вход по аккаунту


Complement activation in the central nervous system following bloodЦbrain barrier damage in man.

код для вставкиСкачать
Corndement Activation in the Central
Nervous System Following Blood-Brain Barrier
Damage in Man
Perttu J. Lindsberg, MD, PhD,*$ Juha Ohman, MD, PhD,? Tim0 Lehto, MS,$
Marja-Liisa Karjalainen-Lindsberg, MD,$ Anders Paetau, MD, PhD,$ Tomi Wuorimaa, MS,$
Olli Carpdn, MD, PhD,§ Markku Kaste, MD, PhD,* and Seppo Meri, MD, PhD$
The central nervous system (CNS) is virtually isolated from circulating immunological factors such as complement (C),
an important mediator of humoral immunity and inflammation. In circulation, C is constantly inhibited to prevent
attack on host cells. Since a host of diseases produce an abnormal blood-braidcerebrospinal fluid (blood-brainlCSF)
permeability allowing C protein extravasation, we investigated if C activation occurs in CSF in vitro and in CNS in
vivo during subarachnoid hemorrhage (SAH) or brain infarction. After SAH (n = 15), the terminal complement complex
(TCC) concentration on days 0 to 2 was higher in the CSF, 210 +- 61 ng/ml, than in the plasma, 63 f 17 ng/ml, but
null in the CSF of controls (n = 8 ) or patients with an ischemic stroke (n = 7). TCC was eliminated from the CSF
after SAH (24 10 nglml on days 7 to 10). Incubation of normal human CSF with serum in vitro also activated the
terminal C pathway. In 10 fatal ischemic brain infarctions, immunohistochemical techniques demonstrated neuronal
fragment-associated deposition of C9 accompanied by neutrophil infiltration. We conclude that the C system becomes
activated intrathecally in SAH and focally in the brain parenchyma in ischemic stroke. By promoting chemotaxis and
vascular perturbation, C activation may instigate nonimmune inflammation and aggravate CNS damage in diseases
associated with plasma extravasation.
Lindsberg PJ, Ohman J, Lehto T, Karjalainen-Lindsberg M-L, Paetau A, Wuorimaa T, Carpkn 0,
Kaste M, Meri S. Complement activation in the central nervous system following
blood-brain barrier damage in man. Ann Neurol 1996;40:587-596
The blood-brain barrier (BBB) largely secludes the
mammalian central nervous system (CNS) from circulating components of the immune system. Complement (C) proteins participate in both immune and
nonimmune responses of the host. They have a limited
access to the CNS and are synthesized only at low levels within the CNS itself. The C system is composed
of at least 20 plasma proteins [I], the concerted action
of which leads to an assembly of terminal complement
complexes (TCCs) in the form of fluid-phase products (SC5b-9) or cytolytic membrane attack complexes
(MACs) on cell membranes. During a physiologic
steady-state blood-cerebrospinal fluid (blood-CSF)
barrier integrity, the level of C proteins (C3, C4) is
typically several hundredfold lower in CSF than in serum [2]. Isolation of the CNS from C proteins is appropriate, since neural elements are vulnerable to complement lysis and oligodendrocytes and myelin can
activate C via the classical pathway in an antibodyindependent manner [3-71.
Apart from homeostatic functions during microbial
infections and removal of foreign or necrotic cells, the
C system can aggravate inflammation and tissue damage, eg, in vasculitic and autoimmune syndromes, experimental allergic encephalomyelitis [S], and myocardial ischemia-reperfusion injury [9- 121. Recently, it
was demonstrated that the pathophysiologic consequences of myocardial ischemia-reperfusion can be diminished by C-inhibiting treatment [ 131, an effect possibly mediated by reduced tissue inflammation and
neutrophil infiltration [ 11, 141. The C system has
emerged as a therapeutically interesting, multifaceted
cascade of factors acting during aggravation of tissue
injury by host response [ 151.
In circulating plasma, C is constantly under active
control by several humoral regulators that include C 1
inhibitor, C4b binding protein, factor H, clusterin, and
vitronectin. Furthermore, host cells normally exposed
to C, such as blood cells and vascular endothelial cells,
are protected from C lysis by membrane proteins decay
From the Departments of "Neurology, ?Neurosurgery, $Bacterialogy and Immunology, and $Pathology, University of Helsinki, Helsinki, Finland.
Received Oct 16, 1995, and in revised form Mar 13, 1996. Accepted for publication Apr 10, 1996.
Address correspondence to Dr Lindsherg, Department of Neurology, University of Helsinki, Haartmaninkatu 4, FIN-00290 Helsinki, Finland.
