You are on page 1of 20

Neuroscience 140 (2006) 1415–1434

NEURAL–IMMUNE INTERFACE IN THE RAT AREA POSTREMA


L. E. GOEHLER,* A. ERISIR AND R. P. A. GAYKEMA nicate (Blalock, 1994; Salzet et al., 2000). Subsequent
Program in Sensory and Systems Neuroscience, Department of Psy- findings revealed a close functional relationship between
chology and Neuroscience Graduate Program, University of Virginia, the immune and nervous systems, indicating that these
Charlottesville, VA 22904, USA two systems are acting in concert to regulate neural, en-
docrine, and immune functioning. Notably, immune activa-
Abstract—The area postrema functions as one interface be- tion can provoke dramatic changes in nervous system
tween the immune system and the brain. Immune cells within functioning, leading to an adaptive reorganization (Hart,
the area postrema express immunoreactivity for the pro-in- 1988) that include the familiar symptoms of sickness, in-
flammatory cytokine, interleukin-1␤ following challenge with cluding fever and social withdrawal (Elmquist et al., 1997b;
immune stimulants, including lipopolysaccharide (from bac- Dantzer, 2004), as well as alterations in mood (anxiety or
terial cell walls). As a circumventricular organ, the area pos-
depression) and cognition (Maier and Watkins, 1998;
trema accesses circulating immune-derived mediators, but
also receives direct primary viscerosensory signals via the Dantzer, 2004). As yet, potential mechanisms that underlie
vagus nerve. Neurons in the area postrema contribute to these brain-mediated effects continue to fuel debate.
central autonomic network neurocircuitry implicated in brain- The brain has traditionally been considered as an im-
mediated host defense responses. These experiments were mune-privileged structure, based on the lack of a well-
directed toward clarifying relationships between immune defined lymphatic system or dedicated immune structures
cells and neurons in the area postrema, with a view toward such as lymph nodes, and the presence of barrier func-
potential mechanisms by which they may communicate. We
used antisera directed toward markers indicating microglia
tions (blood– brain and blood– cerebrospinal fluid barriers)
(CR3/CD11b; OX-42), resident macrophages (CD163; ED-2), that exclude most hydrophilic and large blood-borne mol-
or dendritic cell-like phenotypes (major histocompability ecules from access. This belies the fact that the brain is
complex class II; OX-6), in area postrema sections from li- host to rich and diverse populations of immune cells, in-
popolysaccharide-treated rats processed for light, laser cluding microglia within the brain parenchyma (Gehrmann
scanning confocal, and electron microscopy. Lipopolysac- et al., 1995), perivascular cells associated with blood ves-
charide treatment induced interleukin-1␤-like immunoreactiv-
sels throughout the brain parenchyma (Williams et al.,
ity in immune cells that either associated with the vasculature
(perivascular cells, a subtype of macrophage) or associated 2001), as well as the macrophages and dendritic-like cells
with neuronal elements (dendritic-like, and unknown pheno- that populate the choroid plexus, meninges, and circum-
type). Electron microscopic analysis revealed that some im- ventricular organs “outside” the blood– brain barrier (Mc-
mune cells, including interleukin-1␤-positive cells, evinced Menamin, 1999; McMenamin et al., 2003). These myeloid-
membrane apposition with neuronal elements, including den- derived cells contribute to a variety of functions, including
drites and terminals, that could derive from inputs to the area immune surveillance, as well as the production of neuro-
postrema such as vagal sensory fibers, or intrinsic area pos-
trophic factors supporting development, wound repair, and
trema neurons. This arrangement provides an anatomical
substrate by which immune cells could directly and specifi- plasticity (Batchelor et al., 2002; Boulanger and Shatz,
cally influence individual neurons in the area postrema, that 2004; Roumier et al., 2004). In addition, a subpopulation of
may support the induction and/or maintenance of brain re- brain immune cells, including perivascular cells as well as
sponses to inflammation. © 2006 IBRO. Published by Elsevier other, less well-characterized cells (macrophage/micro-
Ltd. All rights reserved. glia-like) responds to early phases of peripheral immune
activation with the expression of pro-inflammatory media-
Key words: interleukin-1, lipopolysaccharide, microglia, den-
dritic cell, perivascular cell, macrophage. tors, such as prostaglandins (Elmquist et al., 1997a;
Schiltz and Sawchenko, 2002) and cytokines (Nakamori et
al., 1994; Konsman et al., 1999; Van Dam et al., 1995,
The discovery of neuropeptides and similar “neuroactive” 2000). Thus, these cells may play an immune sensory role,
substances in immune cells initiated a new perspective on i.e. contribute to the induction and modulation of brain-
the potential ways that immune and neural cells commu- mediated host defense responses (as above).
*Correspondence to: L. E. Goehler, Department of Psychology, P.O. Whereas brain-derived prostaglandins seem to play a
Box 400400, University of Virginia, Charlottesville, VA 22904, USA. role in the induction of thermogenic and neuroendocrine
Tel: ⫹1-434-243-3547; fax: ⫹1-434-982-4785. responses (Ericsson et al., 1997; Roth et al., 2002), among
E-mail address: goehler@virginia.edu (L. E. Goehler).
Abbreviations: ABC, avidin– biotin peroxidase complex; CR3, comple- the array of signals generated in response to immune
ment receptor 3; DAB, 3,3=-diaminobenzidine; GFAP, glial fibrillary activation or inflammation, the induction of IL-1␤ within the
protein; IL-1␤, interleukin-1␤; IR, immunoreactivity; LPS, lipopolysac- brain is believed to constitute a critical point in the cascade
charide; MHC, major histocompability complex; PB, phosphate buffer;
PBS, phosphate-buffered saline; PBS-T, phosphate-buffered saline with of events leading to behavioral responses including social
0.5% Triton X-100; TH, tyrosine hydroxylase; TLR, Toll-like receptor. withdrawal and psychomotor retardation associated with
0306-4522/06$30.00⫹0.00 © 2006 IBRO. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.neuroscience.2006.03.048

1415
1416 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

behavioral depression (Kent et al., 1992; Bluthe et al., lipopolysaccharide (LPS). Based on ultrastructural charac-
1995; Nadjar et al., 2005). Within the timeframe during teristics and expression of cell marker proteins, we were
which behavioral responses become evident (one to two able to identify both perivascular and dendritic-like immune
hours after immune challenge), immune cells associated cells that produce IL-1␤ following immune challenge, and
with the meninges, choroid plexus, and circumventricular demonstrate that some of these cells make direct physical
organs are the predominant, if not only, sources of inter- contact with neuronal elements within the area postrema.
leukin-1␤ (IL-1␤) in the brain (Nakamori et al., 1994; Van
Dam et al., 1995, 2000; Vernet-der Garabedian et al., EXPERIMENTAL PROCEDURES
2000). Taken together, these findings indicate that CNS
sources of IL-1␤ that are effectively outside the blood– Animals
brain barrier appear critical for the induction and/or main- The animals used for these studies were 48 Sprague–Dawley rats
tenance of sickness behaviors such as social withdrawal. of either sex weighing between 300 and 400 g (Taconic Labora-
In circumventricular organs (e.g. subfornical organ, or- tories, Germantown, NY, USA). The animals were housed in pairs
ganum vasculosum lamina terminalis, area postrema), un- in polypropylene cages within a barrier cage rack (Allentown
like other barrier-associated tissues, immune cells coexist Caging Smart Bio-pak, Allentown, NJ, USA) under standard hous-
ing conditions (12-h h light/dark cycle, lights on at 7 AM) with food
with neurons. This arrangement might allow for a potential and water available ad libitum. All experiments and procedures
direct and specific (cell-to-cell) interaction between im- were carried out in accordance with the National Institutes of
mune cells and neurons, in contrast to the paracrine-type Health Guide for the Care and Use of Laboratory Animals (NIH
action by which cytokines may act via volume transmis- Publications No. 80 –23; revised 1996) and were in accordance
sion. However, evidence for such an interaction is lacking. with protocols approved by the University of Virginia Institutional
The demonstration that cells apparently resident in these Animal Care and Use Committee. All efforts were made to mini-
mize number of animals used and their suffering.
structures respond to immune activation with the induction
of immunoreactivity (IR) for the pro-inflammatory cytokine, Injection of LPS
IL-1␤ or the uncleaved pro-IL-1␤ (Konsman et al., 1999;
Van Dam et al., 2000) and IL-1␤ mRNA (Quan et al., 1998, Expression of the cytokine IL-1␤ within 2– 4 h following peripheral
1999) provided strong evidence that these cells are playing pro-inflammatory challenge is a hallmark of cerebral innate im-
mune cells that may serve a sentinel function in the brain. Be-
a sensory role, but detailed information regarding specific cause immunohistochemically detectable IL-1␤ protein is reliably
cell types expressing (pro-)IL-1␤ IR, or their physical rela- induced by a pro-inflammatory stimulus, e.g. bacterial endotoxin
tionships to neurons is as yet, lacking. Whereas Konsman (LPS) derived from gram-negative bacteria, animals received ei-
et al. (1999) demonstrated colocalization of IL-1␤ IR in ther i.p. injection of LPS (n⫽25) or saline/no injection (n⫽23) prior
cells with isolectin B4 (indicating macrophage/microglia to kill and tissue dissection. LPS (from E. coli, serotype 0111:B4,
lineage) in the area postrema, isolectins nonspecifically Sigma, St. Louis, MO, USA) was dissolved in sterile 0.9% NaCl
and injected i.p. at doses of either 0.1 (n⫽20) or 0.4 (n⫽5) mg/kg.
label various cell types, including the central processes of
The animals were deeply anesthetized either 120 (n⫽6), 150
peripheral nerves. From this work it is not possible to know (n⫽2), or 240 (n⫽10) minutes later with an i.p. injection of Nem-
whether IL-1␤ IR derives from microglia, perivascular cells butal (60 mg/kg) and perfused transcardially briefly with Tyrode’s
or other subtype of macrophages, dendritic-like cells, or solution followed by fixative containing 4% paraformaldehyde,
other cells. Such information would be valuable as specific 15% (v/v) saturated picric acid, and with 0.25% glutaraldehyde
populations of ED2-positive perivascular and meningeal added, in 0.1 M sodium phosphate buffer (PB; pH 7.4). The brains
were removed and kept in cold fixative without glutaraldehyde
macrophages express distinct cell membrane proteins
overnight, and stored the next day in cold PB containing 0.1%
such as mannose receptors that can be specifically tar- sodium azide as preservative.
geted in drug therapy (Polfliet et al., 2001). Further, it is not
known what type of signaling mechanisms (paracrine vs. Tissue processing
direct, cell-cell) might obtain. Thus, the precise mecha-
Antibodies. Non-neural cells in the area postrema were
nisms by which immune cells in circumventricular organs
labeled using antibodies directed toward proteins expressed by
signal neurons are unknown. Nonetheless, this previous different cell phenotypes. For myeloid immune-derived cells we
work has underlined a potentially important immune sen- assessed IR (1) for major histocompability complex (MHC) class II, a
sory role for cells in circumventricular organs. marker for antigen presenting cells, most notably, but not exclusively,
The objectives of the studies reported here were to dendritic cells (Banchereau and Steinman, 1998), using the mouse
explore potential mechanisms by which brain-derived monoclonal OX-6, (2) for microsialin (CD68), a marker for peripheral
and meningeal macrophages, using the monoclonal ED1, (3) for the
IL-1␤ can access or influence neuronal circuits that may
scavenger receptor CD163-like antigen, using ED2, and (4) for mi-
mediate sickness behavior, by 1) identifying phenotypes of croglia, using the OX-42 monoclonal directed toward complement
the immune cells cohabiting the area postrema that ex- receptor 3 (CR3 or CD11b). These monoclonals were purchased
press IL-1␤ IR upon immune challenge, and 2) determining from Biosource (OX-6: cat. # ARU0111, ED1: cat. # ARU0151),
whether direct interactions, via physical contact, occur be- BD-Pharmingen (biotinylated OX-6: cat. # 554927, OX-42: cat. #
tween IL-1-expressing, or other immune cells, and neu- 554859) and Serotec (ED2: cat. # MCA342R). We also labeled
astrocytes using a rabbit polyclonal antibody directed toward glial
rons. Double-label light, confocal laser-scanning, and
fibrillary protein (GFAP, from Chemicon, cat. # AB5804), because of
transmission electron microscopy were used to distinguish their ramified morphology similar to microglia and some reports
different immune cell types and their relationships to neu- regarding their ability to express IL-1␤ (Dong and Benveniste,
rons, in brain sections from rats challenged in vivo with i.p. 2001). Polyclonal affinity-purified anti-rat IL-1␤ antibodies, ob-
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1417

