See the Editorial and the Response in this issue, pp 985–987.

J Neurosurg 96:1094–1102, 2002

Heme oxygenase-1 gene therapy for prevention of

vasospasm in rats

SHIGEKI ONO, M.D., TARO KOMURO, M.D., PH.D., AND R. LOCH MACDONALD, M.D., PH.D.
Department of Neurological Surgery, Okayama University Medical School, Okayama; Department of
Neurosurgery, Osaka Red Cross Hospital, Osaka, Japan; and Section of Neurosurgery, Department of
Surgery, Pritzker School of Medicine, University of Chicago Medical Center, Chicago, Illinois

Object. Hemoglobin causes contraction of cerebral arteries and is also believed to cause vasospasm after subarach-
noid hemorrhage (SAH). The goal in this study was to determine if overexpression of heme oxygenase-1 (HO-1), the
principal enzyme involved in the metabolism of hemoglobin, would reduce contractions of cerebral arteries brought on
by hemoglobin and decrease vasospasm after experimental SAH.
Methods. Injection of adenovirus expressing HO-1 (Ad5HO-1) into the cisterna magna of rats produced a significant
increase in expression of HO-1 messenger RNA, and protein and HO-1 activity in the basilar artery ([BA]; p  0.05
for each measure compared with vehicle and/or control virus, according to analysis of variance or unpaired t-test). In-
jection of adenovirus expressing
-galactosidase (Ad-
Gal) produced only mild, statistically nonsignificant increases.
The HO-1 immunoreactivity was localized to the BA adventitia after injection of Ad5HO-1 or Ad-
Gal. Injection of
Ad5HO-1 and Ad-
Gal increased the baseline diameter of the BA (measured directly via a transclival window) and
brainstem cerebral blood flow (CBF), measured by laser Doppler flowmetry, compared with vehicle. Contraction of the
BA after addition of hemoglobin was significantly inhibited, reduction in brainstem CBF was significantly prevented,
and carboxyhemoglobin concentration was significantly increased in rats injected with Ad5HO-1 compared with Ad-

Gal and vehicle. Vasospasm was significantly ameliorated in rats in which Ad5HO-1 was injected into the cisterna
magna at the time of SAH in a double-hemorrhage model.
Conclusions. These results show that overexpression of HO-1 inhibits arterial contractions induced by hemoglobin
and can reduce vasospasm after experimental SAH.

KEY WORDS • subarachnoid hemorrhage • vasospasm • basilar artery • hemoglobin •

adenovirus • rat

appears to play an important role in the cluding neurons.28 Induction of HO-1 after SAH in rats was

H
EMOGLOBIN
pathogenesis of vasospasm after SAH. Modifica-
tion of the metabolism of hemoglobin and heme
may be important in the prevention of vasospasm. Oxida-
tion of the ferrous iron in oxy- or deoxyhemoglobin produc-
es methemoglobin, which more readily releases its heme
groups from the globin chains.5 Globin chains are probably
degraded by intra- and extracellular proteases. The heme
group is metabolized into biliverdin, CO, and iron by HOs
that consist of at least three isozymes: the oxidative stress-
inducible protein HO-1, constitutively expressed HO-2, and
HO-3.28,31 Heme oxygenase-1 (also known as heat shock
protein 32) usually is not expressed except in response to
stimuli such as hemin, heavy metals including iron, ultra-
violet light, H2O2, sodium arsenite, lipopolysaccharide, and
heat shock in rats but not in humans. In contrast, HO-2 and
HO-3 are constitutively expressed in many cell types, in-
Abbreviations used in this paper: ANOVA = analysis of variance;
BA = basilar artery; CBF = cerebral blood flow; CMV = cytomeg-
alovirus; CO = carbon monoxide; CSF = cerebrospinal fluid; HO =
heme oxygenase; LDF = laser Doppler flowmetry; mRNA = messen-
ger RNA; NOS = nitric oxide synthase; PBS = phosphate-buffered
saline; pfu = plaque-forming units; RT-PCR = reverse transcriptase–
polymerase chain reaction; SAH = subarachnoid hemorrhage.
suggested to have neuroprotective effects in the brain,30 and
inhibition of HO-1 expression in these arteries was report-
ed to increase vasospasm in cerebral arteries.47 It has never
been directly demonstrated, however, whether induction of
HO-1 may be used to modify cerebral vasospasm. In these
experiments we present direct evidence of a vasodilatory
contribution of the HO-1–CO system in the major cerebral
arteries of rats, as well as evidence that overexpression of
HO-1 in cerebral arteries can reduce vasospasm after exper-
imental SAH.
Materials and Methods
Adenovirus Preparation, Subarachnoid Virus Injection, and
Cranial Window Model
Synthesis of a replication-defective adenovirus expressing rat HO-
1 under control of the CMV immediate early promoter was per-
formed as described earlier.42 A recombinant, replication-defective
adenovirus containing the Escherichia coli
-galactosidase gene un-
der control of the CMV promoter (Ad-
Gal) was also prepared using
standard methods.14,20,44
Animal Preparation
All procedures performed in animals were approved by the Insti-

