вход по аккаунту


Cyclooxygenase-2 is induced globally in infarcted human brain.

код для вставкиСкачать
Cyclooxygenase-2 Is Induced Globally in
Infarcted Human Brain
Tiina Sairanen, MD,*t Ari Ristimaki, MD, PhD,f$ Marja-Liisa Karjalainen-Lindsberg, MD, PhD,?
Anders Paetau, MD, PhD,t Markku Kaste, MD, PhD,* and Perttu J. Lindsberg, MD, PhD"7
Cyclooxygenase (COX) catalyzes synthesis of prostanoids after liberation of arachidonic acid, an important biochemical
sequela of cerebral ischemia that aggravates brain injury. We investigated expression of inducible COX-2 in infarcted
human brains (symptom duration, 15 hours to 18 days) and found that COX-2 protein was present in both neuronal and
glial cells throughout the brain in accord with infarct topography and duration. These results emphasize the global yet
temporally regulated nature of COX-2 induction during focal ischemia in humans, clearly different from the circumscribed acute expression reported in experimental animal models. We speculate that early induction of COX-2 may fuel
tissue damage through prostanoids and free radicals, and delayed induction in remote brain areas may promote reconstitutive processes in the face of tissue scarring and remodeling of the surviving neural networks.
Sairanen T, Ristimaki A, Karjalainen-Lindsberg M-L, Paetau A, Kaste M, Lindsberg PJ. Cyclooxygenase-2 is
induced globally in infarcted human brain. Ann Neurol 1998;43:738-747
The rate-limiting enzyme in the conversion of arachidonic acid to prostanoids is cyclooxygenase (COX),
also referred to as prostaglandin H (PGH) synthase.
Two isoforms of COX are known, COX-1 as the constitutive and COX-2 as the inducible form of the enzyme.Iz2 Prostaglandins (PGs) such as PGD,, PGE,,
and PGF,, have been identified as major prostanoids
produced by brain tissue of several mammalian species
including humans:' Cloning of COX-2 and its localization in rat brain neurons launched hrther studies on
this enzyme that suggested it to be the main isoform in
the central nervous system (CNS).4,5It is interesting
that Yamagata and colleagues4 also showed that
COX-2 expression was regulated by N-methybaspartate-dependent synaptic activity, thus promoting a role
for prostanoid signaling in activity-dependent neural
plasticity. Furthermore, an immediate-early gene (IEG)
function was suggested for COX-2, based on inducibility by natural stress stimulus,* a character further confirmed by parallel induction with a known ieg, c-fos,
after excitotoxin injection.6
Production of vasoactive prostanoids after liberation
of arachidonic acid from membrane phospholipids is a
well-known biochemical sequela of cerebral ischemia,
edema, and other types of CNS
also eventually fuels generation of reactive oxygen species.' Because attenuation of postischemic brain hypo-
From the Departments of *Clinical Neurosciences, ?Pathology,
$Bacteriology and Immunology, and SObstetrics and Gynecology,
University of Helsinki, and SNeuroscience Program, Biomedicum
Helsinki, Helsinki, Finland.
perfusion, reduction in blood-brain barrier permeability,
and reduced infarct volume have been reported after
COX inhibition in animal model^,^^,'^ we studied
whether the COX-2 gene is induced in human infarcted
brain. T o characterize the COX-2 response, it was compared with a class of gene products known to be induced
in rodents during cellular stress in cerebral ischemia, the
heat shock proteins 72/73 (HSP 72/73).I4,l5This article
describes the spatial and temporal evolution of global
COX-2 response to ischemic stroke in humans during
infarction maturation in consecutive postischemic
phases. Patient characteristics are given in the Table.
Materials and Methods
Postmortem Specimens
Autopsy specimens from 16 cases of fatal ischemic stroke,
treated at the Department of Neurology, were studied. Three
patients who died of a nonneurological cause were used as
controls. The mean autopsy delay for cases was 17 hours
(range, 3.5-45 hours) and for controls 16.5 hours (range,
14-2 1 hours). Additional control specimens were collected
from patients (n = 3) going through epilepsy surgery,
whereby postmortem delay was avoided. The study was approved by the ethics committee at the Helsinki University
Central Hospital.
On autopsy, the infarcted brain areas were identified during macroscopic examination, in comparison with the most
Address correspondence to Dr Sairanen, Department of Clinical
Neurosciences, University of Helsinki, Haartmaninkatu 4, FIN00290 Helsinki. Finland.
Received Jun 25, 1997, and in revised form Nov 11 and Dec 17.
Accepted for publication Dec 19, 1997.
Copyright 0 1998 by the American Neurological Association
Table. Cbaracteristics of Deceased Stroke Patients Studied Post Mortem
Age (yr)
Risk Factors
Cause of Death
AF, AS, DM, H , and H F
AF, CAD, DM, and H
CAD, HC, and IE
AF and AS
AS and CAD
CAD, H , and H F
AS (EA) and CAD
VF and stroke
PE (AM1 and VF) and stroke
Stroke and PE
Stroke and PE
PE and stroke
Stroke and PE
Duodenal ulcer
CAD, H, and HF
AF, AS, CAD, DM, and H
AF, CAD, H, and H F
DM, H , and H F
AS, CAD, HF, and IE
15 hr
23 hr
1 day 4 hr
1 day 14 hr
1 day 19 hr
2 days 12 hr
2 days 12 hr
-3 days
3 days
3 days 6 hr
4 days 12 hr
5 days 9 hr
6 days 7 hr
8 days 12 hr
17 days
18 days
”Cases ,4, B, and C are control patients who died suddenly without a neurological cause.
bThe diagnosis was based only on autopsy findings.
