Pyrogenic cytokines injected into the rat cerebral ventricle induce cyclooxygenase-2 in brain endothelial cells and also upregulate their receptors

Pyrogenic cytokines injected into the rat cerebral ventricleinduce cyclooxygenase-2 in brain endothelial cells andalso upregulate their receptors

Chunyu Cao,1 Kiyoshi Matsumura,1,2 Noriyuki Shirakawa,1 Mitsuyo Maeda,3 Ikuyo Jikihara,3 Shigeo Kobayashi2 andYasuyoshi Watanabe1,4

1Department of Neuroscience, Osaka Bioscience Institute, 6-2-4 Furuedai, Suita, Osaka 565-0874, Japan2Department of Intelligence Science and Technology, Graduate School of Informatics, Kyoto University, Yoshida-honmachi,

Sakyo-ku, Kyoto 606-8501, Japan3Department of Anatomy, Graduate School of Medicine, Osaka City University, Abeno-ku, Osaka 545-8585, Japan4Department of System Neuroscience, Graduate School of Medicine, Osaka City University, Abeno-ku, Osaka 545-8585, Japan

Keywords: cyclooxygenase, cytokine interaction, endothelial cells, fever, prostaglandin

Abstract

Peripheral immunological insults induce interleukin (IL)-1b and IL-6 in the brain. To elucidate the mechanism(s) of fever evokedby these brain-derived cytokines, and possible interactions between them, we examined in rats: (i) whether cyclooxygenase-2 is

responsible for fever evoked by central injection of these cytokines; (ii) if so, where in the brain cyclooxygenase-2 is induced;

(iii) where the receptors for these cytokines are located; and (iv) how the expression of these receptors is in¯uenced by thecytokines. Intracerebroventricular injection of these cytokines evoked fever that was suppressed by a cyclooxygenase-2 inhibitor.

Brain endothelium was the site of cyclooxygenase-2 induction by these cytokines. IL-1 receptor (IL-1R) was constitutively

expressed in brain endothelium, and its mRNA was further upregulated by either cytokine. IL-6R mRNA was constitutively

expressed in the cerebral cortex, and was newly induced in as yet unidenti®ed cells in brain blood vessels by either cytokine.Messenger RNAs for cyclooxygenase-2, IL-1R, and IL-6R were often observed in the same blood vessels. These results suggest

that COX-2 induced in brain endothelium is, at least in part, involved in the fever evoked by these cytokines, and that one

possible interaction between these two cytokines is mutual upregulation of their receptors in the endothelium or perivascularcells, resulting in augmentation of their actions.

Introduction

Fever, a common symptom of various infectious diseases, is brought

about as a consequence of immune-brain signalling. The molecular

cascade in this signalling system is now well documented as the

following: peripheral monocytic cells phagocytose exogenous pyro-

gens and subsequently release endogenous pyrogens that act on the

brain to activate the arachidonic acid cascade, a product of which,

prostaglandin E2 (PGE2), in turn acts on the thermoregulatory

neuronal network to elevate body temperature (Stitt, 1986). In this

cascade, it is important to note that multiple endogenous pyrogens

exist that are now identi®ed to be cytokines, including interleukin-

1a/b (IL-1a/b), IL-6, interferons, and, possibly, tumour necrosis

factor-a (TNF-a) (Kluger, 1991; Dinarello, 1999). These cytokines

act on distinct types of receptors, and, as a consequence, activate

distinct intracellular signalling cascades. Nevertheless, all the

cytokine signals ®nally converge onto a single febrile mediator,

PGE2, to develop fever. This fact raises the question as to how each

of these pyrogenic cytokines works to enhance the PGE2 production

in the fever process.

One possible answer to this question is that each cytokine acts on a

distinct cell group independently to enhance PGE2 biosynthesis; and

as a whole, the total amount of PGE2 is increased. If this were the

case, we would expect speci®c receptors for each cytokine to be

distributed differentially among multiple cell groups in which the

activation of the arachidonic acid cascade takes place independently.

Another possibility is that the cytokines act on a common cell group

to enhance PGE2 biosynthesis either additively or synergistically. In

this case, the receptors for the cytokines should be colocalized in the

same type of cells where the arachidonic acid cascade is activated.

We previously showed that peripherally injected IL-1b (Cao et al.,

1996) and TNF-a (Cao et al., 1998) induced fever and the

concomitant expression of cyclooxygenase-2 (COX-2), a rate-limit-

ing enzyme of the arachidonic acid cascade, in brain endothelial cells,

supporting the latter possibility. However, this possibility has not yet

been fully tested for other cytokines, nor for different routes of

cytokine administration.

Especially, IL-6 possesses pyrogenic characteristics apparently

different from those of other cytokines. IL-6 is generally less

pyrogenic than the other cytokines (Dinarello, 1999), although it is

certainly an indispensable molecule in fever, as shown by studies

using IL-6 gene knockout mice (Chai et al., 1996; Sundgren-

Andersson et al., 1998). In the rat, IL-6 evokes fever only when it is

injected into the brain (LeMay et al., 1990), whereas IL-1b or TNF-a

Correspondence: Dr Kiyoshi Matsumura, 2Department of Intelligence Scienceand Technology, as above.E-mail: [email protected]

Received 31 October 2000, revised 6 March 2001, accepted 7 March 2001

European Journal of Neuroscience, Vol. 13, pp. 1781±1790, 2001 ã Federation of European Neuroscience Societies

does so when administered either peripherally or centrally (Kluger,

1991; Dinarello, 1999). This may indicate that action sites of IL-6 are

distinct from those of other pyrogenic cytokines. In the present study,

we employed a rat model of fever in which either IL-1b or IL-6 was

injected via the intracerebroventricular (i.c.v.) route. Using this

preparation, we examined whether the fever evoked by each cytokine

was mediated via the COX-2 pathway, and, if so, in what cells

COX-2 was induced by these cytokines. Furthermore, localization of

receptors for these cytokines, i.e. IL-1 type-1 receptor (IL-1R1) and

IL-6 receptor (IL-6R), was examined in the same preparation by

in situ hybridization and/or immuno-electron microscopy; and these

sites were compared with those of COX-2 induction.

