Gelatinase B and TIMP-1 are regulated in a cell- and time-dependent manner in association with...

Gelatinase B and TIMP-1 are regulated in a cell- andtime-dependent manner in association with neuronaldeath and glial reactivity after global forebrain ischemia

Santiago Rivera,1 Crystel Ogier,1 JeÂroÃme Jourquin,1 Serge Timsit,2,* Arkadius W. Szklarczyk,3 Karen Miller,4

Andrew J. H. Gearing,4 Leszek Kaczmarek3 and Michel Khrestchatisky1

1IFR Jean Roche. FaculteÂ de MeÂdecine Nord. Bd Pierre Dramard 13916 Marseille Cedex 20, France2INSERM U29, Bd de Port Royal, 75014 Paris, France3Department of Molecular & Cellular Neurobiology, Nencki Institute, 02±093 Warsaw, Pasteura 3, Poland4British Biotech Pharmaceuticals Ltd, Department of Biology, Watlington Road, Cowley, Oxford, OX4 5LY, UK

Keywords: matrix metalloproteinase (MMP), neurodegeneration, in¯ammation, hippocampus

Abstract

Matrix metalloproteinases (MMPs) belong to a large family of endopeptidases that regulate the pericellular environment through

the cleavage of protein components of the extracellular matrix, membrane receptors and cytokines. MMP activity is controlled by

the multifunctional tissue inhibitors of metalloproteinases (TIMPs). Proteases and their inhibitors are critically involved indevelopmental and pathological processes in numerous organs, including the brain. Global transient cerebral ischemia induces

selective delayed neuronal death and neuroin¯ammation. We compared, in discrete vulnerable and resistant areas of the

ischemic rat hippocampus, the kinetics and cellular distribution of gelatinase B and its principal inhibitor TIMP-1 and we assessed

by in situ zymography, the net gelatinolytic activity at the cellular level. We show that gelatinases are expressed and active inneurons, suggesting that MMPs play a role in maintaining neural homeostasis. In the ischemic rat brain, expression and activity

of gelatinase B, and expression of TIMP-1 are altered in a time-, region- and cell-dependent manner. Gelatinase B is induced

®rst in reactive microglia and subsequently in reactive astrocytes. In situ, increases in gelatinase activity accompanied theprogression of neuronal death and glial reactivity. Our results suggest that MMPs and TIMPs are involved in cell viability and

tissue remodelling in the ischemic brain, and reinforces the idea that the MMP/TIMP system contributes both to neuronal demise

and tissue repair in the context of glial reactivity.

Introduction

Matrix metalloproteinases (MMPs) belong to a rapidly growing

family ± to date more than 20 members have been identi®ed ± of

Zn2+-dependent endopeptidases, including, gelatinases, collagenases,

stromelysins and membrane-type MMPs (MT-MMPs) (reviewed in

Murphy et al., 1999; Seiki, 1999). MMPs are expressed as zymogens

that are activated upon proteolytic cleavage of a propeptide region.

MMPs cleave protein components of the extracellular matrix (ECM),

but also process a number of cell surface targets such as receptors,

pro-in¯ammatory cytokines and other soluble proteins. The proteo-

lytic activity of MMPs is controlled by the tissue inhibitors of

metalloproteinases (TIMPs), a family of secreted multifunctional

proteins that comprises four members (TIMP-1 to TIMP-4), with 30±

40% sequence similarity (Edwards et al., 1996; Greene et al., 1996).

All TIMPs inhibit the active forms of most MMPs by forming tight

noncovalent 1 : 1 complexes with them.

TIMPs and MMPs are considered to be instrumental in develop-

mental and pathological/in¯ammatory processes involving cell death

and tissue remodelling (Werb, 1997; Coussens et al., 2001). Recently,

neurobiologists have gained interest in the potential implication of

MMPs and TIMPs in neuroplasticity and neuropathology (reviewed

in Rivera & Khrestchatisky, 1999; Yong et al., 2001). Early work by

Rosenberg and collaborators (reviewed in Mun-Bryce & Rosenberg,

1998) and subsequently by others (Romanic et al., 1998; Gasche et al.,

1999; Planas et al., 2000) demonstrated that MMPs, principally

gelatinases A and B (MMP-2 and MMP-9, respectively), and TIMP-1

(Wang et al., 1998) were up-regulated in rat brain after focal

ischemia, in close association with the breakdown of the blood brain

barrier (BBB) and the neuroin¯ammatory response. Seizures induce

the expression of gelatinase A and B (Zhang et al., 1998) and of

TIMP-1 (Nedivi et al., 1993) in rat brain, while the mRNA levels of

several MMPs and TIMPs are increased in the mouse brain after

experimental autoimmune encephalomyelitis (EAE) (Pagenstecher

et al., 1998). Cell culture studies have demonstrated that MMPs and

TIMPs are induced in glial cells activated by a number of trophic or

in¯ammatory agents (Muir, 1994; Gottschall et al., 1995; Giraudon

et al., 1997). While these studies demonstrate that the expression of

MMPs and TIMPs is regulated by neuronal activity or in¯ammatory-

driven processes, information on their regional and cellular localiza-

tion in the brain remains scarce. We have shown in vivo that kainate-

induced excitotoxic seizures rapidly trigger TIMP-1 expression in

neurons and subsequently in astrocytes, and that TIMP-1 expression

is sustained in areas of neuronal degeneration and tissue remodelling

Correspondence: Dr Santiago Rivera, as aboveE-mail: [email protected]

*Present address, Neurotech 4, rue Pierre Fontaine 91000 Evry, France

Received 30 July 2001, revised 6 November 2001, accepted 16 November 2001

European Journal of Neuroscience, Vol. 15, pp. 19±32, 2002 ã Federation of European Neuroscience Societies

(Rivera et al., 1997). These data highlight the relevance of MMPs and

TIMPs in the pathological brain after excitotoxic insult. In this

context, the induction of global transient cerebral ischemia in rats

induces selective neuronal death and neuroin¯ammatory processes in

brain structures such as the hippocampus. We investigated the

kinetics and cellular distribution of gelatinase B and its endogenous

inhibitor TIMP-1, as well as the net gelatinolytic balance between

proteases and inhibitors by in situ zymography. Our results indicate

that in the ischemic brain proteolytic activity, gelatinase B and TIMP-

1 expression are regulated in a time-, region- and cell-dependent

manner.

Materials and methods

Four vessel occlusion forebrain ischemia and histology

Experiments involving animals were approved by the French Ethical

Committee (statement no. 04223). Transient forebrain ischemia was

induced by four-vessel occlusion, as described previously by

Pulsinelli & Brierley (1979). Male Wistar rats (300±320 g, Charles

River, France) were anaesthetized with intraperitoneal 3% chloral

hydrate (300 mg/kg; Merck, Darmstadt, Germany), vertebral arteries

were electro-coagulated, and clamps were set on both carotid arteries.

Twenty-four hours later, carotid arteries were reversibly clamped for

20 min in spontaneously ventilating rats. Only rats with unrespon-

siveness and loss of the righting re¯ex during carotid clamping were

kept for further analysis and killed at different time points after

reperfusion. Rats with only electrocoagulated vertebral arteries were

used as control (sham operated rats). For immunohistochemistry and

for combined in situ hybridization-immunohistochemistry, rats were

deeply anaesthetized with chloral hydrate, intracardially perfused

with 400 mL of 4% paraformaldehyde (PFA, Sigma, St Louis, MO)

in 100 mM phosphate buffer; brains were removed rapidly, post®xed

for 24 h in 4% PFA and cryoprotected 24 h in 20% sucrose. Twenty-

four micrometer thick coronal sections were prepared with a freezing

microtom and collected into phosphate-buffered saline (PBS) with

0.05% sodium azide (Sigma). For in situ hybridization and in situ

zymography non perfused brains were directly frozen in isopentane at

±70 °C. Cryostat coronal serial sections 15 and 20 mm thick were

collected for in situ hybridization and for in situ zymography,

respectively, and stored at ±80 °C. We used cresyl violet (Sigma)

staining to evaluate neurodegeneration and gliosis in sections

adjacent to those used for in situ hybridization or immunohistochem-

istry studies.

