NIMH Conte Center for the Neuroscience of Fear and Anxiety Centeror Neural Science, 4 Washington Place, New York University, Nework, NY 10003, USA
Center Neural Science, New York University, New York, NY 10003,SA
Department of Neurobiology, Mount Sinai School of Medicine, Nework, NY, 10029, USA
Laboratory for Neuroendocrinology, The Rockefeller University, Nework, NY, 10021, USA
bstract—Glucocorticoids, released in high concentrationsrom the adrenal cortex during stressful experiences, bind tolucocorticoid receptors in nuclear and peri-nuclear sites ineuronal somata. Their classically known mode of action iso induce gene promoter receptors to alter gene transcrip-ion. Nuclear glucocorticoid receptors are particularly densen brain regions crucial for memory, including memory oftressful experiences, such as the hippocampus and amyg-ala. While it has been proposed that glucocorticoids maylso act via membrane bound receptors, the existence of the
atter remains controversial. Using electron microscopy, weound glucocorticoid receptors localized to non-genomic sitesn rat lateral amygdala, glia processes, presynaptic terminals,euronal dendrites, and dendritic spines including spine or-anelles and postsynaptic membrane densities. The lateralucleus of the amygdala is a region specifically implicated inhe formation of memories for stressful experiences. Theseewly observed glucocorticoid receptor immunoreactiveites were in addition to glucocorticoid receptor immunore-ctive signals observed using electron and confocal micros-opy in lateral amygdala principal neuron and GABA neuronoma and nuclei, cellular domains traditionally associatedith glucocorticoid immunoreactivity. In lateral amygdala,lucocorticoid receptors are thus also localized to non-nu-lear-membrane translocation sites, particularly dendriticpines, where they show an affinity for postsynaptic mem-rane densities, and may have a specialized role in modulat-
ey words: fear, postsynaptic density, non-genomic, plasticity.
Correspondence to: L. R. Johnson, Center for Neural Science, Nework University, 4 Washington Place, New York, NY 10003, USA. Tel:1-212-998-3945; fax: �1-212-995-4704.-mail address: [email protected] (L. Johnson).bbreviations: CS, conditioned stimuli; DAB, 3,3=-diaminobenzidine;AA, excitatory amino acid; ER, estrogen receptor; GABA, gamma-minobutyric acid; GR, glucocorticoid receptor; ir, immunoreactive;A, lateral nucleus of the amygdala; NE, norepinephrine; NMDA,-methyl-D-aspartic acid; PSD, postsynaptic membrane density;
he hypothalamic pituitary adrenal (HPA) axis is activateduring stressful experiences, leading to the release into theloodstream of adrenal cortical hormones (cortisol in hu-ans, corticosterone in rodents). These, in turn, bind toineralocorticoid receptors (MR) and at higher concentra-
ions of circulating corticosterone, to glucocorticoid recep-ors (GRs) in the brain (Reul and de Kloet, 1985). Chronictress affects neuronal structure and survival, leading toendritic atrophy in hippocampal principal neurons and den-ritic hypertrophy in amygdala principal neurons (Sapolsky,992; de Kloet et al., 1999; Kim and Diamond, 2002; Vyast al., 2002; Sapolsky, 2003).
Stress can significantly modulate memory consolidationDiamond et al., 1996; Kim et al., 1996; McGaugh, 2000; Kimnd Diamond, 2002; McGaugh and Roozendaal, 2002) andetrieval (de Quervain et al., 1998). Memories susceptible toodulation by stress include conditioned fear-related mem-ries, including memories of both contextual stimuli (Conradt al., 1999; Radulovic et al., 1999; Sandi et al., 2001; Cord-ro et al., 2003a,b) and specific cues (Conrad et al., 1999;adulovic et al., 1999). Considerable evidence has alsoemonstrated that corticosterone activated GRs can di-ectly modulate memory in both animal modelsRoozendaal, 2000; Kim and Diamond, 2002; McGaughnd Roozendaal, 2002) and in humans (de Quervain et al.,000; Buchanan et al., 2001). Oral cortisol impairs theetrieval of negative but not neutral words in human sub-ects (Kuhlmann et al., 2005). In agreement with stress
odels, manipulation of endogenous levels of corticoste-one either by direct injection or oral administration oforticosterone, or with GR pharmacological manipulation,esults in an enhancement and/or inhibition of severalorms of fear-related learning. These include passivevoidance learning (McGaugh, 2000; Roozendaal, 2000)nd Pavlovian conditioning to discrete stimuli (Corodimast al., 1994; Hui et al., 2004) and context (Pugh et al.,997; Conrad et al., 2004). In addition Pavlovian appetitiveonditioning is also modulated by GR agonists (Zorawskind Killcross, 2003).
