?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
Endothelial sulfonylurea receptor 1–
regulated NCCa-ATP channels mediate
progressive hemorrhagic necrosis
following spinal cord injury
J. Marc Simard,1,2,3 Orest Tsymbalyuk,1 Alexander Ivanov,1 Svetlana Ivanova,1
Sergei Bhatta,1 Zhihua Geng,1 S. Kyoon Woo,1 and Volodymyr Gerzanich1
1Department of Neurosurgery, 2Department of Pathology, and
3Department of Physiology, School of Medicine, University of Maryland at Baltimore, Baltimore, Maryland, USA.
Acute spinal cord injury (SCI) results in physical disruption of spi-
nal cord neurons and axons leading to deficits in motor, sensory,
and autonomic function. SCI is a debilitating neurological disor-
der common in young adults that often requires lifelong therapy
and rehabilitative care, placing significant burdens on health care
systems. Although many patients exhibit neuropathologically and
clinically complete cord injuries following SCI, many others have
neuropathologically incomplete lesions (1, 2), giving hope that
proper treatment to minimize secondary injury may reduce the
functional impact of SCI.
The concept of secondary injury in SCI arises from the observa-
tion that the lesion expands and evolves over time (2, 3). Whereas
primary injured tissues are irrevocably damaged at the time of
impact, tissues that are destined to become secondarily injured
are considered to be potentially salvageable. Older observations,
based on histological studies, that gave rise to the concept of
lesion evolution have been confirmed with noninvasive MRI (4).
Several mechanisms of secondary injury have been postulated,
including ischemia/hypoxia, oxidative stress, and inflamma-
tion, all of which have been considered to be responsible for the
devastating process of progressive hemorrhagic necrosis (PHN)
(2, 5–9). PHN is a mysterious condition, first recognized over 3
decades ago, that has thus far eluded understanding and treat-
ment. Shortly after injury (10–15 min), a small hemorrhagic
lesion involving primarily the capillary-rich central gray matter
is observed, but over the following 3–24 hours, petechial hemor-
rhages emerge in more distant tissues, eventually coalescing into
the characteristic lesion of hemorrhagic necrosis (10, 11). The
white matter surrounding the hemorrhagic gray matter shows
a variety of abnormalities, including decreased H&E staining,
disrupted myelin, and axonal and periaxonal swelling. White
matter lesions extend far from the injury site, especially in the
posterior columns (8). The evolution of hemorrhage and necro-
sis has been referred to as “autodestruction.” PHN results in
loss of vital spinal cord tissue and, in some species, including
humans, leads to post-traumatic cystic cavitation surrounded
by glial scar tissue.
The mechanism responsible for PHN is not known. Tator and
Koyanagi (8) speculated that obstruction of small intramedullary
vessels by the initial mechanical stress or secondary injury might
be responsible for PHN, whereas Kawata and colleagues (11) attrib-
uted the progressive changes to leukocyte infiltration around the
injured area leading to plugging of capillaries. Given that petechial
hemorrhages, the pathognomonic feature of PHN, form as a result
of catastrophic failure of vascular integrity, damage to the endothe-
lium of spinal cord capillaries and postcapillary venules has long
Nonstandard?abbreviations?used: ODN, oligodeoxynucleotide; NCCa-ATP, Ca2+-acti-
vated, [ATP]i-sensitive nonspecific cation (channel); PHN, progressive hemorrhagic
necrosis; SCI, spinal cord injury; SUR1, sulfonylurea receptor 1.
Conflict?of?interest: J.M. Simard has applied for a US patent, “A novel non-selective
cation channel in neural cells and methods for treating brain swelling” (application no.
10/391,561). The remaining authors have declared that no conflict of interest exists.
Citation?for?this?article: J. Clin. Invest. 117:2105–2113 (2007). doi:10.1172/JCI32041.
2106? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
been regarded as a major factor in the pathogenesis of PHN (5, 12,
13). However, to our knowledge, no molecular mechanism for pro-
gressive dysfunction of endothelium has yet been identified.
The sulfonylurea receptor 1–regulated (SUR1-regulated) Ca2+-
activated, [ATP]i-sensitive nonspecific cation (NCCa-ATP) channel is
a nonselective cation channel that is not constitutively expressed,
but is transcriptionally upregulated in astrocytes and neurons fol-
lowing hypoxic or ischemic insult (14–16). The channel is inactive
when expressed but becomes activated when intracellular ATP is
depleted, with activation leading to cell depolarization, cytotoxic
edema, and oncotic cell death. Block of the channel in vitro by the
sulfonylurea glibenclamide prevents cell depolarization, cytotoxic
edema, and oncotic cell death induced by ATP depletion. In rodent
models of ischemic stroke, treatment with glibenclamide results in
significant improvements in edema, lesion volume, and mortality
(16). In humans with diabetes mellitus, use of sulfonylureas before
and during hospitalization for stroke is associated with signifi-
cantly better stroke outcomes (17).
We hypothesized that NCCa-ATP channels might also be involved
in PHN in SCI. Although endothelial dysfunction has been
implicated in PHN, SUR1-regulated NCCa-ATP channels have not
previously been shown to our knowledge in capillary endothe-
lium. Here, we used a rodent model of unilateral cervical SCI and
endothelial cell cultures to evaluate our hypothesis. We found
SUR1 is upregulated in SCI. (A) Immunohistochemical localization of SUR1 in control rats (CTR) and at different times after SCI as indicated,
with montages constructed from multiple individual images and positive labeling shown in black pseudocolor. (B) Magnified views of SUR1-
immunolabeled sections taken from control and from the core (heavily labeled area in A at 6 hours). (C and D) Immunolabeling of capillaries with
vimentin (Vim) and colabeling with SUR1 in control rats (C) and from the penumbra of SCI rats (tissue adjacent to the heavily labeled core in A,
6 hours) (D). (E) Western blots for SUR1 of spinal cord tissue from control rats (50 μg protein), from rats 6 hours after SCI (50 μg protein), and
from an equivalent amount of blood (BL; 2 μl) as is present in the injured cord. Blots are representative of 5–6 control and SCI rats. (F and G) In
situ hybridization for Abcc8 in control rats and in whole cords (F) or in the penumbra (G) 6 hours after SCI using antisense (AS) and sense (SE)
as indicated. Immunohistochemistry and in situ hybridization images are representative of findings in 3–5 rats per group. Scale bars: 1 mm (A);
100 μM (B–D and G, top panels and bottom left panel); 50 μM (G, bottom right panel).