Gel zymography

Tissue extractions for gel zymography were performed according to

the method previously reported by Weeks et al. (1976) with minor

modi®cations. Animals were anaesthetized with pentobarbital

(Sigma), decapitated, brains were rapidly removed and hippocampi

dissected on a cold plate. Samples were homogenized in a buffer

(1 : 19 w/vol) containing 10 mM CaCl2 (Sigma) 0.25% Triton X-100

(Sigma), and were centrifuged at 6000 3 g for 30 min; the super-

natant (triton soluble fraction) contained essentially cytoplasmic

proteins. The pellet, containing ECM proteins was resuspended in a

buffer containing 50 mM Tris pH, 7.4 and 100 mM CaCl2 and heated

for 15 min at 60 °C in order to extract proteinases from their bound

proteins, then centrifuged at 10 000 3 g for 30 min at 4 °C. The

supernatant, enriched in solubilized metalloproteinases, and the triton

soluble fraction were washed in 60% ethanol, centrifuged at

15 000 3 g for 5 min and the pellet was solubilized in 2% SDS

sample buffer. These fractions were subjected to electrophoresis

using a MiniBlot system (Bio-Rad, Hercules CA), in sodium dodecyl

sulphate (SDS)-polyacrylamide (Bio-Rad) gels 7.5% containing 0.5%

gelatine (Sigma) in nondenaturing, nonreducing conditions. Gels

were washed twice for 15 min in 2.5% Triton X-100 to remove SDS,

and incubated for 72 h in 50 mM Tris pH 7.5, 10 mM CaCl2, 1 mM

ZnCl2, 1% Triton X-100 and 0.02% sodium azide at 37 °C. Gels

were then stained with 0.1% Coomassie blue G-250 (Bio-Rad) for 3 h

in 40% 2-propanol and destained with a solution containing 5% acetic

acid until clear bands of gelatinolysis appeared on a dark background.

Gels were dried, digitized using a Samba/200 S image analysis

system (TITN Alcatel, France) and optical densities assessed with the

NIH Image software. Some zymogram gels were incubated with

1 mM 1,10-0-phenanthroline, a broad spectrum inhibitor of metallo-

proteinases. For each zymogram, equal volumes of the same samples

were subjected in parallel to SDS-PAGE and gels were stained with

Coomassie blue as a loading control.

In situ hybridization

In vitro transcription of 35S-UTP (Amersham Pharmacia, Saclay,

France) labelled TIMP-1 riboprobes was performed from linearized

pBluescript plasmids using an in vitro transcription kit (Promega

Madison, WI) with T7 (Amersham Pharmacia) and T3 (Boheringer

Mannheim, Mannheim, Germany) RNA polymerases for sense and

antisense probes, respectively. The plasmid (kindly provided by the

late Dr Yoav Citri) carried a 700-bp rat cDNA containing part of the 5¢untranslated region and the entire coding region TIMP-1. On Northern

blots with brain RNA, this probe recognized only one band corres-

ponding to the molecular size of TIMP-1 mRNA (Nedivi et al., 1993).

Frozen sections were brought to room temperature and ®xed in cold

4% PFA in 100 mM phosphate buffer for 30 min, then rinsed in

glycine-phosphate buffer, acetylated, rinsed, dehydrated, delipidated,

and incubated overnight at 50 °C with 1 3 106 cpm/100 mL of

antisense or sense riboprobes. Tissue was rinsed twice for 30 min at

60 °C in 4 3 saline±sodium citrate (SSC), then treated with 20 mg/mL

RNAse A (Boehringer Mannheim), and ®nally washed in increasingly

stringent conditions up to 0.1 3 SSC at 60 °C during 30 min. All

tissue sections were processed for both ®lm (Biomax, Kodak,

Rochester, NY) and emulsion autoradiography (NTB2, Kodak), with

exposure times of 3±5 days and 3±4 weeks, respectively. Following

development of emulsion autoradiograms, the sections were counter-

stained with cresyl violet and mounted with Permount.

Densitometric analysis of mRNA levels on ®lm autoradiograms

were performed using the Samba/200 S image analysis system.

Multiple measures were obtained from at least four sections per brain.

Double labelling procedures combining in situ hybridizationand immunohistochemistry

In order to identify precisely the cell types expressing TIMP-1

mRNA, we combined in situ hybridization with a TIMP-1 35S-cRNA

probe and immunohistochemistry, with either a rabbit polyclonal

antiglial ®brillary acidic protein (GFAP, DAKO, Trappes, France) or

a mouse monoclonal antimicrotubule associated protein 5 (MAP-5,

Sigma) antibody that recognizes astrocytes and neurons, respectively.

Using microtome-cut ¯oating sections, in situ hybridization was

performed as indicated above except that the dehydration and

delipidation steps were carried out after the immunohistochemistry

procedure. After the last wash in 0.1 3 SSC, tissue was rinsed

3 3 10 min in PBS pH 7.4, preincubated for 1 h in a PBS blocking

solution containing 0.2% gelatin, 3% normal goat serum (Vector

Laboratories, Burlingame, CA) and 0.1% Triton X-100, followed by

overnight incubation at 4 °C with either an anti-GFAP (1 : 300) or an

anti-MAP-5 (1 : 2000) in PBS containing 0.2% gelatin and 1%

20 S. Rivera et al.

ã 2002 Federation of European Neuroscience Societies, European Journal of Neuroscience, 15, 19±32

normal goat serum. The following day, tissue was rinsed in PBS,

incubated 1 h at room temperature with a biotinylated goat antirabbit

or goat antimouse antibody (1 : 400, Vector), rinsed in PBS,

incubated 1 h in avidin-biotin-peroxidase (1 : 200, ABC kit,

Vector), developed in 0.05 M Tris buffer containing 0.05%

diaminobenzidine (Sigma) and 0.1% H2O2 (Sigma). Sections were

mounted, dehydrated, delipidated and, ®nally, processed for emulsion

autoradiography as indicated above.

Immunohistochemistry

For gelatinase B and TIMP-1 immunohistochemistry, PFA ®xed

¯oating sections were rinsed in PBS. Endogenous peroxidase activity

was inhibited with 0.1% H2O2 for 20 min. Non-speci®c binding sites

were blocked by preincubating tissue for 1 h at room temperature in

PBS containing 0.1% Triton X-100, 0.2% gelatin and 3% normal

goat serum. Sections were then incubated at 4 °C overnight in a PBS

solution containing either a mouse monoclonal antigelatinase B

(clone 2C10 from British Biotech Pharmaceuticals, Oxford, UK) or a

rabbit polyclonal anti-TIMP-1 (Chemicon, Temecula, CA) diluted

1 : 300 in PBS, 0.1% Triton, 0.2% gelatin, 1% normal goat serum.

Tissue was processed as described above with goat antimouse or goat

antirabbit secondary antibodies, developed with diaminobenzidine

(DAB), intensi®ed with 0.2% ammonium nickel (II) sulphate

(Sigma), and ®nally dehydrated and mounted.

For astrocyte and microglia immunostaining, sections were incu-

bated with a rabbit polyclonal anti-GFAP (1 : 300, DAKO) or a

mouse monoclonal anti-OX-42 (1 : 300, Serotec, Varilhes, France),

respectively. Staining with TIMP-1, MMPs and OX-42 antibodies

was revealed with nickel intensi®cation.