1094 J. Neurosurg. / Volume 96 / June, 2002

Prevention of vasospasm with HO-1 gene therapy
tutional Animal Care and Use Committee. Male Sprague–Dawley
rats weighing 250 to 300 g were used to examine the efficiency, loca-
tion, and function of viral transgene expression in the subarachnoid
space and BA. Methods for cisterna magna injection and creation
of a cranial window have been described.38 Briefly, purified adenovi-
rus (0.1 ml of Ad5HO-1 or Ad-
Gal containing 1010 pfu) or 0.1 ml
of 10% glycerol in saline (vehicle) were injected under sterile condi-
tions into the cisterna magna of anesthetized rats. One day after the
injection, rats were reanesthetized and placed supine in a stereotactic
frame. The BA was exposed transclivally and its diameter was mon-
itored using a surgical microscope equipped with a charge-coupled
device camera, video monitor, and calibrated optical measuring de-
vice. Body temperature, blood pressure, and PaO2 and PaCO2 were
measured using a rectal thermometer and a catheter inserted into the
femoral artery, respectively, and values were maintained in the phys-
iological range. The BA diameter was measured at baseline and after
application of pure human ferrous hemoglobin A0 in physiological
phosphate buffer. Applications were made so that the final concen-
tration in the window was 107 to 104 mol/L oxyhemoglobin, based
on a volume of the rat subarachnoid space of 0.3 ml.38 Brainstem
blood flow was measured using laser Doppler flowmetry 1 mm later-
al to the BA.
Assessment of Transgene Expression
The HO-1 mRNA was measured using RT-PCR 1 day after in-
jection of Ad5HO-1, Ad-
Gal, or vehicle into the cisterna mag-
na. One day after virus injections, the rats were anesthetized and
decapitated. The BAs were removed and placed in liquid N2. The
RNA was extracted and amplified using methods described previous
ly.29 Sequences of HO-1 and
-actin (internal control) primers were
5-AGAACCCAGTCTATGCCCCG-3 (sense primer), and 5-
TTGTCGATGCTCGGGAAGGTG-3 (antisense primer), 5-AGA-
TCATGTTTGAGACCTTCAACA-3 (sense primer) and 5-GCA-
CAGCTTCTCCTTAATGTC-3 (antisense primer), respectively.
The PCRs were performed in triplicate and radioactivity was deter-
mined directly by using commercially available software.
Translation of viral transgene products into protein was assessed
by immunoblotting of protein extracted from BAs used for PCR.
This was compared with a positive control obtained by incubation
of cultured rat BA smooth-muscle cells with 106 mol/L pure he-
moglobin.29 Immunoblotting was performed according to previously
published methods by using rabbit polyclonal antibody against rat
HO-1. Signal detection was performed with an enhanced chemilu-
minescence detection kit. Quantification of the density of specific
bands of HO-1 and internal standards was done using the National
Institutes of Health Gel scan program and standardized as density per
microgram protein measured using a protein assay.
For immunohistochemical localization of transgene expression,
BAs from rats injected with Ad5HO-1, Ad-
Gal, or vehicle were
harvested 1 day postinjection, after transcardiac perfusion of the ani-
mals at physiological pressure with PBS, followed by 4% parafor-
maldehyde in PBS. Cross-sections of the brainstem were immersed
in a graded series of sucrose solutions (up to 30%) and frozen
sections were cut on a cryostat. Immunohistochemical studies were
performed according to methods described previously, by using the
same HO-1 antibody as for immunoblotting.37,42 Alternate control
sections were incubated without primary antibody or with primary
antibody alone. Slides were counterstained with hematoxylin.
Assay of HO Activity and Measurement of Carboxyhemoglobin
To determine whether viral transfections altered HO activity, as-
says were performed in transfected BAs.33 The BAs were harvested
1 day after cisternal injection of Ad5HO-1, Ad-
Gal, or vehicle, or
in normal control rats after transcardiac perfusion of the animals
at physiological blood pressure with PBS containing (in mM): 137
NaCl; 8.1 Na2HPO4; 2.7 KCl; and 1.5 KH2PO4; pH 7.4. The BAs
were ground in a Dounce homogenizer in PBS and then assayed for
HO activity by using a previously described spectrophotometric as-
say.33 To further assess HO function, a semiquantitative calculation
of the concentration of carboxyhemoglobin as a product of CO re-
action with hemoglobin was determined in additional experiments
in which hemoglobin was applied to the BA of untreated rats or after
J. Neurosurg. / Volume 96 / June, 2002
cisternal injection of vehicle, Ad5HO-1, or Ad-
Gal. The CO con-
tent in the rat CSF was measured using spectrophotometry of hemo-
globin and carboxyhemoglobin, according to established methods.43
Pure human ferrous hemoglobin A0 (final concentration 1 mmol/L)
was applied to the BA, which was exposed through a cranial window
as described earlier. The CSF was aspirated from the window after
30 minutes and the concentration of carboxyhemoglobin was mea-
sured as absorbance at 421 nm minus absorbance at the isobestic
point (480 nm). The established extinction coefficient of 58.7 mmol/
L/cm was used to calculate the concentration of carboxyhemoglobin
in CSF.
Experimental Model of SAH
A rat double-hemorrhage model of SAH was used to assess
whether adenovirus-mediated HO-1 gene transfection prevents vaso-
spasm in vivo.48 Male Sprague–Dawley rats were randomly assigned
to three groups to receive intracisternal Ad5HO-1, Ad-
Gal, or vehi-
cle. The investigator performing injections and analyzing data did so
in a blinded fashion. On Day 0, animals were anesthetized and al-
lowed to breathe spontaneously, after which 0.25 ml of arterial blood
was mixed with 0.1 ml of Ad5HO-1, Ad-
Gal (1010 pfu), or vehicle
and injected into the cisterna magna over 5 minutes. Rats were re-
anesthetized 2 days after the initial injection (Day 2) and given a
seond injection of 0.3 ml of autologous arterial blood. Seven days af-
ter the first injection, rats were killed with overdoses of anesthetic
agents and their tissues were harvested, or the animals were fixed by
perfusion with PBS, followed by 4% paraformaldehyde in PBS at
physiological blood pressure. Frozen sections of BA and brainstem
were cut 10 m thick on the cryostat, and BA diameters were mea-
sured using a light microscope equipped with a micrometer. Cross-
sections of BA were obtained for measurement at three points: 200
m above the union of the vertebral arteries, just below the anteri-
or inferior cerebellar arteries, and 200 m below the BA bifurcation.
The mean of the three points was used as the diameter of the BA.
Data Analysis
All data analysis was conducted in a blinded fashion, and data are
expressed as the mean  standard deviation. Comparisons of levels
of mRNA and protein, and arterial diameter and blood flow within
and between groups, were made using ANOVA for multiple com-
parisons, followed by pairwise comparisons with the Fisher test if
significant variance was found. Paired or unpaired t-tests were used
for comparisons between two measurements. A probability value of
less than 0.05 was considered statistically significant.
Sources of Supplies and Equipment
The charge-coupled device camera and video monitor were pur-
chased from Hitachi, Yokohama, Japan. The optical measuring de-
vice and the micrometer-equipped light microscope were acquired
from Fisher Scientific, Pittsburgh, PA. The blood pressure moni-
toring device was obtained from Stoelting, Wood Dale, IL, and the
PaO2 and PaCO2 monitoring device (i-STAT System) was obtained
from Sensor Devices, Waukesha, WI. The purified hemoglobin was
supplied by Hemosol, Etobicoke, ON. The LDF device was pur-
chased from Transonic Systems, Ithaca, NY. The Phosphor Imager
and Image Quant software were acquired from Molecular Dynamics,
Sunnyvale, CA. The rabbit polyclonal antibody against rat HO-1 was
obtained from StressGen, Victoria, BC. The chemiluminescence de-
tection kit was supplied by New Life Science Products, Boston, MA.
The BCA protein assay kit was purchased from Pierce Chemical Co.,
Rockford, IL. The spectrophotometric device (Spectronic Genesys
5) was obtained from Milton Roy, Ivyland, PA.
Results
Expression of Recombinant HO-1 Adenovirus in the BA
The BAs that were removed 1 day after injection of
Ad5HO-1 into the cisterna magna contained increased
amounts of HO-1 mRNA (Fig. 1), whereas injection of Ad-
1095