Materials and methods

Materials

Male Wistar rats of 7 weeks of age were purchased from Shizuoka

Laboratory Animal Cooperative (Shizuoka, Japan). They were

housed four or ®ve to a cage in a room at 26 6 2 °C, with a

standard 12 h light : 12 h dark cycle. Other materials used were as

follows: [35S]UTP and X-ray ®lms (bmax, Amersham Life Sciences);

emulsion for micro-autoradiography (NTB2, Kodak); recombinant

human IL-1b (a gift from Otsuka Pharmaceutical Co. Ltd,

Tokushima, Japan); recombinant human IL-6 (a gift from Dr A.

Okano of Ajinomoto Co., Ltd, Yokohama, Japan); RNA transcription

kit (Stratagene); NS-398, a COX-2-speci®c inhibitor (a gift from Dr

S. Higuchi of Taisho Pharmaceutical Co., Tokyo, Japan); rabbit

polyclonal antibody against murine COX-2 (Cayman Chemical);

sheep polyclonal antibody against von Willebrand factor (Dako);

antimouse IL-1R1 monoclonal antibody (Genzyme); rat IL-1R1

cDNA (a gift from Dr R.P. Hart of Rutgers University); rat COX-2

cDNA (a gift from Dr K. Yamagata of Tokyo Metropolitan Institute

for Neurosciences); rat IL-6R cDNA (a gift from Dr L. Vallieres of

Laval University).

Surgery

Under pentobarbital anaesthesia, temperature transmitters (Mini-

Mitter, Sunriver, OR, USA) were implanted in the abdominal cavity

of the rats. Their heads were then ®xed in a stereotaxic apparatus

(Narishige, Tokyo, Japan) according to the rat brain atlas (Paxinos &

Watson, 1986). The skin of the skull was incised in the midline, and

two holes (0.8 mm in diameter) for i.c.v. injection were drilled into

the skull so that they would be located above the right and left lateral

ventricles (0.8 mm posterior and 1.2 mm lateral to the bregma). The

incision of the skin was closed, and the rats were housed individually

in cages placed on the receivers of the temperature telemetry system

(Data Sciences, St Paul, MN, USA).

Experimental protocol

Intracerebroventricular injections were made after a recovery period of

at least 1 week. Between 09.00 h and 09.30 h on the day of the

experiment, the rats were lightly anaesthetized with halothane (3% in

the air) and then received an i.c.v. injection of either IL-1b (25 ng per

rat in 15 mL saline, n = 6), IL-6 (250 ng or 1 mg per rat in 15 mL

saline, n = 5±6) or saline alone (n = 4). The injections were made

through one of the holes in the skull with a stainless steel needle (27

gauge) that was connected to a 1-mL disposable syringe via

polyethylene (PE) tubing (PE20, Cray Adams) of 50-cm length. The

syringe, PE tubing and needle were ®lled with either saline or the

cytokine solution. Once the needle had been inserted into the brain

4 mm from the skull, the syringe was removed from the PE tubing and

the free end of the PE tubing was lifted up in such a way that it was

located 50 cm above the rat's head. If the tip of the needle had been

properly located in the lateral ventricle, the solution in the PE tubing

was delivered into the ventricle by hydrostatic pressure. This was

con®rmed by a drop in the level of meniscus of the solution in the PE

tubing. The volume of the solution injected was checked with a

precalibrated scale put on the PE tubing. If the meniscus of the solution

did not drop, the needle was moved slightly until the meniscus started

to drop. After the injection, the rats were returned to their home cages.

The whole process of this injection took approximately 5 min for each

rat. The other hole in the skull was left unused. The abdominal

temperature (Tab) of each rat was monitored every 10 min during a 24-

h period before and after the injection. The averaged Tab during the

30-min period prior to the i.c.v. injection was used as the baseline of

body temperature in each rat.

To investigate the effect of a COX-2 inhibitor on the febrile

response, we injected rats intraperitoneally with either NS-398

(4 mg/kg, i.p.) or its vehicle (500 mL of 50% dimethyl sulphoxide

[DMSO] in saline) 1 h prior to the injection of either IL-1b or IL-6.

NS-398 possesses a high selectivity for COX-2, as shown by several

in vitro and in vivo studies (Futaki et al., 1993a, b, 1994; Masferrer

et al., 1994; Gierse et al., 1995). The averaged Tab during the 30-min

period prior to the NS-398 or DMSO injection was used as the

baseline of body temperature in each rat. All animal experiments

were performed under the guidelines of the Animal Committee of the

Osaka Bioscience Institute.

In situ hybridization

Based on the time-course of fever, the time-point of 3 h after the

injection of IL-1b, IL-6 or saline (control) was chosen for killing the

rats for in situ hybridization to detect COX-2, IL-1R1 or IL-6R

mRNA. Two rats for each cytokine group were anaesthetized with

diethylether and perfused via the left ventricle of the heart with

50 mL of ice-cold saline, followed by 200 mL of 4% paraformalde-

hyde solution (pH 6.5). The brains were put in a solution containing

20% sucrose and 4% paraformaldehyde overnight, frozen in dry-ice

powder, and stored at ±80 °C until sectioned. Brain sections of 16-mm

thickness were made and thaw-mounted on silane-coated glass slides.

In situ hybridization with 35S-labelled riboprobes was conducted as

described in our previous study (Cao et al., 1995). Sense-strand 35S-

labelled riboprobes of identical length were used for negative-control

experiments. When colocalizations of two kinds of mRNA were

examined, the mirror section method was employed (Cao et al.,

1996).

Immunohistochemistry

For immunohistochemical detection of COX-2 protein, rats were

anaesthetized with diethylether 1.5 h or 3.5 h after the i.c.v.

injections of IL-1b or IL-6, and perfused via the left ventricle of

the heart with 100 mL of ice-cold 20 mM phosphate-buffered saline

(pH 7.4). At each time-point, three rats were killed. The brains were

cut at a thickness of 14 mm on a cryostat and thaw-mounted on silane-

coated glass slides. Immunostaining for COX-2 and von Willebrand

(vW) factor (an endothelial marker) were performed as reported

previously (Matsumura et al., 1998).