Sections incubated in the absence of the primary antibody, with

rabbit or mouse IgGs or with the preabsorbed gelatinase B antibody,

were not immunoreactive.

Double labelling TRITC/FITC coupled antibodies

In order to identify the glial cell types expressing TIMP-1 and

gelatinase B, double ¯urorescent immunohistochemistry was carried

out. We used the same primary anti-TIMP-1 and antigelatinase B

antibodies mentioned above. Glial cells were characterized using a

mouse monoclonal anti-GFAP antibody (1 : 500, Sigma) or isolectin

B4 (BSI) from Bandeiraea simplicifolia directly coupled to FITC

¯uorophore (1 : 100, Sigma) for immunodetection of astrocytes and

microglia, respectively. Immunolabelling was revealed using either

antirabbit or antimouse secondary antibodies (1 : 400, Vector)

coupled to either FITC or TRITC ¯uorophores. Sections were rinsed

1 h in PBS, preincubated as indicated above and then incubated

overnight at 4 °C with the primary antibodies. Tissue was rinsed and

incubated for 2 h at room temperature with secondary antibodies

coupled to the ¯uorophores, rinsed and mounted in Mowiol (Sigma).

The OX-42 antibody and BSI recognize resident microglia and

invading macrophages. Hence, in this report the term microglia

includes both cell types.

In situ zymography

We adapted a new in situ zymography method (Oh et al., 1999) to

localize net gelatinolytic activity in brain sections. The modi®cations

introduced in our procedure increase resolution and prevents signal

from rapid fading. Frozen non®xed brain sections were thawed and

incubated for 48 h at 37 °C in a humid chamber in 100 mL of

reaction buffer containing 100 mg/mL of FITC-labelled DQ-gelatin

(EnzCheck collagenase kit, Molecular Probes, Eugene, OR) that is

intramolecularly quenched. Gelatin-FITC is cleaved by tissue

proteases and yields peptides whose ¯uorescence is representative

of net proteolytic activity. Coverslips were removed, sections rinsed

in PBS and ®xed in cold PFA for 30 min, then mounted in Mowiol,

and observed using ¯uorescence microscopy. In each experiment,

some sections contained metalloproteinase inhibitors in the reaction

buffer at the following concentrations: 1 mM phenanthroline

(EnzCheck collagenase kit, Molecular Probes), 100±500 ng/mL

human TIMP-1 or TIMP-2 (Chemicon). Other sections were incuba-

ted with a cocktail of nonmetalloproteinase inhibitors containing

1 mM AEBSF (serine protease inhibitor), 10 mM E64 (cysteine

protease inhibitor), 100 mM leupeptine (serine protease inhibitor)

from ICN (Costa mesa, CA) and pepstatine (acidic protease inhibitor)

from Sigma. Sections incubated without DQ-gelatin were not

¯uorescent.

Statistical analysis

All studies were performed with at least three rats per time point.

Analysis of variance (ANOVA) followed by Student's t-test was used

for statistical comparisons among experimental groups.

Results

Histopathological changes after ischemia

The ®rst signs of neuronal degeneration, characterized by the loss of

the typical patchy pattern of stained Nissl bodies, were observed one

day postischemia in some hilar interneurons. Three days after

ischemia, cell damage extended to the CA1/subiculum sub®eld,

where numerous pyramidal neurons presented pyknotic nuclei and

many others were in an intermediate stage of degeneration with

characteristic loss of Nissl staining and elongation of cell bodies

(Fig. 1C). Six days after ischemia (Fig. 1D), most pyramidal CA1

neurons had disappeared, while no signs of degeneration were

observed in pyramidal CA3 neurons and dentate granule cells.

With respect to glial changes after ischemia, characteristic reactive

microglia, presenting retracted processes and hypertrophic cell bodies

was already present 24 h after ischemia in region CA1 (Fig. 1F), in

the hilus and in the striatum. Increased GFAP immunostaining

associated with reactive astrocytes became apparent 3 days post-

ischemia (Fig. 1K). By this time, glial cells invaded the neuronal cell

layer of region CA1. Both microglial and astrocytic immunostaining

were more intense at 6 days, essentially in vulnerable CA1 (Fig. 1H

and L) and hilar areas.

Gel zymography in the hippocampus after ischemia

The tissue extraction method described here was used previously to

extract collagenases from the uterus with a high yield (Weeks et al.,

1976). We have applied the technique to brain samples in order to

increase sensitivity of MMP detection, as proteinases are enriched in

the triton nonsoluble fraction. Faint bands of gelatinase activity were

detected in the triton soluble fraction (results not shown), but most of

the gelatinolytic activity was found in the triton nonsoluble fraction

(Fig. 2). In the latter, bands of gelatinolytic activity at 99, 95 and

66 kDa were present in the hippocampus of sham operated animals.

The 99 and 95 kDa bands were suggestive of the active and latent

forms of gelatinase B, respectively, while the 66 kDa band matches

the molecular weight of the active form of gelatinase A. Levels of the

latter were nearly twofold the levels of any of the upper bands. After

ischemia, changes in gelatinolytic activity were conspicuous with

respect to controls. Levels of pro-gelatinase B signi®cantly increased

at 3 days postischemia (225%) and reached a maximum value at

6 days (306%), whereas levels of the active form remained

unchanged at all time points. Levels of the 66 kDa band increased

Gelatinase B and TIMP-1 after global brain ischemia 21


161% at 3 and 6 days postischemia. A band of »250 kDa, absent in

control animals, was induced in the ischemic hippocampi at 3 and

6 days postischemia. Diffuse bands of gelatinolysis were observed

slightly above (72 kDa) and below (62 kDa) the gelatinase A band. In

all cases, gelatinolytic activity was abolished by incubating the gels

with 1 mM phenanthroline (not shown).

Gelatinase B immunohistochemistry in the hippocampus afterischemia

As shown in Fig. 3A, in control hippocampi, gelatinase B

immunolabelling was found in pyramidal neurons and granule cells

of the dentate gyrus and only occasionally, cells morphologically

identi®ed as astrocytes appeared weakly immunolabelled.

Microscopic observation at higher magni®cation revealed that

gelatinase B immunoreactivity was preferentially pericellular

(Fig. 3B).

In ischemic animals, a decrease in neuronal gelatinase B

immunoreactivity was observed in the main cell layers of the

hippocampus 24 h after reperfusion. In contrast, increases in

immunoreactivity appeared associated with cells scattered in the

CA1 region and with a diffuse distribution outside the cell bodies

(Fig. 3C and D). Immunoreactivity remained unchanged in neurons

of the neocortex and slightly increased in large neurons of the central

and ventral regions of the striatum (results not shown). In the

CA1 area, 3 days after ischemia, the decrease in gelatinase B

immunoreactivity was clear in subicular and pyramidal neurons, but

FIG. 1. Global ischemia induces delayed neuronal death and glial reactivity in the hippocampus. Bright-®eld photomicrographs showing cresyl violet, OX-42and GFAP labelling in region CA1 of the hippocampus of control animals (CTL, sham operated at 6 days) and ischemic (Isch) rats at 1, 3 and 6 days afterischemia. Note that pyknotic nuclei in the pyramidal neurons (p) appear at 3 days after ischemia, while at 6 days postischemia most neurons have alreadydisappeared (D). Microglial reaction is observed at 1 day after ischemia (F). The density and intensity of OX-42 immunoreactivity increase progressively at 3(G) and 6 (H) days postischemia. Astrocyte reactivity is not apparent until 3 days after ischemia (K), and is higher at 6 days (L). Scale bar, 50 mm.