FIG. 1. A: Representative blots showing RT-PCR amplifica-
tion of rat HO-1 (upper blot) and
-actin (lower blot) mRNA from
BAs in rats in which Ad5HO-1 (first and second lanes), Ad-
Gal
(third and fourth lanes), or 10% glycerol in saline vehicle (fifth and
sixth lanes) had been injected into the cisterna magna 1 day earli-
er, or BAs of untreated control rats (seventh and eighth lanes). B:
Bar graph showing semiquantification of RT-PCR bands for each
group (three rats per bar). There is significantly more HO-1 mRNA
after injection of Ad5HO-1 compared with the other groups (*p 
0.001 according to ANOVA).

Gal or vehicle (10% glycerol in saline) did not result in
substantial expression of HO-1 mRNA (three rats per
group). There was a statistically nonsignificant increase in
expression of HO-1 mRNA in the BAs of rats injected with
Ad-
Gal that was significantly less than in the BAs of
rats injected with Ad5HO-1 (p  0.001, ANOVA). Signif-
icantly increased HO-1 protein, as determined on immu-
noblot analysis, also was observed in the BA 1 day after
intracisternal injection of Ad5HO-1 (three rats per group,
p  0.001 compared with other groups; ANOVA, Fig. 2).
Injection of Ad-
Gal resulted in a low level of expression
of HO-1 protein in the BA, whereas injection of vehicle
did not.
Immunohistochemical studies of BAs obtained 1 day af-
ter injection of Ad5HO-1, Ad-
Gal, or vehicle showed
negligible HO-1 immunoreactivity after injection of vehicle
(three rats per group, Fig. 3). Marked HO-1 immunoreac-
1096
S. Ono, T. Komuro, and R. L. Macdonald
FIG. 2. A: Representative immunoblots of rat HO-1 (upper
blot) and
-actin (lower blot) from control smooth-muscle cells
exposed to hemoglobin, 10 mol/L for 24 hours (first lane) and
from BAs of rats in which Ad5HO-1 (second lane), Ad-
Gal (third
lane), or 10% glycerol in saline vehicle (fourth lane) had been in-
jected into the cisterna magna 1 day earlier, or BAs of untreated
control rats (fifth lane). B: Bar graph showing semiquantification
of immunoblot bands for each group (three rats per bar). There is
significantly more HO-1 protein after injection of Ad5HO-1 com-
pared with groups injected with Ad-
Gal or vehicle, and also sig-
nificantly more after injection of Ad-
Gal compared with vehicle,
but it was less than with Ad5HO-1 injections (*p  0.001 accord-
ing to ANOVA).
tivity was detected in the rat BA 1 day after Ad5HO-1 in-
jection. Immunoreactivity was confined to numerous cells
in the adventitia and to inflammatory cells that had sur-
rounded the BA and that were not seen after injection of ve-
hicle or around normal BAs. There was no detectable stain-
ing in the smooth-muscle cells of the tunica media or in the
tunica intima. Occasional cells showing immunoreactivity
for HO-1 were observed in the adventitia of BAs of rats in-
jected with Ad-
Gal, although these cells were much less
abundant than after injection of Ad5HO-1. Inflammatory
cells containing HO-1 immunoreactivity also were seen
around the BA of rats injected with Ad-
Gal. There was no
qualitative difference in the amount of inflammation ob-
served after injection of Ad5HO-1 or Ad-
Gal.
J. Neurosurg. / Volume 96 / June, 2002
Prevention of vasospasm with HO-1 gene therapy

FIG. 3. Photomicrographs showing immunohistochemical staining for HO-1 in BAs of rats injected with Ad5HO-1 (A
and C), Ad-
Gal (D), or vehicle (B) and counterstained with hematoxylin. Omission of primary antibody resulted in no
nonspecific staining in a rat injected with Ad5HO-1 (A); similarly, there is no HO-1 immunoreactivity after injection with
vehicle (B). After injection of Ad5HO-1 there is intense immunoreactivity in the adventitia and in inflammatory cells
around the BA (C), whereas after injection of Ad-
Gal, there are scattered positive cells in the adventitia and in inflam-
matory cells around the artery (D). Bars = 50 m.