The amygdala has been implicated in the storage andxpression of fear memory in both animal (Fanselow andeDoux, 1999; LeDoux, 2000; Davis and Whalen, 2001;ah et al., 2003; Pare et al., 2004; Dityatev and Bolshakov,005; Maren, 2005) and human studies (Phelps andnderson, 1997; Whalen et al., 2004). The neural basis ofavlovian fear conditioning is especially well understoodnd is known to depend upon the amygdala (Fanselow andeDoux, 1999; LeDoux, 2000; Davis and Whalen, 2001;ah et al., 2003; Pare et al., 2004; Dityatev and Bolshakov,
005; Maren, 2005). In this procedure an emotionally neu-
ved.
teLP2atof
sotsmctiSeeDr(ssi1oc
aMme1cpmoahKenLrcoRptoet
Ma
GIOcp(p
1itcivademaBm1edtdrdpAr
C
T(alsmlsai-Bb
L
Laacwccsl
n(Gs
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299290
ral stimulus, such as a tone, is associated with an aversivevent, such as a foot shock (Fanselow and LeDoux, 1999;eDoux, 2000; Davis and Whalen, 2001; Sah et al., 2003;are et al., 2004; Dityatev and Bolshakov, 2005; Maren,005). Within the amygdala, the lateral nucleus (LA) playsn especially important role. Thus, damage to or inactiva-ion of the LA prevents fear conditioning, and the responsef LA neurons to the tone is enhanced after pairing with theootshock (LeDoux, 2000).
Fear conditioning is believed to be relevant to under-tanding fear disorders, such as post-traumatic stress dis-rder (PTSD), a condition in which the traumatic nature of
he stressful experience enhances the strength and per-istence of memories that then become intrusive into nor-al life (Sullivan and Gorman, 2002). Specifically, insuffi-
ient glucocorticoid levels, possibly mediated by deficits inhe ACTH-feedback circuits, have been associated withncreased risk for PTSD (Yehuda et al., 1998, 2004;chelling et al., 2001, 2004; Bonne et al., 2003; Goenjiant al., 2003; Aerni et al., 2004; Bremner et al., 2004; Duvalt al., 2004; Luecken et al., 2004; Pico-Alfonso et al., 2004;elahanty et al., 2005; Neylan et al., 2005). However,
educed cortisol has not been replicated in all studiesYoung and Breslau, 2004; Young et al., 2004). Given thattress can enhance fear conditioning, it is possible thattress-related alterations in amygdala circuits might partic-pate in PTSD or other stress-related disorders (Bremner,999). Understanding the localization and microstructuralrganization of GRs in the amygdala, and their role in fearonditioning, is thus of great interest.
In light of the fact that fear conditioning involves syn-ptic plasticity in LA (LeDoux, 2000; Blair et al., 2001;aren, 2001) and that glucocorticoid action through GRodulates synaptic plasticity (Pavlides et al., 1995; Karstt al., 2000, 2002; McEwen, 2000), (Coussens et al.,997), it is possible that the effects of stress on fearonditioning are mediated by GR modulation of synapticlasticity in LA circuits. Although the mechanism of GRodulation of synaptic plasticity is still unknown, one modef GR action occurs via excitatory amino acid receptorsnd potentiation of calcium currents in the amygdala andippocampus (Kerr et al., 1989; Coussens et al., 1997;arst et al., 2002). This is relevant given the role of bothxcitatory amino acid (EAA) receptors and calcium chan-els in fear conditioning (Weisskopf and LeDoux, 1999;eDoux, 2000; Karst et al., 2002). Activated GR is alsoequired for serotoninergic inhibition of the glutamate ex-itatory drive on spiny principal neurons, via enhancementf GABA inhibition, in LA circuits (Stutzmann et al., 1998;ainnie, 1999; Stutzmann and LeDoux, 1999). GR thuslays a significant role in the modulation of synaptic plas-icity, dendritic structure, and transmitter systems in LA. Inrder to begin to ascertain how GR may mediate theseffects, we examined the cellular and subcellular localiza-ion of GR in the LA.