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
that SUR1 was prominently upregulated in capillaries in the
region of SCI in rats; that endothelial cells subjected to hypoxic
conditions express SUR1-regulated NCCa-ATP channels; and that
inhibition of SUR1 by a variety of molecularly distinct mecha-
nisms largely eliminated the progressive extravasation of blood
characteristic of PHN, reduced lesion size, and was associated
with marked neurobehavioral functional improvement. These
findings are all consistent with a critical role for SUR1-regulated
NCCa-ATP channels in PHN following SCI.
Upregulation of SUR1 in SCI. We studied SUR1 expression in spinal
cords of uninjured rats and rats after severe SCI (10-g weight dropped
from a 25-mm height; n = 3–5 per group) (18, 19). In controls, low lev-
els of SUR1 expression were found in the dorsal horns (Figure 1A) as
a result of constitutively expressed KATP channels (20).
After unilateral SCI, the lesion itself as well as the pattern of SUR1
expression changed with time and distance from the impact site
(Figure 1A). Early after SCI (45 minutes), the lesion was small and
was not immunolabeled by anti-SUR1 antibody (data not shown).
At 6 hours, a necrotic lesion was apparent as a void in the ipsilateral
cord, and SUR1 upregulation was prominent in tissues surround-
ing the void. At 24 hours, as the necrotic lesion had enlarged (5,
21), SUR1 upregulation was still apparent in the rim of the necrotic
lesion, but extended to tissues more distant from the impact site,
including into the contralateral hemicord. Immunolabeling for
SUR2 was detected only in vascular smooth muscle cells of pial
arterioles, both before and after SCI (data not shown).
In the core of the lesion (heavily labeled area in Figure 1A, cen-
ter), SUR1 upregulation was present in various cells and structures,
including large ballooned neuron-like cells and capillary-like elon-
gated structures (Figure 1B). In the penumbra (tissue adjacent to
the lesion core), SUR1 upregulation was associated predominantly
with capillaries (Figure 1, C and D).
Upregulation of SUR1 was confirmed with immunoblots. With
the amount of protein loaded, SUR1 was not detectable in normal
cords, whereas a prominent single band at about 190 kDa (16) was
observed 6 hours after SCI (Figure 1E). The blood introduced into
the tissues by the injury did not account for the increase in SUR1
(Figure 1E). In situ hybridization confirmed widespread expression
of Abcc8, which encoded for SUR1, after injury, especially in capillar-
ies and postcapillary venules in the penumbra (Figure 1, F and G).
SUR1 in endothelium is associated with the NCCa-ATP channel. SUR1
forms the regulatory subunit of NCCa-ATP channels and some KATP
channels (15). Our previous work demonstrated that following
exposure to hypoxia or ischemia in vivo, upregulation of SUR1
in astrocytes and neurons is associated with expression of func-
tional NCCa-ATP channels, not KATP channels (15, 16). The same
reports also showed upregulation of SUR1 in capillaries, as we
found here with SCI, but the associated channel was not identi-
fied. Endothelial cells may normally express KATP channels, but
the regulatory subunit of cardiovascular KATP channels is generally
SUR2, not SUR1 (22). Nevertheless, it was important to determine
with which of the 2 channels, KATP or NCCa-ATP, the newly expressed
SUR1 was associated in capillary endothelium.
Endothelial cell cultures from 3 sources — human brain micro-
vascular, human aorta, and murine brain microvascular — were used
to assess SUR1 expression and characterize channel properties fol-
lowing exposure to hypoxia, with the same results observed for all
3 cultures. Control cultures showed little expression of SUR1, but
exposure to hypoxia for 24 hours caused substantial upregulation
of SUR1 (Figure 2A). Insulinoma cells, which constitutively express
The SUR1-regulated NCCa-ATP channel is upregulated in endothelial
cells by hypoxia. (A) Immunolabeling (scale bar: 50 μm) and Western
blots for SUR1 in human aortic endothelial cells (ENDO) cultured under
normoxic (N) or hypoxic (H) conditions as indicated, as well as Western
blots for SUR1 of rat insulinoma RIN-m5F cells (INSUL) cultured under
normoxic or hypoxic condition, with β-actin also shown. (B and C)
Whole-cell currents during ramp pulses (4 per minute; holding poten-
tial [HP], –50 mV) or at the holding potential of –50 mV, before and
after application of diazoxide (B) or Na azide (C), in endothelial cells
exposed to normoxic or hypoxic conditions; the difference currents are
also shown in red. Erev, reversal potential; GLIB, glibenclamide. Data
are representative of 7–15 recordings from human aortic endothelial
cells (B) or bEnd.3 cells (C) for each condition. (D) Single-channel
recordings of inside-out patches with Cs+ as the principal cation, with
channel openings inhibited by ATP on the cytoplasmic side; channel
amplitude at various potentials indicated a slope conductance of 37
pS (data from 7 patches) from human brain microvascular endothelial
cells. Error bars indicate SEM.
2108?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
SUR1-regulated KATP channels, showed no upregulation of SUR1
when exposed to the same hypoxic conditions (Figure 2A).
Patch clamp of endothelial cells was performed using a nystat-
in-perforated patch technique in order to maintain the metabolic
integrity of the cells. The identity of the activated channel can be
assessed by measurement of the reversal potential, the potential
at which an ion channel current reverses from inward to outward.
With physiologically relevant concentrations of ions intracellularly
and extracellularly (high K inside, high Na outside), the reversal
potential can unambiguously distinguish between a K+ channel
current such as KATP, which reverses at a potential negative to –50
mV, and a nonselective cation channel current such as NCCa-ATP,
which reverses near 0 mV.
We studied channel activation by diazoxide, which opens SUR-
regulated channels without ATP depletion and, of SUR activa-
tors, is the most selective for SUR1 over SUR2 (15). Patch clamp
of endothelial cells cultured under normoxic conditions showed
that diazoxide had no effect or, in half of the cells, activated an
outwardly rectifying current that reversed at potentials more nega-
tive than –50 mV, consistent with a KATP channel (Figure 2B) (23).
By contrast, in most endothelial cells cultured under hypoxic con-
ditions, diazoxide activated an ohmic current that reversed near 0
mV and was inward at –50 mV (Figure 2B), which is incompatible
with KATP, but consistent with NCCa-ATP channels (14–16).