Copyright 0 1996 by the American Neurological Association 587
Churucteristics and Results of the lnzmunohistochemicuI Examination of the Bruins of
Succumbed Stroke Putients Studied Postmortem
1 IM
Cause of
Risk Factorsh
PE, stroke
PE, stroke
PE, stroke
PE, stroke
Duodenal ulcer
15 hr
1 day 4 hr
1 day 19 hr
2 days 5 hr
3 days 6 hr
4 days 12 hr
6 days 7 hr
8 days
17 days
18 days
'Cases A and B represent control patients who died suddenly of a nonneurological cause.
bAbbreviations for risk factors: H = hypertension; DM = diabetes mellitus; AF = atrial fibrillation; HF = heart failure; CAD = coronary
artery disease; AS = generalized arteriosclerosis; LVH = left ventricular hypertrophy.
'Abbreviations for cause of death: VF = ventricular fibrillation; CA = cardiac arrest; AM1 = acute myocardial infarction; PE = pulmonary
"Abbreviations for occluded vessels: ICA = internal carotid artery; MCA = middle cerebral artery; BA = basilar artery; OLA = occipital
lobe artery; T = thrombosis; E = thromboembolisni.
'Grading of C3 immunohistochemistry: (-) = positive staining only on the basal endothelial membranes; (2)= enhanced staining on the
vascular wall structures but no definite neuropil staining; (+) = enhanced vascular staining and slightly enhanced staining on neuropil
structures in some infarct lesions; (+ +) = clearly increased diffuse staining in infarct foci and specific neuronal structures such as axons in
the vicinity of these areas.
'Granulocyte (GR) count per millimeter squared in the infarcted brain area. M D = data missing for technical reasons.
phagocyte (MO) count per millimeter squared in the infarcted brain area.
accelerating factor (DAF, CD5 5 ) , membrane cofactor
protein (MCP, CD46), C3b receptor (CR1, CD35),
and protectin (CD59) (reviewed in reference 16). Active suppression of C is crucial to prevent its spontaneous or surface-induced activation. The CSF and brain
interstitial fluid are normally not exposed to C. This
led us to hypothesize that a contact between human
plasma and CSF or CNS tissue might trigger C activation and assembly of TCC. Generation of anaphylatoxins and lytic or sublytic MAC attack against autologous
tissue have the potential to mount a nonimmune inflammatory response within the CNS. In this study we
show the relevance of this hypothesis to events following BBB damage caused by aneurysmal subarachnoid
hemorrhage (SAH) and ischemic brain infarction in
Materials and Methods
The group with SAH, treated at the Department of Neurosurgery, contained 15 consecutive patients, age 19 to 63 years
(median, 49 years) of either sex (6 males, 9 females), who
had suffered a rupture of an aneurysm within the previous
72 hours. In each patient, computed tomography (CT) had
shown blood in the basal cisterns (Fisher score [17]ranging
from 2 to 4 , median 3). and The cause of SAH was verified
by angiography to be an aneurysm. The clinical condition
of the patients was scored according to Hunt and Hess [18]
Annals of Neurology Vol 40
No 4
October 1996
and ranged from I (asymptomatic or minimal symptoms) to
4 (stupor). The majority of patients (12) had scores of 2 or
3. Each patient undenvent intracranial surgery to occlude
the aneurysm within a mean of 2.1 days after the rupture.
All received nimodipine intravenously at a rate of 0.5 F g l
kg/min for 7 to 10 days after surgery. Betamethasone ( 4 mg)
was given intramuscularly four times a day. The eventual
outcome was determined using the Glasgow Outcome Scale
[19], and 9 patients recovered completely (score I), whereas
in 2 cases the SAH was fatal (score 5). The appropriate ethics
committee at the Helsinki University Central Hospital had
accepted the study protocol.
Two groups of patients with stroke were studied. The first
group was a postmortem series of 10 patients, treated at the
Department of Neurology for acute ischemic stroke until
death. Autopsies were performed within 15 -+ 4 (mean ?
SD) hours after death. In addition, 2 patients who died suddenly due to a nonneurological cause were included as controls 14 and 14.5 hours postmortem, respectively. Details of
these patients are given in the Table.
The second stroke group consisted of 7 consecutive patients, who had suffered a severe hemiparesis and had a lumbar puncture done within the next 24 hours. In these cases,
the initial CT scan had revealed a nonhemorrhagic, hemispheric infarction.
Histological and Irnmunohistocbernical Methods
At autopsy, the infarcted areas were localized on the basis of
the most recent C T scan and the macroscopical appearance
of the brain parenchyma and cerebrovasculature. Approximately 1-cm' cortical samples including subcortical white
matter were dissected out, frozen in liquid nitrogen, and
stored at -70°C until analyzed. Similar specimens were routinely fixed with formaldehyde, embedded in paraffin, and
cut for hematoxylin-eosin staining. From the frozen tissue
blocks, fresh-frozen sections (5 pm) were cut, stained with
hematoxylin-eosin, and studied microscopically to confirm
ischemic neuronal changes. Samples from the corresponding
areas of the contralateral, noninfarcted hemispheres and from
the 2 control patients, who died without a neurological disease, were processed in a similar way. Fresh-frozen sections
were fixed with cold acetone (-20°C) for 10 minutes and
rinsed with phosphate-buffered saline, p H 7.4 (PBS). The
sections were incubated for 30 minutes at +22"C with polyclonal rabbit antibodies against C3d and C 9 (Behringwerke
AG, Marburg, Germany) at concentrations of 50 pgiml.