Immuno-electron microscopy for IL-1R1

Adult untreated rats were anaesthetized with sodium pentobarbital

and perfused with Zamboni's solution containing 0.05% glutaralde-

hyde. Sections of 50-mm thickness were made with a microslicer

(Dosaka EM). The sections were incubated at 4 °C with rat antimouse

IL-1R1 monoclonal antibody diluted 1 : 100 in the PBS overnight

1782 C. Cao et al.

ã 2001 Federation of European Neuroscience Societies, European Journal of Neuroscience, 13, 1781±1790

and processed for further immunohistochemical procedures. The

immunoreactivity was visualized with a Vectastain Elite ABC kit.

The sections were post®xed with 1% osmium tetroxide, dehydrated

with alcohol, treated with propylene oxide, and then embedded in

Durcupan (Fluka). Ultrathin sections were cut on an ultramicrotome

(Reichert-Jung), and stained with uranyl acetate.

Analysis and statistics

The numbers of COX-2-immunoreactive (COX-2-ir) cells were

counted in regions around the third ventricle, including the medial

preoptic area and cisterna chiasmatis, and around the lateral ventricle,

including the septum and striatum in coronal sections corresponding

to the plane of bregma around 0 mm in the stereotaxic atlas (Paxinos

& Watson, 1986). The counting was made in ®ve consecutive

sections from each of three rats killed at two time-points after i.c.v.

injection of IL-1b or IL-6, approximately corresponding to the rising

phase (1.5 h) and the maximally febrile point (3.5 h), respectively,

and the results from three rats were averaged. Data were expressed as

means 6 SEM. Mann±Whitney U-test was used to evaluate the

signi®cance of differences.

Results

Fever induced by intracerebroventricular injection of IL-1b or IL-6

Figure 1A shows the time-course of fever induced by the i.c.v.

injection of IL-1b (25 ng per rat). The average Tab at the time of

injection (time-point 0) was 37.0 6 0.17 °C (mean 6 SEM) for the

saline-injected group (n = 4), and 36.9 6 0.09 °C for the IL-1b-

injected group. Due to the light halothane anaesthesia, the Tabs of both

IL-1b-injected rats and saline-injected rats showed a slight decrease.

The Tab of the IL-1b-injected rat started to rise from 10 min after the

injection, peaked at 3.5 h with a 1.1 °C increase above the pre-

injection value, and then gradually declined. On the other hand, the

Tab of the saline-injected rats showed only a small increase. The Tab

of both rats treated with IL-1b and those injected with saline increased

signi®cantly 9 h after the injections due to the circadian ¯uctuation of

body temperature, at which time the Tab of untreated rats also showed

a similar increase (Cao et al., 1999). Figure 1B shows the time-course

of fever induced by IL-6 (250 ng or 1 mg per rat, i.c.v.). The average

Tab at the time of injection was 36.7 6 0.47 °C for the group injected

with 250 ng IL-6 (n = 5) and 36.6 6 0.12 °C for the group injected

with 1 mg IL-6 (n = 5). The curve for the Tab of the rats injected with

saline is the same as that shown in Fig. 1A. After either a 250-ng or a 1-

mg IL-6 injection, the Tab started to rise between 10 and 20 min. The

degrees of Tab increase were almost the same between the two dosage

groups until 4 h after the IL-6 injection. After this time-point, the Tab

of the rats that had received 250 ng of IL-6 declined gradually, while

that of the rats received 1 mg of IL-6 stayed at a higher level. In both

cases, however, the degree of fever was smaller than that induced by

the low dose of IL-1b.

Effect of a COX-2 inhibitor on the fever

Rats were injected intraperitoneally with either a COX-2 inhibitor,

NS-398 (4 mg/kg, i.p.), or its vehicle (500 mL of 50% DMSO) 1 h

FIG. 1. (A and B) Changes in abdominal temperature (Tab) after intracerebroventricular (i.c.v.) injection of IL-1b (25 ng per rat, A) or IL-6 (250 ng or 1 mgper rat, B). The injections were made at time-point 0. Each value represents the mean 6 SEM of 4±6 rats. The change in Tab of the cytokine-injected ratswas signi®cantly different from that of the saline-injected ones during the time-period denoted by the asterisk (P < 0.05). (C and D) Changes in Tab of NS-398- or vehicle-pretreated rats after the i.c.v. injection of IL-1b or IL-6. NS-398 (4 mg/kg) or its vehicle (500 mL of 50% DMSO) was injectedintraperitoneally at time-point 0. IL-1b (60 ng per rat, C) or IL-6 (1 mg per rat, D) was injected at the 60-min time-point. Solid line in D indicates the meanTab change after NS-398 injection and anaesthesia without i.c.v. injection. Each value represents the mean 6 SEM of 4±6 rats. The change in Tab wassigni®cantly different between NS-398- and vehicle-pretreated rats during the time-period denoted by the asterisk (P < 0.05).

Brain cytokines induce COX-2 and fever 1783


before the i.c.v. injection of IL-1b or IL-6. This pretreatment

slightly increased the Tab due to the stress associated with handling

and injection (Fig. 1C and D). Pretreatment with NS-398, but not

that with the vehicle, almost completely suppressed the fever

induced by either IL-1b (60 ng, i.c.v.; Fig. 1C) or IL-6 (1 mg,

i.c.v.; Fig. 1D). We also con®rmed that NS-398 completely sup-

pressed the fever induced by a lower dose of IL-1b (25 ng, i.c.v.; data

not shown). Neither NS-398 (solid line in Fig. 1D) nor its vehicle

(data not shown) exerted a profound effect on the basal body

temperature.