We also studied channel activation induced by Na azide, a mito-
chondrial uncoupler that depletes cellular ATP (14). In most
endothelial cells exposed to hypoxic conditions, Na azide–induced
ATP depletion activated an ohmic current that was inward at
–50 mV, reversed near 0 mV, and was blocked by 1 μM gliben-
clamide (Figure 2C), again consistent with NCCa-ATP channels.
Single-channel recordings were performed using inside-out
patches, with Cs+ as the only permeant cation. This confirmed the
presence of a channel that was sensitive to block by ATP on the
cytoplasmic side and that had a single channel conductance of
37 pS (Figure 2D). These findings are incompatible with KATP
channels, which are not permeable to Cs+ and have slope conduc-
tance of ~75 pS, but are consistent with NCCa-ATP channels.
The characteristics of the channel identified in endothelial cells
from both aorta and brain capillaries from 2 species — including
expression only after exposure to hypoxia, activation by deple-
tion of cellular ATP or diazoxide, a reversal potential near 0 mV,
conductance of Cs+, and single channel conductance of 37 pS —
Block of SUR1 reduces hemorrhage after SCI. (A) Whole cords and longitudinal sections of cords 24 hours after SCI, from vehicle-treated con-
trol and glibenclamide-treated rats. White circles indicate site of impact; arrows denote petechial hemorrhages. (B) Cord homogenates in test
tubes at 24 hours (inset) and spectrophotometric measurements of blood in cord homogenates at various times after SCI from vehicle-treated
(n = 66) and glibenclamide-treated (n = 62) rats. *P < 0.05, **P < 0.01, ***P < 0.001 versus control. (C) Cord sections immunolabeled for vimentin
to show capillaries from SCI rats treated with vehicle or glibenclamide; arrows indicate the central canal. Right panels are higher-magnification
images of boxed areas in left panels. Images are representative of findings in 6 rats per group. Asterisks indicate lesion core at impact site. DH,
dorsal horn. (D) Zymography of recombinant MMP-2 and MMP-9 performed under control conditions, in the presence of glibenclamide (10 μM),
and in the presence of MMP inhibitor II (MMP inhib; 300 nM). (E) Bleeding times in uninjured rats infused with vehicle or glibenclamide (n = 3 per
group). Error bars indicate SEM. Scale bars: 1 mm (A); 0.3 mm (C, left panels). Original magnification, ×40 (C, right panels).
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
reproduce exactly our previous findings with NCCa-ATP channels
in astrocytes and neurons (14–16) and reaffirm that the NCCa-ATP
channel is not constitutively expressed, is upregulated only with
an appropriate insult, and, when expressed, is inactive until
intracellular ATP is depleted.
Glibenclamide block of SUR1: extravasation of blood. To assess the
role of SUR1 in SCI, we studied the effect of glibenclamide, a sul-
fonylurea inhibitor that binds with subnanomolar or nanomo-
lar affinity (0.4–4.0 nM) to SUR1 (24). Immediately after injury,
animals were implanted with mini-osmotic pumps that delivered
either vehicle or glibenclamide (200 ng/h) s.c. We used constant
infusion of a low dose of drug to achieve sustained occupancy of
only high-affinity receptors.
Cords of vehicle-treated animals examined 24 hours after SCI
showed prominent bleeding at the surface and internally, with
internal bleeding consisting of a central region of hemorrhage
plus numerous distinct petechial hemorrhages at the periphery
(Figure 3A, arrows). By contrast, cords of glibenclamide-treated
animals showed visibly less hemorrhage, and it was largely con-
fined to the site of impact, with fewer petechial hemorrhages in
surrounding tissues (Figure 3A).
We quantified the amount of extravasated blood in tissue
homogenates at different times after SCI, after first removing
intravascular blood (Figure 3B). In cords from vehicle-treated ani-
mals, measurements showed a progressive increase in the amount
of blood, with a maximum reached about 12 hours after SCI (Fig-
ure 3B). By contrast, cords from glibenclamide-treated animals
showed little increase in extravasated blood during the 24 hours
after injury, with most of the blood present at 24 hours attribut-
able to the initial impact (Figure 3B).
Formation of petechial hemorrhages implies catastrophic fail-
ure of capillary integrity. We examined capillaries in the region of
injury by immunolabeling with vimentin, which is upregulated in
endothelium following injury (25). In control animals, vimentin-
positive capillaries appeared foreshortened or fragmented, where-
as in glibenclamide-treated animals, the capillaries were elongated
and appeared more normal (Figure 3C).
In postischemic reperfusion of CNS tissues, catastrophic failure
of capillary integrity has been attributed to the action of MMPs
(26). We assessed whether glibenclamide might have an effect on
MMP activity by using zymography to measure gelatinase activity
of recombinant MMP. Gelatinase activity was not affected by gliben-
clamide, although it was strongly inhibited by a specific MMP inhib-
itor (Figure 3D), indicating that the reduction in hemorrhage with
glibenclamide could not be attributed to MMP inhibition. Gliben-
clamide did not affect bleeding time (Figure 3E), suggesting that
the reduction in hemorrhage with glibenclamide following SCI was
unlikely to be the result of an effect on coagulation or platelet func-
tion (27). The dose of glibenclamide used caused a small decrease in
serum glucose levels, from 236 ± 15 mg/dl to 201 ± 20 mg/dl (n = 5
per group; P = 0.19), measured 3 hours after SCI.
Glibenclamide block of SUR1: lesion size. Labeling of longitudinal sec-
tions for the astrocyte marker glial fibrillary acidic protein and for
myelin revealed that glibenclamide treatment was associated with
smaller lesions, less reactive gliosis, and better myelin preservation
24 hours after SCI compared with controls (Figure 4, A and B). Sim-
ilarly, H&E staining of cross sections showed that glibenclamide
treatment was associated with smaller lesions at 7 days after SCI
compared with controls (Figure 4C). In vehicle-treated controls at
both 1 and 7 days after SCI, the lesions incorporated large voids of
necrotic tissue that involved most of the hemicord ipsilateral to the
impact site and typically extended to the contralateral hemicord.
White matter tracts of the contralateral hemicord were typically dis-
rupted. By contrast, lesions in glibenclamide-treated animals were
smaller and typically did not cross the midline, and contralateral
as well as portions of ipsilateral white matter tracts were spared.