After washing three times with PBS, the sections were treated
with fluorescein isothiocyanate (F1TC)-conjugated antibodies against rabbit immunoglobulins (Dakopatts, Glostrup,
Denmark). Immunofluorescence stainings were controlled by
replacing the primary antibody with PBS, normal rabbit serum, or rabbit antibody against human IgG (Behringwerke
AG). After covering the samples with a mounting medium
and coverslips they were examined on a Zeiss standard fluorescence microscope. The intensity of immunofluorescence
was examined without knowledge of the sample localization
and finally cross-checked during morphologic identification
of the stained structures by a neuropathologist (A.P.). Some
sections were stained using a monoclonal antibody against
C5b-9 neoepitope (aEl1, a kind gift from Dr T. E. Mollnes,
Department of Immunology and Transfusion Medicine,
Notdland Central Hospital, Bod0, Norway) as the primary
antibody. Hematoxylin-eosin-stained sections from the same
tissue blocks were used for reference. Immunostaining of
phagocytes was performed with the immunoperoxidase technique, using an antibody against the CD15 epitope on granulocytes (Dakopatts) and another recognizing the mononuclear phagocytes (HAM56) (Dakopatts). Quantitation of the
phagocyte infiltration was performed using sections cut from
the same tissue blocks and studied with a light microscope.
An average of the phagocyte counts in five representative
fields of 1.31 or 0.32 mm2 were taken as the phagocyte count
in each case.
Sample Collection
CSF samples from the SAH patients (n = 15) were collected
by lumbar punctures on admission (day 0-2) and on day 7
to 10 after the SAH. Blood samples were collected at the
same time into tubes with and without 10 m M EDTA. Control CSF and blood samples were obtained from patients
(n = 8) undergoing lumbar myelography for the investigation of radiculopathy. The CSF samples were initially collected into cold sterile plastic tubes containing no additives.
Samples were frozen immediately and stored at -7OOC until
analyzed. Upon thawing, the samples were diluted into a
buffer containing 10 mM EDTA. CSF was collected from
three additional SAH patients (not included in data analysis)
into tubes containing 10 m M EDTA, and the levels of
SC5b-9 in these samples were found to be similar to those
in the patients from whom samples were taken into empty
tubes. For in vitro admixture experiments, samples of CSF
and serum were collected into plastic tubes with no additives
and frozen (-70°C) until used. CSF was obtained after an
informed consent from patients (n = 5) who were examined
at the neurological outpatient clinic for various symptoms
and eventually not found to have a neurological disease affecting the BBB integrity.
Assay of SC5b-9 Complexes
A commercially available enzyme-linked immunosorbent
assay (ELISA) kit (Quidel Corp, La Jolla, CA) was used for
the quantitation of SC5b-9 in CSF and plasma. Samples
were diluted into PBS containing 0.05% Tween 20 and
added to wells coated with a mouse monoclonal antibody
against a neoepitope of human SC5b-9. After a I-hour incubation period, the wells were washed five times and horseradish peroxidase-conjugated goat polyclonal antibodies to C6
and C7 were added. Afier another 1-hour incubation, the
wells were washed five times and a chromogen was added.
Absorbance at 405 nm was measured after a 30-minute incubation at room temperature. A standard curve for SC5b-9
(ng/ml) versus absorbance was generated using specimens
with known amounts of SC5b-9. The lower detection limit
for the assay was 10 ngiml.
Examination of SC5b-9 Generation In Vitro by an
Admixture of CSF and Serum
Varying amounts of normal CSF were diluted into isotonic
PBS and mixed with 150 ~1 of serum from the same subject
(autologous admixture) or from other normal controls (homologous admixture) to obtain CSF/serum ratios 64 : 1, 16 : 1,
4 : 1, and 1 : 1 in a final volume of 300 pl. The series was
continued by mixing a constant amount of CSF (150 pl) with
150 pI of various dilutions of normal human serum (NHS)
(into PBS) to obtain CSF/serum ratios 1 : 4 , 1 : 16, and 1 :64,
respectively. Control experiments were performed with NHS
diluted into isotonic PBS in the absence of CSF, with CSF
in PBS without N H S and with EDTA plasma. All admixtures
were incubated for 30 minutes at +37"C. Background control
contained 150 pI of N H S and 150 p1 of PBS and was kept
at 0°C. The SC5b-9 content of the samples was determined
by ELISA as described.