AF = atrial fibrillation; AS = generalized arteriosclerosis; CAD = coronary artery disease; DM = diabetes mellitus; FA = carotid endarterectomy; H = hypertension; HC = hypercholesterolemia; HF = heart failure; IE = chronic ischemic encephalopathy; VF = ventricular
fibrillation; CA = cardiac arrest; AM1 = acute myocardial infarction; PE = pulmonary embolism; ICA = internal carotid artery; MCA =
middle cerebral artery; BA = basilar artery; T = thrombosis; TE = thromboembolism.
recent computed tomographic scan. Approximately 1.an3
cortical samples, including subcortical white matter, were
dissected and frozen in liquid nitrogen. Samples from the
corresponding areas of the contralateral or noninfarcted
hemispheres and from the control brains were processed in a
similar manner. Fresh-frozen samples were collected from patients undergoing epilepsy surgery, where normal brain tissue
(including cortical and subcortical tissue) from frontal or
temporal lobes encompassing the epileptogenic focus was
Although tissue sampling was macroscopically targeted to
catch tissue from the infarct core, the periinfarct region, and
the homologous brain areas of the contralateral hemisphere,
microscopic determination of signs of neuronal damage was
the basis for systematic immunohistochemical analysis. Estimation of the severity of neuronal ischemic changes was performed (by A.P.) on hematoxylin-eosin staining as described
previously, 16,’’ and grading from 1 (largely normal morphology) to 4 (irreversible changes) was used as reference when
presenting results from peroxidase-stained sections. Interrelated topographic areas of the ipsilateral infarct core (score
4),periinfarct area (scores 3-2), and contralateral or control
areas (scores 1--0; 0 = no ischemic changes) were used in
the presentation of results from quantitative analysis of
COX-2 immunohistochemistry.
On acetone-fixed fresh-frozen sections, immunohistochemical staining was performed with a polyclonal anti-human
PGH synthase 2 antibody (Cayman Chemical, Ann Arbor,
MI) for detection of COX-2 ( 1 : l O O dilution for 1 hour) or
with normal rabbit serum as control, using the three-step
avidin-biotin complexlhorseradish peroxidase method (ABC,
Vectastain, Vector Laboratories, Burlingame, CA) with the
chromogen aminoethylcarbazole and counterstaining with
hematoxylin. Omission of the primary antibody resulted in
abolition of the staining (data not shown). Peroxidase staining of adjacent sections with a monoclonal antibody to HSP
72173 (Boehringer Mannheim Biochemica, Germany), in
1 :100 dilution, was performed similarly.
Double-fluorescent labeling with cellular markers for neuronal filament NF-200 (mouse monoclonal anti-neurofilament of 200 kd; Boehringer Mannheim GmbH, Germany), in 1:50 dilution, and astrocytes (mouse monoclonal
anti-glial fibrillary acidic protein, anti-GFAP; Dako AIS,
Glostrup, Denmark), in 1:25 dilution, was combined by primary incubation with COX-2 antibody (1:50 dilution) for 1
hour, which was succeeded by biotinylared anti-rabbit antibody and subsequent detection with Neutralite avididTexas
Red conjugate (Molecular Probes Europe BV, Leiden, The
Netherlands) in 1:200 dilution for 30 minutes. The second
primary antibody was visualized with fluorescein isothiocyanate-conjugated anti-mouse antibody (Dako AJS) in 1:20 dilution.
The specificity of the COX-2 antibody was tested by western blotting. Fibroblasts were incubated either with or without interleukin-1 (10 nglml) for 6 hours before lysis in fivefold volume of modified radioimmunoprecipitation assay
buffer containing 150 nM NaCl, 1% Nonidet P-40, 0.1%
sodium dodecyl sulfate (SDS), 50 nM Tris (pH 8.0), 0.5
mM phenylmethylsulfonyl fluoride, 0.5% Triton X-100, and
1 m M EDTA. The lysates were boiled for 5 minutes and
Sairanen et al: COX-2 in Infarcted Human Brain
centrifuged at 2,000 g for 30 minutes at +4"C. Supernatants
were diluted with reducing Laemmli buffer for separation on
an 8% SDS-polyacrylamide gel and subsequent transfer to
nitrocellulose filter. Enhanced chemiluminescence detection
was performed after overnight incubation with COX-2 antibody in 1:1,000 dilution (f4"C) followed by anti-rabbit
IgG peroxidase-conjugated secondary antibody (Boehringer
Mannheim). The blotting experiment was duplicated. Densitometric study of the autoradiograms by N I H Image Analysis revealed a 50-fold increase in COX-2 signal between the
unstimulated and stimulated fibroblasts, whereas control incubation of a parallel filter with normal rabbit serum remained negative (data not shown).