tained from Abcam (raised in rabbit, cat. # ab9787, lot 80734) and (1:1000), rabbit anti-IL-1␤ (1:2000), goat anti-IL-1␤ (1:2000) over-
from R&D Systems (raised in goat, cat. # AF-501-NA, lot night at room temperature (RT). The next day the slides were
YR052091) were used to identify LPS-responsive cells in the area washed and incubated in biotin-conjugated secondary antibody
postrema in endotoxin-challenged rats. Finally, we used a rabbit (goat anti-mouse IgG or goat anti-rabbit IgG (1:500; overnight at
polyclonal raised against the catecholamine synthesizing enzyme 4 °C). All antibodies were diluted in phosphate-buffered saline with
tyrosine hydroxylase (TH), from Chemicon (cat. # AB152), to label 0.5% Triton X-100 (PBS-T) containing 0.1% sodium azide and
a major subset of area postrema neurons in conjunction with 10% NGS. Finally, sections were incubated in avidin– biotin per-
double labeling with OX-6. The pan-neuronal marker PGP9.5 oxidase complex (ABC, Vector Elite kit, 1:100 in PBS-T, 2 h;
(Chemicon, cat. # AB1761) was applied in combination with the Vector Laboratories, Burlingame, CA, USA) and rinsed in Tris–
anti-IL-1 antibodies to explore possible neuronal origin of IL-1␤ HCl (pH 7.6). Peroxidase staining was performed using 3,3=-
expression. The anti-IL-1␤ antibodies do not show cross-reactivity diaminobenzidine (DAB, Sigma) as a chromogen dissolved in
to other cytokines (GM-CSF, IL-1␣, IL-2, IL-4, IL-10, TNF-␣) ac- Tris–HCl (40 mg/100 ml), glucose oxidase (Sigma; type V-S,
cording to the vendors’ specifications. The specificity of the anti- 0.02%) and ␤-D-glucose (0.1% w/v) to generate hydrogen perox-
IL-1␤ antisera was further assessed by preabsorption of the incu- ide (modified after Shu et al., 1988), yielding a reddish– brown
bation solution with the immunogen recombinant mature rat IL-1␤ precipitate. In some instances, nickelous ammonium sulfate was
(R&D Systems, cat # 501-RL) and by Western blotting of rat brain added (1.50 mg/ml) to the DAB (20 mg/100 ml) solution to inten-
(area postrema) tissue as described below. sify the staining yielding a purplish– black color. One series of
sections was processed as above except that the primary anti-
General pre-incubation treatments. The brainstems con- body was omitted from the incubation buffer (omission controls).
taining the area postrema were dissected and sectioned on a The slides were dried overnight, cleared in ethyl alcohol and
vibratome at a thickness of 50 ␮m. Sections were collected seri- Histoclear, and coverslipped with DPX prior to viewing with an
ally in ice-cold PB (pH 7.4) containing 0.1% sodium azide. Sec- Olympus BX-51 brightfield microscope. Images were collected
tions containing the area postrema were then selected (approxi- with a Magnafire digital camera (Optronics) coupled to an Apple
mately six to eight per rat) and either assigned to procedures for PowerMac G4 equipped. The obtained micrographs were slightly
electron microscopic analysis or for light microscopic analysis. adjusted for brightness and contrast and labeled using Adobe
Sections for electron microscopy underwent the following treat- Photoshop CS2 (version 9, MacOS 10.3). Except for these minor
ments prior to any of the various immunohistochemical detec- adjustments, the images were not altered.
tion protocols as described below. First, the sections were
immersed in 20% sucrose in PB, flash-frozen on dry ice-chilled Double labeling immunoperoxidase. Selected sections were
isopentane while flattened in aluminum weighing boats, and stained for two different antigens by following above-described pro-
subsequently warmed back to room temperature. This freeze– cedure twice in succession, the first reacted in nickel-enhanced
thaw procedure was repeated one more time to improve antibody DAB to yield a blue/purplish– gray to black staining, the second
penetration without seriously compromising ultrastructural fea- reacted in regular DAB to yield a reddish– brown staining. Follow-
tures. Second, sections were transferred to phosphate-buffered ing the first reaction, sections were treated with the hydrogen
saline (PBS; 0.145 M NaCl, 0.01 M PB), and treated with freshly peroxidase–sodium azide mix (0.3% and 0.1%) again to eliminate
dissolved sodium borohydrate (0.5%) for 30 min, followed by three residual peroxidase activity before starting the second procedure.
rinses in PBS. Thereafter the sections were treated with 0.3% The following combinations of primary antibodies were applied: (1)
hydrogen peroxide and 0.1% sodium azide in PBS for 30 min to OX-6 and TH to assess the fine anatomical relationship between
quench endogenous peroxidase activity. This latter procedure the MHC class II-expressing cells and the abundant TH-positive
was omitted if sections were not subjected to any peroxidase AP neurons; (2) OX-42 and OX-6 to address the possible co-
staining procedure later on. The third treatment involved the avi- expression of both immune cell markers; and (3) IL-1␤ and OX-6
din– biotin blocking procedure using Vector’s blocking kit, leading to determine whether the constitutive MHC class II-expressing
to a drastically reduced “noise” in staining methods utilizing biotin- immune cells are among those that respond to LPS challenge with
assisted detection. Sections were incubated in avidin and biotin the production of IL-1␤ protein. Care was taken to avoid intense
(four drops per ml PBS) for 1 h each interspersed with rinses in precipitation of either reaction product as that hinders the reliable
PBS. Fourth, nonspecific antibody binding was reduced by incu- detection of both in the cells. Colocalization was indicated by
bating the sections overnight in Fab= fragments of anti-rat IgG or blending of the gray and reddish– brown colors or by the presence
anti-mouse IgG (1:200, raised in either goat or donkey to match of both colors in different compartments of the same cell. Although
normal serum and secondary antibody host), diluted in incubation this method carries limitations due to possible visual interference
buffer modified for EM (PBS containing 0.1% sodium azide and due to the nature of the reaction products, it has proven to be
3% normal serum, of which the host is the same as the secondary superior in showing the fine morphological features of the cells
antibody utilized, e.g. goat or donkey). All incubations lasting and their fine ramifications at the light microscopic level.
overnight or longer were kept at low temperature (4 °C) under Because OX-6 and OX-42 are both mouse monoclonal anti-
gentle agitation. The sections selected for light and confocal laser bodies, we adapted the protocol to avoid or minimize cross-
scanning microscopy were treated as described above with the reactivity. First sections were incubated in OX42, followed by
following notable exception: instead of undergoing the flash freeze– biotinylated Fab= fragment of goat anti-mouse IgG (1:500; Jack-
thaw procedure they were to incubation buffer to which 0.5% son Immunoresearch) and ABC. After staining with nickel-en-
Triton X-100 was added for all pre-incubations (e.g. avidin– biotin hanced DAB, sections were treated with the avidin– biotin blocking kit
block, normal sera and Fab= anti-rat/mouse IgG). For IL-1␤ immu- and subsequently incubated overnight in biotinylated OX-6 (Pharm-
nostaining, non-specific background was reduced by pre-incubat- ingen, 1:400), followed by ABC and staining in regular DAB.
ing anti-IL-1␤ antisera overnight with brain sections from rats that
received saline injections or were untreated (which tissue was not Double-labeling immunofluorescence. The phenotypic char-
further used in these experiments), before being applied to tissue acteristics of the cells in the area postrema that express IL-1␤
sections as described above. protein in response to systemic endotoxin challenge were estab-
lished by concurrent immunofluorescent detection of IL-1␤ protein
Immunoperoxidase histochemistry for light microscopy. Fol- and either PGP9.5, a pan-neuronal marker, or each of the innate
lowing the pre-embedding steps, selected sections were washed and immune cell markers detectable with the monoclonals ED2, OX-
incubated in one of the following primary antibodies: the mono- 42, and OX-6. All antibodies were diluted in incubation buffer
clonals OX-6, OX-42, ED1, ED2 (1:300 –1:500), rabbit anti-GFAP containing 0.5% Triton X-100, 0.1% sodium azide, and 3% normal
1418 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