Analysis of C3b Deposition on the Membranes of
Subarachnoid Red Blood Cells
Subarachnoid red blood cells (RBCs) obtained from SAH
patients within 2 days after the bleeding were washed with
PBS into a suspension of 108/ml in 50 11. T o examine C3b
deposition, the cells were incubated for 30 minutes at
+23"C with FITC-conjugated antibody against C3c (Behringwerke) diluted into 0.5% bovine serum albumin (BSA)/
PBS. After washing three times with 0.5% BSAIPBS, the
cells were fixed with 2.5% glutaraldehyde and analyzed by
flow cytometry (FACScan, Becton Dickinson, Palo Alto,
CA). Positive controls were prepared by incubating rabbit
RBCs with human serum deficient in C7 (C7DS) for 20
minutes at +37"C. Washed RBCs obtained by venipuncture
from healthy donors served as negative controls.
Lindsberg et al: Complement Activation in BBB Damage
Fig 1. Terminal complement (C) complex (TCC) concentration in the cerebrospinul fluid (CSF) of patients with subarachnoid hemorrhage (SAH) and controls without a central
nervous system disorder. The CSF of controls and patients
with severe ischemic stroke contuined no detectable TCC,
which indicates thut the in vivo admixture of CSF and
serum was responsiblefor the increased level of TCC on
admission day (SAHl = 0-2 daysfion? bleedind and 7 to
10 days after bleeding (SAH2). Statistical sign$cances:
**7, < 0.001 and 7, < 0.05, in a comparison between
SAH patients und controls, und 'p = 0.01, in a comparison
between the CSF and plasma of the same SAH patients.
Statistical Metbods
All data are expressed as mean 2 SE values for the indicated
number of experiments. The paired Wilcoxon test was used
to compare the concentrations of SC5b-9 in the CSF and
plasma of SAH patients. The Mann-Whitney U test was
used to compare the Concentrations of SC5b-9 in the CSF
of SAH patients and control subjects. The p values were
adjusted according to the Bonferroni procedure; p values of
<0.05 were considered significant.
To investigate whether C activation had occurred intrathecally in SAH, we measured the concentration of
SC5b-9, the soluble terminal C complexes (TCC) in
the CSF and plasma of patients and controls. Controls
had no detectable TCC in their CSF. While approximately equal amounts of soluble SC5b-9 were observed
in the plasma of SAH patients and healthy controls,
the CSF of SAH patients at days 0 to 2 after the bleeding contained substantially more soluble TCC than
their plasma at the same time (Fig 1). CSF obtained
within 24 hours after a severe ischemic stroke (n = 7)
contained no detectable TCC. At days 7 to 10 after
SAH. the level of soluble TCC in the CSF had de-
590 Annals of Neurology
Vol 40
October 1996
Fig 2. Generation of terminal complement (C) complex
(TCC) in admixtures of normal cerebrospinal j u i d (CSF)
and normal human serum (NHS) in vitro during u 30minute incubation at 37°C. The solid burs in the right
punel illustrate TCC generation when va ying amounts of
CSF diluted in phosphate-buffered saline were mixed with
150 p l of NHS. The hatched bars in the left panel show
TCC generation when varying amounts of NHS were mixed
with 150 p l of CSF. The empty bars in the right punel ure
controls and show TCC levels in NHS in the ubsence of
CSF. Since the CSF has virtually no C proteins, the C proteins necessary for de novo TCC generation must come fiom
the NHS. It should therefore be noted that the admixture
where 4 purts of CSF is mixed into 1 purt of NHS (bur
4/1) muy stimulate TCC generation even more potently thun
the admixture 1/1, since it generated more than 40% o f the
maximal observed TCC level and bud only 25% of the complement proteins uvailable for TCC assembly in the admixture 1/1, which produced the maximal TCC amount. The
spinal fluids exumined bud un average protein concentration
of 531 mg/L and leukocyte count of 0.25/~1.The dutu are
expressed as mean 2 SE values ofjive separate experiments
with different spinul juids.
creased below the plasma level observed at days 0 to
2 or days 7 to 10 after the SAH but was still significantly higher than in control patients or in patients
with acute ischemic stroke ( < I 0 nglml).
The evidence that TCC formation had occurred in
the CSF in vivo during SAH led us next to examine
if CSF stimulated SC5b-9 generation in NHS in vitro.
When normal CSF and N H S from healthy individuals
were incubated together at +37"C for 30 minutes, the
C cascade became activated as indicated by de novo
generation of SC5b-9 (Fig 2). Admixtures of CSF and
NHS at ratios I : 1 and 4:1 generated TCC up to five
times more (adjusted for the amount of NHS in each
admixture) than in the controls without CSF. Homologous and autologous admixtures of NHS and normal
CSF generated TCC to a similar extent. No detectable
TCC was generated in EDTA plasma.