Induction of COX-2 in the brain

Figure 2 shows autoradiographic images of COX-2 mRNA in two

representative coronal brain sections from rats 3 h after the i.c.v.

injection of IL-1b (25 ng; Fig. 2A1±A3 and D), IL-6 (1 mg;

Fig. 2B1±B3 and E) or saline (Fig. 2C1±C3). These treatments

induced COX-2 mRNA in the brain in two distinct manners. As the

®rst one, all the injections increased COX-2 mRNA in the cerebral

cortex on the side where the injection needle had penetrated (right

side in the ®gures). This type of response was neuronal in origin, as

shown in a previous study (Cao et al., 1999), and seemed not to be

essential for fever as the saline injection also elevated COX-2 mRNA

without inducing fever. On the other hand, the second type of COX-2

mRNA induction was seen only after the injection of IL-1b or IL-6.

In this case, the COX-2 mRNA signals appeared in a spot-like

manner (Fig. 2A1±A3 and B1±B3). Although IL-6-induced COX-2

mRNA was hard to see in the macroscopic views (Fig. 2B1 and B3),

it was clearly seen in the dark-®eld view of emulsion-coated samples

(Fig. 2B2). Apparently, IL-1b induced this type of COX-2 mRNA

more abundantly than did IL-6, in spite of the lower dose of IL-1b.

Light microscopic observation of emulsion-coated samples revealed

that the COX-2 mRNA signals induced by the cytokines were

associated with blood vessels near the cerebral ventricles (Fig. 2A2

and B2) or in the subarachnoidal space (Fig. 2D and E).

Immunohistochemical studies also demonstrated that COX-2-like

immunoreactivity was induced in some cells associated with blood

vessels after i.c.v. injection of IL-1b (25 ng; Fig. 3B±D) or IL-6

(1 mg; Fig. 3F and G) but not after that of saline (Fig. 3E). In

FIG. 2. Expression of COX-2 mRNA 3 h after intracerebroventricular (i.c.v.) injection of IL-1b (A1±A3), IL-6 (B1±B3) or saline (C1±C3). The brain sectionsin the left column (A1±C1) correspond to bregma 0 mm of the stereotaxic coordinate (Paxinos & Watson, 1986), containing the rostral part of the preopticarea, the organum vasculosum laminae terminalis (OVLT), cerebral cortex (ctx), optic chiasma (OX), and lateral ventricle (LV). In the middle column (A2±C2), dark-®eld views of the rostral part of the preoptic area and the third ventricle (3V) in emulsion-coated specimens are shown. The bright dots representsilver grains corresponding to COX-2 mRNA signals. The brain sections in the right column (A3±C3) correspond to bregma ±3.5 mm of the stereotaxiccoordinates, containing the dorsal hippocampus (DG, dentate gyrus, and CA3 in Ammon's horn), thalamus and mediobasal hypothalamus. The sections arearranged in such a way that the side in which the i.c.v. injection was made is on the right. The mRNA signals were enhanced on the injection side of thecerebral cortex regardless of the substance injected. Spot-like COX-2 mRNA signals (arrows) appeared in the brain parenchyma near the cerebral ventriclesand the OVLT region robustly after IL-1b injection (A1±A3) and faintly after IL-6 (B1±B3). Microscopically, the spot-like signals induced by IL-1b (D) andIL-6 (E) were associated with blood vessels (BV) in the subarachnoidal space (cisterna chiasmatis indicated as area D in Fig. 4A). The scale bars indicate100 mm.

1784 C. Cao et al.


addition, no COX-2-like (COX-2)-ir cells were found in brain blood

vessels of untreated rats (data not shown). After i.c.v. injection of IL-

1b, COX-2-ir cells appeared by 1.5 h in the parenchymal and

subarachnoidal blood vessels, and the number and intensity of the

staining further increased by 3.5 h (Fig. 3B±D). COX-2-ir cells were

in particular abundance in the blood vessels of the cisterna chiasmatis

(Fig. 3D), which forms the subarachnoidal space lateral to the optic

chiasma, and in the rostral part of the preoptic area (Fig. 3B).

Figure 4A summarizes the number of COX-2-ir cells 1.5 h and 3.5 h

after the IL-1b injection. After i.c.v. injection of IL-6 (1 mg), COX-2-

ir cells also appeared in blood vessels in the subarachnoidal space

(Fig. 3F) and near the cerebral ventricles (Fig. 3G). However, as

compared with the response to IL-1b, the intensity of COX-2 staining

after IL-6 injection was weaker (Fig. 3F and G) and the number of

COX-2-ir cells was smaller (Fig. 4B), being in line with COX-2

mRNA induction.

Identi®cation of COX-2-immunoreactive cells in the bloodvessels

Figure 5 shows double-immunostaining of COX-2 (red) and vW

factor (green) in blood vessels 3.5 h after the i.c.v. injection of IL-1b(Fig. 5A1±A3) or IL-6 (Fig. 5B1±B3). In these preparations, COX-2-

FIG. 4. Number of COX-2-immunoreactive (COX-2-ir) cells at 1.5 h and 3.5 h after IL-1b (25 ng per rat, A) and IL-6 (1 mg per rat, B). The numbers ofCOX-2-ir cells were counted in regions around the third ventricle, including the medial preoptic area and cisterna chiasmatis, and around the lateral ventricleincluding the septum and striatum. The counting was made in ®ve consecutive sections from one rat, and the results from three rats were averaged. Eachvalue represents the mean 6 SEM.

FIG. 3. COX-2-like-immunoreactive (COX-2-ir) cells in brain blood vessels after intracerebroventricular (i.c.v.) injection of IL-1b (25 ng per rat) (B±D) orIL-6 (1 mg per rat) (F,G). The area for each microscopic view is illustrated in (A). COX-2-ir cells were observed in the blood vessels near the third ventricle(3V) (B), and lateral ventricle (C), and those in the cisterna chiasmatis (D) 3.5 h after IL-1b injection. Similarly, after i.c.v. injection of IL-6, COX-2-ir cellswere found in blood vessels in the cisterna chiasmatis (F) and near the third ventricle (arrows in G), although the intensity of the staining was lower than thatafter IL-1b injection. OX, optic chiasma; 3V, third ventricle; OVLT, organum vasculosum laminae terminalis. Scale bar, 100 mm (B), 50 mm (C and D),200 mm (E), and 75 mm (F and G).



ir structures (red) were surrounded by vW factor-positive ones

(green). vW factor is a protein speci®cally expressed in the

endothelial cytosol (Wagner & Marder, 1983). Thus, the results

indicate that COX-2 was predominantly expressed in the endothelial

cells after i.c.v. injections of the cytokines. Double-immunostaining

of COX-2 and ED2 antigen, the latter being a marker for microglia

and tissue macrophages, provided little evidence for the expression of

COX-2 in these phagocytic cells.