donkey or goat serum (host for serum is the same as that for the particles (British Biocell international, Ted Pella, Inc.) or 4 nm
secondary antibody). One selection of sections was incubated in colloidal gold particles (Jackson ImmunoResearch). The sections
goat anti-rat IL-1␤ for 24 – 48 h, followed by Cy3-conjugated don- were rinsed in PBS, fixed in 1% glutaraldehyde for 5 min, and
key anti-goat IgG (Jackson Immunoresearch) overnight at 4 °C. transferred to sodium citrate buffer (0.1 M, pH 6.5) prior to silver
Subsequently the sections were incubated in either of the follow- enhancement. Silver enhancement was carried out according to
ing antibodies diluted in the same incubation buffer: PGP9.5 (1: the manufacturer’s instructions (IntenSE kit, Amersham), and the
1000), OX-6 (1:500), OX-42 (1:250), and ED2 (1:400) for 24 h development was terminated upon visual confirmation of staining
followed by FITC-conjugated F(ab)2 fragment of donkey anti-rab- in the light microscope. The sections were washed in citrate
bit or mouse IgG (1:200) overnight. Finally the sections were buffer, transferred to 0.1 M PB before osmification and resin
mounted and coverslipped in Vectashield anti-fading medium embedding.
(Vector Laboratories). Another selection of sections was incu- Another selection of sections through the area postrema un-
bated in rabbit anti-rat IL-1␤ antiserum followed by Alexa Fluor derwent single immunoperoxidase staining for rat IL-1␤ with DAB
488-labeled goat anti-rabbit IgG (1:200). Antibodies were diluted as the chromogen. Sections passed through the following incuba-
in incubation buffer containing 3% normal goat serum), with incu- tions rabbit anti-IL-1␤ (1:2000, diluted in EM incubation buffer with
bation times as described above. The following sequence involved 3% NGS for 24 – 48 h), biotinylated goat anti-rabbit IgG (1:500;
either ED2, OX-6, or OX-42 (dilution as above) followed by Alexa overnight at 4 °C), and ABC (Vector, 1:500 in PBS). The sections
Fluor 546-conjugated goat anti mouse IgG (1:200). These sec- were rinses in Tris–HCl, and reacted in DAB. Alternatively, sec-
tions were mounted on microscope slides from PBS and cover- tions were incubated in goat anti IL-1␤ (1:2000, in incubation
slipped in glycerol/PBS mixture. The immunofluorescence was solution containing NDS) and biotinylated donkey anti-goat IgG
confirmed using an Olympus BX51 equipped with epi-fluores- (1:500) before proceeding to ABC and DAB. Following completion
cence filter sets (Chroma) specific for each of the fluorochromes. of the staining, sections underwent osmification and resin embed-
In order to obtain sharp optical focus at different levels through the ding as described below.
sections, colocalization of the fluorescent markers was estab-
lished with the aid of a laser scanning confocal microscope (Olym- Dual immunogold/peroxidase labeling for electron microscopy.
pus Fluoview B2111) with 40⫻ objectives. Krypton/argon lasers of Sections through the area postrema were first processed for IR for
488 nm (green) and 543 nm (red) wavelengths were used in OX-6, ED2, and OX-42 with immunogold labeling using the pro-
combination with barrier filters, set to excite the specimen with cedure described above. Following the silver intensification, sec-
only one laser line at a time. Optical sections were generated at a tions were washed in PBS and subjected to a second, immuno-
resolution of 1024⫻1024 pixels for each fluorochrome by separate peroxidase staining procedure involving immunoperoxidase stain-
scans of the specimens in subsequent planes that were either 2 or ing for TH to label catecholaminergic neurons that are abundant in
3 ␮m apart, which is well within the diameter of most cells. The the area postrema. The second staining sequence included incu-
excitation emitted from the specimens was captured and visual- bation in the primary antibody for 24 – 48 h at 4 °C, followed by
ized using FluoView FV300 software version 3.3 (Olympus Optical biotinylated goat anti-rabbit IgG (overnight), and ABC (4 h, over-
Co., Ltd.). For semi-quantitative assessment of colocalization of night). The sections were then stained with DAB. Thereafter the
IL-1␤ with each of the immune cell phenotypes labeled with ED2, sections were transferred to 0.1 M PB and prepared for osmifica-
OX-6 and OX-42, one section per rat through the rostral third of tion and resin embedding.
the area postrema was selected for each combination of double
labeling from the four LPS-treated rats used (0.4 mg/kg i.p., two at Osmification, embedding and ultrathin sectioning for EM.
2.5 and two at 4 h). The sections were scanned at multiple z axis The sections designated for EM examination were postfixed with
levels separated by 3 ␮m and the cells fluorescing for ED2, OX-6 osmium tetroxide and embedded in resin using routine protocols.
or OX-42 were counted, as well as those that show IL-1␤ immu- In short, the sections were fixed in 1% osmium tetroxide (in 0.1 M
nofluorescence, and those co-expressing both fluorophores. From PB) for 1 h, dehydrated in increasing concentrations of ethyl
these numbers the percentage of colocalization for each rat was alcohol, during which they were incubated en bloc in 4% uranyl
calculated. As no differences were noted between the two time acetate (in 70% ethyl alcohol) for increased contrast for a mini-
points, data from all four animals were pooled. Sections from a mum of 2 h. Then the sections were transferred into 90 and 100%
fifth saline-treated rat were included and showed fluorescence for ethyl alcohol, rinsed in acetone twice, and left in a 1:1 mixture of
the ED2, OX-6 and OX-42 indistinguishable from the endotoxin acetone and EPON resin (EMBED 812; EMS) overnight, and
treated animals, but completely lacked immunofluorescence for transferred to full resin. Sections were flat embedded between two
IL-1␤. sheets of Aclar, and the resin allowed to polymerize overnight at
60 °C. Small pieces containing the area postrema were excised
Electron microscopy and laid flat on the inner surfaces of BEEM capsule caps. The
capsules were filled with EPON resin and polymerized overnight
Single immunolabeling of innate immune cell markers and at 60 °C. Ultrathin sections were cut at a thickness of 90 –100 nm
IL-1␤ for electron microscopy. Selected sections through the on an ultramicrotome (Leica), placed on copper mesh grids, and
area postrema were stained for the three immune cell markers examined on a JEOL 1010 transmission electron microscope.
detected by ED2, OX-42 and OX-6 using a pre-embedding immu- Electron micrographs were captured on film (Kodak 4489) that
nogold and silver enhancement protocol. Immunogold labeling was developed with Kodak D-19 developer and commercial fixer
combined with silver enhancement provides a sensitive technique (Eastman Kodak Co., Rochester, NY, USA). Each negative was
for both light and electron microscopy in particular for antigens then digitized using an optical scanner (Epson Perfection 1200
present on the outer cell surface, which is the case for those Photo) at a pixel density of 600 d.p.i. In addition, images were
identified with the ED2, OX-6 and OX-42 monoclonal antibodies captured with the use of a SIA-12C digital camera (siacam.com).
(phenotypical markers for resident tissue macrophages, putative Images were taken using a 2⫻2 binning yielding a density of
dendritic sentinel cells, and microglia, respectively). All incuba- 2048⫻2048 pixels. The digitized images were imported with the
tions were done at 4 °C. Sections were incubated for 24 – 48 h in MaxIm DL CCD software and stored as eight bit TIFF files after
either ED2, OX-6, or OX-42 diluted 400⫻ in incubation solution “stretch” procedure to yield optimal brightness and contrast
adapted for EM (PBS containing 3% NGS and 0.1% sodium azide; properties.
Triton X-100 was omitted). Following three rinses in PBS, the
sections were transferred to and incubated overnight in solution Omission and pre-absorption controls. Because non-spe-
containing goat anti-mouse IgG conjugated with 1 nm size gold cific or spurious staining can emerge especially when using poly-
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1419

clonal antibodies (Saper and Sawchenko, 2003), we applied sev- to-dorsal central region (Fig. 1A), and, albeit fewer in num-
eral procedures to test the specificity of the anti-IL-1b antibodies in ber, along the lateral borders with the subjacent nTS. The
the brain tissue used in this study. Selected sections through the MHC class II-IR cells were morphologically highly hetero-
area postrema were subjected to pre-incubation treatments to
reduce non-specific immunolabeling as described above. These
geneous. Most cells were quite large with somata ranging
sections, from both saline- and LPS-treated rats were then incu- from 15 ␮m to 30 ␮m. These cells exhibited two to three
bated in non-specific rabbit or mouse IgG to replace the primary dendritic-like processes and were either spindle shaped,
antisera. In addition, sets of consecutive sections were incubated with prominent processes on either end, or shaped rather
in two solutions with the anti-rat IL-1␤ antisera (1:2000), of which like a rocket (Fig. 2A) with a tuft of processes emanating
one was first adsorbed overnight at 4 °C with an excess of the
from one end and a single, sometimes branching process
immunogen rat recombinant IL-1␤ (R&D Systems, 2 ␮g/0.5 ␮g
primary antibody). The sections were then processed for immu- extending opposite. Other cells exhibited a marked pyra-
noperoxidase labeling in an identical way as described above for midal shape (Fig. 2A). The dendritic processes varied in
light microscopy. length and number of ramifications, and some cells pos-
sessed an extensive network of branches. A few cells were
Western blot analysis. Two male rats were used for this
purpose, one of which was challenged with LPS (0.4 mg/kg i.p.) oval or round and lacking obvious processes, and others,
and the other with sterile saline. Four hours later, they were more rarely, were highly ramified with a smaller soma, and
perfused transcardially with Tyrode’s solution, after which the resembled microglia in morphology (Fig. 2B). The number
brains and pituitary glands were rapidly removed. Selected brain of MHC class II-immunoreactive cells varied among indi-
regions were dissected on an ice-cold surface, weighed, and vidual rats and their distribution was often skewed toward
flash-frozen in liquid nitrogen after which they were stored at
⫺80 °C until use. The dorsal medulla, including the area pos-
the dorsal aspect of the area postrema in some cases and
trema, and pituitary glands (weighing 10 –12 mg) were homoge- along the lateral border in others. Cells labeled constitu-
nized in 25 ␮l 20 mM HEPES buffered saline (pH 7.8) containing tively with OX-6 were not encountered within the nTS, but
1 mM EDTA, 1 mM EGTA, and protease inhibitors, and centri- consistently present in adjacent choroid plexus and, when
fuged at 14,000 r.c.f. The supernatant was derived from each still attached, also on the leptomeninges (pia mater).
sample and mixed with Laemmli loading buffer (Bio-Rad) and 5%
␤-mercaptoethanol, brought to 100 °C for 2 min and loaded onto OX-42: CD11b (CR3). As they do within the brain
a polyacrylamide minigel (Bio-Rad) together with recombinant rat parenchyma, the OX-42-labeled cells were distributed
IL-1␤ (10 ng/lane), and a prestained precision protein ladder for throughout the area postrema in a more or less uniform
the estimation of protein size. Proteins were separated at 150 V
and transferred onto a PVDF membrane (Bio-Rad) overnight at 30
fashion that was virtually the same among the different
V. The membrane was subsequently incubated in blocking solu- animals (Fig. 1B). In contrast to the rather idiosyncratic
tion (5% nonfat milk in PBS and 0.1% Tween-20) for 30 min, in distribution and morphological features of the cells labeled
rabbit anti-IL-1␤ (1:5000, previously pre-absorbed with normal with OX-6, those expressing CR3/CD11b evinced classic
brain sections, as above for immunohistochemistry) overnight at microglial morphology, with small oval or elongated so-
4 °C, followed by biotinylated goat anti-rabbit (1:1000) for 1 h and
mata and numerous branching processes that showed a
streptavidin–alkaline phosphatase (Jackson ImmunoResearch,
1 ␮g/ml Tris-buffered saline, pH 8.0). The membrane was stained classic ramified pattern (Fig. 2D).
with a BCIP/NBT substrate kit (Vector). ED2: CD163-like antigen (scavenger receptor). ED2-
positive cells generally showed an even distribution
RESULTS
throughout the entire area postrema with no apparent pref-
The experiments reported here were directed toward iden- erence to certain portion (Fig. 1C). In addition they were
tifying subtypes of immune cells resident in the area pos- encountered adjacent to the fourth ventricle, and also oc-
trema, and determining whether, given the cohabitation of cupied meninges. ED2, like OX-6, labeled a population of
immune cells and neurons in this structure, any direct cells characterized by large somata. A ramified appear-
interactions occur (appositions with possible synaptic spe- ance was rare among ED2-positive cells, but otherwise
cializations in cell membranes). The results revealed mul- these cells were morphologically similar to the spindle,
tiple immune cell phenotypes, some of which abutted neu- rocket and pyramidal shaped OX-6 positive cells (Fig. 2E).
ronal elements, and respond to peripheral immune chal- However, the branches appear to be generally shorter and
lenge by inducing expression of IL-1␤-like IR. lack the extensive ramification characteristic for many
OX-6 labeled cells. A microglia-like morphology was never
Expression patterns of immune cell markers in the
seen among ED2-positive cells. In contrast to cells ex-
area postrema
pressing OX-6 and OX-42, ED2-positive cells associate
Light microscopic analysis revealed several, slightly over- almost exclusively with vasculature, and often curve into
lapping, populations of immune cells, which differed pri- crescent-moon-like shapes as they wrap around the ves-
marily in their distribution throughout the area postrema sels. In this way, ED-2 positive cells were highly similar to
(Fig. 1), the type of marker they expressed, and their perivascular macrophages, or perivascular cells.
response to peripheral immune challenge with LPS.
GFAP. The GFAP antiserum labeled classic astro-
OX-6: MHC class II. Cells immunoreactive for the cytic cells with very small somata from which multiple long
MHC class II protein (labeled with OX-6) were distributed smooth and delicate processes emanate. These cells were
throughout the area postrema, but were particularly nu- numerous and uniformly distributed in both the area pos-
merous along the dorsal aspect, and enriched in the mid- trema and brainstem parenchyma (Fig. 1D). The morphol-
1420 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