Microscopic Quantz$cation of Neuronal
COX-2 Immunoreactivity
COX-2-immunoreactive cell processes identical to NF-200 positive processes were counted, in five random consecutive
fields of 0.152 mm2 from each slide, by an investigator
(T.S.) blinded to the region of brain section. The brain sections analyzed included control cases and the infarction core,
the periinfarct region, and the contralateral or noninfarcted
hemisphere from each stroke case. The sections were submitted for a systematic immunohistochemical analysis based on
the histological grading of neuronal damage. The exact number (three to five) of tissue sections available for this analysis
was determined by individual infarct topography and distribution of obtained tissue samples in each case, according to
their assignment to different grades of neuronal damage. A
few (eight sections) technically substandard immunohistochemical stainings were excluded. Data are presented as
mean t SEM values (Fig 1).
Northern Blotting
Cytoplasmic total RNA was isolated from brain tissues by
the guanidine isothiocyanate-cesium chloride method.I8
Three to 17 p g of RNA was denatured in 1 M glyoxal, 50%
dimethyl sulfoxide, and 10 m M phosphate buffer at 50°C
for GO minutes and then electrophoresed through 1.2% agarose gel before transferal onto Hybond-N nylon membranes
(Amersham International, Aylesbury, UK), which were then
baked for 1 hour at 80°C and cross-linked by ultraviolet
light for 6 minutes. Human COX-2 and glyceraldehyde-3phosphate dehydrogenase cDNAs" were labeled by using
[ C X - ~ ~ P ] ~ C(Du
T P Pont-New England Nuclear, Boston,
MA) and the Prime-a-Gene kit (Promega, Madison, WI).
Probes were purified with nick columns (Pharmacia) and
used at 1 X lo6 cpm/ml. After hybridization at 42°C for 16
hours in solution containing 50% formamide, 6X salinesodium citrate, 0.1 Yo Ficoll, 0.1% polyvinylpyrrolidone,
0.1% bovine serum albumin, 100 pg/ml herring sperm
DNA, 100 pg yeast FWA, and 0.5% SDS, membranes were
washed three times (50°C) for 15 minutes with 0.1 X salinesodium citrate and 0.1 X SDS. Northern blots were visualized by autoradiography and quantitated with N I H Image
Analysis software.
Statistical Methods
The mean values of COX-2-immunoreactive neuronal processes in the different topographic areas (periinfarct area, in-
740 Annals of Neurology
Vol 43
No 6
June 1998
Infarct Core
Peri-infarct Area
and Noninfarcted
1000 800
600 7
Controls @=a)
0.6-1 d (n=2)
1.2-6.3 d (n=11)
8.5-18 d In=3)
Survival Time (d)
Fig 1. Cyclooxygenase-2 (COX-2)-immunoreactive neuronal
processes in control brains, in the ipsilateral and in the contralateral and noninfarcted (basilar artevy occlusions) hemispheres of stroke cases. Microscopic quantification of neuronal
C O X 2 was performed by counting COX-2-immunoreactive
cell processes identical to NF-200-positive processesfiom J;ve
jelds of 0.152 mm2 fiom each slide. Three to jive slides per
case were counted. An interrelated topographic grouping to
periinfarct and infarct core areas was based on the scoring of
the relative severity of neuronal ischemic damuge. (A) Mean 2
SEM values of COX-2-immunoreactive neuronal processes in
control brains (n = 6)and in dtfferent brain regions of infarcted brains (n = 1@, including the infarction core, the periinfarct region, and the contralateral and noninfarcted hemisphere. Statistical comparison of the COX-2-positive neuronal
processes with Kruskal-Wallis one-way analysis of variance on
ranks suggested high& signifkant dzfferences in the means of the
mean values (p = 0.008). Post hoc Dunni test suggested sign;Jcant elevation of C O X 2 expression in each locution of the
infarcted brain but no signijcant dzfferences among them.
*p < 0.01; p < 0.05, compared with controlr (KmskalWallis plus Dunn j multiple comparison test). (B) Mean L
SEM values of COX-2-immunoreactive neuronal processes in
the dfferent brain regions during subsequent maturation phases
of the infarction, including acute ( 0 . 6 1 days; n = 2), subacute (1.2-63 hys; n = 11), and chronic (8.5-18 Ays; n =
3) infarctions. A comparison between the controllc and subacute
(1.2- 63 Ays) infarctionsyielhd statistically signgiant differ=
ences in the mean values of all brain regions (p < 0.05).
periinfarct urea; 0 = infarct core area; 0 = contralateral und
noninfarcted hemisphere.
farction core, and contralateral and noninfarcted hemisphere)
within an infarct group (acute, subacute, or chronic) and the
differences between the mean values and controls were analyzed by the Kruskal-Wallis test followed by Dunn's multiple comparison test. p < 0.05 was considered statistically
significant. Data are presented as mean ? standard error
(SEM) values.
IrnrnunohistochernicalStaining of COX-2
Neuronal processes were chosen for quantification because COX-2
immunoreactivity was recently reported in neuronal
dendrites and dendritic spines.20A topographic grouping of periinfarct and infarct core areas was achieved
based on the scoring of the relative severity of neuronal
damage as described in Materials and Methods.