22 S. Rivera et al.


dramatically increased in glial cells within the neuronal layer and in

both strata radiatum and oriens. At this time point, the reduction in

gelatinase B immunostaining in the CA1 pyramidal cell layer

contrasted with the strong labelling observed in the CA3 and the

granule cell layer (Fig. 3E). Six days after ischemia, immunostaining

remained strong in the CA3 and granule cell layers, but also in glial

cells of astrocytic morphology. In these cells, labelling was most

intense in the lesioned subiculum/CA1 areas and decreased gradually

towards the resistant CA3 sub®eld (Fig. 3F). Double labelling

experiments (Fig. 4) revealed that at least three different cell types,

including neurons, astrocytes and microglia, were immunoreactive

with the gelatinase B antibody in a time- and region-dependent

manner. Three days after ischemia, most gelatinase B immuno-

reactivity was associated with resistant neurons and with BSI positive

cells in the CA1 area (Fig. 4A and B). The intensity of labelling and

density of gelatinase B-positive cells was as follows:

CA1 = hilus > CA3 > stratum granulosum and molecular layer of

the dentate gyrus. Six days after ischemia (Fig. 4C and D), gelatinase

B immunoreactivity was no longer associated with BSI positive cells;

instead, gelatinase B immunostaining was associated with cells

morphologically identi®ed as astrocytes.

TIMP-1 mRNA expression in the ischemic hippocampus

TIMP-1 mRNA was constitutively expressed at low levels in control

brains, mainly in the pyramidal and granule cell layers of the

hippocampus (Fig. 5A). During the ®rst hours postischemia, we

observed a marked but transient increase in radiolabelling, essentially

con®ned to the granule cell layer. A slight increase in mRNA levels

was detected in the stratum granulosum of the dentate gyrus as early

as 1 h postischemia. Levels of labelling were highest at 4 h, with a

®vefold increase compared to sham operated controls (Fig. 5B), and

declined to baseline levels by 24 h (Fig. 5C). Three days after

ischemia, a second wave of TIMP-1 mRNA induction was restricted

to the vulnerable CA1 sub®eld (Fig. 5D). The signal was intense

within the pyramidal cell body layer and sparse and diffuse in the

strata oriens and radiatum (Fig. 5G). Six days after ischemia,

increased TIMP-1 mRNA levels were observed in the entire

pyramidal cell layer and in the hilus (Fig. 5E). Besides neuronal

and glial elements, strong labelling was observed 4 h after ischemia

(Fig. 5B) in the ependymal cell layer of the third and lateral

ventricles. Radiolabelling in these regions remained elevated as

compared to controls at all times studied. As shown in Fig. 5H,

combined in situ hybridization and immunohistochemistry demon-

strated that at 3 days postischemia, GFAP positive cells identi®ed as

reactive astrocytes were labelled with the TIMP-1 probe. These cells

were found within the pyramidal cell layer and in the strata radiatum

and oriens of the CA1 region, accounting for the sparse and diffuse

labelling mentioned above. All silver grain clusters outside the

pyramidal cell layer were associated with GFAP positive cells. Due to

the loss of speci®c neuronal markers in lesioned neurons at this stage,

we were not able to associate TIMP-1 mRNA labelling with MAP-5

immunoreactivity in the CA1 area.

TIMP-1 immunoreactivity in the ischemic hippocampus

TIMP-1 immunoreactivity in control brains was mostly associated

with the main neuronal layers of the hippocampus, and to a lesser

extent with their areas of projection in the outer third of the

dentate molecular layer and in the strata oriens, radiatum and

lucidum (Fig. 6A). One day after ischemia, immunolabelling

increased in the entire hippocampus, in particular in the molecular

layer and in the stratum lucidum (Fig. 6B). Three days after

ischemia, the increase in immunoreactivity was prominent in the

pyramidal CA3 neurons and in the stratum granulosum. At this

time point, immunoreactivity decreased in the pyramidal neurons

of CA1, while glial cells appeared immunoreactive (Fig. 6C).

TIMP-1 immunolabelling was maintained 6 days after ischemia in

both resistant CA3 and dentate neurons, and in glial cells across

the entire hippocampus (Fig. 6B±D). The highest intensity of

FIG. 2. Gelatinolytic activity increases after ischemia. (A) Gel zymography represents changes in gelatinase B and gelatinase A activity 1, 3 and 6 days afterglobal ischemia in the hippocampus. The »250 kDa band induced 3 days after ischemia could correspond to multimeric complexes of gelatinases. (B)Quanti®cation of gelatinolytic activity in the hippocampus shows that gelatinase B and gelatinase A activity increases at 3 and 6 days post ischemia. Valuesrepresent the mean of optical density measurements obtained from three animals per time point and are expressed as a percentage of paired control rats. TheANOVA test, followed by the Student t-test (*P < 0.05) was used.



immunostaining was found in glial cells of the CA1 sub®eld

(Fig. 6E), identi®ed as astrocytes in double labelling experiments

(Fig. 6G and H). In contrast, we found no colocalization between

TIMP-1 and BSI labelling (not shown).

24 S. Rivera et al.


Net in situ gelatinolytic activity in hippocampus after ischemia

In order to determine whether net gelatinolytic activity was also

altered in the ischemic brain, we used in situ zymography in ex vivo

brain slices. We observed ¯uorescence signal in control brains,

principally associated with neurons and blood vessels in different

brain areas. In the hippocampus (Fig. 7A), labelling was important in

the pyramidal and granule cell layers, weak in interneurons and white

matter, and virtually absent in glial cells.

One day after ischemia, strong increases in ¯uorescence signal

were detected in hilar cells clearly identi®ed as interneurons

(Fig. 7F). In contrast, no differences were observed in ischemic-

resistant interneurons of other hippocampal regions. Incipient

increases in ¯uorescence were also detected in blood vessels of all

hippocampal sub®elds, in pyramidal cells of CA1 region and in small

cells sparsely distributed in the strata radiatum and oriens, the albium

and the corpus callosum. Three days after ischemia (Fig. 7C), we

observed a high increase in the number and intensity of ¯uorescent

cells within the pyramidal cell layer of region CA1 and in scattered

cells outside the main cell layer, reminiscent of glial distribution.

Labelling became prominent in blood vessels across the hippocam-

pus, and was no longer present in hilar interneurons. This pattern of

labelling was accentuated six days after ischemia (Fig. 7D), with a

more accentuated ¯uorescence signal in the corpus callosum.

In all cases, ¯uorescence associated with neurons and glia was

virtually abolished by phenanthroline (Fig. 8B), and clearly reduced

by TIMP-1 and TIMP-2 (Fig. 8C±E). The effects of the TIMPs were

dose-dependent, as they were more ef®cient at 500 ng/mL than at

100 ng/mL. Phenanthroline and TIMPs showed a rather poor

FIG. 4. Gelatinase B is sequentially expressed in microglia and astroglia after ischemia. Fluorescent photomicrographs showing double labelling combining agelatinase B speci®c antibody (A, C) and lectin B labelling for microglia (B, D), 3 (A, B) and 6 (C, D) days after ischemia in the hippocampus. Note that3 days after ischemia (A, B), most cells are positive for gelatinase B and lectin B (arrows) in the CA1 region; the white asterisk indicates a neuronexpressing gelatinase B that is not labelled by lectin B. At 6 days (C, D), there is no colocalization of these two markers: white arrows indicate gelatinase Bpositive cells with astrocytic morphology in the CA1 layer (C) that are not positive for lectin B (open circles in D). Photomicrographs are representative ofobservations made from three animals per time point. Scale bar, 20 mm.