Because SAH is characterized by exposure of the
subarachnoid space to RBCs, which become lyzed in
a few days, we also examined the possibility that the
RBC membranes become a focus of C activation. Subarachnoid RBCs obtained from patients who had suffered an SAH up to 2 days earlier (n = 3) were immunostained to detect the presence of C3 breakdown
products. Flow cytometry of these RBCs, unlike rabbit
RBCs treated with human serum deficient in C7 (control), demonstrated no C3 deposition on their cell
membranes (not shown).
The findings that SC5b-9 can be detected in the
CSF of SAH patients in vivo and that its generation
in serum is enhanced by CSF in vitro suggests that the
terminal C pathway is activated in the fluid phase. To
see if C activation takes place in the brain extracellular
space when leakage of plasma occurs during ischemic
brain infarction, we examined brain samples from patients who died after a nonhemorrhagic stroke. In 10
such patients, immunohistochemical staining demonstrated that cerebral infarction and brain edema were
associated with a varying intensity of focal C9 deposition, whereas there was not an increased staining in
the noninfarcted hemispheres (see Table). The staining
for C3d mostly correlated with C9 deposition (Fig
3a, b), but no similar deposition of human IgG was
detected. Staining with the anti-C5b-9 neoantigen
monoclonal antibody (mAb) showed consistent deposition within areas of infarction necrosis (Fig 4a). The
topography of this deposition was less widespread than
that for C9 and seemed to focus in defined zones on
the outer edges of the infarct lesions. There was no
such deposition in samples taken from the noninfarcted hemispheres.
These results suggested that BBB damage had occurred and resulted in C activation. Since C activation
causes potent chemoattraction, we performed additional immunohistochemical stainings to measure
phagocyte (granulocytes and macrophages) infiltration
in the same tissue areas where C activation was examined. Both phagocyte counts were below l/ mm2 in the
brains of 2 control subjects who had died from extracerebral causes. In these control patients, linear deposition of C9 was observed in the basement membranes
of blood vessels (Fig 5b). In Patient 1, who died only
Fig 3. Deposition of C3d and C9 in a 63-day-old focal
brain stem infarct lesion (Case 7). Cyostat sections were
treated with rabbit polyclonal antibodies against C3d (a) or
0 (b, c). Bound antibody was detected with a jluorescein
isothiocyanate-conjugated seconday antibody. (a) Staining f i r
C3d indicates extravasation and covalent attachment of
blood-borne C3 to infdrcted brain tissue (X 75). (6) Deposition of C9firther demonstrates BBB damage and assembly of
C5b-9 in infarcted brain (X 75). (c) C9 deposition in
injured neuronal processes (X 350).
Lindsberg et al: Complement Activation in BBB Damage 591
Fig 4. Detection of membrane attack complex (MAC) with a monoclonal antibody against C5b-9 neoantigen. Bound antibody
was detected with a juorescein isothiocyanate-conjugated secondd y antibody (Case 8). (a) A zone of MAC deposition within an
infarct lesion. (6) A control staining without the prima y antibody.
15 hours after a stroke, C9 deposition was already enhanced in the vascular wall and dispersed in the neuropil (Fig 5a), while granulocytes were found to marginate at the endothelium (Fig 5c). In patients who
died between days 1 and 2 after a stroke, increasing
numbers of granulocytes and C9 deposition were observed in the extravascular compartment of the brain.
C9 deposition in infarcted brain was apparently increased in patients who died during days 2 to 4 after
a stroke (Cases 4 and 5 ) (Fig 6), at which stage the
infiltration was granulocytic and macrophages were not
detected (Case 5 ) (see Table). Structures with C9 deposits included neuropilic threads and streaks, which
we considered to be axons exposed by stripping of their
covering myelin sheath (Figs 3c, Gc). C9 deposits were
seen also in axon baskets surrounding the Purkinje cell
bodies in infarcted cerebellum (Fig 7a). In some cases,
C9 staining was widely increased around the infarct
(Fig 6d). Occasionally, enhanced C9 staining was observed also in the endothelia and basement membranes
of blood vessels in sites not associated with the infarct.
Infarctions caused by basilar artery occlusion (Cases 5,
7, and 8) had less granulocyte invasion than the hemispheric infarctions.
The experiments of the present study show that the
human C cascade becomes activated upon contact with
CSF and CNS under both in vitro and in vivo conditions. This supports the hypothesis that the normal
immunologically privileged environments of the CNS
and CSF alter the regulatory balance of the plasma C
system. In health, the CNS contains three fluid compartments with interposed barriers of varying integrity,
the intravascular, the intracellular, and the extracellular
spaces. The CSF space, the ventricles and the subarach-
Annals of Neurolow
Vol 40 No 4
October 1996
noid space, is actually a lacuna of the extracellular space
of the CNS, because the ependyma lining the ventricles
does not act as a barrier to solutes [20].Activation of
the plasma C cascade in the spaces of CSF may be
relevant to many human diseases where there is extravasation of plasma into the brain, eg, ischemic, hemorrhagic and traumatic brain injury, cerebral vasculitis,
and multiple sclerosis (MS).