Effect of HO-1 Transfection on BA Diameter and CBF
The baseline diameter of the BA and its diameter after
exposure to increasing doses of ferrous hemoglobin were
examined 1 day after intracisternal injection of Ad5HO-1
(1010 pfu, five rats), Ad-
Gal (1010 pfu, five rats), or an
equivalent volume of vehicle (six rats). Physiological pa-
rameters were maintained within the normal range (Table
1). The diameter of the BA at baseline was significant-
ly greater in the Ad5HO-1 (423  17 m) and Ad-
Gal
(419  13 m) groups compared with vehicle (367  11
m, p  0.001, ANOVA, Fig. 4), although there was no
significant difference in diameter between the Ad5HO-1
and Ad-
Gal groups. Application of 107 to 104 mol/L
of pure hemoglobin led to contractions that started within
minutes and that reached a maximum in 10 minutes. Wash-
out of the hemoglobin with PBS reversed the contraction to
baseline diameter in 50 minutes. Hemoglobin caused a con-
centration-dependent contraction of the BA in each group.
To allow for differences in the baseline diameter between
groups, the contractions to hemoglobin were expressed as a
percentage of the baseline diameter. This did not change the
overall pattern, however, which was that contractions of ar-
teries from animals transfected with Ad5HO-1 were always
less. The maximal percentage of contraction (Fig. 4) was
significantly less (p  0.05, ANOVA) in the group injected
0.001, ANOVA; Fig. 5). For 104 mol/L hemoglobin, the
CBF was significantly higher in the Ad5HO-1 group com-
pared with the other groups (p  0.001, ANOVA).
Activity of HO and Carboxyhemoglobin Formation
To determine whether adenovirus-mediated HO-1 trans-
fection produces active HO protein and CO, we examined
HO activity and the carboxyhemoglobin concentration in
untreated control rats (four for each factor) and in rats 1 day
after injection of Ad5HO-1 (six rats each), Ad-
Gal (HO
activity in six and carboxyhemoglobin in four), or vehicle
(HO activity in five and carboxyhemoglobin in three) into
the cisterna magna. The HO activity was increased sig-
nificantly in the Ad5HO-1 group compared with the ve-
TABLE 1
Physiological variables for effect of viral transfection and
hemoglobin on BA diameter*
Group
Ad5HO-1 Ad-
Gal Vehicle
Variable (5 rats) (5 rats) (6 rats)

with Ad5HO-1 compared with Ad-
Gal. blood pressure (mm Hg) 132  13 127  16 130  17
For LDF-measured blood flow, the baseline CBF and PaCO2 (mm Hg) 39  2 42  1 42  3
that recorded after application of 107 to 104 mol/L he- PaO2 (mm Hg) 88  5 87  5 85  6
moglobin were significantly higher in both groups inject- pH 7.40  0.02 7.41  0.02 7.41  0.06
ed with virus than in vehicle-injected control animals (p  * All values are expressed as the means  standard deviations.

J. Neurosurg. / Volume 96 / June, 2002 1097


FIG. 4. A: Bar graph of baseline BA diameter 1 day after in-
jection of Ad5HO-1 (five rats), Ad-Gal (five rats), or vehicle
(six rats), showing significantly increased diameter after injection
of either virus compared with vehicle (*p
0.001 according to
ANOVA). B: Graph showing concentration-contraction curves
generated by application of pure hemoglobin to the BA of rats in
which Ad5HO-1, Ad-Gal, or vehicle was injected into the cister-
na magna 1 day earlier. Data are expressed as the percentage of
change from baseline. Injection of Ad5HO-1 was associated with
less contraction to hemoglobin (107–104 mol/L hemoglobin, p