EXPERIMENTAL PROCEDURES
ale Sprague–Dawley rats (n�16) were anesthetized, perfused
nd processed as previously described (Farb and LeDoux, 1997). G
oat anti-rabbit IgG conjugated to Alexa-543 and goat anti-mousegG conjugated to Alexa-488 (1:200; Molecular Probes, Eugene,R, USA) were used. LA tissue was prepared for electron-micros-opy as previously described (Farb and LeDoux, 1997). Immuno-ositive structures were labeled with 3,3=-diaminobenzidine (DAB)Sigma, St. Louis, MO, USA) and silver intensified colloidal-goldarticles (Amersham, Piscataway, NJ, USA).
In contrast to previous studies of GR (Liposits and Bohn,993), we specifically carried out our analysis on adrenal gland
ntact rats in order to obtain a non-manipulated subcellular ana-omical profile of GR. We hypothesized that basal levels of corti-osterone, free of manipulations, may reveal discrete GR local-zation not previously identified. However, as natural circadianariations in the levels and localization of GR may occur, intactnimals were killed, at the same time, during their light cycle. Theistribution and subcellular localization of GR in the LA werexamined by light, confocal and electron microscopy using aouse monoclonal anti-GR-2 antibody and a rabbit polyclonalnti-human GR57 anti-serum (5 mg/ml for both antisera; Affinityioreagents, Golden, CO, USA). Since monoclonal antibodiesay prefer the conformed liganded-form of GR (Cintra et al.,994), we compared the cellular and subcellular distribution ofach antibody. Dual-label confocal microscopy was performed toetermine the proportion of GR expressed in GABAergic (inhibi-ory) neurons in the LA, using a mouse monoclonal antibodyirected against GABA (1:400; ICN, Costa Mesa, CA, USA) andabbit anti-GR57. All animal procedures were performed in accor-ance with the National Institutes of Health Guide and with ap-roval of the New York Uniersity Animal Welfare Committee.nimal numbers used were the minimum needed for replication of
esults and incorporation of extensive controls.
ontrols
hree different control studies were performed. For single labelingGR-immunoreactive (ir) only) the primary antibody was omittednd specificity of the secondary antibody was tested. For dual
abeling experiments (GR-ir and GABA-ir) primary antibodies andecondary antibodies were mismatched and the specificity of theouse and rabbit IgG’s were tested for appearance of fluorescent
abels in the inappropriate confocal channel. Finally, we tested thepecificity of the primary antibody labeled with DAB by pre-bsorbing the rabbit polyclonal anti-human GR57 anti-serum with
ts control peptide (amino to carboxyl terminus D - Q - K - P - I - FN - V - I - P - P - I - P - V - G - S - E - N - W - N - R - C; Affinityioreagents, Co.). Control and experimental sections were incu-ated and processed for all steps in parallel.
RESULTS
ight microscopy
ight microscopy revealed GR57-labeled cells throughoutmygdala sub-regions, including LA, the basal nucleusnd the central nucleus, as well as in the endopiriformortex adjacent to the amygdala (Fig. 1). Reaction productas visualized in the nuclei and in the somatic peri-nucleusytoplasm of some cell bodies (Fig. 1a). This finding isonsistent with classical immunocytochemical studieshowing GR-ir in the presence of the endogenous GR
igand (Cintra et al., 1994).To ascertain whether the GR-ir cells included GABAergic-
eurons, we performed dual-label immunofluorescenceFig. 1b). Confocal analysis of the LA revealed that someR-ir cells are also ir for GABA (1.5�2.7% n�36/2406,ections n�12, rats n�3). GABA neurons that expressed
R-ir formed a sub-population of GABA neurons
(sTeaas
C
Csslofawntpmcm
E
UsnTnbswo
apScnaoa
icvtocP1r
sdllpe1
t16w1s
F(lii �m for Ai al (dorsa
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299 291
13�3.8%, n�36/262, sections n�12, rats n�3) (Fig. 1b),uggesting that most GABAergic cells do not express GR.hus neurons and cells that do not express GABA-ir areither spiny principal neurons that use the excitatory aminocid, glutamate, as a transmitter or glial cells (Beinfeld etl., 1984; McDonald, 1985). EM results described nexthow that GR-ir is present in both.