One of 3 control brains showed COX-2-positive
neuronal perikarya and processes (see Fig l), which
may have been provoked by a transient hemodynamic
collapse in the agonal phase in this individual who died
of acute myocardial infarction (see Table). This may
also naturally reflect interindividual differences. Moreover, in the normal brain tissue obtained during epilepsy surgery, no COX-2-immunoreactive neuronal
processes were detected in any of the 3 cases. In the
infarcted brains (n = 16), the number of COX-2immunoreactive neuronal processes in the different topographic brain regions, including the infarction core
(522 f- 195, p < 0.05), the periinfarct area (462 t
94, p < O.O1), and the contralateral and the noninfarcted hemisphere (370 ? 83, p < 0.05), was significantly higher, compared with controls (39 2 39) (see
Fig 1A). During the acute phase (0.6-1 days), COX-2
immunoreactivity was concentrated in the neuronal
perikarya and processes in the core and peripheral infarct areas but was also seen already in the contralateral
and noninfarcted (basilar artery occlusions) hemispheres (Fig 2A-C; see Fig 1B). The number of COX2-immunoreactive neuronal processes tended to be
higher in the periinfarct region (475 2 99/mm2), compared with control brains (39 -t 39/mm2), but due to
the small number of cases is not amenable to statistical
analysis. In the subacute phase (1.2-6.3 days), the
number of COX-2-positive neuronal processes was increased in the periinfarct region in comparison with
controls (529 2 125 vs 39 -t 39, p < 0.05), and increased numbers were seen in the infarct core as well
(709 f- 303, p < 0.05; considerable heterogeneity in
protein expression between infarct core areas of different cases was evident and likely underrates the statistical significance). In the contralateral and noninfarcted
hemispheres, a significant rise above the control level
was also evident (378 +- 110, p < 0.05). Along with
infarct maturation (chronic phase, 8.5-1 8 days), the
number of immunoreactive neuronal processes de-
creased in the ipsilateral hemisphere, but a rising pattern in the number of COX-2-stained processes was
discovered in the contralateral and noninfarcted hemispheres (see ~i~ 1 ~ ) .
Results from the contralateral and noninfarcted
hemispheres were grouped together because no statistical differences were detected between the mean values
of COX-2-immunoreactive neuronal processes in the
regions contralateral to either the infarction core or the
periinfarct area (in the internal carotid or middle cerebra1 artery occlusions).
Our present finding of neuronal expression dominating in
COX-2 immunoreactivity of postischemic brains is
in accord with previous observations of the cellular
sources of COX-2 in the brain.21p24The more variable
and less frequent COX-2 expression in glial and vascular structures was not suitable for quantitative analysis.
Already in the most acute cases, small cells and processes consistent with microglial morphology were seen
in the infarct core and homologous region contralaterally. A perinuclear rim of COX-2 immunoreactivity
was detected in cells with a rounded nucleus, consistent with oligodendrocyte morphology in both the core
and the periinfarct regions and contralaterally. COX-2
immunoreactivity was occasionally observed in vascular
structures outside the infarct core in some subacute
cases (see Fig 2D). The microglial COX-2 positivity
was accompanied by increased astroglial immunoreactivity in the periphery of infarcts aged more than 3
days. In the core region, COX-2-positive inflammatory
cells were also detected from the third day on. Together with increased neuronal expression, a strong
glial COX-2 response was detected (Fig 3A see Fig
2E). In chronic infarctions, endothelial immunoreactivity was detected in the ipsilateral cerebellum and contralateral brain areas, where it associated closely with
neuronal COX-2 immunoreactivity. In the control
brains (n = 6), only sporadic cells consistent with the
morphology of astrocytes and oligodendrocytes showed
COX-2 immunoreactivity.
In sections from the infarction core of acute infarctions, COX-2 fluorescence in cell processes colocalized
with a great number of structures fluorescent for NF200, a marker for neuronal processes. Colocalization of
GFAP-positive astrocytes and COX-2 fluorescent cell
processes was considerably less frequent, which confirmed that most COX-2 producing cellular extensions
were neuronal processes (see Fig 3A-D). Similar distribution was detected in the more matured infarctions
investigated (see Fig 3E and F).
Sairanen et al: COX-2 in Infarcted Human Brain
Fig 2. Cyclooxygenase-2 (COX-2) and heat shock proteins 72/73 (HSP 72/73) immunohistochemisty. (A) Microscopic field from a
periinfarct region of an acute infarction (Case 2) showing an increased number of COX-2-positive neuronal processes (arrowheads).
(B) A field from the infarct core region reveals a smaller number of COX-2-positive neuronal processes. The high-power inset shows
cytoplasmic COX-2 in a cortical neuron. (C) An area homologous to that in A fiom the contralateral hemisphere hurbors neuronal
COX-2 (arrowheads), although in less cellular processes. (0)
COX-2 immunoreactivity in structures representing capillaries in the
infarct core area o f a subacute infarction (Case 8). (E) COX-2 immunoreactivity in uctivuted microglial cells (arrows) from ipsilatera1 pontine region of a case of basilar artery thrombosis (Case 7). (F) A negative cortical section fiom a control brain stained with
HSP 72/73. (G) Neuronal positivity f . r HSP 72/73 in the periinfarct region fiom an acute infarction of 23 hours ’ ymptom duration before death (Case 2). The insert shows in a high power the neuronal cytoplasmic distribution of HSP 72/73 immunoreactivity. (Magnification X 400, before 3 0 % reduction; insets: X G30, bgore 3 0 % reduction; hematoxylin counterstuining) Black
bar = 20 pm.