The deposition of activated or nonactivated C components on cellular structures has not previously been
reported after stroke, but deposition of C9, suggesting
activation of the terminal C pathway, has been observed in association with capillaries in MS lesions [21].
The concentration of TCC in the CSF is increased
in Guillain-Bark polyradiculitis [22]. These disorders
characteristically are associated with BBB damage. Although the mechanism of C activation in the aforementioned conditions is not known, our observation
that the TCC is assembled already upon extravasation
of plasma into the CSF could basically explain why C
becomes activated in these states. A similar mechanism
may induce C activation in the subarachnoid space
during SAH. Severe ischemic brain damage per se did
not elicit subarachnoid TCC formation, suggesting
that a considerable mixing of serum with CSF may be
needed to cause subarachnoid fluid-phase C activation.
O n the other hand, C activation took place at intraparenchymal sites within the brain infarcts. After contact
of the plasma C system with the ischemic brain tissue,
C activation products appeared to become deposited
on the neural structures. The presence and colocalization of the blood-borne C products C3d and C9 as
well as extravasated leukocytes indicate that BBB damage had occurred in areas of C activation. Furthermore,
the possibility exists that a proportion of C proteins
are produced locally after an ischemic insult.
Fig 5. C9 deposition and granulocyte adhesion 15 hours afier
brain ischemia (Case 1) and C9 staining of normal brain tissue (Case A). (a) Enhanced 0deposition in the neuropil
and on a thickened endothelium of a blood vessel surrounded
by an enlarged perivascular space (Case I ) (X 75 before 2%
reduction). (b) Lack of C9 deposition in normal brain tissue
(Case A). Slight C9 deposition was obsewed on the vascular
endothelium. The intense punctate autoFuorescence represents
lipofiscin particles (X 75 before 2% reduction). (c) Neutrophi1 margination at vascular endothelium in ischemic brain
tissue. Immunoperoxidase staining of the CD15 epitope on
granulocytes; Case 1 (X 400 b4o.e 2% reduction).
Although clearly a prerequisite, the diffusion of
plasma components across the BBB alone is not sufficient to induce a positive staining reaction for C9,
which results from stable incorporation of C9 into the
TCC [21]. Circulating C9 normally exists as a free
globular molecule and is unable to deposit on membranes without prior formation of the Cjb-8 complex.
Immunofluorescence staining for IgG, which leaks out
from blood vessels during BBB damage, did not show
similar local deposition as seen for C9. Deposited C9
may represent fully or partially assembled MAC that
has inserted into the cell membrane or soluble SCSb9 that has subsequently become deposited on the brain
tissue, eg, through interactions via vitronectin (Sprotein) or clusterin. The positive staining for the C5b9 neoantigen in more limited areas confirms that indeed the terminal C pathway has been activated and
resulted in MAC assembly in discrete infarct regions
(see Fig 4).
Our observations of robust C activation in SAH are
in accordance with a previous study that indicated
cleavage of the earlier components (C4, C3) of C in
the subarachnoid space during the first days after SAH
[23]. The decrease in the SC5b-9 concentration over
the next week indicates that assembly of TCC in the
subarachnoid space occurs shortly after SAH, when the
rate of hemolysis of subarachnoid RBCs also peaks
[24]. The lack of C3b deposition on the subarachnoid
RBCs suggests that erythrocyte membranes neither initiate nor become a target for C activation. It is thus
unlikely that C would directly contribute to the rapid
lysis of the subarachnoid RBCs in SAH, a phenomenon that is not well explained [24J. This does not,
however, exclude the possibility that surface and membrane alterations on aging RBCs could make these cells
more susceptible to C lysis during the later stages after
SAH [25]. In view of the possibility that the release of
free oxyhemoglobin into the subarachnoid space promotes the vasospasm [26, 271, a decisive factor for the
prognosis after SAH, further studies are essential to establish if the rapid intrathecal TCC assembly in SAH
participates in the release of free hemoglobin or in
other clinically important sequelae. Furthermore, comparison of intrathecal and plasma C activation products
may be a useful addition to differentiate SAH from
artifactual “bloody” spinal taps.
We postulate that the C system, activated by BBB
damage, acts in association with other mediator systems to recruit neutrophils and stimulates their diapedesis and migration through the brain interstitiurn.
The observation that the deposition of C9 was associated with dense neutrophil infiltration suggests that C
activation may participate in formation of the gradient
of neutrophil chemotactic factors (Fig 6b). C activation
releases C5a and its derivative CSa-desArg, which are
potent mediators of chemotaxis and activation of in-
Iindsberg et al: Complement Activation in BBB Damage
Fig 6. C9 deposition in ischemic brain tissue 2.2 days afier the onset of stroke in the occipital lobe. Case 4. (a) C9 deposition in
the vasculature and neuropil below the infarct core (X 75). (b) A close-up view of the C9 gradient at a perivascular location.