ontrols and background labeling
ontrol sections were processed in parallel to experimentections. In the first control, sections were incubated in theecondary antibodies and stained with DAB. No DAB-abeled somata and nuclei were observed in the LA (orther brain structures). In the second control experiment,or confocal double labeling (GR-ir and GABA-ir), primarynd secondary antibodies were mismatched. No signalas detected in the opposite channel indicating there waso cross-reactivity between the mouse and rabbit IgG’s. Inhe thirds control experiment, GR peptide fragment wasre-absorption with polyclonal rabbit anti-GR: At the lighticroscope the tissue appeared totally unlabeled and
lear. See electron microscopy control results (below) forore details.
lectron microscopy
ltrastructural analysis of GR-ir structures in the LAhowed that GRs are discretely localized in the cytoplasm,ucleus and postsynaptic membrane of neurons (Fig. 1c).he ultrastructural morphological characteristics of neuro-al somata and nuclei indicated that GR-ir was present inoth (glutamatergic) principal neurons (Fig. 1c) and alsoome (GABA) interneurons (not shown). This conclusionas based on the presence a smooth non-invaginated,
ig. 1. GR-ir labeling in the rat amygdala. (A) GR-ir cells are diarrowheads) and cytoplasmic (arrows) GR-ir labeling. (B) Dual immunabeled cells within the amygdala. Inset: Confocal micrographs show GRn the nucleus and cytoplasm of the LA neurons (closed arrowheads) ans also shown. Asterisks�location expanded in inset. Scale bars�250nset. Amygdala nuclei subdivisions B�basal, Ce�central, LAd�later
val nucleus and thin rim of cytoplasm, in principal neurons t
nd a round invaginated nucleus and thicker rim of cyto-lasm in GABA neurons (Pare and Smith, 1993, 1998;mith et al., 1998). This finding was consistent with ouronfocal analysis (above). GR-ir was also observed withineuronal perikaryal plasmalemma, endoplasmic reticulumnd Golgi apparatus. GR-ir was observed within the nucleif glia (not shown), including microglia, oligodendrocytes,nd protoplasmic and fibrous astrocytes.
Within the extra-somatic neuropil, GR-ir was observedn presynaptic axon terminals, where it was observed tooat presynaptic vesicles (Fig. 2a and b). Often singleesicles were intensely labeled (Fig. 2a and b). Labelederminals formed asymmetric synapses on small dendritesr spines. Since asymmetric synapses in the LA are asso-iated with excitatory transmission (Smith and Pare, 1994;are et al., 1995; Farb and LeDoux, 1997; Pare and Smith,998), these terminals likely arise from glutamatergic neu-ons.
GR-ir was observed in dendritic processes especiallypines (Fig. 2d) and also glial elements (Fig. 2a). Labeledendrites and spines comprised almost one-quarter of all
abeled profiles (see below for quantification). The preva-ence of GRs on dendritic spines likely reflects their ex-ression within projection cells since GABAergic cells areither aspiny or at most sparsely spiny (Pare and Smith,993, 1998; Farb and LeDoux, 1997; Smith et al., 1998).
We compared and quantified the sub-cellular struc-ures labeled with each antibody. BuGR2, monoclonal:094 sub-cellular structures were examined in an area of885 �m2 (n�4 animals). Of all DAB labeled profiles 31%ere anatomically unidentifiable; 22.3% were glia-processes;7.4% were presynaptic terminals; 15.1% were dendriticpines including postsynaptic membrane density (PSD);
throughout the amygdala. Inset: Photomicrograph shows nuclearcent labeling of GR (green) and GABA (red) shows the distribution of(GA), and dually (D)-labeled cells. (C) GR-labeling (circles) is detectedhe nuclear membrane (open arrowheads). A labeled glial (LG) process
and B, 10 �m for A and B insets, and 700 nm for C and 1 �m for Cl division), Lvl�lateral (ventral–lateral), Lvm�lateral (ventral–medial).
stributedofluores, GABA
d along t
he remaining 14.2% composed of large and small den-
dcw3pdpgtastpt
lra1wldcgPt
aP2GlPesbte
C
Il
l(lsuoddislrAaalSE
Fspsglial process (LG) is also shown. Arrowheads point to immunoperoxidase
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299292
rites and both neuron and glia somata. For GR-57 poly-lonal: 985 sub-cellular structures in an area of 9173 �m2
ere examined (n�3 animals). Of all DAB-labeled profiles5% were anatomically unidentifiable; 21.7% were glia-rocesses; 19% were presynaptic terminals; 16.6% wereendritic spines including PSD; the remaining 7.7% com-osed of large and small dendrites and both neuron andlia somata. (See Fig. 3 for summary.) These data suggesthat the monoclonal (BuGR2) and the polyclonal (GR-57)ntibodies labeled the same population of subcellulartructures, and support the findings of extra-somatic GR inhe LA. Labeled cellular structures composed only a pro-ortion of all structures (see Fig. 2c for example and con-rol results (below) for quantification).