0.05 according to ANOVA) compared with the other groups.
hicle and untreated control groups (p
0.001, ANOVA;
Fig. 6). The HO activity was also higher in the Ad5HO-
1 than in the Ad-Gal group, although this was significant
only according to the unpaired t-test (p
0.05). Similar-
ly, for carboxyhemoglobin, there was significant variance,
with Ad5HO-1 treatment associated with significantly
higher concentrations than in the vehicle or untreated con-
trol groups (p
0.001, ANOVA) and significantly higher
concentrations than in the Ad-Gal–injected group accord-
ing to the unpaired t-test (p
0.05, Fig. 6).
Effect of Ad5HO-1 on Vasospasm After SAH
Five of 24 rats died of respiratory arrest after the first
1098
S. Ono, T. Komuro, and R. L. Macdonald
FIG. 5. Graph showing baseline LDF-measured blood flow in
the brainstem 1 day after injection of Ad5HO-1 (five rats), Ad-
Gal (five rats), or vehicle (six rats), and after application of 107
to 104 mol/L hemoglobin. Baseline flow and flow after appli-
cation of hemoglobin were significantly higher in both groups in-
jected with virus than in the vehicle-injected control animals for all
concentrations except 104 mol/L hemoglobin in animals inject
ed with Ad-Gal (*p
0.01 according to ANOVA). For 104 mol/
L hemoglobin, flow was significantly higher in the Ad5HO-1
group compared with the other groups (*p
0.001 according to
ANOVA).
or second cisternal blood injection. The 19 surviving rats
(Ad5HO-1, six rats; Ad-Gal, six rats; vehicle, seven rats)
exhibited normal behavior and appetite after recovery from
SAH and anesthesia. Two additional rats, both in the vehi-
cle group, were killed before Day 7 because of dehydration,
anorexia, or severe decrease in activity. Expression of HO-
1 mRNA in the Ad5HO-1–injected group increased sig-
nificantly 7 days after SAH compared with Ad-Gal– or
vehicle–injected groups (Fig. 7). Immunoblotting showed
that expression of HO-1 protein was significantly greater in
the Ad5HO-1 group compared with the Ad-Gal and vehi-
cle groups on Day 7 (p
0.05, ANOVA). Although HO-1
mRNA was detected on vehicle-injected rats with SAH,
protein was not detected by immunoblotting with methods
that detected protein in the virus-injected groups.
Overexpression of HO-1 by cisternal injection of Ad-
5HO-1 significantly diminished cerebral vasospasm af-
ter SAH (Fig. 8). The diameter of the BA 7 days after
SAH was significantly greater after injection of Ad5HO-1
(380  36 m) compared with Ad-Gal (282  44 m)
and vehicle (287  31 m, p
0.001; ANOVA).
Discussion
The new findings in this experiment are that rat BA di-
ameter and CBF can be increased by transfection of cells in
the adventitia with adenovirus expressing HO-1 or -galac-
tosidase. Overexpression of the HO-1 protein in the BA of
rats specifically increases HO activity, diminishes contrac-
tions of the artery to hemoglobin, prevents the reduction in
blood flow that occurs after exposure to hemoglobin, and
decreases vasospasm after SAH.
Previous investigations have shown that injection of
J. Neurosurg. / Volume 96 / June, 2002
Prevention of vasospasm with HO-1 gene therapy
FIG. 6. Bar graphs showing HO activity (A) and carboxyhe-
moglobin (CO-hemoglobin) concentration (B) in rats that 1 day
earlier had received intracisternal injections of Ad5HO-1 (six rats),
Ad-
Gal (HO activity, six rats; carboxyhemoglobin, four rats), or
vehicle (HO activity, five rats; carboxyhemoglobin, three rats), or
untreated control rats (HO activity and carboxyhemoglobin, four
rats each). The HO activity was increased significantly in the
Ad5HO-1 group compared with the vehicle and untreated control
groups (*p  0.001 according to ANOVA) and compared with the
Ad-
Gal group (*p  0.05 according to the unpaired t-test). Car-
boxyhemoglobin concentrations showed significant variance, with
Ad5HO-1 associated with significantly higher concentrations than
vehicle or untreated controls (*p  0.001 according to ANOVA),
and significantly higher concentrations than Ad-
Gal, according to
the unpaired t-test (p  0.05).
replication-defective adenoviruses expressing
-galactosi-
dase,9,41 NOS,7,8,51 or calcitonin gene–related peptide50 into
the cisterna magna can be used to transfect cells in the ad-
ventitia and arachnoid membrane of dogs, rats, and mice.
The finding that injection of virus mixed with blood does
not appreciably alter the pattern of expression is consistent
with our results.35,40,46 Injection of NOS and calcitonin gene–
related peptide were associated with alteration of vascular
J. Neurosurg. / Volume 96 / June, 2002
FIG. 7. A: Representative blots showing RT-PCR amplifica-
tion of rat HO-1 (upper blot) and
-actin (lower blot) mRNA from
BAs in rats with SAH and injection of Ad5HO-1 (first three lanes),
Ad-
Gal (fourth and fifth lanes), or vehicle (lanes 6–8) into the cis-
terna magna 7 days earlier. There is expression of HO-1 mRNA af-
ter injection of Ad5HO-1 and Ad-
Gal. B: Bar graph showing
semiquantification of RT-PCR bands for each group (three rats per
bar). There is significantly more HO-1 mRNA after injection of
Ad5HO-1 compared with the other groups (p  0.01 according to
ANOVA).
function when the arteries were removed and studied under
isometric tension in vitro.7,8,50,51 In most of these studies vas-
cular function, arterial diameters at baseline, or CBF in vivo
were not assessed.
We previously reported that injection of adenovirus ex-
pressing endothelial NOS increased the baseline BA di-
ameter in dogs;46 arterial diameter was assessed 7 days after
viral injection. In the present study we have shown that
injection of both control adenovirus expressing
-galacto-
sidase and of Ad5HO-1 increased BA diameter and brain-
stem CBF 1 day later. This may be because injection of
virus was associated with acute inflammation, which is
known to increase CBF acutely.15,21 The increase in CBF
may be due to expression of inducible NOS in inflammato-
ry cells, with vasodilation due to the production of nitric
oxide.16 The finding that inhibition of post-SAH inflamma-
tion by intracisternal administration of antisense oligonu-
1099