Fig. 1. Distribution of different immune cell populations in the area postrema characterized by their expression of MHC class II antigen labeled with
OX-6 (A), CD11b/CR3 labeled with OX-42 (B), CD163-like antigen labeled with ED2 (C), GFAP (D), a cytoplasmic lysosomal antigen (CD68) labeled
with ED1 (E), and IL-1␤ IR (F, G). IL-1␤ IR in F was induced by LPS challenge (0.4 mg/kg, 4 h), as the area postrema in saline-treated rats did not
show any IR for IL-1␤ (G). Scale bar⫽100 ␮m in A (applies to all panels).

ogy of these cells was quite distinct from those expressing few somata were seen within the meninges and interface
abovementioned immune-related markers (Fig. 2F) and with the fourth ventricle, as well as numerous processes
those stained for IL-1␤ (see below). For this reason, GFAP stained in a grainy fashion. Most labeled cells were round
was not further included in the analysis. Processes of or oval. IR for ED1 was irregularly distributed within the
GFAP-positive cells in the area postrema encircle blood cytoplasm of these cells (Fig. 2C), a feature also seen in
vessels and were particularly prominent at the margins of ED1-positive putative macrophages in other tissues, in-
the area postrema and subjacent nTS. cluding peripheral nerve (Goehler et al., 1999).
ED1: CD68-like antigen (macrosialin). ED1-positive IL-1␤. The two different polyclonal antibodies gave
cells were extremely sparse in the area postrema com- excellent staining for rat IL-1␤ IR in the area postrema of
pared with those expressing other immune markers, and rats that were treated with bacterial endotoxin (LPS). In
were not seen in the brainstem parenchyma (Fig. 1E). A animals treated i.p. with 0.4 mg/kg LPS, strong IL-1␤-like
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1421

Fig. 2. Micrographs at high magnification showing the cell morphology features of the different cells encountered in the area postrema. (A) OX-6 recognizing
MHC class II antigen strongly labels a variety of cells with pyramidal (p), rocket (r), and spindle (s) appearance. Some cell bodies emanate fine cilia-like
protrusions (asterisk). (B) OX-6-stained cells with extensively ramifying arborizations giving them a microglia-like appearance. (C) ED1 staining of cytoplasmic
lysosomal antigen (CD68) in a subpopulation of immune cells. (D) OX-42 (CR3/CD11b)-positive microglial cells with extensive ramifications. Although they
resemble those in the brain parenchyma proper, the ones in the area postrema appear to have larger somata. (E) ED2 (CD163)-labeled cells with spindle
and crescent appearance. Many are closely associated with vasculature. (F) GFAP IR reveals smooth ramifications emanating from small somata,
characteristic for astroglia. (G) IL-1␤ IR (4 h after 0.4 mg/kg LPS i.p.) in cells with spindle and crescent appearance, some of which are associated with
vasculature (asterisk). (H) Inset, lower magnification image of the area postrema showing total elimination of staining for IL-1 IR when primary antiserum was
preadsorped with recombinant rat IL-1␤ (4:1 ratio w/w). Scale bar⫽25 ␮m in E (applies to panels A—G); 25 ␮m in H.

IR appeared in cells with the area postrema at both 2 and neurons in the area postrema among which the MHC-II-
4 h (Fig. 1F). Cells stained for IL-1␤ IR were consistently labeled cells intermingle (Fig. 3A, B). Closer examination
absent in the area postrema of rats that were untreated or at the light microscopic level indicates that many branches
were injected with sterile saline (Fig. 1G). In LPS-treated emanating from the MHC-II-labeled cells appear to form
rats, IL-1␤-immunoreactive cells were evenly distributed appositions with the TH-positive neuronal elements includ-
across the area postrema (Fig. 1F). IL-1␤ IR appeared ing their dendrites and somata (Fig. 3C, D).
primarily in spindle- and pyramid-shaped cells, as well as
Codistribution of OX-6 and OX-42 labeling. Based on
in some with a rocket-like morphology (Fig. 2G), similar to
morphological similarities between some of the MHC-II
those stained with ED2 and OX-6. Weak IL-1␤ IR was
(OX-6)-labeled cells and the CR3/CD11b (OX-42) -positive
seen in few cells with microglia-like morphology (highly
ones (see Fig. 2B, C), we investigated whether these
ramified with small somata), but never in round cells. The
immune cell markers can be co-expressed. Double-label
morphology of IL-1␤-positive cells was very different from
immunoperoxidase labeling nevertheless clearly re-
those staining for GFAP IR, and was inconsistent with
vealed separate populations of OX-6- and OX-42-la-
possible localization within astrocytes. Many, but not all,
beled cells (Fig. 3E), even when OX-6-labeled cells
IL-1␤-immunoreactive cells were closely associated with
appear in a strongly ramified microglia-like appearance
the vasculature (see also Fig. 3H, and below), some of
(Fig. 3F). Caution, however, is needed as a light staining
which exhibiting the typical crescent moon-shapes as seen
can be masked by the intensity of the other. Attempts to
with the ED2-positive cells.
address this issue using dual labeling with fluoro-
Dual OX-6 and TH staining. Double-labeling for both chromes unfortunately failed due to methodological
MHC-II and TH revealed a dense population of TH-positive problems.
1422 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

Fig. 3. Dual immunoperoxidase staining for MHC class II (with OX-6) and for either TH (in A–D), CD11b/CR3 (with OX-42, in E and F), or IL-1␤
(in F–H). Nickel-enhanced and regular DAB as chromogens yielded distinct purple– gray to black and reddish– brown staining. (A) Characteristic
distribution in the area postrema of constitutive MHC class II expression (stained black) against the background of reddish– brown TH-positive
neurons. (B–D) OX-6 stains cells with multiple dendritic ramifications (black arrow B), many of which make appositions with TH-labeled neuronal
somata and dendrites (arrows in C and D). (E) OX-42 and OX-6 label distinct populations of myeloid cells in the area postrema. Whereas OX-42
reveals the extensively ramified microglia (black arrows), OX-6 labels other cells that are spindle- and rocket-shaped (red arrows). (F) Staining
for IL-1␤ (in black) and OX-6 (reddish– brown) in the AP reveals dendriform and spindle-shaped cells that express either IL-1␤ IR alone (black
arrows) or both IL-1␤ and MHC-II IR (a mixing of gray and brown, indicated with both short black and red arrows). (G–I) Laser scanning confocal
imaging of neuronal elements in the area postrema labeled with Alexa Fluor 488 (green) for PGP9.5 and a totally separate population of cells
emitting Alexa Fluor 546 fluorescence (red) indicative of IL-1␤ IR. Many, but not all, IL-1␤-labeled elements are located in perivascular space, and none
were found to be double-labeled for PGP9.5. The arrow points at the location of an IL-1␤-labeled cell embedded in between PGP9.5-labeled
neurons. In panels F, H, and I, IL-1␤-like IR was induced 4 h after 0.4 mg/kg endotoxin challenge, and occurred in cells with a variety ramifications
and spindle-like extensions. Scale bars⫽100 ␮m in A, 20 ␮m in B, D (applies also to panel C), E, and F; 100 ␮m in I (applies also to panels G
and H).

Double-labeling of IL-1␤ and neurons. Fluorescent labeling indicative of dendritic, axonal and dendritic labeling.
double labeling for IL-1␤ and the pan-neuronal marker In contrast, IL-1␤ immunofluorescence highlighted distinct
PGPG9.5 occurred in strictly separated groups of cells in the cells that were spindle- and rocket-shaped. The latter distrib-
area postrema without any colocalization (Fig. 3G–I). The uted mainly within the perivascular spaces apparent in the
PGP9.5 antibody labeled a dense collection of round and double-labeled image (Fig. 3H), although several were em-
oval cell bodies, and a dense plexus of fibers and punctate bedded within the neuronal plexus (see arrow in Fig. 3H, I).
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1423

Fig. 4. Laser scanning confocal imaging of cells in the area postrema labeled with Alexa Fluor 488 (green) for IL-1␤-like IR and with Alexa FLuor 546
(red) for IR to one of the three immune cells markers: CD163 (ED2 clone, in A–C), MHC class II (OX-6 clone, in D–F), and CR3/CD11b (OX-42 clone,
in G–I). The images were compiled from stacking 12–15 optical scans 3 ␮m apart). All sections were from the same representative male rat treated
with 0.4 mg/kg endotoxin 2.5 h prior to kill. (A–C) Double-labeling reveals that virtually all CD163-positive cells strongly express IL-1␤ (see arrows
aimed at some of them appearing yellow in C). Although there is a high degree of colocalization, there is a portion of IL-1␤-expressing cells that do
not express the CD163 antigen. The asterisk indicates one CD163-positive cell expressing no IL-1␤ IR that is positioned just outside the AP and within
the nTS, associated with a blood vessel. (D–F) Of the MHC-II-positive cells labeled with OX-6, only a subpopulation expresses IL-1␤ (indicated with
arrows), and with varying intensity. (G–I) The OX-42-labeled population of microglia barely overlaps with the IL-1␤-positive cells, only a small number
of double-labeled cells were encountered (the arrow points to one). Double-labeled cells generally showed weak fluorescence for IL-1␤. Scale
bar⫽100 ␮m in A (applies to all panels).