Localization of mRNAs for the cytokine receptors and theirupregulation by the cytokines

To elucidate the receptor sites for the cytokines, we conducted an

in situ hybridization study for the cytokine receptor mRNAs in the

same brain samples as those used for COX-2 mRNA, as well as in the

brains from untreated rats. IL-1R1 mRNA was found to be distributed

in a spot-like fashion (arrows in Fig. 6A1, B1 and C1). Light

microscopic examination revealed that all of these spot-like signals

FIG. 5. Laser confocal immuno¯uorescence microscopic views of COX-2-immunoreactive (COX-2-ir) cells 3.5 h after IL-1b (25 ng per rat, A1±A3) or IL-6(1 mg per rat, B1±B3) injection. In both cases, oval-shaped COX-2-ir structures (red) were overlaid by a space occupied by von Willebrand factor (green), amarker for the cytosol of endothelial cells. In C, double immunostaining was performed for COX-2 (red) and ED2 antigen (green), the latter being a markerfor perivascular macrophages, in a brain section from an IL-1b-treated rat. Although located close to each other, COX-2-ir cells and ED2-positive ones weredistinct. All these blood vessels were located in the cisterna chiasmatis. Scale bars, 100 mm (A), 10 mm (B) and 20 mm (C).

FIG. 6. Expression of mRNAs for IL-1R1 (A1±C2) and IL-6R (A3±C4) in coronal brain sections of rats 3 h after an intracerebro-ventricular (i.c.v.) injection of IL-1b (A1±A4), IL-6 (B1±B4) or saline (C1±C4). In the macroscopic views (i.e. the ®rst and third columns), the sections arearranged so that the side of the i.c.v. injection is on the right. The amount of IL-1R1 mRNA signals (arrows) is apparently larger in the brain treated with IL-1b (A1) or IL-6 (B1), than in that injected with saline (C1). Dark-®eld microscopic observation in emulsion-coated specimens revealed these signals weremainly from blood vessels (A2, B2 and C2). Similarly, spot-like IL-6R mRNA signals are evident near the lateral ventricle, organum vasculosum laminaeterminalis (OVLT) and cisterna chiasmatis in the cytokine-injected (arrows in A3 and B3), but not saline-injected (C3), brains. Again, these signals originatedfrom blood vessels, as shown by dark-®eld microscopy (A4 and B4). Regardless of the substance injected, the IL-6R mRNA level markedly increased in thecerebral cortex on the injected side. Scale bar in C2 indicates 100 mm for the dark-®eld photographs. Abbreviations are the same as those in Fig. 1.

1786 C. Cao et al.


were closely associated with blood vessels (Fig. 6A2, B2 and C2,

Fig. 7A2). The number of IL-1R1 mRNA-positive blood vessels

appeared to be larger in the brains that had been injected with either

IL-1b (Fig. 6A1) or IL-6 (Fig. 6B1) than in those injected with saline

(Fig. 6C1). The average number of positive blood vessels per section

being 204 6 11 after IL-1b, 264 6 20 after IL-6, and 108 6 10 after

saline. These values were obtained from ®ve consecutive coronal

sections from each of two rats. This result indicates that expression of

IL-1R1 was upregulated by both IL-1b and IL-6.

Macroscopically, IL-6R mRNA signals were evident in the

cerebral cortex, hippocampus and amygdala under normal conditions

(data not shown). After the i.c.v. injections, the mRNA signals were

markedly enhanced in the cortex on the side where the injection

needle had penetrated (right side of Fig. 6A3, B3 and C3) irrespective

of the solution injected. When examined carefully, IL-6R mRNA was

also found to have a spot-like distribution near the cerebral ventricles

3 h after the cytokine injections (Fig. 6A3, B3, A4 and B4), but not

after saline injection (Fig. 6C3 and C4). Light microscopic study

revealed that these spot-like signals were all associated with the sites

of blood vessels (Fig. 7A3). The average number of IL-6R mRNA-

positive blood vessels per section was 17.9 6 1.0 after IL-1b,

11.1 6 2.1 after IL-6, and 1.2 6 0.6 after saline. These values were

obtained from ®ve consecutive coronal sections from each of two

rats. As in the case for IL-1R1 mRNA, expression of IL-6R mRNA

was also upregulated by both IL-1b and IL-6.

Possible colocalization of the cytokine receptors and COX-2 inendothelial cells

As described earlier, brain blood vessels were the sites where all the

mRNAs for COX-2, IL-1R1 and IL-6R were expressed in response to

the cytokine stimuli. We then examined whether these mRNAs were

expressed in the same blood vessels. As visualization of multiple

kinds of mRNAs in the same sections was technically dif®cult, we

studied it in adjacent sections. The brain sections in Fig. 7 were made

from a rat injected with IL-1b (25 ng, i.c.v.) 3 h before, and

contained the rostral part of the preoptic area, a presumed

thermoregulatory centre. Messenger RNAs for COX-2 (Fig. 7A1),

IL-1R1 (Fig. 7A2) and IL-6R (Fig. 7A3) were found in the same

blood vessels, as denoted by the same numbers in the ®gures.

Exceptionally, in the organum vasculosum laminae terminalis

FIG. 7. Light microscopic views of mRNAs for COX-2 (A1 and B1), IL-1R1 (A2 and B2) and IL-6R (A3) in serial sections near the organum vasculosumlaminae terminalis (OVLT) 3 h after intracerebroventricular injection of IL-1b (25 ng per rat). The corresponding numbers in these views represent theidentical blood vessel. The arrowhead indicates an identical blood vessel almost negative for all the mRNAs. In the case of B1 and B2, the sections werearranged in a mirror fashion so that the same cutting surface was exposed to the cRNA probes for COX-2 (B1) and IL-1R1 (B2), respectively. Lightmicroscopic view of B2 was reversed to compare the position of mRNA signals between the two sections. The corresponding numbers represent clusters ofmRNA signals located in a similar position in an identical blood vessel. Scale bars, 100 mm (A3) and 50 mm (B2).