We also observed, for the first time, discrete GR-irabeling at the PSD. Synapses observed were asymmet-ical, consistent with these GR-ir synapses being postsyn-ptic to excitatory amino acid release (Smith and Pare,994; Pare et al., 1995; Farb and LeDoux, 1997). GR-iras located at the postsynaptic membrane along the
ength of the PSD (Fig. 2d and Fig. 3a, 3b, 3d). In addition,endritic spines possessing GR at the PSD frequently alsoontained GR-ir spine apparatus (Fig. 2d) and spine or-anelles (Fig. 3a) and spine apparatus. We observed twoSD both postsynaptic to the same bouton with only one of
he PSD GR-ir (Fig. 3b–d).In addition to identifying PSD labeling with both mono
nd polyclonal antibodies, we also identified spine andSD immunoreactivity using both diaminobenzidine (Figs.b and 3a, b, d) and silver-intensified gold (Fig. 2e) asR-ir labels. We found examples of silver intensified-gold
abeling within dendritic spines both at and adjacent to theSD (Fig. 2e). Thus we found GR-ir PSD using two differ-nt antibodies with the same label (DAB) and also with theame antibody (GR-57 polyclonal) using two different la-eling techniques. These data provide the first ultrastruc-ural morphological evidence that GRs are associated withxtranuclear sites, including the PSD.
ontrols and background labeling
n order to assess the degree of potential backgroundabeling we examined tissue in the electron microscope
abeling in terminal and glia. (B) GR is present in axon terminalsasterisks). Both terminals show coated vesicles with some intenselyabeled (arrows). One presynaptic axon terminal forms an asymmetricynapse onto an usp. Also present in micrograph is an example of annlabeled presynaptic terminal (ut) that forms an asymmetric synapsento an usp. (C) An example electron micrograph of the density andistribution of GR-ir structures in the neuropil. Present is a GR-irendritic spine (LSp), a GR-ir axon terminal (arrow) and also two other
dentified structures (asterisks). Also present are unlabeled dendrites,pines, and axons terminals (not labeled). Quantification revealedabeled structures were 0.3552 per �m2. (D) An ut forms an asymmet-ic junction onto a GR-labeled spine (LSp). The PSD is ir (asterisks).rrow points to immunolabel along the spine apparatus (sa) andrrowhead points to a deposit of reaction product. (E) An ut forms ansymmetric synapse (asterisks) onto a dendritic spine (LSP) that is
abeled with GR-ir silver-intensified gold (SIG). Arrowheads point to
ig. 2. Electron micrographs reveal that GR is localized to non-omatic sites in the lateral amygdala including the PSD. (A) GR isresent in a presynaptic axon terminal (LT) that forms an asymmetricynapse (asterisks) onto an unlabeled dendritic spine (usp). A labeled
IG particles on and adjacent to the PSD. Scale bar�250 nm for A and
, 200 nm for B and D, and 500 nm for C.
trasfisa(s((pitlFdlptptias
Itbnd(Wa
as2sasaSnaqfoisfiorssip
nDlrGlbwfaI
F(o( n of GR-ii C, D) 50
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299 293
hat had the primary antibody omitted. As with the controlesults from the transmitted light microscope analysis (seebove), no labeled neuronal or glia somata or neuropiltructures were detected. In order to establish the speci-city of the GR-ir for somatic sites as well as non-somaticites including the PSD we next tested the polyclonalntibody (GR-57) preabsorbed with the control peptidesee Experimental Procedures). Alternative sections wereelected for incubation in either the polyclonal antibody1 �g/ml) or the polyclonal preabsorbed with peptide10 �g/ml) and the tissue was processed in parallel. Ex-eriment and control tissue was assessed and quantified
n the electron microscope for the number of labeled struc-ures. Experimental tissue was found to have 0.3552 DAB-abeled profiles per �m2 (2483.075 �m2 examined, seeig. 2c for example of labeled structure distribution andensity). Control tissue was found to have 0.0163 DAB-
abeled profiles per �m2 (1531.0 �m2 examined). Thus,reabsorbed control tissue showed a 95.4% reduction inhe number of labeled profiles. The reduction in labeledrofiles appeared consistent across all anatomical struc-ures observed. These findings, together with the anatom-cal findings using both the monoclonal and polyclonalntibodies, provide further support that the novel GR-irtructures identified in this study are specific.
