented from an air puffer di-
rected at the tail from 3 cm away. Stimuli were presented with 10- to 20-s
interstimulus intervals for single puff stimuli, 60-s interstimulus interval for
the 10-s stimulations, and 120-s interstimulus interval for the 30-s stimuli.
Control and vibrissa stimuli were randomly interleaved.
For diameter measurements, movies were made from sequences of frames
acquired at 6–30 frames per second. Individual frames were aligned to
remove motion artifacts by calculating the maximum in the cross-correlation
between the 2D Fourier transform of each image, yielding a phase shift that
can be converted back to a translation in real space (38). Diameter meas-
urements from frames that contained motion more than 4 μm over the
entire field were excluded from further analysis. To determine the diameter
of vessels from the movies, a small segment of the vessel, typically 2–5 μm,
was manually selected and averaged along the long axis. The diameter was
determined from the full width at half-maximum (39); we typically resolved
changes down to 0.2 μm. For spontaneous measurements of diameter
oscillations, only traces with more than 500 s of continuous data were used.
Time series of the diameter were filtered with a three-point median filter,
resampled at 25 Hz, and low-pass filtered with a 3-Hz first-order Butter-
worth filter for peak measurements. Multitaper coherence calculations (40,
41) and their error estimates (42, 43) were performed in MATLAB with the
Chronux toolbox (44). Plots of spontaneous dilations were normalized by the
median diameter. Those of sensory evoked changes in diameter were nor-
malized to the value of the baseline diameter, defined as an averaged over
the 2 s before stimulus onset. To estimate the time constant of the change in
vessel diameter, the responses to 30-s stimuli were fit to boxcar functions
convolved with an exponential (14); this yields equal rise and fall times.
For velocity measurements in the capillaries, line-scans were made of one
or two vessels at a time with the use of arbitrary scan paths (45), from which
velocities were calculated using the Radon transform method (46) imple-
mented in MATLAB. Data were windowed into 40-ms windows with 10 ms of
overlap. Exceptional points with separability score (46), a bootstrap measure
of line-scan quality, of <3, or with a velocity that was less than one fourth or
more than four times the mean velocity, were excluded from the averages.
Velocity traces were filtered with a three-point median filter. Heart rate was
determined from the peak in the power spectrum of the velocity within the
5- to 20-Hz band (7), obtained 0.5–1.5 s after the start of stimulation for the
impulse or 0.5–30.5 s for the 30-s prolonged stimulation.
All statistics were performed with a two-sided t test, unless otherwise
indicated, and all numbers are reported as mean ± SD. For peak velocities
and diameters, the population maxima of the 1 s of baseline data were
randomly resampled with replacement to match the duration that the
maximum of the stimulus was measured over. A maximum of the resampled
baseline data was taken and compared with the evoked response using
a Kolmogorov–Smirnov test.
For histological reconstructions, animals were deeply anesthetized and
transcardially perfused with PBS (Sigma), followed by 4% parformaldahyde
vol) gelatin (Sigma) mixture in PBS (47). The brain was extracted, the cortex
flattened, fiduciary marks were made, and fluorescence images were taken to
identify the locations of pial vessels relative to that of the fiduciary marks.
After sinking in 20% (wt/vol) sucrose in PBS, we prepared 100-μm-thick brain
sections tangential to the cortical surface and stained for cytochrome oxidase
(48).Theimagesofthevascularmaps and cytochromeoxidase-stainedsections
were overlaid in Adobe Illustrator, according to the fiduciary marks, to de-
termine the locations of imaged vessels relative to the cortical representation
of the vibrissae (48).
Note Added in Proof. A recent paper has reported vasomotor oscillations in
the intrinsic optical signal of mice (49).
ACKNOWLEDGMENTS. We thank the two anonymous reviewers for insight-
ful comments; Richard Buxton for critical discussions with regard to
stimulation parameters; Seong-Gi Kim and Tae Kim for discussions on
models of the functional MRI signal and for graciously sharing unpublished
data; Michael Fuentes and Carl Petersen for advice on head fixation;
Jonathan Driscoll and Philbert Tsai for assistance with the imaging system;
and Jonathan Driscoll and David Matthews for comments on an early version
of the manuscript. This work was supported by National Institutes of Health
Grants EB003832, MH085499, NS059832, and OD006831 (to D.K.) and by the
American Heart Association (A.Y.S.).
1. Ogawa S, Lee TM, Nayak AS, Glynn P (1990) Oxygenation-sensitive contrast in magnetic
resonance image of rodent brain at high magnetic fields. Magn Reson Med 14:68–78.
2. Grinvald A, Lieke EE, Frostig RD, Gilbert CD, Wiesel TN (1986) Functional architecture
of cortex revealed by optical imaging of intrinsic signals. Nature 324:361–364.