FIG. 8. Light microscopic views of mRNA signals for COX-2 (A), IL-1R1 (B) and IL-6R (C) in three serial sections 3 h after intracerebroventricularinjection of IL-6 (1 mg per rat). The corresponding numbers represent identical blood vessels. Arrowheads indicate blood vessels negative for all the threemRNAs. 3V, the third ventricle. Scale bar, 100 mm.



(OVLT), IL-1R1 mRNA was abundant (Fig. 7A2), whereas COX-2

mRNA (Fig. 7A1) and IL-6R mRNA (Fig. 7A3) were less so or

absent. One blood vessel indicated by the arrowhead was almost

negative for all types of mRNAs (Fig. 7A1±3), indicating that

induction of these mRNAs was restricted to a subset of brain blood

vessels. When mRNAs for COX-2 and IL-1R1 were examined in

adjacent sections arranged in a mirror fashion, they were found in

similar positions within a blood vessel, suggesting that the mRNAs

were expressed even in the same cells (Fig. 7B1 and B2).

Counterstaining of the cellular nuclei clearly showed that the cells

expressing the mRNAs were located on the most luminal part of the

vessels (Fig. 7B1 and B2). These three kinds of mRNAs were also

expressed in the same blood vessels after i.c.v. injection of IL-6

(Fig. 8). Again, there were blood vessels negative for all the mRNAs

(arrowheads in Fig. 8).

Immuno-electron microscopy for IL-1R1 clearly showed that this

receptor was expressed in the endothelial cells within which the

immunoreactivity was localized on both luminal and abluminal sides

of the cell membrane (Fig. 9). Unfortunately, immuno-electron

microscopy for the IL-6 receptor was not successful. As for

COX-2, its expression in the endothelial cells was demonstrated

earlier in Fig. 5. Thus, at least, COX-2 and IL-1R1 were expressed in

the endothelial cells and IL-6R was found in the endothelial cells or

perivascular cells.

Discussion

IL-1b and IL-6 are both pyrogenic cytokines that are produced in the

brain under pathological conditions, although the sites of their

production seem to be distinct. For example, in a model of bacterial

infection, in which lipopolysaccharide (LPS) was injected systemic-

ally, IL-1b was produced in the meningeal/perivascular macrophages

and parenchymal microglia (Van Dam et al., 1992; Nakamori et al.,

1994), whereas IL-6 was produced in the choroid plexus (Vallieres &

Rivest, 1997). The sites of their actions might also be distinct. IL-1bwas pyrogenic when injected either peripherally or centrally (Kluger,

1991; Dinarello, 1999), whereas IL-6 caused fever only when injected

centrally in rats (LeMay et al., 1990). These distinct features between

the two cytokines led us to compare the pyrogenic mechanisms of

centrally injected IL-1b and IL-6, and a possible interaction between

them.

Fever evoked by intracerebroventricular injection of IL-1b orIL-6 is COX-2 dependent

Intracerebroventricular injection of IL-1b or IL-6 evoked fever in

rats, a result consistent with a number of earlier reports (Kluger,

1991; Dinarello, 1999). The present study demonstrated for the ®rst

time that the fever was almost completely suppressed when the

animals had been pretreated with a COX-2-speci®c inhibitor, NS-398,

indicating that the fever was evoked through a COX-2-dependent

mechanism. We have already shown that LPS evoked fever through a

COX-2-dependent mechanism when LPS was injected either peri-

pherally or centrally (Cao et al., 1997; Cao et al., 1999). In addition,

mice with genetically disrupted COX-2 gene did not evoke fever in

response to peripheral LPS injection, whereas those with a disrupted

COX-1 gene did (Li et al., 1999). Thus, COX-2 seems to be an

enzyme common to fever evoked by various in¯ammatory agents.

Intracerebroventricular injection of IL-1b or IL-6 inducesCOX-2 in brain endothelial cells

The present immunohistochemical studies demonstrated clearly that

IL-1b and IL-6 induced COX-2 in the same type of cells, i.e. brain

endothelial cells. Induction of COX-2 in brain endothelial cells

concomitant with fever was also demonstrated in other experimental

models of fever, including the peripheral injections of LPS

(Matsumura et al., 1998; Quan et al., 1998; La¯amme et al., 1999),

IL-1b (Cao et al., 1996), TNF-a (Cao et al., 1998) or turpentine

(La¯amme et al., 1999), and the central injections of LPS (Cao et al.,

1999) or TNF-a (Cao et al., 1998). Taken together, these results

highlight the crucial role of brain endothelial cells in the activation of

the brain arachidonic acid cascade during fever. We assume that the

activated endothelial cells produce PGE2 and secrete it into the brain

parenchyma and cerebrospinal ¯uid, and then, PGE2 acts on the

thermoregulatory neurons in the hypothalamus to increase body

temperature. The COX-2 mRNA level in the cortical neurons was

also elevated ipsilateral to the i.c.v. injection site. This response was,

however, not brought about by the cytokine action as saline injection

also evoked the same response without fever. Thus, the neuronal

induction of COX-2 should have no relevance to the cytokine-

induced fever. The possible mechanism of this neuronal response was

discussed in an earlier paper (Cao et al., 1999).

One critical issue to be discussed here is the temporal relation

between fever and COX-2 induction. The febrile response to IL-1band IL-6 began immediately after the cytokine injections (within

20 min), and even this quick response seemed to be COX-2-

dependent because the COX-2 inhibitor suppressed it. This immedi-

ate febrile response might be too short to be explained by de novo

production of COX-2 protein in the endothelial cells. Although

telencephalic neurons are the only cell group constitutively express-

ing detectable levels of COX-2 in the brain, their involvement in

fever is unlikely, as discussed in the last paragraph. One possible

explanation might be that the early febrile phase was supported by

FIG. 9. Immuno-electron microscopic images for IL-1R1-like immuno-reactivity in the endothelial cells under the basal condition. Bothimmunopositive cells seen are located between the basal lamina (whitearrowhead) and lumen of the blood vessel, indicating them to be endothelialcells. Within these cells, IL-1R1-like immunoreactivity (black arrows andarrowheads) is seen on both the luminal side and brain side. N, nucleus.Area indicated by white box was magni®ed and added as an inset. Electron-dense area in the nucleus does not indicate the presence of IL-1R1immunoreactivity as a similar result was obtained without the primaryantibody. Scale bar, 0.5 mm.