Double-labeling of IL-1␤ and immune cell markers. all of the ED2-labeled cells residing within the borders of
IL-1␤-like immunofluorescence was encountered in loosely the area postrema (Fig. 4A–C) and also about half of the
distributed cells across the entire area postrema in LPS- OX-6-positive cells (Fig. 4D–F) showed IL-1␤-like immu-
treated rats, but was completely absent in saline-treated nofluorescence (see also Fig. 5A). ED2-positive cells that
rats. Cells with IL-1␤-like immunofluorescence evinced a were occasionally found outside the area postrema (i.e. in
morphology highly similar to cells that stain with OX-6 and the subjacent nTS) were not double-labeled. In contrast,
ED2. In accordance with this highly similar appearance, microglia identified with OX-42 rarely co-localized immu-
confocal laser scanning microscopy revealed that virtually nofluorescence for IL-1␤ (Fig. 4G–I). In general, the pop-
1424 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

by extracellular space (Fig. 6A), that is, they did not appear
to be anchored to extracellular matrix. Appositions be-
tween MHC-II-IR cells and neurons were frequently en-
countered (Fig. 6C–F). In sections processed for both
MHC-II and TH-IR, MHC-II-laden ramifications apposed
neuronal elements that were TH-positive (Fig. 7), as well
as some that were not. Appositions occurred on cell so-
mata (Fig. 6C) and dendrites (Figs. 6 D, E, 7) and some
were observed on neuronal axons and terminals as well. A
remarkable aspect of many appositions was the absence
of any MHC-II label along the outer cell membrane lining
up with the neuronal cell membrane (Fig. 6C).

Fig. 5. (A) Bar graphs showing the rate of colocalization with IL-1␤ of OX-42: CR3/CD11b IR
the three immune cell populations identified by labeling with ED2
(CD163 antigen), OX-6 (MHC class II antigen) and OX-42 (CR3/ Silver-enhanced gold particles lining the outer membranes
CD11b antigen) as observed in the laser scanning confocal micro- indicating CR3/CD11b IR were most frequently observed
scope. Data from four male rats were pooled (2.5 [n⫽2] and 4 h [n⫽2]
following 0.4 mg/kg endotoxin) as no differences were apparent be- along processes and around somata (Fig. 8). Occasionally
tween the time points. Whereas almost all ED2-positive cells express CR3/CD11b IR occurred in perivascular-like cells encir-
IL-1␤, about half of the OX-6-positive cells and a small fraction (7%) of cling the endothelium (Fig. 8A). The somata typically were
the OX-42-labeled microglia do so. (B) Among the population of en- small, with little perinuclear cytoplasm (Fig. 8C). OX-42-
dotoxin-responsive cells expressing IL-1␤, many of them show the
ED2 phenotype, but only a small portion of them are OX-6 and OX-
labeled profiles and especially the distal ends were often
42-positive. A significant portion of these IL-1␤-positive cells ex- observed in the vicinity of neuronal processes, which they
presses neither marker. sometimes contacted (Fig. 8D–F).

ulation of generally bright IL-1␤-labeled cells is very dis- ED2: CD163-like antigen IR
tinct from the plexus of OX-42-labeled microglia. Despite
CD163-like IR was found in the outer membranes of cells
the high incidence of colocalization of IL-1␤ within the ED2
associated with the vasculature (Fig. 9). These cells con-
and OX6-labeled populations, a smaller but significant por-
tained many tubules and a relatively electron-dense cyto-
tion of cells showing bright immunofluorescence for IL-1␤
plasm, often including opaque inclusions (possibly lyso-
expressed neither marker (Fig. 5B), indicating another
somes). Many of these cells encircled the endothelium, or
group of LPS-responsive cells that express neither MHC
lurked within the perivascular space, extending one or two
class II nor CD163-like protein recognized by respectively
processes to contact astrocytic or endothelial cells (Fig.
OX-6 and ED2.
9B, C) These processes were characteristically very small
in diameter, and sometimes resemble cilia as they were
Ultrastructural features of immune cells in the area
short and curved slightly. One larger and longer apical
postrema
process usually protruded from the ED-2-labeled cell wrap-
Cells expressing immune-related markers share many ping around the endothelial cell. Although ED-2-labeled
highly similar ultrastructural features. Particularly charac- cells were usually positioned closely to endothelial cells
teristic are the nuclei, which are irregularly shaped and forming the wall of blood vessels, most were in direct
contain large patches of highly condensed chromatin contact with small-diameter extensions of astroglia inter-
(Figs. 6A, B). At rest (in saline-treated rats) these cells spersed in between (Fig. 9A, B, D).
appear remarkably similar, with the most notable distinc-
tion being their location within the area postrema, and IL-1␤-like IR
response to peripheral endotoxin challenge.
IR for IL-1␤ was found in two types of cells. One type asso-
ciated with the vasculature (Fig. 10), and evinced character-
OX-6: MHC class-II IR
istics similar to the ED-2 positive cells (perivascular-like cells).
Membrane-associated MHC-II IR was occasionally de- These cells encircled the vessels (Fig. 10A, C), often making
tected in perivascular cells (Fig. 6B), but most frequently contacts with endothelial cells or astrocytes. IL-1 positive
occurred in cells distributed among the neurons and astro- perivascular-like cells contained many inclusions of various
cytes within the parenchymal area postrema (Fig. 6A). The sizes indicative of phagocytic activity. The other type of IL-
latter type shared characteristics with true dendritic cells, 1␤-positive cell closely associated with neurons (Fig. 11),
including high levels of constitutive MHC-II expression, including somata (Fig. 11A, E) and dendrites (Fig. 11B–D). In
which is dramatically upregulated following systemic en- some cases, IL-1␤-like immunoreactive processes entirely
dotoxin treatment (Fig. 6C, F). MHC-II positive cells con- encircled neuronal dendrites (Fig. 11B, C). This type of inter-
tained a plethora of tubules in their dendritic extensions, action appeared to be rather specific, in that IL-1␤ positive
and contained numerous mitochondria and smooth endo- processes contacted only a subset of neuronal dendrites in
plasmic reticulum. These cells invariably were surrounded their vicinity. Astrocytes were typically also closely associated
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1425

Fig. 6. Electron micrographs of immune cells in the area postrema constitutively expressing the MHC class II antigen, visible by the black granular
deposits along the outer membranes. (A, B) Low magnification power images of OX-6-labeled cells associated with neurons (in A) and vasculature
(in B). Note the characteristically large extracellular space that surrounds these cells in both instances. The cell nuclei of OX-6-labeled immune cells
[indicated with Nu (i)] contain patches of dense chromatin very distinct from neuronal cell nuclei [indicated with Nu (n)]. Arrows in A indicate the outer
membrane dotted with the granular silver deposits. (C) High power electron micrograph showing an OX-6-labeled dendritic profile (indicated with
asterisk) that abuts a neuronal soma at the junction of a neuronal dendrite to the left. The outer membrane lining along the neuronal soma is clear
of OX-6 label (indicated with arrows). (D, E) OX-6-positive profile (indicated with asterisks) with scattered granular label along the outer membrane
(arrows in D) in close proximity to a neuronal dendrite, which is shown in panel E at higher magnification. (F) OX-6-labeled dendritic profile (indicated
with asterisks) running along a neuronal soma (arrows). Sections were derived from endotoxin-treated male rats (0.1 mg/kg, 2 h: A, B, C; 4 h: F) and
saline-treated male (D, E). Scale bars⫽1 ␮m in A, B, and D; 500 nm in C, E, and F.

with both the neuronal immune cells processes sometimes 11D). In line with observations with light and confocal laser
interjected between them (Fig. 11C). IL-1␤ positive pro- scanning microscopy, ultrathin sections through the area
cesses were particularly evident in the neuropil along the postrema from rats that received i.p. injections of sterile
lateral borders of the area postrema, adjacent to the nTS (Fig. saline lacked any discernable IL-1␤ IR.
1426 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

Fig. 7. Close associations between the OX-6-labeled cells and TH positive neuronal elements in the area postrema (2 h following 0.1 mg/kg LPS i.p.).
OX-6 positive ramifications (indicated with asterisks) were identified with silver granule deposits along the outer cell membranes (some of them are
indicated with concave arrows in A), and form close anatomical associations with TH-immunoreactive neuronal dendrites (TH⫹ d) visualized with the
electron-dense DAB precipitate. The close proximity of both profiles is also indicated with arrows in B and C. The membrane apposing the TH-positive
profile often lack OX-6 label (as in A and B) but not always (as in C). D indicates a “threesome” of an OX-labeled profile, a TH-positive, and an
unlabeled neuronal d. Scale bars⫽500 nm in all panels.

Characterization of the anti IL-1␤ antibodies cates the presence of pro-IL-1␤ prior to post-translational
splicing into mature IL-1␤ (17 kDa), as no 17 kDa band
Selected sections through the area postrema from both
(detectable in the lane loaded with the rat recombinant
saline- and LPS-treated rats that were incubated in non- IL-1␤) was present, indicating that the IL-1␤-like IR in the
specific rabbit or mouse IgG to replace the primary anti- area postrema described here predominantly represents
bodies showed no positive staining. Preabsorption with the newly synthesized pro-hormone in response to LPS treat-
immunogen recombinant rat IL-1␤ completely eradicated ment.
the staining for IL-1␤-like IR otherwise seen in the area
postrema from LPS-treated rats (Fig. 2H). No IL-1␤-like
staining appeared in sections through the area postrema DISCUSSION
from saline-treated rats irrespective of preabsorption. The results of these studies have revealed an intimate
Western blot analysis (SDS-PAGE) of IL-1␤ IR when pre- relationship between immune cells and neuronal ele-
viously pre-absorbed with normal rat brain sections re- ments in the area postrema, providing potential for a
vealed a single band with an approximate molecular novel, direct influence of immune cells on neurons within
weight of 31 kDa in the lanes loaded with supernatant of the CNS. Following peripheral immune challenge with
homogenates from dorsal medulla of the LPS-treated, but bacterial endotoxin, robust IR for the pro-inflammatory
not the saline-treated rat (Fig. 12). This observation indi- cytokine IL-1␤ was induced in many innate immune
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1427

Fig. 8. Electron micrographs of OX-42 labeled immune cells (microglia) in the area postrema. Immunoreactivity for the CD11b antigen as silver
granules distributes along the outer cell membranes of the OX-42-labeled cells and their profiles (indicated with asterisks in all panels. Similar staining
was observed following saline (A–C) and endotoxin (D–F) treatment. (A) An OX-42-labeled profile curves along the vasculature. (B) The cytoplasm
of the OX-42-positive ramifications is rich in endoplasmic reticulum, mitochondria and small vesicular inclusions. (C) OX-42-labeled soma containing
a cell nucleus (Nu) with electron-dense patches of chromatin. (D, E) OX-42-labeled ramifications come in close proximity with neuronal dendrites (d),
however do not form close membrane-to-membrane associations. (F) Extensive ramifications in distal ends of microglia. Scale bars⫽1 ␮m in all
panels.