FIG. 8. Bar graph of BA diameter 7 days after SAH in a double-
injection model with injection of Ad5HO-1 (five rats), Ad-
Gal
(five rats), or vehicle (six rats) at the time of initial SAH, showing
significantly increased diameter after injection of either virus com-
pared with vehicle (p  0.001 according to ANOVA).
cleotides to nuclear factor-B prevented vascular contrac-
tion 7 days after SAH argues against this mechanism.36 An-
other explanation is that both viruses were associated with
HO-1 expression and that CO produced by HO-1 led to va-
sodilation; there is evidence that CO dilates systemic arter-
ies.18,23 It has been reported that CO increased CBF in rats,22
although this phenomenon also could be due to the reduced
O2-carrying capacity of carboxyhemoglobin. Brian, et al.,4
reported that CO did not dilate cerebral arteries, but this
finding is at variance with the accepted opinion that CO is
a vasodilator.
It is unlikely that HO-1 expression was a nonspecific re-
sponse to the inflammation in the Ad5HO-1 group, because
the virus is known to produce HO-1 in rats,42 and there was
greater HO-1 expression after injection of this virus than
after injection of Ad-
Gal. It must be acknowledged, how-
ever, that we cannot differentiate endogenous HO-1 from
HO-1 produced by the virus, because both were of rat ori-
gin. On the other hand, HO-1 may suppress inflammation,
so it would be expected that nonspecific induction of HO-1
due to inflammation would be less in this group.6 We ob-
served no differences in the degree of inflammation associ-
ated with each virus. In our previous study we did not find
dilation of the BA in dogs 7 days after intracisternal injec-
tion of Ad-
Gal. One possibility is that the dilatory effects
of inflammation induced by Ad-
Gal are transient. The
CBF is elevated in the acute stages of meningitis, and when
inflammation becomes chronic the CBF is reduced.24 Fur-
thermore, inhibition of inflammation 7 days after SAH re-
duced vasospasm.36 The duration of transgene expression
also must be taken into account, although it was not inves-
tigated fully in this model. Christenson, et al.,9 noted that
transgene expression lasted only a few days when it was
driven by the CMV promoter. There was increased HO-1
expression 7 days post-SAH, after injection of Ad5HO-1,
although this did not appear to be as high as after 1 day.
Production of HO-1 in cells in the adventitia was associ-
ated with an increase in diameter of the BA and with an in-
1100
S. Ono, T. Komuro, and R. L. Macdonald
crease in CBF in the adjacent brainstem. The vasodilation
could be due to several mechanisms. The CO produced by
HO-1 could diffuse from adventitial cells to the smooth-
muscle cells to mediate this effect. This also would be con-
sistent with the selective reduction in hemoglobin-induced
contraction of the BA and the increase in brainstem CBF
that was observed. Expression was achieved only in the ar-
terial adventitia, which is consistent with previous reports
on the results of cisternal injection of viruses.7–9,39,41
We previously reported that hemoglobin mediates vaso-
spasm;27 numerous mechanisms have been proposed, but
most involve the intact hemoglobin molecule.12,26 Thus,
a method for removing hemoglobin more rapidly from
the subarachnoid space after SAH might be efficacious in
preventing vasospasm. These experiments show that over-
expression of HO-1 can reduce the contractile effect of he-
moglobin. Numerous mechanisms are possible: overpro-
duction of CO could mediate vascular relaxation or could
bind to hemoglobin, preventing hemoglobin from scav-
enging nitric oxide. The HO could metabolize hemoglobin,
producing more CO, which would have the same effects,
and reducing the amount of vasoactive hemoglobin. It has
been shown previously that induction of HO-1 may medi-
ate similar effects in the systemic vasculature. Infusion of
hemoglobin in rats usually elevates blood pressure although
this does not occur 1 day after animals are stressed and at a
time when HO-1 is induced in the vasculature.34 Induction
of HO-1 is often associated with increased expression of
ferritin.1–3,6 We did not assess ferritin levels but iron uptake
by ferritin could theoretically be involved in preventing
hemoglobin-related contraction, because iron compounds
may contract cerebral arteries.13 Iron chelators prevent ex-
perimental vasospasm but it seems unlikely that iron re-
leased from hemoglobin would participate in its acute con-
tractile action.17,52
After observing reduced contractions to hemoglobin, we
tested whether Ad5HO-1 could prevent vasospasm after
SAH. The rat model of SAH was chosen, which has the
benefits of simplicity, economy, and production of a de-
layed arterial narrowing that lasts for up to 7 days. Two in-
jections of blood were administered, because it is widely
accepted that just one injection produces vasospasm that
lasts for only 2 days and that is not associated with other ac-
cepted features of vasospasm, such as pathological changes
in the cerebral arteries.32 Overexpression of HO-1 reduced
vasospasm after SAH. In addition to the mechanisms dis-
cussed earlier, HO metabolism of heme produces biliver-
din, which could be converted to bilirubin. The importance
of this compound in vasospasm is uncertain because it has
been reported to cause vasospasm,11,27 but it also is an an-
tioxidant that could reduce oxidative reactions that are sug-
gested to mediate vasospasm.10,25,28 Our finding that over-
expression of HO-1 reduces vasospasm is consistent with
earlier reports of a role for HO-1 in vasospasm.
We have reported that HO-1 and ferritin are increased in
monkey cerebral arteries 7 and 14 days after SAH, at a time
when vasospasm is resolving even in the continued pres-
ence of SAH, indicating that there may be an adaptive re-
sponse of the cerebral arteries that allows them to relax
in the continued presence of the subarachnoid blood that
caused the spasm in the first place.45 Suzuki, et al.,48 showed
that HO-1 was markedly overexpressed in the rat BA after
SAH and that inhibition of HO-1 expression increased va-
J. Neurosurg. / Volume 96 / June, 2002
Prevention of vasospasm with HO-1 gene therapy
sospasm and delayed clearance of hemoglobin from the
subarachnoid space after SAH. Theoretically, an additional
beneficial effect is that HO may be neuroprotective in cere-
bral ischemia which is the mechanism by which vasospasm
would damage the brain.10,19,49 We could not assess neuro-
logical effects of vasospasm in this model because ischemia
is not produced.