CORRELATION OF HSP 72/73 AND COX-2 IMMUNOREACTIVITIES. In the acute phase (0.6-1 days), weak HSP
72/73 immunoreactivity was detected in neuronal
perikarya (see Fig 2G) and in cytoplasms of cells with
elongated nuclei and cell shape, consistent with endothelial cell morphology in the periinfarct areas. Most
severely damaged infarction cores had no HSP-stained
cells, but the surrounding penumbral region contained
scattered HSP-positive neuronal somas and processes.
In contrast to the striking COX-2 immunoreactivity,
HSP 72/73 positivity was faint and observed only in a
small number of cells. HSP immunoreactivity was restricted to the acute phase with the exception of sporadic short, thin glial cell processes and endothelid
cells in the infarcted areas during the subacute and
chronic phases. This was in contrast to the delayed intensification of COX-2 immunoreactivity observed bilaterally and also in neuronal cells.
Fig 3. Double-immuno~uorescencestaining of cyclooxygenase-2 (COX-2) and cellular markers. (A) Microscopic j e l d fiom the infarct core of an acute infarction (Case 2) showing COX-2-positive multiform cellular processes labeled with Texas Red conjugate.
(B) The same j e l d with detection of glial jbrilla y acidic protein (GFAP)-positive astrocytes with fluorescein isothiocyanate (FITC)
shows only partial colocalization with COX-2-positive cellular processes (arrows), thw promoting another cellular origin for C O X 2
protein besides astrocytes. (C) COX-2 visualized by Texas Red in a section from the core area of an infarction o f 5 duys and 9
hours in duration (Case 12) demonstrates a robust response in cellular processes intensely colocalizing with NF-200-positive neuronal extensions detected by FITC labeling (0).
(E) Neuronal processes showing COX-2 positivity (Texas R e d up to extended durations of ischemia (17 &ys; Case 15) in a section from the periinfarct area. (F) Thorough colocalization with the neuronal extensions (NF-200; FITC) is noteworthy, yet at the late time point. (Magnzfication X 400.) Black bur = 20 pm.
742 Annals of Neurology
Vol 43
No 6
June 1998
Sairanen et al: COX-2 in Infarcted Human Brain
sphere, COX-2 protein was equally translated in the
same two areas (645 ? 111 neuronal processes in the
ipsilateral vs 421 ? 92 in the contralateral hemisphere;
Case 8). In another patient, a 30-fold COX-2 mRNA
level detected in the infarction core (see Fig 4) corresponded to protein production that was more than 10fold above control levels.
Fig 4. Cyclooxygenase-2 (COX-2) mRMA in control and
stroke cases. Cortical samples fiom the infarcted areas (Cases
6 and 8; infarct core sampLes), the contralateral hemisphere (Case S), and from a control brain were used for
northern blotting. (A) Autoradiograms of COX-2 (top) and
glyceraldehyde-3-phosphate dehydrogendte (GAPDH) mRNA
(bottom) fiom control and stroke brains. Markers show the
localization of 18s and 2 8 s ribosomal mRNAs. (B) Diagram
showing the ratio of COX-2 mRNA to GAPDH mRNA in
control and stroke cases by densitometric analysis of the autoradiagrams. Bidirectional changes in COX-2 rnWA after ischemia in different brain regions are evident, compared with
control brain level (equds I).
COX-2 mRNA Levels Determined by
Northern Blotting
In control brain, a low basal level of COX-2 mRNA
was evident (Fig 4). In 2 subacute stroke cases investigated, either an increased (30-fold) or a depressed
mRNA level (20% of the control level) was detected in
the core areas of the infarction. In a sample from the
hemisphere contralateral to the infarction, the COX-2
mRNA level was fivefold that of control. Despite depressed mRNA synthesis in the infarction core contrasted with an increase in the contralateral hemi-
744 Annals of Neurology
Vol 43
No 6 June 1998
We have characterized the expression of COX-2 in human brains in response to focal infarction, which provides an explanation for the rapid appearance of COX
products, PGs, and thromboxane, seen in cerebral ischemia. 10,25-27 A spatial and temporal evolution of the
strikingly bilaterally increased COX-2 protein was revealed. Despite the fact that animal studies have suggested a low constitutional expression of COX-2 in discrete brain regions, including n e ~ c o r t e x , our
~ , ~data
supported this only at the mRNA level, whereas at the
protein level neuronal COX-2 was detectable in only
one of the control brains. Our present data on human
brain infarction emphasize the widespread and prolonged nature of COX-2 response in stroke, a phenomenon clearly different from the transient, unilateral pattern reported in experimental animal models of focal
and global i s ~ h e m i a . ~ *This
~ ~ ~study
- ~ ~ also demonstrates that methods used to investigate altered gene expression and utilization of the known markers of cellular stress during cerebral pathophysiology in animal
models can be applied to investigate acute cerebral
ischemia in humans.