1788 C. Cao et al.


constitutively expressed COX-2, the level of which was too low to be

detected immunohistochemically. In fact, Western blot analysis of

isolated brain blood vessels revealed that a small amount of COX-2

protein was constitutively expressed there under the normal condi-

tions (Matsumura et al., 1998).

Localization of receptors for IL-1b and IL-6

In line with endothelial induction of COX-2, the receptor for IL-1bwas found in the endothelial cells, as demonstrated by immuno-

electron microscopy for the ®rst time, and was further upregulated by

either IL-1b or IL-6. The IL-1b receptor protein was located on both

luminal and abluminal surfaces of the endothelial cells. This receptor

localization should make IL-1b able to induce COX-2 and evoke

fever from both peripheral and central sides of the endothelial cells.

As for the IL-6 receptor, there is no doubt about its constitutive

expression in the brain under basal conditions because i.c.v. injection

of IL-6 evoked fever, and induced mRNAs for COX-2 and the

cytokine receptors. However, under the basal conditions, IL-6R

mRNA was detectable only in the cerebral cortex but not in blood

vessels, the results being in agreement with those reported by

Vallieres & Rivest (1997). This fact raises the question as to which

cells were the primary targets of IL-6 when it induced COX-2 in the

brain endothelial cells. There are two possibilities: (i) IL-6 ®rst acts

on the cortical cells to release another mediator that, in turn, acts on

the blood vessels to induce COX-2 and cytokine receptor mRNAs, or

(ii) IL-6 acts directly on the cells in the blood vessels that

constitutively expressed IL-6R protein and only a minimal amount

of its mRNA. Unfortunately, the immunostaining of IL-6R was

unsuccessful in the present study.

Unexpectedly, we found that the IL-6R mRNA level was elevated

in the cerebral cortex ipsilateral to the i.c.v. injection site. As in the

case of COX-2 mRNA, this response was not brought about by the

cytokines themselves, because saline injection also resulted in the

same response. Probably, cortical spreading depression, in which a

depolarizing wave spread over one hemisphere, triggered the

transcription of the IL-6R gene. Although it was quite a dramatic

and interesting response, its physiological signi®cance remains to be

elucidated.

IL-1b and IL-6 mutually upregulate their receptors

An important ®nding in the present study is that the two

pyrogenic cytokines upregulated not only their own receptors but

also those for the others in the brain blood vessels. This seems to

represent one part of the mutual facilitation between the two

cytokine systems. Another part of the mutual facilitation mech-

anism is the well-known fact that IL-1b enhances the biosynthesis

of itself, as well as that of IL-6 (Dinarello, 1992). This mutual

facilitation may explain why a tiny amount of cytokine is so

effective. One consequence of this mutual facilitation should be

the enhanced transcription of the COX-2 gene. The mechanism of

COX-2 gene transcription has been well studied during the last

decade. In the 5¢-upstream region of COX-2 gene exist multiple

binding sites for transcription factors, including nuclear factor-kB

and nuclear factor-IL6 (Herschman, 1996). IL-1b activates both of

these transcription factors, whereas IL-6 activates only the latter

(Bankers-Fulbright et al., 1996). Because IL-1b more or less

induces endogenous IL-6 (Dinarello, 1992), it is impossible to

completely isolate the sole action of IL-1b and to evaluate how

far IL-6 enhances the action of IL-1b. Thus, this possibility

should be tested with gene knockout mice that lack one of the

cytokine systems.

Acknowledgements

The authors would like to thank Dr Larry D. Frye for critical reading of themanuscript. We also thank Dr Yamagata of Tokyo Metropolitan Institute forNeuroscience, Dr Hart of Rutgers University and Dr Vallieres of LavalUniversity for providing us cDNAs of rat COX-2, rat IL-1R1 and rat IL-6R,respectively. This work was supported in part by the Special CoordinationFunds for Promoting Science and Technology from the Science andTechnology Agency, by grants from the programme Grants-in-Aid forScienti®c Research (C) of the Japan Society for the Promotion of Science(to K.M.) and by the Research for the Future Program (RFTF) JSPS-RFTF98L00201 from the Japan Society for the Promotion of Science (JSPS). C.C.was a recipient of an STA fellowship in Japan.

Abbreviations

COX-2, cyclooxygenase-2; DMSO, dimethyl sulphoxide; i.c.v., intracerebro-ventricular; IL, interleukin; IL-1R1, IL-1 type-1 receptor; IL-6R, IL-6receptor; ir, immunoreactive; LPS, lipopolysaccharide; OVLT, organumvasculosum laminae terminalis; PE, polyethylene; PGE2, prostaglandin E2;Tab, abdominal temperature; vW, von Willebrand.

References

Bankers-Fulbright, J.L., Kalli, K.R. & McKean, D.J. (1996) Interleukin-1signal transduction. Life Sci., 59, 61±83.

Cao, C., Matsumura, K., Ozaki, M. & Watanabe, Y. (1999)Lipopolysaccharide injected into the cerebral ventricle evokes feverthrough induction of cyclooxygenase-2 in brain endothelial cells. J.Neurosci., 19, 716±725.

Cao, C., Matsumura, K., Yamagata, K. & Watanabe, Y. (1995) Induction bylipopolysaccharide of cyclooxygenase-2 mRNA in rat brain; its possiblerole in the febrile response. Brain Res., 697, 187±196.

Cao, C., Matsumura, K., Yamagata, K. & Watanabe, Y. (1996) Endothelialcells of the rat brain vasculature express cyclooxygenase-2 mRNA inresponse to systemic interleukin-1b: a possible site of prostaglandinsynthesis responsible for fever. Brain Res., 733, 263±272.