Conclusions
We report successful transfection of the arterial adventitia
of the rat BA with HO-1. Although adenoviral transfection
is associated with a nonspecific increase in vascular diame-
ter and CBF after 1 day, HO-1 overexpression specifical-
ly prevents hemoglobin-induced contraction of the BA and
vasospasm after SAH in a rat double-hemorrhage model.
Further investigations are needed to determine the mecha-
nisms of these effects and whether vasospasm can be pre-
vented in models of this disease in large animals.
Acknowledgments
We thank Hemosol for providing pure hemoglobin, Drs. A. Choi
for supplying Ad5HO-1, and J. Schaak for supplying Ad-
Gal.
References
1. Applegate LA, Scaletta C, Panizzon R, et al: Evidence that ferritin
is UV inducible in human skin: part of a putative defense mech-
anism. J Invest Dermatol 111:159–163, 1998
2. Balla G, Jacob HS, Balla J, et al: Ferritin: a cytoprotective antiox-
idant strategem of endothelium. J Biol Chem 267:18148–18153,
1992
3. Bornman L, Baladi S, Richard MJ, et al: Differential regulation
and expression of stress proteins and ferritin in human mono-
cytes. J Cell Physiol 178:1–8, 1999
4. Brian JE Jr, Heistad DD, Faraci FM: Effect of carbon monoxide
on rabbit cerebral arteries. Stroke 25:639–644, 1994
5. Bunn HF, Jandl JH: Exchange of heme among hemoglobins and
between hemoglobin and albumin. J Biol Chem 243:465–475,
1968
6. Carraway MS, Ghio AJ, Taylor JL, et al: Induction of ferritin and
heme oxygenase-1 by endotoxin in the lung. Am J Physiol 275:
L583–L592, 1998
7. Chen AF, Jiang SW, Crotty TB, et al: Effects of in vivo adven-
titial expression of recombinant endothelial nitric oxide syn-
thase gene in cerebral arteries. Proc Nat Acad Sci USA 94:
12568–12573, 1997
8. Chen AF, O’Brien T, Tsutsui M, et al: Expression and function
of recombinant endothelial nitric oxide synthase gene in canine
basilar artery. Circ Res 80:327–335, 1997
9. Christenson SD, Lake KD, Ooboshi H, et al: Adenovirus-mediat-
ed gene transfer in vivo to cerebral blood vessels and perivascu-
lar tissue in mice. Stroke 29:1411–1416, 1998
10. Dore S, Sampei K, Goto S, et al: Heme oxygenase-2 is neuropro-
tective in cerebral ischemia. Mol Med 56:656–663, 1999
11. Duff TA, Feilbach JA, Yusuf Q, et al: Bilirubin and the induction
of intracranial arterial spasm. J Neurosurg 69:593–598, 1988
12. Everse J, Hsia N: The toxicities of native and modified hemoglo-
bins. Free Radic Biol Med 22:1075–1099, 1997
13. Fox JL: Intracranial vasospasm: a study with iron compounds.
Surg Neurol 11:363–368, 1979
14. Graham FL, Prevec L: Manipulation of adenoviral vectors, in
Murray EJ (ed): Molecular Biology: Gene Transfer and Ex-
pression Protocols. Clifton, NJ: Humana Press, 1991, Vol 7, pp
7109–7128
15. Haberl RL, Anneser F, Kodel U, et al: Is nitric oxide involved as
J. Neurosurg. / Volume 96 / June, 2002
a mediator of cerebrovascular changes in the early phase of ex-
perimental pneumococcal meningitis. Neurol Res 16:108–112,
1994
16. Hartsfield CL, Alam J, Cook JL, et al: Regulation of heme oxy-
genase-1 gene expression in vascular smooth muscle cells by ni-
tric oxide. Am J Physiol 273:L980–L988, 1997
17. Horky LL, Pluta RM, Boock RJ, et al: Role of ferrous iron chela-
tor 2,2-dipyridyl in preventing delayed vasospasm in a primate
model of subarachnoid hemorrhage. J Neurosurg 88:298–303,
1998
18. Johnson RA, Lavesa M, Askari B, et al: A heme oxygenase
product, presumably carbon monoxide, mediates a vasodepressor
function in rats. Hypertension 25:166–169, 1995
19. Kadoya C, Domino EF, Yang GY, et al: Preischemic but not
postischemic zinc protoporphyrin treatment reduces infarct size
and edema accumulation after temporary focal cerebral ischemia
in rats. Stroke 26:1035–1038, 1995
20. Kanegae Y, Makimura M, Saito I: A simple and efficient meth-
od for purification of infectious recombinant adenovirus. Jpn J
Med Sci Biol 47:157–166, 1994
21. Koedel U, Bernatowicz A, Paul R, et al: Experimental pneumo-
coccal meningitis: cerebrovascular alterations, brain edema, and
meningeal inflammation are linked to the production of nitric ox-
ide. Ann Neurol 37:313–323, 1995
22. Koehler RC, Jones MD Jr, Traystman RJ: Cerebral circulatory re-
sponse to carbon monoxide and hypoxic hypoxia in the lamb.
Am J Physiol 43:H27–H32, 1982
23. Kozma F, Johnson RA, Zhang F, et al: Contribution of endog-
enous carbon monoxide to regulation of diameter in resistance
vessels. Am J Physiol 276:R1087–R1094, 1999
24. Leib SL, Täuber MG: Pathogenesis of bacterial meningitis. Infect
Dis Clin North Am 13:527–548, 1999
25. Macdonald RL, Weir BK: Cerebral vasospasm and free radicals.
Free Radic Biol Med 16:633–643, 1994
26. Macdonald RL, Weir BK: A review of hemoglobin and the
pathogenesis of cerebral vasospasm. Stroke 22:971–982, 1991
27. Macdonald RL, Weir BK, Runzer TD, et al: Etiology of cerebral
vasospasm in primates. J Neurosurg 75:415–424, 1991
28. Maines MD: The heme oxygenase system: a regulator of second
messenger gases. Annu Rev Pharmacol Toxicol 37:514–554,
1997
29. Marton LS, Wang X, Kowalczuk A, et al: Effects of hemoglobin
on heme oxygenase gene expression and viability of cultured
smooth muscle cells. Am J Physiol Heart Circ Physiol 279:
H2405–H2413, 2000
30. Matz P, Turner C, Weinstein PR, et al: Heme-oxygenase-1 in-
duction in glia throughout rat brain following experimental sub-
arachnoid hemorrhage. Brain Res 713:211–222, 1996
31. McCoubrey WK Jr, Huang TJ, Maines MD: Isolation and char-
acterization of a cDNA from the rat brain that encodes hemopro-
tein heme oxygenase-3. Eur J Biochem 247:725–732, 1997
32. Megyesi JF, Vollrath B, Cook DA, et al: In vivo animal models of
cerebral vasospasm: a review. Neurosurgery 46:448–461, 2000
33. Motterlini R, Foresti R, Vandegriff K, et al: Oxidative-stress re-
sponse in vascular endothelial cells exposed to acellular hemo-
globin solutions. Am J Physiol 269:H648–H655, 1995
34. Motterlini R, Gonzales A, Foresti R, et al: Heme oxygenase-1-de-
rived carbon monoxide contributes to the suppression of acute
hypertensive responses in vivo. Circ Res 83:568–577, 1998
35. Muhonen MG, Ooboshi H, Welsh MJ, et al: Gene transfer to ce-
rebral blood vessels after subarachnoid hemorrhage. Stroke 28:
822–829, 1997
36. Ono S, Date I, Onoda K, et al: Decoy administration of NF-B
into the subarachnoid space for cerebral angiopathy. Hum Gene
Ther 9:1003–1011, 1998
37. Ono S, Zhang ZD, Marton LS, et al: Heme oxygenase-1 and fer-
ritin are increased in cerebral arteries after subarachnoid hemor-
rhage in monkeys. J Cereb Blood Flow Metab 20:1066–1076,
2000
1101