The bilaterality of COX-2 response to ischemia seen
in our study is in contradiction with results from anima1 s t ~ d i e s . ~ ~Am
, ~ong
~ ,the
~ ~limited
- ~ ~ descriptions
of inflammatory events during ischemia in human
stroke, bilateral intercellular adhesion molecule- 1 expression of microvessels, in separation of granulocyte
infiltration to the hemisphere contralateral to the infarction, has been de~cribed.’~
Although the effect of
humorally distributed inflammatory modulators cannot
be overruled in our cases, other possible explanations
do emerge. It is interesting that COX-2 induction
leading to prostanoid liberation and proinflammatory
changes in cellular adhesion could underlie the changes
in electrical activity, blood flow, and metabolism occurring in the healthy hemisphere during focal ischemia, which is a phenomenon called diaschisis and its
mechanisms are not yet known.31
Transcriptional and posttranscriptional regulation of
COX-2 is a well-known p h e n ~ m e n o n , ~ ~and
~ ’ tran,~~
scriptional regulation has been suggested to also occur
in brain.33 Studies on human autopsy specimens, such
as ours, must be assessed in light of possible postmortem effects.34235Although the mRNA signal may fade
during extended postmortem intervals,36 several studies
imply that the postmortem interval may not be as im-
portant a factor in the maintenance of RNA stability as
are the premorbid agonal state of the patient and the
cellular impact of specific neuropathological disease
pro~ess.~’In brain ischemia, a complex balance between the injury-driven up-regulation of mRNA and
the failing oxygen supply could be influencing COX-2
synthesis at the transcriptional level. Regulation of
COX-2 synthesis could also be influenced by the
relative depth of ischemia, because we could demonstrate clear-cut topographic protein expression patterns.
However, even an infarction core area with “irreversible changes” by hematoxylin and eosin staining, during the subacute phase of infarction, had synthesized
COX-2 protein and immunoreactivity was still seen in
the more matured infarction cores. One possibility is
that this is associated with synaptic activity of processes
of more distant neurons that synthesized COX-2 at
an earlier time point, before the ischemic neuronal
changes had culminated. A more extensive study, combining mRNA and protein data, will allow a better description of the transcriptional, posttranscriptional, and
translational regulation of COX-2 during ischemic
brain damage.
The signaling pathways leading to increased COX-2
expression after brain injury are unclear. Attenuation of
COX-2 expression in experimental epilepsy models, by
a selective platelet-activating factor receptor antagonist,
implies a downstream regulation of COX-2 by plateletactivating factor in brain injury33 and provides a possible pathway between induced platelet-activating factor and prostanoid products, both of which are released
rapidly after brain ischemia.10225p27,38 It is noteworthy
that COX-2 has been suggested to be dependent on
induced phospholipase A, a~tivi$‘,*~ and the stores of
endogenous substrate (arachidonic acid) appear to be
metabolized solely by the inducible form of COX in
fibroblasts and macrophages despite the presence of
COX-1, which supports a pivotal role for COX-2 in
triggering lipid mediator m e t a b o l i ~ m .Due
~ ~ to its role
as an IEG responding to synaptic activity and cellular
tress,^,^ COX-2 could serve as a proximal intracellular
signal regulating cerebral release of vasoactive and injurious lipid mediators in general.
In addition to the well-known efficacy of COX antagonism in limiting ischemic neuronal injury,l ‘-13 a
recent study with a selective COX-2 antagonist showed
infarct volume reduction and attenuation in the elevation of PGE, in the focal ischemia model, where
COX-1 was shown not to be induced.23 COX inhibitors have also been shown to decrease hydroxyl radical
production in global ischemia.42 These data suggest
that prostanoid production is a therapeutically important biochemical sequela of brain ischemia, and our
present results extend the hypothesis to human ischemic stroke. Because a more sustained expression of
COX-2 was evident in human infarctions (0-6 days),
a wider therapeutic window for antagonism of acute
COX-2 production should be considered in clinical situations. However, more detailed experimental study,
based on selective COX-2 inhibition, is required.
HSP 72/73 immunoreactivity as a marker of cellular
stress was detected in neurons, and less so in glial and
endothelial cells of the penumbral areas of infarctions
with less than 24 hours’ duration, which corresponds
to observations in experimental models.’ However,
our material is limited by the lack of even more acute
infarctions and does not reveal immediate changes in
gene expression, which might also involve more proximal induction of COX-2 as an IEG before the phenotypic correlates of cell stress can be observed. The
lack of temporal and topographic correlation between
heat shock response and evolving COX-2 expression
indicates that COX-2 is not merely responding to
injury-related cellular stress, but represents a more globally regulated change in the gene expression in focally
infarcted human brain. The topographic expression
shared some characteristics with IEG induction observed typically for genes such as c-fos in experimental
model^.^' However, COX-2 was not transiently induced, but persisted up to 2.5 weeks and therefore
probably plays a more fundamental role in the face of
graded tissue ischemia as well as remodeling of the injured brain.
To summarize, our data revealed a neuronal and
glial COX induction early (15-24 hours) in the periinfarct areas, which subsequently (1.2-6.3 days) was
elicited prominently in the neurons of the infarct core
and the contralateral hemisphere, and thereafter
(8.5-18 days) persisted predominantly in the contralateral, surviving hemisphere. Because experimental animal studies have suggested that production of prostanoids and free radicals aggravates tissue injury during
cerebral ischemia9,10,13,23,25,27 COX-2 inhibitors may
hold promise for humans during acute stroke. However, in view of the inducibility of COX-2 by synaptic
activity and its possible participation in activitydependent neural pla~ticity,~COX-2 induction observed remotely from the infarct throughout the surviving brain after the acute phase could reflect global
cerebral adaptive and regenerative processes, and thus
suggests a role for prostanoid production in the remodeling of neural networks after focal infarction. It is
noteworthy that a dual role for COX-2 could mean it
functions as a physiological intracellular modulator of
signal transduction, leading to long-term potentiation
and plasticity, and also as a deleterious mediator, under excessive N-methyl-D-aspartate receptor-dependent
glutamatergic stimulation. The latter might aggravate
excitotoxic cell death and coexist with the extracellular
proinflammatory consequences of COX-2 activity.4,21244
Indeed, these data could also bear therapeutic
significance for a broad spectrum of human CNS dis-
Sairanen et al: COX-2 in Infarcted H u m a n Brain
eases, in which COX activation may play a role, including spinal cord injury, l o human immunodeficiency
and primary degenerative
dementias such as Alzheimer’s disease. 1021,22,44346
Financial support came from the Paulo Foundation, the Maire
Taponen Foundation, the Helsinki Universiry Central Hospital
(P.L.), and the Academy of Finland (A.R.). Dr Sairanen is a recipient of PhD graduate school funding by the University of Helsinki,
Helsinki, Finland.