Blocking SUR1 reduces lesion size and improves
neurobehavioral function after SCI. (A–C) Cord
sections immunolabeled for glial fibrillary acidic
protein (A) or stained with eriochrome cyanine
R (B) or H&E (C), 1 day (A and B) or 7 days (C)
after SCI, from vehicle-treated and glibenclamide-
treated rats. Images are representative of findings
in 3 rats per group. Scale bars: 1 mm. (D) Cas-
caded outlines of lesion areas in serial sections
250 μm apart, 7 days after SCI, as well as lesion
volumes from vehicle-treated and glibenclamide-
treated rats (n = 4–6 per group; excludes 2 con-
trol rats that died). (E) Performance on inclined
plane (head up and head down), ipsilateral paw
placement, and rearing in the same vehicle-
treated and glibenclamide-treated rats as in D.
Paw placement was measured 1 day after SCI.
Error bars indicate SEM. *P < 0.05, **P < 0.01,
***P < 0.001 versus control.
2110?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
Lesion volumes at 7 days were approximately 3-fold smaller in
glibenclamide-treated rats compared with controls (Figure 4D).
Notably, the lesion volumes we observed with glibenclamide treat-
ment following severe impact were comparable to those observed
by other investigators in untreated rats using the same cervical con-
tusion model following mild impact (10-g weight dropped from a
6.25-mm height; ref. 19).
Glibenclamide block of SUR1: neurobehavioral function. Vehicle-treated
rats were generally not mobile (18), whereas glibenclamide-treated
rats were typically ambulatory and often exhibited proficient
exploratory behavior. When suspended by their tails, vehicle-treat-
ed rats hung passively with little or no flexion of the trunk, whereas
glibenclamide-treated rats could typically flex their trunks, bring-
ing the snout to the level of the thorax or hindquarters.
We tested the same animals used to assess lesion size on an
inclined plane, a standard test that requires more and more dex-
terous function of the limbs and paws as the angle of the plane is
increased (28). At 1, 3, and 7 days after SCI, glibenclamide treat-
ment was associated with significantly better performance than
vehicle treatment (Figure 4E). We also quantified ipsilateral paw
placement, which is characteristically lost following cervical hemi-
cord transection (29). In the same animals tested 1 day after SCI,
glibenclamide treatment was associated with significantly better
performance than vehicle treatment (Figure 4E).
The Basso, Beattie, Bresnahan scale (BBB scale; ref. 30) is com-
monly used to evaluate neurobehavioral function in rodents after
SCI. However, it was designed for thoracic-level lesions, not cervi-
cal-level lesions, and the highest level of performance that it records
is less than what our glibenclamide-treated rats could achieve. We
therefore quantified vertical exploratory behavior (referred to
herein as rearing), a complex exercise that requires balance, trun-
cal stability, bilateral hind limb dexterity, and strength, and at
least unilateral fore limb dexterity and strength, which together
are excellent markers of cervical spinal cord function. Testing the
same rats as above at 1, 3, and 7 days after SCI showed that gliben-
clamide treatment was associated with significantly better perfor-
mance than vehicle treatment (Figure 4E). In additional groups of
rats tested only at 1 day after SCI, similar differences were observed
(3 ± 1 s versus 42 ± 7 s; P = 0.001; n = 14–15 per group).
Repaglinide block of SUR1. Repaglinide is a member of a distinct
class of insulin secretagogues that are structurally unrelated to
sulfonylureas and whose binding site may differ from that of a
sulfonylurea (31). Like glibenclamide, repaglinide produces high-
affinity block of both native and recombinant β cell KATP channels
(IC50, 0.9–7 nM) and shows higher potency in inhibiting pancre-
atic SUR1-regulated KATP channels than in inhibiting cardiovas-
cular SUR2-regulated channels (32). We examined the effect of
repaglinide on PHN, using the same treatment regimen as used
for glibenclamide. As with glibenclamide, repaglinide treatment
reduced blood in cord homogenates from 1.8 ± 0.2 μl to 1.2 ± 0.1 μl
at 1 day after SCI (P < 0.01; n = 5–8 per group) and was associated
with significantly better performance than vehicle-treated controls
on the inclined plane (head up, 40 ± 4° versus 62 ± 2°, P = 0.01;
head down, 29 ± 4° versus 47 ± 3°, P = 0.03; n = 3–8 per group) and
in rearing (3 ± 2 s versus 27 ± 6 s; P = 0.005; n = 5–6 per group).
Gene suppression of SUR1. We used gene suppression to confirm
involvement of SUR1 in PHN, choosing an antisense oligode-
oxynucleotide (ODN) strategy shown to be effective in vitro (33).
To validate the antisense strategy, we first implemented it in the
model that we previously used for the discovery of the NCCa-ATP
channel, in which a gelatin sponge is implanted into the parietal
lobe to stimulate formation of a gliotic capsule (14). Here, ani-
mals were also fitted with mini-osmotic pumps that delivered
ODNs, either antisense or scrambled, continuously for 7 days
Gene suppression of SUR1 blocks expression of functional NCCa-ATP
channels and improves outcome in SCI. (A) Western blots for SUR1
in gliotic capsule from rats with infusion of scrambled ODN (Scr-ODN)
or antisense ODN (AS-ODN) directly into the brain injury site for 10–12
days prior to tissue harvest. Also shown is densitometric analysis of
Western blots from the same groups of rats (n = 3 per group). (B) Mem-
brane potential of astrocytes from gliotic capsules of the same groups
of rats as in A during application of Na azide to deplete ATP. Mean
depolarization of 3 cells per group is shown. (C) Cord sections immu-
nolabeled for SUR1, 1 day after SCI, from rats treated with i.v. infusion
of scrambled ODN or antisense ODN. Scale bar: 0.5 mm. Also shown
is quantitative immunofluorescence for the same groups of rats (n = 3
per group). ROI, region of interest. (D) Blood in cord homogenates,
performance on angled plane, and rearing, 1 day after SCI, for rats
treated with i.v. infusion of scrambled ODN or antisense ODN. Error
bars indicate SEM. *P < 0.05, **P < 0.01 versus scrambled ODN.
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
into the injury site. Gliotic capsules from rats treated with anti-
sense ODN showed a significant reduction in SUR1 protein com-
pared with scrambled ODN–treated rats (Figure 5A). Patch clamp
of astrocytes from gliotic capsule of rats treated with scrambled
ODN showed that they rapidly depolarized when cellular ATP was
depleted by exposure to Na azide (Figure 5B), an effect that we
previously showed was caused by the opening of NCCa-ATP channels
(15). By contrast, astrocytes from rats treated with antisense ODN
depolarized slightly or not at all (Figure 5B), demonstrating that
SUR1 is required for expression of functional NCCa-ATP channels,
just as with KATP channels (34).