3. Logothetis NK, Wandell BA (2004) Interpreting the BOLD signal. Annu Rev Physiol 66:
4. Buxton RB, Wong EC, Frank LR (1998) Dynamics of blood flow and oxygenation
changes during brain activation: the balloon model. Magn Reson Med 39:855–864.
Table 2. Stimulus-evoked increase in RBC velocity in capillaries
(0.5–1.5 s after onset)
(0.5–30.5 s after onset)
20.6 ± 12.4 (P < 10−4)
30.1 ± 14.6 (P < 10−5)
35.1 ± 15.1 (P < 10−4)
81.8 ± 31.3 (P < 10−5)
7.9 ± 8.4 (NS)
13.8 ± 6.0 (NS)
n = 63
Diameter = 4.9 ± 1.3 μm
Depth range = 0–140 μm
5.1 ± 14.6 (NS)
41.6 ± 38.6 (P < 3 × 10−2)
n = 16
Diameter = 3.2 ± 0.7 μm
Depth range = 20–50 μm
All numbers are mean ± SD.
Drew et al.PNAS
| May 17, 2011
| vol. 108
| no. 20
5. Mandeville JB, et al. (1999) Evidence of a cerebrovascular postarteriole windkessel Download full-text
with delayed compliance. J Cereb Blood Flow Metab 19:679–689.
6. Lee SP, Duong TQ, Yang G, Iadecola C, Kim SG (2001) Relative changes of cerebral
arterial and venous blood volumes during increased cerebral blood flow: Implications
for BOLD fMRI. Magn Reson Med 45:791–800.
7. Drew PJ, et al. (2010) Chronic optical access through a polished and reinforced
thinned skull. Nat Methods 7:981–984.
8. Sohler TP, Lothrop GN, Forbes HS (1941) The pial circulation of normal, non-
anesthetized animals. Part 1. Description of a method of observation. J Pharmacol Exp
9. O’Connor DH, et al. (2010) Vibrissa-based object localization in head-fixed mice.
J Neurosci 30:1947–1967.
10. Devor A, et al. (2007) Suppressed neuronal activity and concurrent arteriolar
vasoconstriction may explain negative blood oxygenation level-dependent signal.
J Neurosci 27:4452–4459.
11. Peppiatt CM, Howarth C, Mobbs P, Attwell D (2006) Bidirectional control of CNS
capillary diameter by pericytes. Nature 443:700–704.
12. Fernández-Klett F, Offenhauser N, Dirnagl U, Priller J, Lindauer U (2010) Pericytes in
capillaries are contractile in vivo, but arterioles mediate functional hyperemia in the
mouse brain. Proc Natl Acad Sci USA 107:22290–22295.
13. Ogawa S, et al. (1992) Intrinsic signal changes accompanying sensory stimulation:
Functional brain mapping with magnetic resonance imaging. Proc Natl Acad Sci USA
14. Kim T, Kim SG (2010) Temporal dynamics and spatial specificity of aterial and venous
blood volume changes during visual stimulation: Implication for BOLD quantification.
J Cereb Blood Flow Metab, 10.1038/jcbfm.2010.226.
15. Kim T, Hendrich KS, Masamoto K, Kim SG (2007) Arterial versus total blood volume
changes during neural activity-induced cerebral blood flow change: Implication for
BOLD fMRI. J Cereb Blood Flow Metab 27:1235–1247.
16. Silva AC, Koretsky AP, Duyn JH (2007) Functional MRI impulse response for BOLD and
CBV contrast in rat somatosensory cortex. Magn Reson Med 57:1110–1118.
17. Hirano Y, Stefanovic B, Silva AC (2011) Spatiotemporal evolution of the functional
magnetic resonance imaging response to ultrashort stimuli. J Neurosci 31:1440–1447.
18. Stefanovic B, et al. (2008) Functional reactivity of cerebral capillaries. J Cereb Blood
Flow Metab 28:961–972.
19. Kobat D, et al. (2009) Deep tissue multiphoton microscopy using longer wavelength
excitation. Opt Express 17:13354–13364.
20. Ji N, Milkie DE, Betzig E (2010) Adaptive optics via pupil segmentation for high-
resolution imaging in biological tissues. Nat Methods 7:141–147.
21. Rueckel M, Mack-Bucher JA, Denk W (2006) Adaptive wavefront correction in two-
photon microscopy using coherence-gated wavefront sensing. Proc Natl Acad Sci USA
22. Tsai PS, et al. (2007) Spherical aberration correction in nonlinear microscopy and
optical ablation using a transparent deformable membrane. Appl Phys Lett 91:
23. Débarre D, et al. (2009) Image-based adaptive optics for two-photon microscopy. Opt
24. Mayhew JEW, et al. (1996) Cerebral vasomotion: A 0.1-Hz oscillation in reflected light
imaging of neural activity. Neuroimage 4:183–193.