cells, which associated either with the vasculature or Complexity of identifying immune cell phenotypes
with neuronal elements. The findings suggest multiple
roles for immune cells in the area postrema, and provide Characterization of immune cell populations in the brain
a potential mechanism by which brain-derived cytokines has been confusing and controversial. The varying mor-
might directly influence constituents of the neurocircuitry phology and expression of functional markers such as
that mediates brain responses to infection and infla- MHC-II (OX-6) imply that these cells may correspond to
mmation. different activation states of the same cell, or they may
1428 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

Fig. 9. ED2-labeled perivascular immune cells showing silver-gold immunoprecipitate along the outer cell membranes. The cells exhibit nuclei (Nu)
contain characteristic electron-dense patches of chromatin. Although the cells are associated with the vasculature, astrocytes (as) often intersperse.
Very thin and short cilia-like ramifications (asterisks in B–D) emanating from the soma are another characteristic feature of these ED2-labeled cells
that often make close contacts with as (arrows in B and C). Electron-dense inclusions seen in the somata in panels A and D indicate phagocytic activity.
Other abbreviations: bv, blood vessel; endo, endothelial cell. Scale bars⫽1 ␮m in all panels.

represent cells of separate lineage. In the area postrema, lar cell membranes, but limited, as the overall distribution
subpopulations of cells can be described based on the 1) of MHC-II-expressing (OX-6) cells differs from that of the
differential expression of CD163, CR3/CD11b, and MHC- CD163-positive (ED-2) ones. That CD163 (ED2) and
II, 2) on morphological differences, and 3) on location MHC-II (OX-6) expression occurs in largely separate pop-
within the area postrema. Nevertheless, it is likely that ulations in the area postrema appears to be in line with
these are overlapping populations, and cells may co-ex- similar observations in dura mater, leptomeninges, and
press two or all of these markers. However, based on choroid plexus isolated from rat brain (McMenamin, 1999).
morphological appearance and distribution characteristics Similarly, CR3/CD11b (OX-42) and MHC-II (OX-6) may
(as shown in Figs. 1 and 2), co-expression of CD163 colocalize in a small subset of cells, as we observed that a
(ED-2) and CR3/CD11b (OX42) seems unlikely, based on small proportion of MHC-II-positive cells displays a mor-
morphological and tissue distribution differences. Some phology highly similar to those CR3/CD11b-labeled cells
overlap between CD163 (ED-2) and MHC-II (OX-6) is likely characteristic of microglia. Although the populations of
based on OX-like IR seen very occasionally on perivascu- immune cells we describe may not represent truly separate
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1429

Fig. 10. IL-1␤-like immunoreactivity in perivascular immune cells following endotoxin treatment (0.4 mg/kg i.p., 4 h). The DAB precipitation
indicative of IL-1␤ IR substantially darkens the cytoplasm. (A) Perivascular IL-1␤-positive cell with two distal ends of the nucleus (Nu) and a
cilia-like extension (indicated with asterisk). Although the cell wraps around the endothelial cell (endo), it is partly interspersed by an astrocytic
ramification (as), a feature also characteristic for ED-2 positive cells. The cell is rich in organelles including mitochondria. (B) IL-1␤-labeled soma
with multiple thin extensions (asterisk). (C, D) IL-1␤-positive profiles (distal somata) associated with blood vessels (bv). (E) An IL-1␤-positive
cell positioned within the area postrema parenchyma that shows partially extruding vesicles that are densely stained (arrow). Scale bars⫽1 ␮m
in all panels.

phenotypes, assessment of the distribution of the different kamori et al. (1994) described IL-1␤ expression in macro-
markers provides information as to the potential functional phages exclusively restricted to the OVLT (a forebrain
roles of each immune cell phenotype (see Guillemin and circumventricular organ) in rabbits, whereas IL-1␤ expres-
Brew, 2004) and provides information relevant for potential sion in cells associated with the vasculature of the brain
intervention with e.g. targeted immunolesion techniques. parenchyma was attributed to microglia. Konsman et al.
(1999) described colocalization of IL-1-like IR with lectin-
Immune cell subpopulations: substrate for immune binding cells, also supporting a myeloid immune cell-like
sensory specialization?
identity. However, lectin-binding does not distinguish be-
The results from these studies confirm and extend earlier tween subtypes (e.g. macrophage, microglial or dendritic-
reports regarding the myeloid immune-like identity of IL- like) of immune cells. Under the conditions during our
1␤-immunoreactive cells in circumventricular organs. Na- study, that is, peripherally administered endotoxin at mod-
1430 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

Fig. 11. IL-1␤-immunoreactive cells associated with neurons in the area postrema following endotoxin treatment (0.4 mg/kg i.p., 2 h: D, E and 4 h:
A–C). (A) Cell soma and proximal extension showing IL-1␤ IR in close proximity to a neuronal cell body. The neuron is characterized by its lighter
stained nucleus [Nu(n)] and the embedded nucleolus. The insets a and b show higher magnification of the image areas indicated with asterisks and
labeled a and b. Here the IL-1␤-labeled profile directly opposes a neuronal dendrite (d in a), and the neuronal soma (n in b). (B) An IL-1␤-labeled profile
(asterisk) wraps around a neuronal d. Dense immunoreactive label along the outer membrane is visible juxtaposing the neuronal membrane (arrow).
(C) An extensively ramified IL-1␤-positive profile that wraps itself around two neuronal d, the larger of the two is completely wrapped (two arrows
indicate the “closure”). The lower arrows and the asterisk indicate a darkening of the neuronal membranes, which could indicate possible released and
bound IL-1␤. (D) A IL-1␤-IR profile (asterisk) in tight association with a neuronal d in the border region of the area postrema and nucleus of the solitary
tract. (E) A distal IL-1␤-IR ramification runs along the surface of a neuronal soma with nucleus [Nu(n)]. Scale bars⫽1 ␮m in A and E; 500 nm in B–D
and insets a, b.

erate doses, with two to four hours’ survival, the most ron nor astroglial cells were found IL-1␤-positive. Thus,
robust induction of IL-1␤-like IR in the area postrema was these findings demonstrate differential responsivity to pe-
contributed by CD163 (ED2)-labeled tissue macrophages/ ripheral LPS challenge among cells within the area pos-
perivascular cells. MHC-II (OX-6)-positive dendritic-like trema.
cells, and another morphologically similar group of cells The differences in responses to peripheral LPS chal-
expressing neither marker also responded. Although some lenge as revealed by IL-1␤ expression between the im-
CR3/CD11b (OX42)-positive cells with a microglia-like mune phenotypes of cells in the area postrema may derive
morphology were also IL-1␤-positive, they were far fewer from functional requirements. A picture is emerging from
in number, at least at the earlier time points chosen in this recent studies (Chakravarty and Herkenham, 2005; Val-
study. In addition, at these early time points, neither neu- lieres and Sawchenko, 2003) that indicate a dichotomy
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1431

nisms by which IL-1␤ can influence area postrema func-


tions. IL-1␤-IR cells that contact neurons could directly
influence specific neural circuits (e.g. those involved in
emesis vs. those involved in viscerosensory relay to the
forebrain). Such a direct action would likely serve to in-
crease the fidelity of immune-neural signaling, by enabling
more rapid and consistent interaction of IL-1␤ with recep-
tors. In contrast, perivascular cells are positioned to influ-
ence functions associated with the vasculature, including
immune cell trafficking (Ching et al., 2005), and could
release IL-1␤ into the abundant extracellular space of the
area postrema to influence neurons, astrocytes, or other
immune-like cells via a paracrine-like mechanisms. Fur-
ther, given the observation that astrocytes were typically
interspersed around and between the IL-1␤-IR immune
processes and the neuronal dendrites and axons, IL-1␤
may also modulate astrocytic functions.
Evidence for physiological effects of IL-1␤ on neurons
derives from reports that IL-1␤ influences intracellular cal-
Fig. 12. Western blot of tissue supernatant from the dorsomedial cium levels in cortical synaptosomes (Campbell and
medulla of a LPS-treated rat (right lane) showing immunoreactive Lynch, 2000) and cultured hippocampal neurons (Viviani et
protein of 31 kDa size as compared with the ladder in the left column, al., 2003). A similar effect of IL-1␤ on neurons in the area
likely to be the pro-IL-1␤. The middle column shows recombinant rat
IL-1␤ (10 ng) at approximately 17.5 kDa.
postrema would be consistent with the findings of Desson
and Ferguson (2003) showing that in another circumven-
between “resident” (microglia) and “immigrant” immune tricular organ, the subfornical organ, IL-1␤ influences neu-
cells in the brain. Microglia enter the brain during prenatal ronal excitability. The effects of IL-1␤ on neurons in the
and early postnatal development, seem to turn over slowly area postrema are likely mediated by the type 1 IL-1 re-
and are not readily replenished (Vallieres and Sawchenko, ceptors, which have been reported to be expressed by
2003). The immigrants, which are primarily perivascular several other neuronal subtypes, including sensory neu-
cells associated with the brain vasculature and (presum- rons (Ek et al., 1998), as well as neurons in the hypothal-
ably) macrophages and dendritic cells outside the blood– amus (Diana et al., 1999), hippocampus (French et al.,
brain barrier (meninges and choroids plexus) or where it is 1999), and amygdala (Ericsson et al., 1995). Thus, find-
weak (circumventricular organs), show, in contrast, a ings from studies using both physiological and anatomical
higher rate of turnover and readily repopulate from hema- approaches support the idea that IL-1␤ can exert its ac-
topoietic precursors (Hickey et al., 1992; Hess et al., tions in the brain via direct effects on neurons.
2004). The apparent bias of earlier and more robust IL-1␤ In addition to a potential immunosensory role, im-
expression toward “immigrant” type cells observed in our mune–neural contact in the area postrema may serve
study implies that these cells constitute the principal immune other functions. For instance, these immune cells may
sensory cell in the brain. Indeed, CD163 (ED2)-positive provide trophic factors specific for neuronal subpopula-
perivascular cells have previously been implicated in im- tions, and/or may play a role in synaptic plasticity, as do
mune-to-brain signaling (Schiltz and Sawchenko, 2002) microglia in other brain regions (Roumier et al., 2004). At
based on their responses to systemic IL-1␤ administration. any rate, immune cells likely play multiple roles in the area
MHC-II expression has been previously described in postrema, related to different challenges of developmental
immune cells of the area postrema (Pedersen et al., 1997), and/or immune status (Boulanger and Shatz, 2004).
however the findings here represent the first report implicat-
ing these cells in contributing to IL-1␤ production. Whereas The area postrema as an immune sensory structure
the function of contact between MHC-II (OX-6)-immunoreac- Evidence of an immunosensory function of the area pos-
tive cells and neurons in the area postrema cannot be deter- trema derives from multiple lines of evidence. Neurons in
mined from this morphological study, the apparently specific the area postrema consistently express c-Fos-IR (an acti-
targeting of immune cells and TH-IR neurons (Fig. 3A–D) vation marker) following both central and peripheral im-
suggests immune influence on area postrema functions, mune challenge (Brady et al., 1994; Goehler et al., 2005).
such as emesis (Miller and Leslie, 1994; Jovanovich-Micic et Toll-like receptors (TLR), which are immune sensory re-
al.,1995) or relay to rostral autonomic regions, notably the ceptors (Beutler et al., 2003), are expressed in the area
parabrachial nucleus (Yamamoto et al., 2003). postrema, including TRL4 (LaFlamme and Rivest, 2001)
and TLR9 (Sako et al., 2005). Further, the area postrema
Potential mechanisms of immune-neural interaction
is one of the structures identified in the rat to express IL-1
in the area postrema
receptor type I mRNA (Cunningham et al., 1992; Ericsson
The finding that IL-1␤-like IR cells are differentially distrib- et al., 1995), providing a substrate by which IL-1␤ pro-
uted within the area postrema suggests multiple mecha- duced within the area postrema (see also Quan et al.,
1432 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