DISCUSSION
n this paper we show the cellular and subcellular localiza-ion of GR in the lateral amygdala of the rat. The LA is aelieved to be a site of convergent US auditory and CSociceptive signaling during Pavlovian auditory fear con-itioning, resulting in plasticity at US afferents to the LAFanselow and LeDoux, 1999; LeDoux, 2000; Davis and
halen, 2001; Sah et al., 2003; Pare et al., 2004; Dityatev
ig. 3. GR immunolabeling of the PSD. (A) GR-ir labeling of the PSDsp). GR-ir spine organelles are also present in the spine head (asterisnto spines (arrows): One spine is GR-ir labeled (lsp) at the PSD whileasterisks) is also present in the lsp. (C, D) Enlargement for comparison B. (D) GR-ir PSD shown in B. Scale bar�(A) 500 nm (B) 200 nm (
nd Bolshakov, 2005; Maren, 2005). Fear learning can be u
stressful experience or it can also be coincident withtressful experiences (McEwen, 2000; Kim and Diamond,002; Sapolsky, 2003; Sullivan et al., 2004). Because thetress response involves activation of the hypothalamicdrenal cortical axis leading to increased plasma cortico-terone and activation of brain GR (McEwen, 2000; Kimnd Diamond, 2002; McGaugh and Roozendaal, 2002;apolsky, 2003), this feedback of stress could influenceeural processing and fear consolidation in the lateralmygdala (McGaugh and Roozendaal, 2002). Thus a keyuestion in understanding the cellular impact of stress onear consolidation is to understand the cellular localizationf GR within the LA network. We used immunocytochem-
cal techniques combined with transmitted light micro-copy, confocal microscopy and electron microscopy. Wend GR-ir is located in the somata and nuclear membranef principal neurons and a subpopulation of GABA neu-ons. We also find, GR-ir located in specific non-somatictructures within the neuropil including glia processes, pre-ynaptic terminals, neuronal dendrites, and dendritic spines
ncluding spine organelles and find GR-ir has an affiliation forostsynaptic membrane densities.
We first asked whether GR are localized in principaleurons (McDonald, 1992) or GABA interneurons (Mc-onald, 1992) of the LA. We found GR are predominantly
ocated in principal neurons. A proportion of GABA neu-ons (13%) also showed GR-ir. The data thus suggest thatR-ir is either not present or not detectible in comparable
evels in most GABA neurons of the LA. The LA, like otherrain regions, contains subpopulations of GABA neuronshich are differentially connected within the network and ir
or different neural peptides (McDonald, 1985; McDonaldnd Betette, 2001; McDonald and Mascagni, 2001, 2002).t would be of great interest therefore to identify the pop-
ads) of an asymmetrical synapse located on the head of a LA spinepresynaptic terminal simultaneously forms two asymmetric synapsesspine PSD (upper spine) is unlabeled (ulsp). A labeled spine organeller labeled and unlabeled PSD’s shown in B. (C) Unlabeled PSD shownnm.
(arrowhek). (B) A
the other
lation of stress responsive (GR-ir) GABA neurons and to
ln
srsssLaDGmwwneF1a(sGtt
ldlAFoimaAn1apotaONscia2mb
T
Irsia
at
Gipwqqltsgs
pissgmt(
FHsns2dsaa
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299294
earn how stress via GR influences their behavior within LAetwork.