Cao, C., Matsumura, K., Yamagata, K. & Watanabe, Y. (1997) Involvement ofcyclooxygenase-2 in LPS-induced fever and regulation of its mRNA in therat brain by LPS. Am. J. Physiol., 272, R1712±R1725.

Cao, C., Matsumura, K., Yamagata, K. & Watanabe, Y. (1998)Cyclooxygenase-2 is induced in brain blood vessels during fever evokedby peripheral or central administration of tumor necrosis factor. Mol. BrainRes., 56, 45±56.

Chai, Z., Gatti, S., Toniatti, C., Poli, V. & Bartfai, T. (1996) Interleukin (IL) -6gene expression in the central nervous system is necessary for feverresponse to lipopolysaccharide or IL-1 beta: a study on IL-6-de®cient mice.J. Exp. Med., 183, 311±316.

Dinarello, C.A. (1992) Role of interleukin-1 in infectious diseases. Immunol.Rev., 127, 119±146.

Dinarello, C.A. (1999) Cytokines as endogenous pyrogen. J. Infect. Dis., 179(Suppl. 2), S294±S304.

Futaki, N., Arai, I., Hamasaka, Y., Takahashi, S., Higuchi, S. & Otomo, S.(1993a) Selective inhibition of NS-398 on prostanoid production inin¯amed tissue in rat carrageenan-air-pouch in¯ammation. J. Pharm.Pharmacol., 45, 753±755.

Futaki, N., Takahashi, S., Yokoyama, M., Arai, I., Higuchi, S. & Otomo, S.(1994) NS-398, a new anti-in¯ammatory agent, selectively inhibitsprostaglandin G/H synthase/cyclooxygenase (COX-2) activity in vitro.Prostaglandins, 47, 55±59.

Futaki, N., Yoshikawa, K., Hamasaka, Y., Arai, I., Higuchi, S., Iizuka, H. &Otomo, S. (1993b) NS-389, a novel non-steroidal anti-in¯ammatory drugwith potent analgesic and antipyretic effects, which causes minimal stomachlesions. Gen. Pharmacol., 24, 105±110.

Gierse, J.K., Hauser, S.D., Creely, D.P., Koboldt, C., Rangwala, S.H., Isakson,P.C. & Seibert, K. (1995) Expression and selective inhibition of theconstitutive and inducible forms of human cyclo-oxygenase. Biochem. J.,305, 479±484.

Herschman, H.R. (1996) Prostaglandin synthase 2. Biochim. Biophys. Acta,1299, 125±140.

Kluger, M.J. (1991) Fever: role of pyrogens and cryogens. Physiol. Rev., 71,93±127.

La¯amme, N., Lacroix, S. & Rivest, S. (1999) An essential role of interleukin-1beta in mediating NF-kappaB activity and COX-2 transcription in cells of



the blood±brain barrier in response to a systemic and localized in¯ammationbut not during endotoxemia. J. Neurosci., 19, 10923±10930.

LeMay, L.G., Vander, A.J. & Kluger, M.J. (1990) Role of interleukin 6 infever in rats. Am. J. Physiol., 258, R798±R803.

Li, S., Wand, Y., Matsumura, K.R.B.L., Morham, S.G. & Blatteis, C.M.(1999) The febrile response to lipopolysaccharide is blocked incyclooxygenase-2 (±/±), but not in cyclooxygenase-1 (±/±) mice. BrainRes., 825, 86±94.

Masferrer, J.L., Zweifel, B.S., Manning, P.T., Hauser, S.D., Leahy, K.M.,Simth, W.G., Isakson, P.C. & Seibert, K. (1994) Selective inhibition ofinducible cyclooxygenase 2 in vivo is antiin¯ammatory and nonulcerogenic.Proc. Natl. Acad. Sci. USA, 91, 3228±3232.

Matsumura, K., Cao, C., Ozaki, M., Morii, H., Nakadate, K. & Watanabe, Y.(1998) Brain endothelial cells express cyclooxygenase-2 duringlipopolysaccharide-induced fever: Light and electron microscopicimmunocytochemical studies. J. Neurosci., 18, 6279±6289.

Nakamori, T., Morimoto, A., Yamaguchi, K., Watanabe, T. & Murakami, N.(1994) Interleukin-1 beta production in the rabbit brain during endotoxin-induced fever. J. Physiol. (Lond.), 476, 177±186.

Paxinos, G. & Watson, C. (1986) The Rat Brain in Stereotaxic Coordinates.Academic Press, London.

Quan, N., Whiteside, M. & Herkenham, M. (1998) Cyclooxygenase 2 mRNAexpression in rat brain after peripheral injection of lipopolysaccharide.Brain Res., 802, 189±197.

Stitt, J.T. (1986) Prostaglandin E as the neural mediator of the febrileresponse. Yale J. Biol. Med., 59, 137±149.

Sundgren-Andersson, A.K., Ostlund, P. & Bartfai, T. (1998) IL-6 is essentialin TNF-alpha-induced fever. Am. J. Physiol., 275, R2028±R2034.

Vallieres, L. & Rivest, S. (1997) Regulation of the genes encoding interleukin-6, its receptor, and gp130 in the rat brain in response to the immuneactivator lipopolysaccharide and the proin¯ammatory cytokine interleukin-1beta. J. Neurochem., 69, 1668±1683.

Van Dam, A.-M., Brouns, M., Louisse, S. & Berkenbosch, F. (1992)Appearance of interleukin-1 in macrophages and in rami®ed microglia inthe brain of endotoxin-treated rats: a pathway for the induction of non-speci®c symptoms of sickness? Brain Res., 588, 291±296.

Wagner, D.D. & Marder, V.J. (1983) Biosynthesis of von Willebrand proteinby human endothelial cells. J. Biol. Chem., 258, 2065±2067.

1790 C. Cao et al.


Documents

Pyrogenic cytokines injected into the rat cerebral ventricle induce cyclooxygenase-2 in brain endothelial cells and also upregulate their receptors