FIG. 3. The cellular distribution of gelatinase B immunoreactivity changes after ischemia. Bright-®eld photomicrographs showing immunoreactivity to agelatinase B speci®c antibody in coronal brain sections from a control animal (A, B) and from ischemic animals at 1 (C, D), 3 (E, F) and 6 (G, H) days afterischemia in the hippocampus. Note the constitutive expression of gelatinase B in the main neuronal layers of the hippocampus (A). Higher magni®cationmicrophotograph (B) shows that in pyramidal neurons (p), immunoreactivity is periplasmic. At 1 day postischemia, glial and neuronal immunostainingdecreases (C), while cells outside the CA1 pyramidal cell layer become immunoreactive (D). At 3 and 6 days postischemia, CA3 and dentate gyrus (DG)neurons recover control levels of immunostaining (E, G). At these time points, glial cells are strongly labelled and invade the lesioned CA1 and subiculum(arrows, E±H). Photomicrographs are representative of observations made from three animals per time point. Scale bars, 500 mm (A, C, E, F); 50 mm (B, D,F, H).



inhibitory effect on blood vessel gelatinolytic activity. Inversely, a

cocktail of nonmetalloproteinase inhibitors highly reduced labelling

in blood vessels with no signi®cant effects on neuronal or glial cell

¯uorescence (Fig. 8F).

26 S. Rivera et al.


Discussion

This study demonstrates that after transient global forebrain ischemia,

the expression and activity of gelatinases A and B, and the expression

of their physiological inhibitor, TIMP-1 are signi®cantly altered in

different brain areas. Modulation of gelatinase B and TIMP-1

expression occurred in a time-, region- and cell-dependent manner,

suggesting that these regulators of the pericellular environment play

an important role in cell viability and/or tissue remodelling in the

ischemic brain. This contention is further supported by the ®nding

that increases in net gelatinase activity accompanied the progression

of neuronal death and glial reactivity. In addition, the ®nding that

gelatinases are expressed and active in the normal brain reinforces the

idea that MMPs may also play a role in maintaining neural

homeostasis.

Global ischemia induces delayed neuronal death and glialreactivity

As expected, three days after transient global ischemia, irreversible

neuronal damage was observed in the subiculum and in the CA1

sub®elds of the hippocampus. Pathology followed a gradient of

severity that increased along the temporoseptal axis of the brain, in

agreement with previous reports (Schmidt-Kastner & Hossmann,

1988; Timsit et al., 1999). Incipient microglial reactivity was

observed as early as 1 day after ischemia and preceded astroglial

reactivity and neuronal death in the hippocampus and other brain

structures. Three days after ischemia, reactive astrocytes were clearly

observed in the lesioned areas, but also, to a lesser extent, in the

resistant CA3 sub®eld. This spatio-temporal sequence of glial

response is in agreement with the idea that in the early stages of

pathology, activated microglial cells act as cytotoxic effectors that

may contribute to neuronal degeneration (Kreutzberg, 1996), whereas

astrocytes could be involved in subsequent tissue remodelling and

repair (Dusart et al, 1991).

Gelatinase A and gelatinase B activities are increased in theischemic hippocampus

Gel zymography led to the detection of gelatinolytic bands inhibited

by phenanthroline, thereby con®rming metalloproteinase activity.

The molecular weights of these bands, a doublet of »100 kDa and

single band at 66 kDa, suggest they correspond to gelatinase B and

gelatinase A, respectively, in agreement with previous results

(Rosenberg et al., 1994; Zhang et al., 1998). The constitutive

expression of gelatinases in the matrix-enriched fraction suggests that

in the CNS, they are involved in physiological regulation of the

pericellular environment. After ischemia, levels of gelatinases A and

B and of a novel gelatinolytic band around 250 kDa increase

progressively between days 3 and 6, concomitant with the progres-

sion of neuronal lesions. Similar gelatinolytic activity »200 kDa has

been reported by other groups in mouse brain after LPS treatment

(Pagenstecher et al., 2000). These high molecular weight bands may

correspond to complexes of dimerized gelatinase B, favoured by

increases in the gelatinase B/TIMP-1 ratio (Goldberg et al, 1992;

Dubois et al., 1998), or to complexes of gelatinase B and integrins

(Brooks et al., 1996). The temporal sequence of changes in

proteolytic activity suggest that all these forms of gelatinase

contribute both to pathogenesis and to subsequent healing and

remodelling at later stages. A different interpretation of the role of

gelatinases has been proposed in focal ischemia (Rosenberg et al.,

1996), where the sequential expression of gelatinase B and gelatinase

A were associated with early brain damage and with delayed tissue

repair, respectively. The partial disparity observed between both

ischemia models illustrates that the regulation of MMP expression,

and perhaps function, is speci®c of each pathological condition.

Gelatinolytic activity is consistent with a sequential expressionof gelatinase B, ®rst in neurons, then in glia

Immunohistochemistry was used in an attempt to correlate increased

gelatinase B activity with its cellular distribution. The constitutive

expression of gelatinase B in hippocampal neurons further supports

previous experimental data indicating that central neurons are an

important source of MMP-9 in the normal brain (Backstrom et al.,

1996; Vaillant et al., 1999; Vecil et al., 2000). One day after

ischemia, neuronal gelatinase B expression decreased in the

hippocampus, a probable consequence of the transient protein

synthesis impairment that precedes neuronal death in discrete

neuronal populations (Thilmann et al., 1986). However, this decrease

does not translate in decreased activity of gelatinase B in zymograms

perhaps as a result of the incipient induction of gelatinase B in

microglia. Double labelling experiments revealed that gelatinase B

was further increased in reactive microglia in the lesioned CA1 area

3 days after ischemia. Interestingly, at 6 days, gelatinase B was no

longer expressed by microglial cells, but by astrocytes. In the early

stages of neuronal degeneration, microglial gelatinase B could

contribute to the in¯ammatory reaction. The latter hypothesis is

supported by experimental evidence indicating that MMPs transform

immature TNFa into its active form (Gearing et al., 1994; English

et al., 2000). Moreover, MMPs may activate IL-1b by cleavage of its

pro-peptide and subsequently degrade the mature cytokine

(Schonbeck et al., 1998; Ito et al., 1996). In this context, it is

possible that MMPs both amplify and attenuate the in¯ammatory

response depending on the cell type and stage of the pathological

process. Pro-in¯ammatory cytokines such as IL-1b, IL-6 and TNFaregulate the production of MMPs and TIMPs in cultured microglia

and astrocytes (Gottschall & Deb, 1996; Giraudon et al., 1997).

Expression of these modulators of the MMP/TIMP system is

increased in neural cells after global ischemia (Saito et al., 1996).

Taken together, these data strongly suggest that cross-regulation

between MMPs and cytokines linked to ischemic insult contribute to

the neuroin¯ammatory response, which may accentuate excitotoxic

and metabolic insults.

FIG. 5. TIMP-1 mRNA is induced in the adult rat brain after global ischemia. Autoradiographs of coronal brain sections showing in situ hybridization to anantisense (35S)-labelled TIMP-1 cRNA probe in a sham operated control animal (A) and at different times after ischemia: 4 h (B), 1 d (C), 3 d (D, G), 6 d(E); F shows the absence of labelling with the sense probe. Hybridization increases markedly at 4 h postischemia in the stratum granulosum (sg) of thedentate gyrus (DG) and in the ependymal lining of the third and lateral ventricle (3v and lv, respectively). By 3 d postischemia (D), a second wave of TIMP-1 induction occurs in the CA1 region and spreads to the entire hippocampus by 6 d postischemia (E). Note in G the high density of emulsion grains in theCA1 pyramidal layer (p), but also in the strata oriens (so) and radiatum (sr). (H) Bright ®eld photomicrograph showing combined in situ hybridization with(35S)-labelled TIMP-1 cRNA probe and immunohistochemistry with an astroglial speci®c GFAP antibody. Because immunostaining and emulsion grains arein different plans, only the emulsion grains are in focus in this photomicrograph. Note that clusters of emulsion grains in the CA1 area colocalize withastrocytes. Scale bars, 500 mm (A±F); 50 mm (G); 10 mm (H).