We thank Ritva Javanainen for excellent technical assistance.
This study is one of a series of postmortem stroke studies, designated collectively as the Helsinki Stroke Study (HSS).
1. Herschnian HR. Regulation of prostaglandin synthase-1 and
prostaglandin synthase-2. Cancer Metastasis Rev 1994;13:241256
2. Smith WL, Garavito RM, DeWit DL. Minireview: prostaglandin endoperoxidase H synthases (cyclooxygenases) -1 and -2.
J Biol Chem 1996;271:33157-33160
3. Ogorochi T, Narumiya S, Mizuno N, ec al. Regional distribution of prostaglandins D,, E,, and F,, and related enzymes in
postmortem human brain. J Neurochem 1984;43:71-82
4. Yamagata K, Andreasson KI, Kaufmann WE, et al. Expression
of mitogen-inducible cyclooxygenase in brain neurons: regulation by synaptic activity and glucocorticoids. Neuron 1993;ll:
5. Peri KG, Hardy P, Li DY, et al. Prostaglandin G/H synthase-2
is a major contributor of brain prostaglandins in the newborn.
J Biol Chem 1995;270:24615-24620
6. Adams J, CollaFo-Moraes Y, de Belleroche J. Cyclooxygenase-2
induction in cerebral cortex: an intracellular response to synaptic excitation. J Neurochem 1996;66:6-13
7. Gaudet RJ, Alam I, Levine L. Accumulation of cyclooxygenase
products of arachidonic acid metabolism in gerbil brain during
reperfusion after bilateral common carotid artery occlusion.
J Neurochem 1980;35:653-658
8. Maier-Hauff K, Lange M, Schurer L, et al. Mediator compounds in CSF of neurosurgical patients with raised intracranial
pressure. Adv Neurosurg 1984;12:302-306
9. Baethmann A, Maier-Hauff K, Schurer L, et al. Release of glutamate and free fatry acids in vasogenic brain edema. J Neurosurg 1989;70:578-591
10. Farooqui AA, Horroch LA. Excitatory amino acid receptors,
neural membrane phospholipid metabolism and neurological
disorders. Brain Res Rev 1991;16:171-191
11. Kontos HA, Wei EP, Povlishock JT, et al. Cerebral arteriolar
damage by arachidonic acid and prostaglandin G,. Science
12. Hallenbeck JM, Furlow TW. Prostaglandin I, and indomethacin prevent impairment of post-ischemic brain reperfusion in
the dog. Stroke 1979;10:629-637
13. Chen J, Weinstein PR, Graham S. Attenuation of postischemic
brain hypoperfusion and reperfusion injury by the cyclooxygenase-lipoxygenase inhibitor BW755C. J Neurosurg 1995;83:
14. Kawagoe J, Abe K, Sat0 S, et al. Distribution of heat shock
protein-70 mRNAs and heat shock cognate protein-70 after
transient global ischemia in gerbil brain. J Cereb BIood Flow
Metab 1992;12:794- 80 1
15. Kinouchi H, Sharp FR, Hill MP, et al. Induction of 70 kDa
746 Annals of Neurology Vol 43
No 6
June 1998
heat shock protein and hsp70 mRNA following transient focal
cerebral ischemia in the rat. J Cereb Blood Flow Metab 1993;
Eke A, Conger KA, Anderson M, et al. Histological assessment
of neurons in rat models of cerebral ischemia. Stroke 1990;21:
Lindsberg PJ, CarpCn 0, Paetau A, et al. Endothelial ICAM-1
expression associated with inflammatory cell response in human
stroke. Circulation 1996;94:939-945
Chirgwin JM, Przybyla AE, MacDonald RJ, et al. Isolation of
biologically active ribonucleic acid from sources enriched in ribonuclease. Biochemistry 1979;18:5294-5299
Ristimaki A, Garfinkel S, Wessendorf J, et al. Induction of
the cyclooxygenase-2 by interleukin-la: evidence for a posttranscriptional mechanism. J Biol Chem 1994;269:117691 1775
Kaufmann W, Worley PF, Pegg J, et al. COX-2, a synaptically
induced enzyme, is expressed by excitatory neurons at postsynaptic sites in rat cerebral cortex. Proc Natl Acad Sci USA 1996;
93:23 17-2321
Chang JW, Coleman PD, O’Banion MK. Prostaglandin G / H
synthase-2 (cyclooxygenase-2) mRNA expression is decreased in
Alzheimer’s disease. Neurobiol Aging 1997;17:801-808
Oka A, Takashima S. Induction of cyclo-oxygenase 2 in brains
of patients with Down’s syndrome and dementia of Alzheimer
type: specific localization in affected neurons and axons. Neuroreport 1997;5:1161-1164
Nogawa S, Zhang F, Ross ME, Iadecola C. Cyclo-oxygenase-2
gene expression in neurons contributes to ischemic brain damage. J Neurosci 1997;17:2746-2755
Miettinen S, Fusco FR, Yrjanheikki J, et al. Spreading depression and focal brain ischemia induce cyclooxygenase-2 in cortical neurons through N-methyhaspartic acid receptors and