The white corpuscles in the parenchyma represent emigrated injammatoy cells (X 350). (c) Exposed elongated neuron axons
showing C9 deposition in the proximity of the infarct lesion (X 350). (d) Widely increased C9 staining within a focal infarct
lesion surrounding a larger vessel (X 75).
flammatory cells [8, 281. Deposition of C3 cleavage
products on cell membranes can mark them for recognition by receptors on phagocytes. It is important that
many experimental studies have shown that pharmacologic inhibition of C activation decreases both the extent of tissue injury and neutrophil infiltration in myocardial infarction [ I l , 13, 141. Viewed as a harmful
event also in acute brain infarction [29], infiltration of
neutrophils in the CNS might be inhibited by interventions blocking C activation. In SAH, C activation
coincides temporally with the development of' the socalled aseptic hemogenic meningitis [30],and may be
a factor in the inflammatory response as well as in the
vasospasm or ischemic deterioration in SAH. Although
594 Annals of Neurology
Vol 40
No 4 October 1936
stimulation of the tissue tropism of leukocytes and
phagocytosis may be the most important effects of C
activation in cerebrovascular disorders, additional
pathophysiologic consequences could include damage
or stimulation of neuronal and glial cells by direct or
sublytic MAC attack.
The mechanism of' C activation following BBB damage remains the subject of further studies. It is possible
that C activation occurs by default through the alternative C pathway when the contact of plasma with CSF
of neural elements diverts the balance between activation and inhibition toward activation. Direct contact
of plasma with CNS myelin has earlier been shown to
lead to antibody-independent activation of the classical
both ischemic and hemorrhagic human cerebrovascular
diseases, where extravasation typically occurs and leads
to C activation in vivo. Our observations suggest that
nonimmune activation of C and the inflammatory response it generates are important also in a variety of
diseases in which the barrier between plasma and the
extracellular spaces of the CNS is broken. Chemotaxis
and migration of phagocytes may be the major inflammatory events induced by activation of C in injured brain. The potential consequences and clinical
significance of our findings should stimulate further
studies of the role of complement in CNS diseases where
the host response aggravates the extent of tissue injury.
This study belongs to a series of postmortem stroke studies designated collectively the Helsinki Stroke Study (HSS). This study was
financially supported by the Academy of Finland, the Sigrid Juselius
Foundation, the Paulo Foundation, the Maire Taponen Foundation, the University of Helsinki, and the Helsinki Universicy Central Hospital.
Fig 7. C9 deposition in infai*cted cerebellar cortex 8 days
ajer the onset of stroke. Case 8. (a) A distinct ldyer o f C9
deposition in a laminar infarction zone at the junction
between the molecular and granular cell layers (X 75 befDre
2% reduction). The densely C9 positive structures are axon
baskets surrounding the Purkinje cell bodies (inset, X 350
befare 2% reduction). (6) Contralateral cerebellum devoid of
infarction and specijic C9 staining (X 75 be$re 2% reduction).
pathway [6, 71.It is possible that the regulatory balance of the alternative pathway amplification system
is altered in conditions of plasma extravasation into
the CNS compartments. The cell-surface regulators
(CD46, CD55, CD59) are absent or radically diminished, and the fluid phase control (factor H) under
conditions of altered pH or ionic strength may decline
as well. These changes would lead to increased assembly of C3/C5 convertase. Other possible mechanisms
leading to C activation include activation of the bloodclotting cascade by exposure of plasma to tissue collagen, fibrinolysis, and generation or release of proteolytic enzymes from the affected tissue.