For experiments with SCI, we used antisense ODNs and scrambled
ODNs that were phosphorothioated at 4 distal bonds to protect
against endogenous nucleases (35); ODNs were administered i.v.
starting immediately after injury. At 6 hours after SCI, cords from
rats treated with antisense ODN showed significantly less immu-
nolabeling for SUR1 than did controls (Figure 5C). With scrambled
ODN, the necrotic void beneath the impact site was surrounded
by an SUR1-positive shell of tissue, similar to our observations in
untreated animals (Figure 1A). With antisense ODN, however, only
the small volume of tissue immediately beneath the impact site was
labeled for SUR1, and no necrotic void was evident (Figure 5C).
Antisense ODN did not affect normal expression of SUR1 in dorsal
horn cells (Figure 5C). At 1 day after SCI, treatment with antisense
ODN reduced blood in cord homogenates and was associated with
significantly better performance on the inclined plane and in rear-
ing compared with scrambled ODN treatment (Figure 5D).
Here, we report what we believe to be the novel findings that SUR1
is strongly upregulated following SCI and that block of SUR1 is
associated with significant improvements in all of the characteris-
tic manifestations of PHN, including hemorrhage, tissue necrosis,
lesion evolution, and neurological dysfunction. Although our focus
in this report was on SUR1 and NCCa-ATP channels in capillary endo-
thelium, our data also showed early (<6 h) upregulation of SUR1 in
large neuron-like cells in the core near the impact site, and in other
experiments, we observed late (12–24 h) upregulation of SUR1 in
reactive astrocytes (our unpublished observations). These responses
to SCI may be compared with findings we previously reported for
ischemic stroke, in which there is early upregulation of SUR1 in
neurons and capillaries in the core and later upregulation of SUR1
in capillaries and astrocytes in penumbral tissues (16).
PHN has been linked to tissue ischemia (5, 21), but to our knowl-
edge, it has not previously been characterized at a molecular level.
PHN is probably a variant of hemorrhagic conversion, a mecha-
nism of secondary injury in the CNS, wherein capillaries or post-
capillary venules undergo delayed catastrophic failure that allows
extravasation of blood to form petechial hemorrhages, which in
turn coalesce into a unified region of hemorrhagic necrosis or
infarction (36). Hemorrhagic conversion is common in traumatic
brain injury (37) and following postischemic reperfusion (26),
with hypoxia and active perfusion being important antecedents
(36). The molecular pathology involved in hemorrhagic conver-
sion has not been fully elucidated, but work in ischemic stroke has
implicated enzymatic destruction of capillaries by MMPs (26, 38).
MMPs have been implicated in SCI (39, 40), but not in PHN.
The work reported here implicates endothelial SUR1-regulated
NCCa-ATP channels in PHN. Our data show that PHN was associated
with upregulation of SUR1 in capillaries and postcapillary venules,
structures long held to be responsible for PHN (12, 13). Moreover,
our data show that block of SUR1 by 3 molecularly distinct agents,
glibenclamide, repaglinide, and antisense ODN, significantly
reduced PHN. The remarkably similar outcomes obtained with
highly selective agents that act via distinct molecular mechanisms
underscore the important role of SUR1. These data also provide
evidence that de novo expression of SUR1 is necessary and suffi-
cient for development of PHN. Use of a knockdown strategy using
antisense ODN appears to have been more informative than a gene-
knockout strategy, because the latter would not have distinguished
between constitutive and de novo expression of SUR1.
SUR1 forms the regulatory subunit of NCCa-ATP channels and
some KATP channels (15, 16). Here, we showed that upregulation of
SUR1 in endothelial cells was associated with expression of func-
tional NCCa-ATP channels, which we previously implicated in edema
formation and cell death in CNS ischemia/hypoxia (16, 36). Our
patch-clamp recordings confirmed the presence of a nonselective
cation channel that was activated by diazoxide and ATP depletion,
blocked by glibenclamide and cytoplasmic ATP, conducted Cs+,
and had a single channel conductance of about 35 pS, all of which
are characteristic of the NCCa-ATP channel (14, 15). We previously
showed that this channel conducts only monovalent cations, not
divalent cations (14), but this was not tested here. The experiments
reported here showing upregulation of functional NCCa-ATP chan-
nels were performed using endothelial cells from CNS as well as
non-CNS sources from 2 species suggesting a certain degree of
generality of the phenomenon. In our patch-clamp experiments,
we did not explicitly study endothelial cells from the spinal cord,
which could potentially differ from those in the brain. However, it
seems unlikely that the upregulation of SUR1 in spinal cord capil-
laries that we observed was associated with a different channel,
such as KATP. Sulfonylurea block of KATP would not be expected to
be neuroprotective (41), whereas block of NCCa-ATP is highly neu-
roprotective in both rodents and humans (16, 17).
Of the numerous treatments assessed in SCI, few have been
shown to actually decrease the hemorrhage and tissue loss associ-
ated with PHN. Methylprednisolone, the only approved therapy
for SCI, improves edema but does not alter the development of
PHN (42). A number of compounds have shown beneficial effects
related to tissue sparing, including the N-methyl-d-aspartic acid
antagonist MK801 (43), the α-amino-3-hydroxy-5-methyl-4-isox-
azole propionic acid antagonist GYKI 52466 (44), Na+ channel
blockers (45), and minocycline (46). Overall, to our knowledge no
treatment has been reported that reduces PHN and lesion volume
and improves neurobehavioral function to the extent observed
here with glibenclamide, repaglinide, and antisense ODN.
There are 2 mechanisms by which glibenclamide can antagonize
SUR1-regulated NCCa-ATP channels: by blocking the channel once
it is expressed and subsequently opened by ATP depletion (15),
and by interfering with trafficking of SUR1 to the cell membrane,
a process that is required for expression of functional channels
(47). Both blockade of open channels (16) and SUR1 binding (48)
needed to inhibit the trafficking of SUR1 to the cell membrane,
are increased an order of magnitude or more at the low pH of isch-
emic tissues. Block of open channels, interference with trafficking,
or both, coupled with the augmented efficacy imparted by low pH,
likely account for the high efficacy of glibenclamide found previ-
ously with stroke (16) and here with SCI.