phospholipase A,. Proc Natl Acad Sci USA 1997;94:65006505
Moskowitz MA, Kiwak KJ, Hekimian K, et al. Synthesis of
compounds with properties of leukotrienes C, and D, in gerbil
brains after ischemia and reperfusion. Science 1984;224:886 889
Kempski 0, Shohami E, von Lubitz D, et al. Postischemic production of eicosanoids in gerbil brain. Stroke 1987;18:111-119
Stevens MK, Yaksh TL. Time course of release in vivo of
PGE,, PGF,,, 6-keto-PGF,,, and TxB, into the brain extracellular space after 15 min of complete global ischemia in the
presence and absence of cyclooxygenase inhibition. J Cereb
Blood Flow Metab 1988;8:790-798
Planas AM, Soriano MA, Rodriguez-Fad, et al. Induction of
cyclooxygenase-2 mRNA and protein following transient focal
ischemia in the rat brain. Neurosci Lett 1995;200:187-190
Colla~o-Moraes Y, Aspey B, Harrison M, et al. Cyclooxygenase-2 messenger RNA induction in focal cerebral ischemia. J Cereb Blood Flow Metab 1996;16:1366-1372
Ohtsuki T, Kitagawa K, Yamagata K, et al. Induction of
cyclooxygenase-2 mRNA in gerbil hippocampal neurons after
transient forebrain ischemia. Brain Res 1996;736:353-356
Andrews RJ. Transhemispheric diaschisis. A review and comment. Stroke 1991;22:943-949
Schmetdje JF Jr, Ji Y-S, Liu W-L, et al. Hypoxia induces
cyclooxygenase-2 via the NF-KB p65 transcription factor in human vascular endothelial cells. J Biol Chem 1997:272:601-608
Marcheselli VL, Bazan, NG. Sustained induction of prostaglandin endoperoxidase synthase-2 by seizures in hippocampus. Inhibition by a platelet-activating factor antagonist. J Biol Chem
Hayes TL, Cameron JL, Fernstrom JD, et al. A comparative
analysis of the distribution of prosomatostatin-derived peptides
in human and monkey neocortex. J Comp Neurol 1991;303:
Knudsen LM, Pallesen G. The preservation and loss of various
non-hematopoietic antigens in human post-mortem tissues as
demonstrated by monoclonal antibody immunohistological
staining. Histopathology 1986;lO:1007-1014
Jackson AC, Rintoul NE. Effects of post-mortem autolysis on
the detection of rabies virus genomic RNA and mRNA in
mouse brain by using in situ hybridization. Mol Cell Probes
Higgins GA, Mah VH. In situ hybridization approaches to human neurological diseases. In: Conn PM, ed. Methods in
neurosciences: gene probes. San Diego, C A Academic Press,
1989: 183-196
Iindsberg PJ, Hallenbeck JM, Feuerstein G. Platelet-activating
factor in stroke and brain injury. Ann Neurol 1991;30:117-129
Reddy ST, Herschman HR. Prostaglandin synthase-1 and prostaglandin synthase-2 are coupled to distinct phospholipases for
the generation of prostaglandin D, in activated mast cells.
J Biol Chem 1997;272:3231-3237
Chen Q-R, Miyaura C, Higashi S, et al. Activation of cytosolic
phospholipase AL by platelet-derived growth factor is essential
for cyclooxygenase-2-dependent prostaglandin E, synthesis in
mouse osteoblasts cultured with interleuhn-1. J Biol Chem
Reddy ST, Herschman HR. Ligand-induced prostaglandin synthesis requires expression of the TIS 1OIPGS-2 prostaglandin
synthase gene in murine fibroblasts and macrophages. J Biol
Chem 1994;269:15473-15480
Hall ED, Andrus PK, Althaus JS, er al. Hydroxyl radical production and lipid peroxidation parallels selective post-ischemic
vulnerability in gerbil brain. J Neurosci Res 1993;34:107-112
Akins PT, Liu PK, Hsu CY. Immediate early gene expression in
response to cerebral ischemia: friend or foe? Stroke 1996;27:
Breitner JCS. Inflammatory processes and antiinflammatory
drugs in Alzheimer’s disease: a current appraisal. Neurobiol Aging 1996;17:789-794
Griffin DE, Wessenlingh SL, McArthur JC. Elevated central
nervous system prostaglandins in human immunodeficiency
virus-associated dementia. Ann Neurol 1994;35:592-597
McGeer PL, Schulzer M, McGeer EG. Arthritis and antiinflammatory agents as possible protective factors for Alzheimer’s disease: a review of 17 epidemiologic studies. Neurology
Sairanen et al: COX-2 in Infarcted Human Brain
Без категории
Размер файла
2 398 Кб
cyclooxygenase, infarcted, induced, brain, human, globally
Пожаловаться на содержимое документа