25. Vincent JL, et al. (2007) Intrinsic functional architecture in the anaesthetized monkey
brain. Nature 447:83–86.
26. Biswal B, Yetkin FZ, Haughton VM, Hyde JS (1995) Functional connectivity in the
motor cortex of resting human brain using echo-planar MRI. Magn Reson Med 34:
27. Fox MD, Raichle ME (2007) Spontaneous fluctuations in brain activity observed with
functional magnetic resonance imaging. Nat Rev Neurosci 8:700–711.
28. Komiyama T, et al. (2010) Learning-related fine-scale specificity imaged in motor
cortex circuits of behaving mice. Nature 464:1182–1186.
29. Dombeck DA, Khabbaz AN, Collman F, Adelman TL, Tank DW (2007) Imaging large-
scale neural activity with cellular resolution in awake, mobile mice. Neuron 56:43–57.
30. Kalatsky VA, Stryker MP (2003) New paradigm for optical imaging: Temporally
encoded maps of intrinsic signal. Neuron 38:529–545.
31. Lee EJ, Woodske ME, Zou B, O’Donnell CP (2009) Dynamic arterial blood gas analysis
in conscious, unrestrained C57BL/6J mice during exposure to intermittent hypoxia.
J Appl Physiol 107:290–294.
32. Helmchen F, Kleinfeld D (2008) Chapter 10. In vivo measurements of blood flow and
glial cell function with two-photon laser-scanning microscopy. Methods Enzymol 444:
33. Tsai PS, Kleinfeld D (2009) In vivo two-photon laser scanning microscopy with
concurrent plasma-mediated ablation: Principles and hardware realization. Methods
for Vivo Optical Imaging, ed Frostig RD (CRC Press, Boca Raton, FL), 2nd Ed, pp 59–115.
34. Driscoll JD, et al. (2011) Photon counting, censor corrections, and lifetime imaging for
improved detection in two-photon microscopy. J Neurophysiol, in press.
35. Nguyen QT, Tsai PS, Kleinfeld D (2006) MPScope: A versatile software suite for
multiphoton microscopy. J Neurosci Methods 156:351–359.
36. Kleinfeld D, Mitra PP, Helmchen F, Denk W (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 USA 95:15741–15746.
37. Drew PJ, Feldman DE (2007) Representation of moving wavefronts of whisker
deflection in rat somatosensory cortex. J Neurophysiol 98:1566–1580.
38. Guizar-Sicairos M, Thurman ST, Fienup JR (2008) Efficient subpixel image registration
algorithms. Opt Lett 33:156–158.
39. Shih AY, et al. (2009) Active dilation of penetrating arterioles restores red blood cell
flux to penumbral neocortex after focal stroke. J Cereb Blood Flow Metab 29:
40. Mitra PP, Ogawa S, Hu X, U? gurbil K (1997) The nature of spatiotemporal changes in
cerebral hemodynamics as manifested in functional magnetic resonance imaging.
Magn Reson Med 37:511–518.
41. Thomson DJ (1982) Spectral estimation and harmonic analysis. Proc IEEE 70:
42. Kleinfeld D, Mitra PP (2011) Applications of spectral methods in functional brain
imaging. Imaging: A Laboratory Manual, ed Yuste R (Cold Spring Harbor Lab Press,
Cold Spring Harbor, NY), Vol 1, pp 12.11–12.17.
43. Fisher NI (1993) Statistical Analysis of Circular Data (Cambridge Press, Cambridge,
44. Mitra PP, Bokil HS (2008) Observed Brain Dynamics (Oxford Univ Press, New York).
45. Valmianski I, et al. (2010) Automatic identification of fluorescently labeled brain cells
for rapid functional imaging. J Neurophysiol 104:1803–1811.
46. Drew PJ, Blinder P, Cauwenberghs G, Shih AY, Kleinfeld D (2010) Rapid determination
of particle velocity from space-time images using the Radon transform. J Comput
47. Tsai PS, et al. (2009) Correlations of neuronal and microvascular densities in murine
cortex revealed by direct counting and colocalization of cell nuclei and microvessels.
J Neurosci 18:14553–14570.
48. Drew PJ, Feldman DE (2009) Intrinsic signal imaging of deprivation-induced
contraction of whisker representations in rat somatosensory cortex. Cereb Cortex 19:
49. White BR, et al. (2011) Imaging of functional connectivity in the mouse brain. PLoS
| www.pnas.org/cgi/doi/10.1073/pnas.1100428108Drew et al.