FIG. 6. TIMP-1 immunoreactivity increases after ischemia. Bright-®eld photomicrographs showing immunoreactivity to a TIMP-1 speci®c antibody in coronalhippocampal sections (A±F) from a control animal (A) and from ischemic animals at 1 (B), 3 (C) and 6 (D±F) days after ischemia. G and H represent double¯uorescent labelling using TIMP-1 (G) and GFAP (H) speci®c antibodies 6 days after ischemia. TIMP-1 is constitutively expressed in the pyramidal celllayer (p) and the stratum granulosum (sg) of the dentate gyrus (DG), but also in the stratum radiatum (sr) and in the molecular layer (ml). TIMP-1immunostaining increases 1 day after ischemia in the entire hippocampus, in particular in the distal part of the dentate gyrus, the molecular layer and thestratum lucidum (sl). At 3 days postischemia, immunostaining is most intense in the molecular layer and the stratum lucidum. At 6 days, the increase is stillelevated when compared with control in the resistant CA3 area, and in the molecular layer. At this time point, strong TIMP-1 labelling is observed in thestratum granulosum and in glial cells scattered across the hippocampus (arrows: E, F), identi®ed as astrocytes in double labelling experiments (arrows: G andH). Scale bars, 500 mm (A±D); 50 mm (E±H).

28 S. Rivera et al.


TIMP-1 expression increases in the ischemic brain afterischemia

TIMP-1 mRNA induction has been previously observed by us and

by other groups following various neuronal insults, including

seizures (Nedivi et al., 1993; Rivera et al., 1997; Jaworski et al.,

1999), EAE (Pagenstecher et al., 1998) and focal ischemia (Wang

et al., 1998). Our results show two phases of TIMP-1 mRNA

expression clearly differentiated in space and time, reminiscent of

observations after excitotoxic kainate-induced seizures (Rivera

et al., 1997): (i) an early phase with a rapid TIMP-1 mRNA

increase in the stratum granulosum, the ependymal cell layer of

the ventricles and the meninges (pia matter); (ii) a delayed phase

characterized by a strong induction of TIMP-1 mRNA in

astrocytes in the areas of lesion. While neuronal activity is likely

to induce TIMP-1 mRNA in neurons, circulating or resident

cytokines and trophic factors might account for rapid ventricular

and delayed astrocytic TIMP-1 expression. Double labelling

experiments con®rmed that besides neurons, TIMP-1 protein is

expressed by reactive astrocytes after ischemia, but not by

microglia. Put together, our observations indicate that in vivo,

gelatinase B and TIMP-1 colocalize in neurons and astrocytes, but

not in microglial cells. Similar cellular distribution has been found

after seizures (Rivera et al., 1997) or EAE (Pagenstecher et al.,

1998). Twenty-four hours after ischemia, increased TIMP-1

FIG. 7. Net in situ gelatinolytic activity increases after ischemia. Fluoresecence photomicrographs of brain sections showing in situ zymography in thehippocampus (A±F), in control (A, E) and 1 (B, F), 3 (C) and 6 (D) days after global ischemia. Fluorescence signal slightly increases at 1 day in thepyramidal cell layer (p) of CA1/subiculum (Sub), in scattered cells of the strata radiatum (sr) and oriens (so), and in blood vessels (arrowheads). Fluorescenceprogressively increases across time in cells and blood vessels of the CA1-subicular area (C and D) and in the corpus callosum (cc) (D). In the hilus (h)¯uorescence increases 1 day after ischemia (F) in interneurons (arrows) as compared with control (E). Scale bars, 100 mm (A±D); 80 mm (E, F).



combined with decreased gelatinase B may imply reduced

proteolytic activity in resistant CA3 and granule neurons. In

contrast, at 3 days postischemia, gelatinase B is probably not

inhibited by TIMP-1 in activated microglia, at least in an

autocrine fashion.

Increased net gelatinolytic activity correlates with neuronallesions and glial reactivity

Constitutive gelatinolytic activity in neuronal layers con®rms that

neurons are a major source of active MMPs in the normal brain. This

is in keeping with the idea that controlled proteolysis of the neuronal

environment is crucial to maintaining neuronal homeostasis and

promoting plasticity in the CNS (Lynch & Baudry, 1984; Monard,

1988; Frey et al., 1996; Rivera & Khrestchatisky, 1999).

One day after ischemia, gelatinolysis increased strongly in hilar

cells, preceding the death of the somatostatinergic interneurons

(Johansen et al., 1987), whereas CA1 interneurons, that do not

degenerate after ischemia, showed no changes in proteolytic activity.

A correlation between increased proteolysis and the onset of neuronal

death was also observed in the lesioned CA1 area 3 days

postischemia. Exacerbated proteolysis by MMPs in the ischemic

brain may impair ECM stability leading to neuronal demise, as

demonstrated for serine proteinases after excitotoxic seizures (Chen

FIG. 8. MMPs contribute mainly to gelatinolytic activity in neural cells but not in blood vessels. Fluorescence photomicrographs of coronal brain sectionsshowing in situ zymography in the hippocampus 6 days after global ischemia (A) in the presence of 1 mM phenanthroline (B), 100 ng/mL (C) and 500 ng/mL(D) of TIMP-1, 100 ng/mL of TIMP-2 (E), and a cocktail of nonmetalloproteinase inhibitors (F). Gelatinolysis is highly reduced by phenanthroline and byTIMPs in neural cells. The metalloproteinase inhibitors are less ef®cient in reducing gelatinolysis associated with blood vessels (arrowheads). In contrast, acocktail of nonmetalloproteinase inhibitors does not inhibit gelatinolysis in neural cells, while it completely abolishes that in blood vessels. Abbreviations: p,pyramidal cell layer; so, stratum oriens; sr, stratum radiatum; CA1, area 1 of Ammon's horn. Scale bar, 200 mm.

30 S. Rivera et al.


& Strickland, 1997). The implication of gelatinases in neuronal death

is further supported by reduced neuronal lesions in gelatinase B

knock-out mice after focal ischemia (Asahi et al., 2000) or traumatic

brain injury (Wang et al., 2000).

The pattern of glial reactivity and gelatinase B expression suggest

that gelatinolysis in CA1 is predominantly associated ®rst with

microglial cells and then with both microglia and astrocytes.

Inhibition of neuronal and glial gelatinolytic activity with phenan-

throline and TIMPs indicates that metalloproteinases, and more

precisely MMPs, contribute to this activity.

Gelatinolytic activity increased progressively after ischemia in

blood vessels, presumably in cells associated with the BBB.

Interestingly, whereas metalloproteinase inhibitors had poor inhibi-

tory effect on the proteolysis associated with blood vessels, a cocktail

of nonmetalloproteinase inhibitors abolished it, indicating the pres-

ence of gelatinases of a nonmetalloproteinase family. We have

evidence that these may be serine proteases, as their inhibitors are

ef®cient in blocking blood vessel proteolytic activities (Jourquin

et al., unpublished observations). Extracellular enzymes of the serine

protease family, such as plasmin ± which have gelatinase activity ± or

plasminogen activators, have been implicated in the opening of the

BBB after ischemia (Mun-Bryce & Rosenberg, 1998). These authors

have also demonstrated in a model of focal ischemia that synthetic

MMP inhibitors ef®ciently block the intial opening of the BBB after

reperfusion but cannot block the delayed opening (Rosenberg et al.,

1998).

In conclusion, our results support the idea that MMPs are set in

motion in the ischemic brain and can contribute to the mechanisms of

brain injury and repair in a cell- and time-dependent manner.