In conclusion, we have provided evidence that the
terminal pathway of the blood C system becomes activated following damage of the BBB. This basic phenomenon is likely to be a pathophysiologic factor in
1. Miiller-Eberhard HJ. The membrane attack complex of complement. Annu Rev Immunol 1986;4:503-528
2. Brooks BR, Hirsch RL, Coyle PK. Cellular and humoral immune responses in human cerebrospinal fluid. In: Wood J H ,
ed. Neurobiology of cerebrospinal fluid. New York: Plenum,
1983:300-30 1
3. Silverberg D H , Manning MC, Schreiber AD. Tissue culture
demyeiination by normal human serum. Ann Neurol 1984;15:
575 -5 80
4. Scolding NJ, Morgan BP, Houston A, et al. Normal rat serum
cytotoxicity against syngeneic oligodendrocytes. J Neurol Sci
5. Wren DR, Noble M. Oligodendrocytes and oligodendrocytel
type-2 astrocyte progenitor cells of adult rats are specifically
susceptible to the lytic effects of complement in the absence
of antibody. Proc Natl Acad Sci USA 1989;86:9025-9029
6. Cyong J-C, Witkin SS, Reiger B, et al. Antibody-independent
complement activation by myelin via the classical complement
pathway. J Exp Med 1982;155:587-598
7. Vanguri P, Koski CL, Silverman B, et al. Complement activation by isolated myelin: activation of the classical pathway in
the absence of myelin-specific antibodies. Proc Natl Acad Sci
USA 1982;79:3290-3294
8. Morgan BP. Complement. Clinical aspects and relevance to
disease. San Diego: Academic Press, 199O:l-185
9. Schafer H, Mathey D, Hugo F, et al. Deposition of the terminal C5b-9 complement complex in infarcted areas of human
myocardium. J Immunol 1986;137:1945-1949
10. Rus HG, Niculescu F, Vlaicu R. Presence of C5b-9 complement complex and S-protein in human myocardial areas with
necrosis and sclerosis. Immunol Lett 1987;16:15-20
11. Crawford MH, Grover FL, Kolb WP, et al. Complement and
neutrophil activation in the pathogenesis of ischemic myocardial injury. Circulation 1988;78:1449-1458
12. Vakeva A, Laurila P, Meri S. Loss of expression of protectin
(CD59) is associated with complement membrane attack complex deposition in myocardial infarction. Lab Invest 1992;67:
608-61 6
13. Weisman HF, Barrow T, Leppo MK, et al. Soluble human
complement receptor cype 1: in vivo inhibitor of complement
Lindsberg et al: Complement Activation in BBB Damage 595
suppressing post-ischemic myocardial inflammation and necrosis. Science 1990;249:146-151
14. Smith EF 111, Griswold DE, Egan JW, et al. Reduction of
myocardial reperfusion injury with human soluble complement
receptor type 1 (BRL 55730). Eur J Pharmacol 1993;236:47748 1
15. Kalli KR, Hsu P, Fearon DT. Therapeutic uses of recombinant
complement protein inhibitors. Springer Semin Immunopathol
1994;15:417-43 1
16. Morgan BP, Meri S. Membrane proteins that protect against
complement lysis. Springer Semin Immunopathol 1994;15:
17. Fisher CM, Kistler JP, Davis JM. Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized
tomographic scanning. Neurosurgery 1980;6:1-9
18. Hunt WE, Hess RM. Surgical risk as related to time of intervention in the repair of intracranial aneurysms. J Neurosurg
19. Jennett B, Bond M. Assessment of outcome after severe brain
damage. Lancet 1975;1:480-484
20. Tourtellatre WW, Shorr KJ. Cerebrospinal fluid. In: Youmans
JR, ed. Neurological surgery. Philadelphia: WB Saunders,
21. Compston DAS, Morgan BP, Campbell AK, et al. Immunocytochemical localization of the terminal complement complex
in multiple sclerosis. Nruropathol Appl Neurobiol 1989;15:
596 Annals of Neurology Vol 40 No 4 October 1996
22. Sanders ME, Koski CL, Robbins D, et al. Activated terminal
complement in CSF in Guillain-Barre syndrome and multiple
sclerosis. J Immunol 1986;1364456-4459
23. Kasuya H, Shimizu T. Activated complement components C3a
and C4a in cerebrospinal fluid and plasma following subarachnoid hemorrhage. J Neurosurg 1989;71:741-746
24. Fishman RA. Cerebrospinal fluid in cerebrovascular disorders. In:
Barnett HJM, ed. Stroke: Pathophysiology, diagnosis and management. New York: Churchill Livingsrone, 1992: 103- 1 10
25. Peterson JM, Kwun B-D, Teramura A, et al. Immunological
reaction against the aging human subarachnoid erythrocyte. J
Neurosurg 1989;71:718-726
26. Osaka K, Handa H, Okamoto S. Subarachnoid erythrocytes
and their contribution to cerebral vasospasm. In: Wood J H ,
ed. Neurobiology of cerebrospinal fluid. New York: Plenum,
27. Takenaka K , Kassell N F , Foley PL, et al. Oxyhemoglobininduced cytotoxicity and arachidonic acid release in cultured
bovine endocheld cells. Stroke 1993;24:839-846
28. Kajita T, Hugli TE. C5a-induced neutrophilia. A primary humoral mechanism for recruitment of neutrophils. Am J Pathol
29. Kohaneck P M , Hallenbeck JM. Polymorphonuclear leukocytes
and monocyreslmacrophages in the pathogenesis of cerebral
ischemia and stroke. Stroke 1992;23:1367-1379
30. Jackson IJ. Aseptic hemogenic meningitis. Arch Neurol Psychiatry 1949;62:572-589
Без категории
Размер файла
1 383 Кб
central, complement, nervous, following, bloodцbrain, activation, damage, system, barriers, man
Пожаловаться на содержимое документа