Half of patients with SCI initially present with an incomplete
lesion (49), making it important to identify therapeutic strategies
2112?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
to inhibit secondary injury mechanisms. Glibenclamide has been
used safely in humans for several decades for treatment of type 2
diabetes, with no untoward side effects except hypoglycemia, and
its continued use immediately after stroke improves outcome in
patients with type 2 diabetes (17). The safety of glibenclamide,
together with its unique mechanism of action in targeting the
capillary failure that leads to PHN, indicate that this drug may be
especially attractive for translational use in human SCI.
SCI injury model. This study was approved by and performed in accordance
with the guidelines of the Institutional Animal Care and Use Commit-
tee of the University of Maryland at Baltimore. Adult female Long-Evans
rats (275–350 g) were anesthetized (60 mg/kg ketamine plus 7.5 mg/kg
xylazine i.p.). The dura at C4-5 was exposed via a left hemilaminectomy.
A hemicervical spinal cord contusion was created using a blunt-force
impactor (1.3-mm impactor head driven by a 10-g weight dropped verti-
cally from a 25-mm height) (18, 19). After SCI, animals were given 10 ml
of glucose-free normal saline s.c. Rectal temperature was maintained at
approximately 37°C using a servo-controlled warming blanket. Blood
gases and serum glucose were determined 10–15 minutes after SCI in
control (pO2, 95 ± 6 mmHg; pCO2, 46 ± 3 mmHg; pH, 7.33 ± 0.01; glucose,
258 ± 17 mg/dl) and glibenclamide-treated animals (pO2, 96 ± 7 mmHg;
pCO2, 45 ± 2 mmHg; pH, 7.37 ± 0.01; glucose, 242 ± 14 mg/dl).
Drug delivery. Within 2–3 minutes of SCI induction, mini-osmotic
pumps (Alzet 2002, 0.5 μl/h; Durect Corp.) were implanted that deliv-
ered either vehicle (saline plus DMSO), glibenclamide (Sigma-Aldrich) in
vehicle, or repaglinide (Sigma-Aldrich) in vehicle s.c. During the course of
the study, slightly different formulations of drug were used, with the best
results obtained using stock solutions made by placing 50 mg (or 25 mg)
of drug into 10 ml DMSO and infusion solutions made by placing 400 μl
(or 800 μl) stock into 4.6 ml (or 4.2 ml) unbuffered saline (0.9% NaCl) and
adjusting the pH to approximately 8.5 using 0.1 N NaOH. Infusion solu-
tions of glibenclamide and repaglinide were delivered at 0.5 μl/h, yielding
infusion doses of 200 ng/h.
For in vivo gene suppression of Abcc8, we used ODNs that were phos-
phorothioated at 4 distal bonds to protect against endogenous nucle-
ases (35). Within a few minutes of SCI, mini-osmotic pumps (Alzet 2002,
0.5 μl/h; Durect Corp.) with jugular vein catheters were implanted that
delivered either scrambled ODN (5′-TGCCTGAGGCGTGGCTGT-3′) or
antisense ODN (5′-GGCCGAGTGGTTCTCGGT-3′) (33) in PBS at a rate
of 1 mg/rat/24 h.
Tissue blood. Rats were sacrificed at various times after SCI (n = 5–11 per
group) and perfused with heparinized saline to remove intravascular blood,
and 5-mm segments of cord encompassing the lesion were homogenized
and processed as described previously (50).
Lesion size. At 7 days after SCI, cords were paraffin sectioned and stained
with H&E. Lesion volumes were calculated from lesion areas measured on
serial sections every 250 μm.
Neurobehavioral assessment. All measurements were performed by evalu-
ators blinded to group. Performance on the inclined plane was evaluated
as described previously (28). To assess paw placement and rearing (29),
animals were videotaped while in a translucent cylinder (19 × 20 cm). Rear-
ing was quantified as the number of seconds spent with both front paws
elevated above shoulder height during a 3-minute period of observation.
Bleeding times. Bleeding times were measured using tail-tip bleeding as
described previously (51).
Zymography. Zymography of recombinant MMP-2 and MMP-9 (Sigma-
Aldrich) was performed as described previously (52). MMP inhibitor II was
Cell culture. Endothelial cell cultures from human brain microvessels and
human aorta (ScienCell Research Laboratories) as well as murine brain
microvessels (bEnd.3; ATCC) were grown at low density using media and
supplements recommended by the suppliers.
Abcc8 knockdown in astrocytes. Abcc8 knockdown in astrocytes was per-
formed in triplicate by implanting rats with gelatin sponges in the parietal
lobe to induce formation of a gliotic capsule containing reactive astro-
cytes that express the SUR1-regulated NCCa-ATP channel (14, 15). At the
same time, they were implanted with mini-osmotic pumps (Alzet 2002,
14-day pump; Durect Corp.) placed in the dorsal thoracolumbar region
that contained ODN (711 μg/ml delivered at a rate of 0.5 μl/h, yielding
1,500 pmol/d), with the delivery catheter placed directly into the site of the
gelatin sponge implant in the brain. Animals were infused with scrambled
ODN or antisense ODN as described above but not phosphorothioated.
After 10–14 days, the gelatin sponge plus encapsulating gliotic tissues were
harvested and processed either for Western immunoblots or to obtain fresh
reactive astrocytes for patch-clamp electrophysiology.
Patch-clamp electrophysiology. Patch-clamp electrophysiology for the NCCa-ATP
channel in this lab has been described previously (14, 15).
Immunohistochemistry. Cryosections were immunolabeled (15, 16) using
primary antibodies directed against SUR1 (C-16; diluted 1:200; 1 hour at
room temperature, 48 hours at 4°C; Santa Cruz Biotechnology Inc.), SUR2
(H-80; diluted 1:200; 1 hour at room temperature, 48 hours at 4°C; Santa
Cruz Biotechnology Inc.), glial fibrillary acidic protein (C-9205; diluted
1:500; Sigma-Aldrich), and vimentin (monoclonal CY3 conjugated; diluted
1:100; Sigma-Aldrich). Quantitative immunofluorescence was performed
as described previously (53).
Eriochrome staining. Visualization of myelin was obtained using erio-
chrome cyanine R (Sigma-Aldrich).
Immunoblots. Immunoblots were prepared using antibodies directed
against SUR1. The specificity of the antibody (16) was demonstrated by
the results of the knockdown experiments shown in Figure 5.