However, it is clear that much work remains to be done to unravel the

complex interplay between MMPs and their inhibitors, modulators

and substrates in CNS physiopathology. The development of more

speci®c substrates and inhibitors for these proteases should help to

accomplish this task and to develop rational therapeutic strategies.

Acknowledgements

Part of this work was carried out at INSERM U-29. This research wassupported in part by funding from INSERM contract 4E003 (ReÂseau Est-Ouest) to MK and by funding from the FaculteÂ de MeÂdecine de Marseille(BQR 2001). SR was granted fellowships from the Fyssen and the SingerFoundations. JJ and CO are supported by fellowships from the French Ministryof Research and from La Ligue Nationale Contre le Cancer, respectively. SRand CO are equal ®rst authors. We gratefully acknowledge Pr. DominiqueBernard and Dr JoseÂ Boucraut for supporting our group in their laboratory.

Abbreviations

BBB, blood±brain-barrier; BSI, isolectin B4; DAB, diaminobenzidine; EAE,experimental autoimmune encephalomyelitis; ECM, extracellular matrix;GFAP, glial ®brillary acidic protein; IL, interleukin; LPS, bacteriallipopolysacharide; MAP, microtubule associated protein; MMP, matrixmetalloproteinase; MT-MMP, membrane type matrix metalloproteinase;PFA, paraformaldehyde; SDS, sodium dodecyl sulphate; TIMP, tissueinhibitor of metalloproteinases; TNF, tumor necrosis factor.

References

Asahi, M., Asahi, K., Jung, J.C., del Zoppo, G.J., Fini, M.E. & Lo, E.H. (2000)Role for matrix metalloproteinase-9 after focal cerebral ischemia: effects ofgene knockout and enzyme inhibition with BB-94. J. Cer Blood FlowMetab., 20, 1681±1689.

Backstrom, J.R., Lim, G.P., Cullen, M.J. & Tokes, Z.A. (1996) Matrixmetalloproteinase-9 (MMP-9) is synthesized in neurons of the human

hippocampus and is capable of degrading the amyloid-beta peptide (1±40).J. Neurosci., 16, 7910±7919.

Brooks, P.C., Stromblad, S., Sanders, L.C., von Schalscha, T.L., Aimes, R.T.,Stetler-Stevenson, W.G., Quigley, J.P. & Cheresh, D.A. (1996) Localizationof matrix metalloproteinase MMP-2 to the surface of invasive cells byinteraction with integrin alpha v beta 3. Cell, 85, 683±693.

Chen, Z.L. & Strickland, S. (1997) Neuronal death in the hippocampus ispromoted by plasmin-catalyzed degradation of laminin. Cell, 91, 917±925.

Coussens, L.M., Shapiro, S.D., Soloway, P.D. & Werb, Z. (2001) Models forgain-of-function and loss-of-function of MMPs: Transgenic and genetargeted mice. Meth. Mol. Biol., 151, 149±179.

Dubois, B., Peumans, W.J., Van Damme, E.J., Van Damme, J. & Opdenakker,G. (1998) Regulation of gelatinase B (MMP-9) in leukocytes by plantlectins. FEBS Lett., 427, 275±278.

Dusart, I., Marty, S. & Peschanski, M. (1991) Glial changes following anexcitotoxic lesion in the CNS ± II Astrocytes. Neuroscience, 45, 541±549.

Edwards, D.R., Beaudry, P.P., Laing, T.D., Kowal, V., Leco, K.J., Leco, P.A.& Lim, M.S. (1996) The roles of tissue inhibitors of metalloproteinases intissue remodelling and cell growth. Int. J. Obesity, 20, S9±S15.

English, W.R., Puente, X.S., Freije, J.M., Knauper, V., Amour, A.,Merryweather, A., Lopez-Otin, C. & Murphy, G. (2000) Membrane type4 matrix metalloproteinase (MMP17) has tumor necrosis factor-alphaconvertase activity but does not activate pro-MMP2. J. Biol. Chem., 275,14046±14055.

Frey, U., Muller, M. & Kuhl, D. (1996) A different form of long-lastingpotentiation revealed in tissue plasminogen activator mutant mice. J.Neurosci., 16, 2057±2063.

Gasche, Y., Fujimura, M., Morita-Fujimura, Y., Copin, J.C., Kawase, M.,Massengale, J. & Chan, P.H. (1999) Early appearance of activated matrixmetalloproteinase-9 after focal cerebral ischemia in mice: a possible role inblood±brain barrier dysfunction. J. Cereb. Blood Flow Metab., 19, 1020±1028.

Gearing, A.J.H., Beckett, P., Christodoulou, M., Churchill, M., Clements, J.,Davidson, A.H., Drummond, A.H., Galloway, W.A., Gilbert, R., Gordon,J.L., Leber, T.M., Mangan, M., Miller, K., Nayee, P., Owen, K., Patel, S.,Thomas, W., Wells, G., Wood, L.M. & Wooley, K. (1994) Processing oftumour necrosis factor-a precursor by metalloproteinases. Nature, 370,555±557.

Giraudon, P., Buart, S., Bernard, A. & Belin, M.F. (1997) Cytokines secretedby glial cells infected with HTLV-I modulate the expression of matrixmetalloproteinases (MMPs) and their natural inhibitor (TIMPs): possibleinvolvement in neurodegenerative processes. Mol. Psychiatry, 2, 107±110.

Goldberg, G.I., Strongin, A., Coolier, I.E., Genrich, L.T. & Marmer, B.L.(1992) Interaction of 92-kDa type IV collagenase with the tissue inhibitor ofmetalloproteases prevents dimerization, complex formation with interstitialcollagenase, and activation of the proenzyme with stromelysin. J. Biol.Chem., 267, 4583±4591.

Gottschall, P.E. & Deb, S. (1996) Regulation of matrix metalloproteinaseexpressions in astrocytes, microglia and neurons. Neuroimmunomodulation,3, 69±75.

Gottschall, P.E., Yu, X. & Bing, B. (1995) Increased production of gelatinaseB (matrix metalloproteinase-9) and interleukin-6 by activated rat microgliain culture. J. Neurosci. Res., 42, 335±342.

Greene, J., Wang, M., Liu, Y.E., Raymond, L.A., Rosen, C. & Shi, Y.E.(1996) Molecular cloning and characterization of human tissue inhibitor ofmetalloproteinase 4. J. Biol. Chem., 29, 30375±30380.

Ito, A., Mukaiyama, A., Itoh, Y., Nagase, H., Thogersen, I.B., Enghild, J.J.,Sasaguri, Y. & Mori, Y. (1996) Degradation of interleukin 1b by matrixmetalloproteinases. J. Biol. Chem., 271, 14657±14660.

Jaworski, J., Biedermann, I.W., Lapinska, J., Szklarczyk, A., Figiel, I.,Konopka, D., Nowicka, D., Filipkowski, R.K., Hetman, M., Kowalczyk, A.& Kaczmarek, L. (1999) Neuronal excitation-driven and AP-1-dependentactivation of tissue inhibitor of metalloproteinases-1 gene expression inrodent hippocampus. J. Biol. Chem., 274, 28106±28112.

Johansen, F.F., Zimmer, J. & Diemer, N.H. (1987) Early loss of somatostatinneurons in dentate hilus after cerebral ischemia in the rat precedes CA-1pyramidal cell loss. Acta Neuropatho.l (Berl.), 73, 110±114.

Kreutzberg, G.W. (1996) Microglia: a sensor for pathological events in theCNS. Trends Neurosci., 19, 312±318.

Lynch, G. & Baudry, M. (1984) The biochemistry of memory: a new andspeci®c hypothesis. Science, 224, 1057±1063.