1998; Konsman et al., 1999; Van Dam et al., 2000) likely CONCLUSION
influences neurons cohabiting the area postrema, and
The use of immunoelectron and confocal microscopy has
from there on to neuronal targets in other parts of the brain
revealed multiple phenotypes of immune cells in the brain-
(see below).
stem area postrema and demonstrated that some of these,
A functional role of the area postrema in brain-medi-
including some that express IL-1␤-like IR following periph-
ated host defense responses is supported by findings that
eral immune challenge, make significant physical contact
lesions of the area postrema can attentuate HPA axis
with neurons. Thus the resolution afforded by this tech-
responses or hypothalamic norepinephrine increase to
nique has revealed a potentially direct mechanism by
systemic immune activation (Ishizuka et al., 1997; Lee et
which neurons and immune cells may interact within brain
al., 1998), and inactivation of the dorsal vagal complex,
circumventricular organs.
including the area postrema, blocks sickness-induced be- The diversity of immune cells’ phenotypes in the area
havioral depression and c-Fos expression in the brain postrema, compared with the normal brain parenchyma,
(Marvel et al., 2004). On the other hand, Ericsson et al. may well follow from the fact that circumventricular organs
(1997) did not find impairment of either HPA response or occupy a position at the interface between the CNS and
medullary and hypothalamic c-Fos protein expression fol- the periphery. Although the blood– brain barrier is weak in
lowing i.v. IL-1␤ administration. The reasons for this dis- these structures, they contain neurons typical of brain
crepancy are not evident, but underline the fact that im- parenchyma rather than those associated with the periph-
mune signals influence the brain via multiple pathways and eral nervous system. This arrangement allows for the po-
mechanisms that are as yet incompletely understood. tential for interaction of immune sentinel cells with brain
The area postrema is well suited for a sensory role sig- neurons, either directly via membrane apposition, or via
naling both peripheral immune activation as well as inflam- local paracrine-type mechanism, and supports the idea
mation within the nervous system (e.g. meningitis). In addi- that circumventricular organs serve as immune sensory
tion to possessing a weak blood– brain barrier, the area pos- structures.
trema protrudes into the fourth ventricle, allowing it access to
substances (including pathogens) in cerebrospinal fluid not Acknowledgments—The authors would like to thanks Ms. Bonnie
normally found in brain parenchyma. Circumventricular or- Shepard for excellent and invaluable technical assistance. This
gans contain abundant extracellular space, which results in work was supported by NIH grants MH 64648, MH 68834 and
fluid dynamics that favor slower throughput of substances, EY12138.
facilitating their function as sensory structures (Gross, 1991).
Moreover, the area postrema is distinct from other circum-
ventricular organs in that it receives direct input from periph-
REFERENCES
eral sensory nerves, via terminations of vagal sensory fibers Banchereau J, Steinman RM (1998) Dendritic cells and the control of
within its lateral margins (Beckstead and Norgren, 1979; immunity. Nature 392:245–252.
Batchelor PE, Porritt MJ, Martinello O, Parish CL, Liberatore GT,
Chernicky et al., 1984; Strain et al., 1990; Sykes et al., 1997; Donnan GA, Howells DW (2002) Macrophages and microglia pro-
Zheng et al., 2002). Vagal sensory fibers have been impli- duce local trophic gradients that stimulate axonal spouting toward
cated as immunosensory conduits (Watkins et al., 1995; Ek but not beyond the wound edge. Mol Cell Neurosci 21:436 – 453.
et al., 1998; Gaykema et al., 1998; Goehler et al., 1998) that Beckstead RM, Norgren R (1979) An autoradiographic examination of
likely play a role in signaling immune activation in internal the central distribution of the trigeminal, facial, glossopharyngeal,
and vagal nerves in the monkey. J Comp Neurol 184:455– 472.
visceral structures, such as the gut (Dantzer, 2004; Goehler
Beutler B, Hoebe K, Du X, Ulevitch RJ (2003) How we detect microbes
et al., 2005). Targets of the area postrema projections are and respond to them: the Toll-like receptors and their transducers.
restricted to specific components of the ascending visceros- J Leukocyte Biol 74:479 – 485.
ensory pathway, notably the nTS and external lateral division Blalock JE (1994) The syntax of immune-neuroendocrine communi-
of the lateral parabrachial nucleus of the pons (Shapiro and cation. Immunol Today 15:504 –511.
Bluthe RM, Beaudu C, Kelley KW, Dantzer R (1995) Differential effects
Miselis, 1985; Herbert et al., 1990; Rogers and McCann,
of IL-1ra on sickness behavior and weight loss induced by IL-1 in
1993; Cunningham et al., 1994). Previously reported findings rats. Brain Res 677:171–176.
that neurons of the area postrema reliably respond to both Boulanger LM, Shatz CJ (2004) Immune signaling in neural develop-
peripherally and centrally administered immune stimulants ment, synaptic plasticity and disease. Nat Neurosci Rev 5:521–
(Wan et al., 1993; Elmquist et al., 1996; Konsman et al., 531.
1999) or live bacteria (Goehler et al., 2005) support the idea Brady LS, Lynn AB, Herkenham M, Gottesfeld Z (1994) Systemic
interleukin-1 induces early and late patterns of c-fos mRNA ex-
that the area postrema indeed contributes to brain responses
pression in brain. J Neurosci 14:4951– 4964.
to peripheral immune activation. Taken together with the Campbell V, Lynch M (2000) The role of ceramide in the modulation of
findings presented here, the close relationships between cy- intracellular Ca2⫹ levels by interleukin-1␤ in rat cortical synapto-
tokine-producing immune cells and neurons observed in the somes. Cytokine 12:487– 490.
area postrema are therefore consistent with a mechanism by Chakravarty S, Herkenham M (2005) Toll-like receptor 4 on nonhe-
matopoietic cells sustains CNS inflammation during endotoxemia,
which circulating pathogens or pathogen products, via influ-
independent of systemic cytokines. J Neurosci 25:1788 –1796.
ence upon specific neural circuits originating in the area Chernicky CL, Barnes KL, Ferrario CM, Conomy JP (1984) Afferent
postrema, may induce, modulate, or maintain brain-mediated projections of the cervical vagus and nodose ganglion in the dog.
host defense responses. Brain Res Bull 13:401– 411.
L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434 1433