38. Onoda K, Ono S, Ogihara K, et al: Inhibition of vascular contrac-
tion by intracisternal administration of preproendothelin-1 mRNA
antisense oligoDNA in a rat experimental vasospasm model. J
Neurosurg 85:846–852, 1996
39. Onoue H, Katusic ZS: Subarachnoid hemorrhage and the role of
potassium channels in relaxations of canine basilar artery to ni-
trovasodilators. J Cereb Blood Flow Metab 18:186–195, 1998
40. Onoue H, Tsutsui M, Smith L, et al: Expression and function of
recombinant endothelial nitric oxide synthase gene in canine bas-
ilar artery after experimental subarachnoid hemorrhage. Stroke
29:1959–1966, 1998
41. Ooboshi H, Welsh MJ, Rios CD, et al: Adenovirus-mediated
gene transfer in vivo to cerebral blood vessels and perivascular
tissue. Circ Res 77:7–13, 1995
42. Otterbein LE, Kolls JK, Mantell LL, et al: Exogenous admin-
istration of heme oxygenase-1 by gene transfer provides protec-
tion against hyperoxia-induced lung injury. J Clin Invest 103:
1047–1054, 1999
43. Rodkey FL, Hill TA, Pitts LL, et al: Spectrophotometric measure-
ment of carboxyhemoglobin and methemoglobin in blood. Clin
Chem 25:1388–1393, 1979
44. Schaack J, Langer S, Guo X: Efficient selection of recombinant
adenoviruses by vectors that express beta-galactosidase. J Virol
69:3920–3923, 1995
45. Stoodley M, Macdonald RL, Weir B, et al: Subarachnoid hem-
orrhage as a cause of an adaptive response in cerebral arteries. J
Neurosurg 93:463–470, 2000
46. Stoodley M, Weihl CC, Zhang ZD, et al: Effect of adenoviral-
mediated nitric oxide synthase gene transfer on vasospasm af-
ter experimental subarachnoid hemorrhage. Neurosurgery 46:
1193–2003, 2000
47. Suzuki H, Kanamaru K, Kuroki M, et al: Effects of tirilazad mes-
1102
S. Ono, T. Komuro, and R. L. Macdonald
ylate on vasospasm and phospholipid hydroperoxides in a pri-
mate model of subarachnoid hemorrhage. Stroke 30:450–456,
1999
48. Suzuki H, Kanamaru K, Tsunoda H, et al: Heme oxygenase-1
gene induction as an intrinsic regulation against delayed cerebral
vasospasm in rats. J Clin Invest 104:59–66, 1999
49. Takizawa S, Hirabayashi H, Matsushima K, et al: Induction of
heme oxygenase protein protects neurons in cortex and striatum,
but not in hippocampus, against transient forebrain ischemia. J
Cereb Blood Flow Metab 18:559–569, 1998
50. Toyoda K, Faraci FM, Russo AF, et al: Gene transfer of calcito-
nin gene-related peptide to cerebral arteries. Am J Physiol Heart
Circ Physiol 278:H586–H594, 2000
51. Tsutsui M, Onoue H, Iida Y, et al: Adventitia-dependent relax-
ations of canine basilar arteries transduced with recombinant
eNOS gene. Am J Physiol 276:H1846–H1852, 1999
52. Vollmer DG, Hongo K, Ogawa H, et al: A study of the effective-
ness of the iron-chelating agent deferoxamine as vasospasm pro-
phylaxis in a rabbit model of subarachnoid hemorrhage. Neuro-
surgery 28:27–32, 1991
Manuscript received August 13, 2001.
Accepted in final form January 15, 2002.
This study was supported by Grant No. NS01831 to Dr. Macdon-
ald from the National Institutes of Health and the Brain Research
Foundation.
Address reprint requests to: R. Loch Macdonald, M.D., Ph.D.,
Section of Neurosurgery, MC3026, University of Chicago Medi-
cal Center, 5841 South Maryland Avenue, Chicago, Illinois 60637.
email: lmacdona@surgery.bsd.uchicago.edu.
J. Neurosurg. / Volume 96 / June, 2002
Title

J. Neurosurg. / Volume 96 / June, 2002 a1