Monard, D. (1988) Cell-derived proteases and protease inhibitors as regulatorsof neurite outgrowth. Trends Neurosci., 11, 541±544.

Muir, D. (1994) Metalloproteinase-dependent neurite outgrowth within a



synthetic extracellular matrix is induced by nerve growth factor. Exp. CellRes., 210, 243±252.

Mun-Bryce, S. & Rosenberg, G.A. (1998) Matrix metalloproteinases incerebrovascular disease. J. Cereb. Blood Flow Metab., 18, 1163±1172.

Murphy, G., Knauper, V., Cowell, S., Hembry, R., Stanton, H., Butler, G.,Freije, J., Pendas, A.M. & Lopez-Otin, C. (1999) Evaluation of some newermatrix metalloproteinases. Ann. N Y Acad. Sci., 878, 25±39.

Nedivi, E., Hevroni, D., Naot, D., Israeli, D. & Citri, Y. (1993) Numerouscandidate plasticity-related genes revealed by differential cDNA cloning.Nature, 363, 718±722.

Oh, L.Y., Larsen, P.H., Krekoski, C.A., Edwards, D.R., Donovan, F., Werb, Z.& Yong, V.W. (1999) Matrix metalloproteinase-9/gelatinase B is requiredfor process outgrowth by oligodendrocytes. J. Neurosci., 19, 8464±8475.

Pagenstecher, A., Stalder, A.K., Kincaid, C.L., Shapiro, S.D. & Campbell, I.L.(1998) Differential expression of matrix metalloproteinase and tissueinhibitor of matrix metalloproteinase genes in the mouse central nervoussystem in normal and in¯ammatory states. Am. J. Pathol., 152, 729±741.

Pagenstecher, A., Stalder, A.K., Kincaid, C.L., Volk, B. & Campbell, I.L.(2000) Regulation of matrix metalloproteinases and their inhibitor genes inlipopolysaccharide-induced endotoxemia in mice. Am. J. Pathol., 157, 197±210.

Planas, A.M., SoleÂ, S., Justicia, C. & RodrõÂguez-FarreÂ, E. (2000) Estimation ofgelatinase content in rat brain: effect of focal ischemia. Biochem. Biophys.Res. Comm., 278, 803±807.

Pulsinelli, W.A. & Brierley, J.B. (1979) A new model of bilateral hemisphericischemia in the unanesthetized rat. Stroke, 10, 267±272.

Rivera, S. & Khrestchatisky, M. (1999). Matrix Metalloproteinases and TissueInhibitors of Metalloproteinases in Neuronal Plasticity and Pathology. InBaudry, M., Davis, J.L. and Thompson, R.F. (eds), Advances in SynapticPlasticity. The MIT Press Cambridge, Massachusetts, pp. 53±86.

Rivera, S., Tremblay, E., Timsit, S., Canals, O., Ben-Ari, Y. & Khrestchatisky,M. (1997) Tissue inhibitor of metalloproteinases-1 (TIMP-1) isdifferentially induced in neurons and astrocytes after seizures: evidencefor developmental, immediate early gene, and lesion response. J. Neurosci.,17, 4223±4235.

Romanic, A.M., White, R.F., Arleth, A.J., Ohlstein, E.H. & Barone, F.C.(1998) Matrix metalloproteinase expression increases after cerebral focalischemia in rats. Stroke, 29, 1020±1030.

Rosenberg, G.A., Dencoff, J.E., McGuire, P.G., Liotta, L.A. & Stetler-Stevenson, W.G. (1994) Injury-induced 92-kilodalton gelatinase andurokinase expression in rat brain. Lab. Invest., 71, 417±422.

Rosenberg, G.A., Estrada, E.Y. & Dencoff, J.E. (1998) Matrixmetalloproteinases and TIMPs are associated with blood±brain barrieropening after reperfusion in rat brain. Stroke, 29, 2189±2195.

Rosenberg, G.A., Navratil, L., Barone, F. & Feurstein, G. (1996) Proteolytic

cascade enzymes increase in focal cerebral ischemia in rat. J. Cer. BloodFlow Metab., 16, 360±366.

Saito, K., Suyama, K., Nishida, K., Sei, Y. & Basile, A.S. (1996) Earlyincreases in TNF-alpha, IL-6 and IL-1 beta levels following transientcerebral ischemia in gerbil brain. Neurosci. Lett., 206, 149±152.

Schmidt-Kastner, R. & Hossmann, K.A. (1988) Distribution of ischemicneuronal damage in the dorsal hippocampus of rat. Acta Neuropathol.(Berl.), 76, 411±421.

Schonbeck, U., Mach, F. & Libby, P. (1998) Generation of biologically activeIL-1 beta by matrix metalloproteinases: a novel caspase-1-independentpathway of IL-1 beta processing. J. Immunol., 161, 3340±3346.

Seiki, M. (1999) Membrane-type matrix metalloproteinases. Apmis, 107, 137±143.

Thilmann, R., Xie, Y., Kleihues, P. & Kiessling, M. (1986) Persistentinhibition of protein synthesis precedes delayed neuronal death inpostischemic gerbil hippocampus. Acta Neuropathol. (Berl.), 71, 88±93.

Timsit, S., Rivera, S., Ouaghi, P., Guischard, F., Tremblay, E., Ben-Ari, Y. &Khrestchatisky, M. (1999) Increased cyclin D1 in vulnerable neurons in thehippocampus after ischaemia and epilepsy: a modulator of in vivoprogrammed cell death? Eur. J. Neurosci., 11, 263±278.

Vaillant, C., Didier-Bazes, M., Hutter, A., Belin, M.F. & Thomasset, N. (1999)Spatiotemporal expression patterns of metalloproteinases and theirinhibitors in the postnatal developing rat cerebellum. J. Neurosci., 19,4994±5004.

Vecil, G.G., Larsen, P.H., Corley, S.M., Herx, L.M., Besson, A., Goodyer,C.G. & Yong, V.W. (2000) Interleukin±1 is a key regulator of matrixmetalloproteinase-9 expression in human neurons in culture and followingmouse brain trauma in vivo. J. Neurosci. Res., 61, 212±224.

Wang, X., Barone, F.C., White, R.F. & Feuerstein, G.Z. (1998) Subtractivecloning identi®es tissue inhibitor of matrix metalloproteinase-1 (TIMP-1)increased gene expression following focal stroke. Stroke, 29, 516±520.

Wang, X., Jung, J.C., Asahi, M., Chwang, W., Russo, L., Moskowitz, M.A.,Dixon, C.E., Fini, E. & Lo, E.H. (2000) Effects of matrix metalloproteinase-9 gene knock-out on morphological and motor outcomes after traumaticbrain injury. J. Neurosci., 20, 7037±7042.

Weeks, J.G., Halme, J. & Woessner, J.F. Jr (1976) Extraction of collagenasefrom the involuting rat uterus. Biochim. Biophys. Acta, 445, 205±214.

Werb, Z. (1997) ECM and cell surface proteolysis: regulating cellular ecology.Cell, 91, 439±442.

Yong, V.W., Power, C., Forsyth, P. & Edwards, D.R. (2001)Metalloproteinases in biology and pathology of the nervous system. Nat.Rev. Neurosci., 2, 502±511.

Zhang, J.W., Deb, S. & Gottschall, P.E. (1998) Regional and differentialexpression of gelatinases in rat brain after systemic kainic acid orbicuculline administration. Eur. J. Neurosci., 10, 3358±3368.

32 S. Rivera et al.


Gelatinase B and TIMP-1 are regulated in a cell- and time-dependent manner in association with...

Documents

Transcript of Gelatinase B and TIMP-1 are regulated in a cell- and time-dependent manner in association with...