Ching S, He L, Lai W, Quan N (2005) IL-1 type 1 receptor plays a key Gross PM (1991) Morphology and physiology of capillary systems in
role in mediating the recruitment of leukocytes into the central subregions of the subfornical organ and area postrema. Can
nervous system. Brain Behav Immunol 19:127–137. J Physiol Pharmacol 69:1010 –1025.
Cunningham ET Jr, Wada E, Carter DB, Tracey DE, Battey JF, De- Guillemin GJ, Brew BJ (2004) Microglia, macrophages, perivascular
Souza EB (1992) In situ histochemical localization of type 1 IL-1 macrophages, and pericytes: a review of function and identifica-
receptor RNA in the central nervous system, pituitary, and adrenal tion. J Leukoc Biol 75:388 –397.
gland of the mouse. J Neurosci 12:1101–1114. Hart BL (1988) Biological basis of the behavior of sick animals. Neu-
Cunningham ET Jr, Miselis RR, Sawchenko PE (1994) The relation- rosci Biobehav Rev 12:123–137.
ship of efferent projections from the area postrema to vagal motor Herbert H, Moga MM, Saper CB (1990) Connections of the parabra-
groups and brain stem catecholamine-containing cell groups: an chial nucleus with the nucleus of the solitary tract and the medul-
axonal transport and immunohistochemical study in the rat. Neu- lary reticular formation in the rat. J Comp Neurol 293:540 –580.
roscience 58:635– 648. Hess DC, Abe T, Hill WD, Studdard AD, Carothers J, Masuya M,
Dantzer R (2004) Cytokine-induced sickness behavior: a neuroim- Fleming PA, Drake CJ, Ogawa M (2004) Hematopoietic origin
mune response to activation of innate immunity. Eur J Pharmacol of microglial and perivascular cells in brain. Exp Neurol 186:
500:399 – 411. 134 –144.
Desson DL, Ferguson AV (2003) Interleukin-1␤ modulates rat subfor- Hickey WF, Vass K, Lassmann H (1992) Bone marrow-derived ele-
nical organ neurons as a result of activation of a non-selective ments in the central nervous system: an immunohistochemical and
cationic conductance. J Physiol 550:113–122. ultrastructural survey of rat chimeras. J Neuropathol Exp Neurol
Diana A, Van Dam A-M, Winblad B, Schultzberg M (1999) Co-local- 51:246 –256.
ization of interleukin-1 receptor type 1 and interleukin-1 receptor Ishizuka Y, Ishida Y, Kunitake T, Kato K, Hanamori T, Mitsuyama Y,
antagonist with vasopressin in magnocellular neurons of the para- Kannan H (1997) Effects of area postrema lesion and abdominal
ventricular and supraoptic nuclei of the rat hypothalamus. Neuro- vagotomy on interleukin-1␤ induced norepinephrine release in the
science 89:137–147. hypothalamic paraventricular nucleus region in the rat. Neurosci
Dong Y, Benveniste EN (2001) Immune function of astrocytes. Glia Lett 223:57– 60.
36:180 –190. Jovanovich-Micic D, Samardzic R, Beleslin DB (1995) The role of
Ek M, Kurosawa M, Lundeberg T, Ericsson A (1998) Activation of alpha-adrenergic mechanisms within the area postrema in dopam-
vagal afferents after intravenous injection of interleukin1␤: role of ine-induced emesis. Eur J Pharmacol 272:21–30.
endogenous prostaglandins. J Neurosci 18:9471–9479. Kent S, Bluthe RM, Dantzer R, Hardwick AJ, Kelley KW, Rothwell NJ,
Elmquist JK, Scammell TE, Jacobson CD, Saper CB (1996) Distribu- Vannice JL (1992) Different receptor mechanisms mediate the
tion of Fos-like immunoreactivity in the rat brain following intrave- pyrogenic and behavioral effects of interleukin 1. Proc Natl Acad
nous lipopolysaccharide administration. J Comp Neurol 371: Sci U S A 89:9117–9120.
85–103. Konsman JP, Kelley K, Dantzer R (1999) Temporal and spatial rela-
Elmquist JK, Breder CD, Sherin JE, Scammell TE, Hickey WF, DeWitt D, tionships between lipopolysaccharide-induced expression of Fos,
Saper CB (1997a) Intravenous lipopolysaccharide induces cyclooxy- interleukin-1␤ and inducible nitric oxide synthase in rat brain. Neu-
genase 2-like immunoreactivity in rat brain perivascular microglia and rosci 89:535–548.
meningeal macrophages. J Comp Neurol 381:119–129. LaFlamme N, Rivest S (2001) Toll-like receptor 4: the missing link of
Elmquist JK, Scammell TE, Saper CB (1997b) Mechanisms of CNS the cerebral innate immune response triggered by circulating
response to systemic immune challenge: the febrile response. gram-negative bacterial cell wall components. FASEB J 15:155–
Trends Neurosci 20:565–570. 163.
Ericsson A, Liu C, Hart RP, Sawchenko PE (1995) Type 1 interleukin-1 Lee HY, Whiteside MB, Herkenham M (1998) Area postrema removal
receptor in the rat brain: distribution, regulation, and relationship to abolishes stimulatory effects of intravenous interleukin-1␤ on hy-
sites of IL-1-induced cellular activation. J Comp Neurol 361:6681– pothalamic-pituitary-adrenal axis activity and c-fos mRNA in the
6698. hypothalamic paraventricular nucleus. Brain Res Bull 46:495–503.
Ericsson A, Arias C, Sawchenko PE (1997) Evidence for an intramed- Maier SF, Watkins LR (1998) Cytokines for psychologists: implications
ullary prostaglandin-dependent mechanism in the activation of of bidirectional immune-to-brain communication for understanding
stress-related neuroendocrine circuitry by intravenous interleu- behavior, mood, and cognition. Psychol Rev 10:83–107.
kin-1. J Neurosci 17:7166 –7179. Marvel FA, Chen C-C, Badr NA, Gaykema RPA, Goehler LE (2004)
French RA, VanHoy RW, Chizzonite R, Zachary J, Dantzer R, Parnet Reversible inactivation of the dorsal vagal complex blocks lipopo-
P, Bluthe R-M, Kelley KW (1999) Expression and localization of lysaccharide-induced social withdrawal and c-Fos expression in
p80 and p68 interleukin-1 receptor proteins in the brain of adult central autonomic nuclei. Brain Behav Immunol 18:123–143.
mice. J Neuroimmunol 93:194 –202. McMenamin PG (1999) Distribution and phenotype of dendritic cells
Gaykema RPA, Goehler LE, Tilders FJH, Bol JGM, McGorry MM, and resident tissue macrophages in the dura mater, leptomeninges
Maier SF, Watkins LR (1998) Bacterial endotoxin induces Fos and choroid plexus of the rat brain as demonstrated in wholemount
immunoreactivity in primary afferent neurons of the vagus nerve. preparations. J Comp Neurol 405:553–562.
Neuroimmunomodulation 5:234 –240. McMenamin PG, Wealthall RJ, Deverall M, Cooper SJ, Griffin B (2003)
Gehrmann J, Matsumoto Y, Kreutzberg GW (1995) Microglia: intrinsic Macrophages and dendritic cells in the rat meninges and choroid
immuneffector cell of the brain. Brain Res Rev 20:269 –287. plexus: three-dimensional localization by environmental scanning
Goehler LE, Gaykema RPA, Hammack SE, Maier SF, Watkins LR (1998) electron microscopy and confocal microscopy. Cell Tissue Res
Interleukin-1 induces c-Fos immunoreactivity in primary afferent neu- 313:259 –269.
rons of the vagus nerve. Brain Res 804:306–310. Miller AD, Leslie RA (1994) The area postrema and vomiting. Front
Goehler LE, Gaykema RPA, Nguyen KT, Lee JL, Tilders FJH, Maier Neuroendocrinol 15:301–320.
SF, Watkins LR (1999) Interleukin-1␤ in immune cells of the ab- Nadjar A, Bluthe RM, May MJ, Dantzer R, Parnet P (2005) Inactivation
dominal vagus nerve: an immune to nervous system link? J Neu- of the cerebral NF␬B pathway inhibits interleukin-1␤-induced sick-
rosci 19:2799 –2806. ness behavior and c-Fos expression in various brain nuclei. Neu-
Goehler LE, Gaykema RPA, Opitz N, Reddaway R, Badr NA, Lyte M ropsychopharmacology 30:1492–1499.
(2005) Activation in vagal afferents and central autonomic Nakamori T, Morimoto A, Yamaguchi K, Watanabe T, Murakami N
pathways: early responses to intestinal infection with Campy- (1994) Interleukin-1␤ production in the rabbit brain during endo-
lobacter jejuni. Brain Behav Immunol 19:334 –344. toxin-induced fever. J Physiol 476:177–186.
1434 L. E. Goehler et al. / Neuroscience 140 (2006) 1415–1434

Pedersen EB, McNulty JA, Castro AJ, Fox LM, Zimmer J, Finsen B Sykes RM, Spyer KM, Izzo PN (1997) Demonstration of glutamate
(1997) Enriched immune-environment of blood-brain barrier defi- immunoreactivity in vagal sensory afferents in the nucleus tractus
cient areas of normal adult rats. J Neuroimmunol 76:117–131. solitarius of the rat. Brain Res 762:1–11.
Polfliet MMJ, Goede PH, van Kesteren-Hendrikx EML, van Rooijen N, Vallieres L, Sawchenko PE (2003) Bone marrow-derived cells that
Dijkstra CD, van den Berg TK (2001) A method for the selective populate the adult mouse brain preserve their hematopoietic iden-
depletion of perivascular and meningeal macrophages in the cen- tity. J Neurosci 23:5197–5207.
tral nervous system. J Neuroimmunol 116:188 –195. Van Dam A-M, Bauer J, Tilders FJH, Berkenbosch F (1995) Endotoxin-
Quan N, Whiteside M, Herkenham M (1998) Time course and local- induced appearance of immunoreactive interleukin-1␤ in ramified
ization patterns of interleukin-1␤ messenger RNA expression in microglia in rat brain: a light and electron microscopic study. Neu-
brain and pituitary after peripheral administration of lipopolysac- rosci 65:815– 826.
charide. Neuroscience 83:281–293. Van Dam A-M, Bol JGJM, Gaykema RPA, Goehler LE, Maier SF,
Quan N, Stern EL, Whiteside MB, Herkenham M (1999) Induction of Watkins LR, Tilders FJH (2000) Vagotomy does not inhibit high
pro-inflammatory cytokine mRNAs in the brain after peripheral
dose LPS-induced interleukin-1␤ immunoreactivity in rat brain and
injection of subseptic doses of lipopolysaccharide in the rat. J Neu-
pituitary gland. Neurosci Lett 285:169 –172.
roimmunol 93:72– 80.
Vernet-der Garabedian B, Lemaigre-Dubreuil Y, Mariani J (2000) Cen-
Rogers RC, McCann MJ (1993) Intramedullary connections of the
tral origin of IL-1␤ produced during peripheral inflammation: role of
gastric region in the solitary nucleus: a biocytin histochemical
meninges. Mol Brain Res 75:259 –263.
tracing study in the rat. J Auton Nerv Syst 42:119 –130.
Roth J, Hubschle T, Pehl U, Ross G, Gerstberger R (2002) Influence Viviani B, Bartesaghi S, Gardoni F, Vezzani A, Behrens MM, Bartfai T,
of systemic treatment with cyclooxygenase inhibitors on lipopo- Binaglia M, Corsini E, Di Luca M, Galli CL, Marinovich M (2003)
lysaccharide-induced fever and circulating levels of cytokines and Interleukin-1␤ enhances NMDA receptor-mediated intracellular
cortisol in guinea pigs. Pflugers Arch 443:411– 417. calcium increase through activation of the Src family of kinases.
Roumier A, Bechade C, Poncer J-C, Smalla K-H, Tomasella E, Vivier J Neurosci 23:8692– 8700.
E, Gundelfinger ED, Triller A, Bessis A (2004) Impaired synaptic Wan W, Janz L, Vriend CY, Sorensen CM, Greenberg AH, Nance DM
function in the microglial KARAP/DAP12-deficient mouse. J Neu- (1993) Differential induction of c-Fos immunoreactivity in hypothal-
rosci 24:11421–11428. amus and brain stem nuclei following central and peripheral ad-
Sako K, Okuma Y, Hosoi T, Nomura Y (2005) STAT3 activation and ministration of endotoxin. Brain Res Bull 32:581–587.
c-FOS expression in the brain following peripheral administration Watkins LR, Maier SF, Goehler LE (1995) Cytokine-to-brain communi-
of bacterial DNA. J Neuroimmunol 158:40 – 49. cation: A review and analysis of alternative mechanisms. Life Sci
Salzet M, Vieau D, Day RD (2000) Crosstalk between nervous and 57:1011–1026.
immune systems through the animal kingdom: focus on opioids. Williams K, Alvarez X, Lackner AA (2001) Central nervous system
Trends Neurosci 23:550 –555. perivascular cells are immunoregulatory cells that connect the
Saper CB, Sawchenko PE (2003) Magic peptides, magic antibodies: CNS with the peripheral immune system. Glia 36:156 –164.
guidelines for appropriate controls for immunohistochemistry. Yamamoto H, Kishi T, Lee CE, Choi BJ, Fang H, Hollenberg AN,
J Comp Neurol 465:161–163. Drucker DJ, Elmquist JK (2003) Glucagon-like peptide-1 respon-
Schiltz JC, Sawchenko PE (2002) Distinct brain vascular cell types sive catecholamine neurons in the area postrema link peripheral
manifest inducible cyclooxygenase expression as a function of the glucagons-like peptide-1 with central autonomic control sites.
strength and nature of immune insults. J Neurosci 22:5606 –5619. J Neurosci 23:2939 –2946.
Shapiro RE, Miselis RR (1985) The central neural connections of the Zheng H, Patterson LM, Berthoud HR (2002) CART in the dorsal vagal
area postrema. J Comp Neurol 234:344 –364. complex: sources of immunoreactivity and effects on Fos expres-
Shu S, Ju G, Fan L (1988) The glucose oxidase-DAB-nickel method in sion and food intake. Brain Res 957:298 –310.
peroxidase histochemistry of the nervous system. Neurosci Lett
85:169 –171.
Strain SM, Gwyn DG, Rutherford JG, Losier BJ (1990) Direct vagal APPENDIX
input to neurons in the area postrema which project to the para-
brachial nucleus: an electron microscopic-HRP study in the cat. Supplementary data associated with this article can be found, in
Brain Res Bull 24:457– 463. the online version, at doi:10.1016/j.neuroscience.2006.03.048.

(Accepted 6 March 2006)


(Available online 2 May 2006)

You might also like