In situ hybridization. Fresh-frozen cord sections were fixed in 5% form-
aldehyde for 5 minutes. Digoxigenin-labeled probes (sense, 5′-GCCC-
3′) were designed and supplied by GeneDetect, and hybridization was
performed according to the manufacturer’s protocol.
Statistics. Statistical significance was evaluated using ANOVA. A P
value of less than 0.05 was considered significant.
This work was supported by grants to J.M. Simard from the Mary-
land Department of Veterans Affairs; the National Institute of
Neurological Disorders and Stroke, NIH (NS048260), the Nation-
al Heart, Lung, and Blood Institute, NIH (HL082517); and the
Christopher and Dana Reeve Foundation.
Received for publication March 6, 2007, and accepted in revised
form May 9, 2007.
Address correspondence to: J. Marc Simard, Department of Neu-
rosurgery, 22 S. Greene Street, Suite 12SD, Baltimore, Maryland
21201-1595, USA. Phone: (410) 328-0850; Fax: (410) 328-0756;
1. Hayes, K.C., and Kakulas, B.A. 1997. Neuropatholo-
gy of human spinal cord injury sustained in sports-
related activities. J. Neurotrauma. 14:235–248.
2. Tator, C.H., and Fehlings, M.G. 1991. Review of
the secondary injury theory of acute spinal cord
trauma with emphasis on vascular mechanisms.
J. Neurosurg. 75:15–26.
3. Kwon, B.K., Tetzlaff, W., Grauer, J.N., Beiner, J., and
Vaccaro, A.R. 2004. Pathophysiology and pharma-
?The?Journal?of?Clinical?Investigation http://www.jci.org Volume 117 Number 8 August 2007
cologic treatment of acute spinal cord injury. Spine J.
4. Bilgen, M., Abbe, R., Liu, S.J., and Narayana, P.A.
2000. Spatial and temporal evolution of hemor-
rhage in the hyperacute phase of experimental spi-
nal cord injury: in vivo magnetic resonance imaging.
Magn. Reson. Med. 43:594–600.
5. Nelson, E., Gertz, S.D., Rennels, M.L., Ducker,
T.B., and Blaumanis, O.R. 1977. Spinal cord inju-
ry. The role of vascular damage in the pathogen-
esis of central hemorrhagic necrosis. Arch. Neurol.
6. Tator, C.H. 1991. Review of experimental spinal
cord injury with emphasis on the local and system-
ic circulatory effects. Neurochirurgie. 37:291–302.
7. Fitch, M.T., Doller, C., Combs, C.K., Landreth,
G.E., and Silver, J. 1999. Cellular and molecular
mechanisms of glial scarring and progressive cavi-
tation: in vivo and in vitro analysis of inflamma-
tion-induced secondary injury after CNS trauma.
J. Neurosci. 19:8182–8198.
8. Tator, C.H., and Koyanagi, I. 1997. Vascular mech-
anisms in the pathophysiology of human spinal
cord injury. J. Neurosurg. 86:483–492.
9. Kraus, K.H. 1996. The pathophysiology of spinal
cord injury and its clinical implications. Semin. Vet.
Med. Surg. Small Anim. 11:201–207.
10. Balentine, J.D. 1978. Pathology of experimental
spinal cord trauma. I. The necrotic lesion as a func-
tion of vascular injury. Lab. Invest. 39:236–253.
11. Kawata, K., et al. 1993. Experimental study of acute
spinal cord injury: a histopathological study [In
Japanese]. No Shinkei Geka. 21:45–51.
12. Griffiths, I.R., Burns, N., and Crawford, A.R. 1978.
Early vascular changes in the spinal grey matter fol-
lowing impact injury. Acta Neuropathol. 41:33–39.
13. Kapadia, S.E. 1984. Ultrastructural alterations in
blood vessels of the white matter after experimen-
tal spinal cord trauma. J. Neurosurg. 61:539–544.
14. Chen, M., and Simard, J.M. 2001. Cell swelling and
a nonselective cation channel regulated by internal
Ca2+ and ATP in native reactive astrocytes from
adult rat brain. J. Neurosci. 21:6512–6521.
15. Chen, M., Dong, Y., and Simard, J.M. 2003. Func-
tional coupling between sulfonylurea receptor
type 1 and a nonselective cation channel in reac-
tive astrocytes from adult rat brain. J. Neurosci.
16. Simard, J.M., et al. 2006. Newly expressed SUR1-
regulated NC(Ca-ATP) channel mediates cerebral
edema after ischemic stroke. Nat. Med. 12:433–440.
17. Kunte, H., et al. 2007. Sulfonylureas improve out-
come in patients with type 2 diabetes and acute
ischemic stroke. Stroke. In press.
18. Soblosky, J.S., Song, J.H., and Dinh, D.H. 2001.
Graded unilateral cervical spinal cord injury in the
rat: evaluation of forelimb recovery and histologi-
cal effects. Behav. Brain Res. 119:1–13.
19. Gensel, J.C., et al. 2006. Behavioral and histological
characterization of unilateral cervical spinal cord
contusion injury in rats. J. Neurotrauma. 23:36–54.
20. Yamashita, S., et al. 1994. Possible presence of
the ATP-sensitive K+ channel in isolated spinal
dorsal horn neurons of the rat. Neurosci. Lett.
21. Tator, C.H. 1995. Update on the pathophysiology
and pathology of acute spinal cord injury. Brain
22. Jansen-Olesen, I., Mortensen, C.H., El-Bariaki, N., and
Ploug, K.B. 2005. Characterization of K(ATP)-chan-
nels in rat basilar and middle cerebral arteries: stud-
ies of vasomotor responses and mRNA expression.
Eur. J. Pharmacol. 523:109–118.
23. Seino, S. 1999. ATP-sensitive potassium chan-
nels: a model of heteromultimeric potassium
channel/receptor assemblies. Annu. Rev. Physiol.
24. Aguilar-Bryan, L., Nelson, D.A., Vu, Q.A., Hum-
phrey, M.B., and Boyd, A.E., 3rd. 1990. Photoaf-
finity labeling and partial purification of the
beta cell sulfonylurea receptor using a novel,
biologically active glyburide analog. J. Biol. Chem.
25. Haseloff, R.F., et al. 2006. Differential protein
expression in brain capillary endothelial cells
induced by hypoxia and posthypoxic reoxygen-
ation. Proteomics. 6:1803–1809.
26. Wang, X., et al. 2004. Mechanisms of hemorrhagic
transformation after tissue plasminogen activator
reperfusion therapy for ischemic stroke. Stroke.
27. Chan, T.K., Chan, V., Teng, C.S., and Yeung, R.T.
1982. Effects of gliclazide and glibenclamide on
platelet function, fibrinolysis and metabolic con-
trol in diabetic patients with retinopathy. Sem. Hop.
28. Rivlin, A.S., and Tator, C.H. 1977. Objective clini-
cal assessment of motor function after experi-
mental spinal cord injury in the rat. J. Neurosurg.
29. Nikulina, E., Tidwell, J.L., Dai, H.N., Bregman, B.S.,
and Filbin, M.T. 2004. The phosphodiesterase inhib-
itor rolipram delivered after a spinal cord lesion pro-
motes axonal regeneration and functional recovery.
Proc. Natl. Acad. Sci. U. S. A. 101:8786–8790.
30. Basso, D.M., Beattie, M.S., and Bresnahan, J.C.
1995. A sensitive and reliable locomotor rating
scale for open field testing in rats. J. Neurotrauma.
31. Hansen, A.M., et al. 2002. Differential interactions of
nateglinide and repaglinide on the human beta-cell
sulphonylurea receptor 1. Diabetes. 51:2789–2795.
32. Stephan, D., Winkler, M., Kuhner, P., Russ, U., and
Quast, U. 2006. Selectivity of repaglinide and glib-
enclamide for the pancreatic over the cardiovascu-
lar K(ATP) channels. Diabetologia. 49:2039–2048.
33. Yokoshiki, H., Sunagawa, M., Seki, T., and Sperela-
kis, N. 1999. Antisense oligodeoxynucleotides of
sulfonylurea receptors inhibit ATP-sensitive K+
channels in cultured neonatal rat ventricular cells.
Pflugers Arch. 437:400–408.
34. Sharma, N., et al. 1999. The C terminus of SUR1 is
required for trafficking of KATP channels. J. Biol.
35. Galderisi, U., Cascino, A., and Giordano, A. 1999.
Antisense oligonucleotides as therapeutic agents.
J. Cell. Physiol. 181:251–257.
36. Simard, J.M., Kent, T.A., Chen, M., Tarasov, K.V.,
and Gerzanich, V. 2007. Brain oedema in focal isch-
aemia: molecular pathophysiology and theoretical
implications. Lancet Neurol. 6:258–268.
37. Oertel, M., et al. 2002. Progressive hemorrhage
after head trauma: predictors and consequences of
the evolving injury. J. Neurosurg. 96:109–116.
38. Gidday, J.M., et al. 2005. Leukocyte-derived matrix
metalloproteinase-9 mediates blood-brain barrier
breakdown and is proinflammatory after transient
focal cerebral ischemia. Am. J. Physiol. Heart Circ.
39. Noble, L.J., Donovan, F., Igarashi, T., Goussev,
S., and Werb, Z. 2002. Matrix metalloproteinases
limit functional recovery after spinal cord injury
by modulation of early vascular events. J. Neurosci.
40. Pannu, R., Christie, D.K., Barbosa, E., Singh, I.,
and Singh, A.K. 2007. Post-trauma Lipitor treat-
ment prevents endothelial dysfunction, facili-
tates neuroprotection, and promotes locomotor
recovery following spinal cord injury. J. Neurochem.
41. Sun, H.S., Feng, Z.P., Barber, P.A., Buchan, A.M.,
and French, R.J. 2007. Kir6.2-containing ATP-sen-
sitive potassium channels protect cortical neurons
from ischemic/anoxic injury in vitro and in vivo.
42. Merola, A., et al. 2002. Histologic characteriza-
tion of acute spinal cord injury treated with intra-
venous methylprednisolone. J. Orthop. Trauma.
43. Faden, A.I., Lemke, M., Simon, R.P., and Noble, L.J.
1988. N-methyl-D-aspartate antagonist MK801
improves outcome following traumatic spinal
cord injury in rats: behavioral, anatomic, and neu-
rochemical studies. J. Neurotrauma. 5:33–45.
44. Colak, A., et al. 2003. Neuroprotective effects of
GYKI 52466 on experimental spinal cord injury in
rats. J. Neurosurg. 98:275–281.
45. Schwartz, G., and Fehlings, M.G. 2001. Evaluation
of the neuroprotective effects of sodium channel
blockers after spinal cord injury: improved behav-
ioral and neuroanatomical recovery with riluzole.
J. Neurosurg. 94:245–256.
46. Teng, Y.D., et al. 2004. Minocycline inhibits contu-
sion-triggered mitochondrial cytochrome c release
and mitigates functional deficits after spinal cord
injury. Proc. Natl. Acad. Sci. U. S. A. 101:3071–3076.
47. Partridge, C.J., Beech, D.J., and Sivaprasadarao,
A. 2001. Identification and pharmacological cor-
rection of a membrane trafficking defect associ-
ated with a mutation in the sulfonylurea receptor
causing familial hyperinsulinism. J. Biol. Chem.
48. Nelson, D.A., Bryan, J., Wechsler, S., Clement, J.P.,
4th, and Aguilar-Bryan, L. 1996. The high-affinity
sulfonylurea receptor: distribution, glycosylation,
purification, and immunoprecipitation of two
forms from endocrine and neuroendocrine cell
lines. Biochemistry. 35:14793–14799.
49. Bracken, M.B., et al. 1990. A randomized, con-
trolled trial of methylprednisolone or naloxone in
the treatment of acute spinal-cord injury. Results
of the Second National Acute Spinal Cord Injury
Study. N. Engl. J. Med. 322:1405–1411.
50. Choudhri, T.F., Hoh, B.L., Solomon, R.A., Con-
nolly, E.S., Jr., and Pinsky, D.J. 1997. Use of a spec-
trophotometric hemoglobin assay to objectively
quantify intracerebral hemorrhage in mice. Stroke.
51. Lorrain, J., et al. 2003. Antithrombotic properties of
SSR182289A, a new, orally active thrombin inhibitor.
J. Pharmacol. Exp. Ther. 304:567–574.
52. Sumii, T., and Lo, E.H. 2002. Involvement of matrix
metalloproteinase in thrombolysis-associated
hemorrhagic transformation after embolic focal
ischemia in rats. Stroke. 33:831–836.
53. Gerzanich, V., et al. 2003. Alternative splicing of
cGMP-dependent protein kinase I in angioten-
sin-hypertension: novel mechanism for nitrate
tolerance in vascular smooth muscle. Circ. Res.