We next asked how GR-ir is distributed within theubcellular space of the LA. GR-ir was found in both neu-onal and glial structures. These structures include gliaomata and glia processes, axonal terminals, dendriticpines, and dendrites. The presence of GR-ir in dendriticpines indicates that GR is most likely present in spines ofA principal neurons (McDonald, 1982) as GABA neuronsnd found to be either non-spiny or sparsely-spiny (Mc-onald, 1982; Smith et al., 2000). Similarly the presence ofR-ir in presynaptic terminals was found in terminals thatade asymmetrical synapses. While it is not knownhether the GR-ir terminals are originating from outside orithin the nucleus the data indicate that the terminals areot GABAergic and are possibly glutamatergic (McDonaldt al., 1989; LeDoux and Farb, 1991; LeDoux et al., 1991;arb et al., 1992; Smith and Pare, 1994; Pare and Smith,998; Smith et al., 2000) or a neuromodulator that makessymmetric synaptic contacts such as norepinephrineNE) (Milner et al., 1998). Taken together there is thusome correlation between the confocal dual label (GR andABA) study and the GR-ir electron microscopy results in
he finding that GR-ir appears to be preferentially localizedo principal neurons.
The third finding in this study is that GR-ir is alsoocalized to the postsynaptic membrane density in den-ritic spines. PSD’s on spines of LA principal neurons are
ikely postsynaptic to glutamatergic afferents and containMPA and NMDA receptors (LeDoux and Farb, 1991;arb et al., 1995; Farb and LeDoux, 1997). Major sourcesf afferents to these spines are from thalamus and cortex,
ncluding auditory afferents from the medial geniculateedial division (MGm) and cortical area TE3 (LeDoux etl., 1991; Farb et al., 1992; Farb and LeDoux, 1997).nother significant source of glutamatergic input to LAeurons is from intrinsic connections (Smith and Pare,994; Pare and Smith, 1998; Smith et al., 1998). Thus byprocess of elimination it is likely the GR-ir PSD are
ostsynaptic to one of these major sources. However,ther sources cannot be excluded. Moreover, the findinghat not all PSD are GR-ir suggests that the presynapticfferents could be from a relatively discrete input to the LA.ne possible source is the brain stem neuromodulator NE.E is a candidate for two reasons. It makes asymmetricalynaptic contacts (Milner et al., 1998) and secondly be-ause co-activation of NE and GR has been extensively
mplicated in amygdala dependent memory consolidationnd retrieval (Roozendaal, 2000; McGaugh and Roozendaal,002; Roozendaal et al., 2002, 2004). Below we discuss inore detail the functional consequences of these findingsut first we consider the technical limitation of the study.
echnical considerations
n this study we show the novel localization of GRs. Ineporting novel localization of classical receptors, there areeveral technical considerations which could impact on thenterpretation of the results. These are the specificity of the
ntibody and the possibility of antibody cross reactivity,
ms
nd also non-specific background labeling. We addressedhese issues using four control strategies.
First, we used two different antibodies directed againstR. We selected a polyclonal and a monoclonal antibody
n order to contrast the results and minimize any falseositive results due to cross-reactivity of the antibodiesith non-specific antigens in the LA. To compliment ourualitative analysis of GR-ir with each antibody we alsouantified the results (Fig. 4). Given that our extra-somatic
abeling profiling is strikingly similar with the two antibodiesested (Fig. 4) it is unlikely that both antibodies show theame non-specific labeling for the same neuronal (andlia) components. Therefore results from this first technicaltrategy suggest the GR-ir may be specific.
The second strategy employed was to preabsorb theolyclonal antibody with the control peptide antigen prior to
ncubation and to quantify the results. Serial LA brainections were processed in parallel and DAB-labeledtructures were quantified for control and experimentroups. Preabsorption of the antibody eliminated the vastajority of DAB-labeled structures (see Results for quan-
ification). However some labeling albeit limited remainedsee Results for quantification). There was no structural
ig. 4. Distribution of sub-cellular GR immunolabeling (GR-ir). (A)istogram compares GR-ir structures for each antibody in identifiabletructures. BuGR2 monoclonal (white bars): 1094 strucutres/6885 �m2,�4 animals (31% unidentifiable); GR-57 polyclonal (black bars): 985trucutres/9173 �m2, n�3 animals (35% unidentified). Glia-processes2.3% mc, 21.7% pc; presynaptic terminals: 17.4% mc, 19% pc;endritic spines including PSD: 15.1% mc, 16.6% pc. (B) Schematichowing ultra-structural distribution of GR-ir in a neuron of the lateralmygdala. GRs are located in the somatic cytoplasm and nucleus andlso in the cytoplasm including dendritic spine apparatus and at the
embrane in the PSD. GR-ir PSDs are a population of asymmetrical
ynapses located on the heads of dendritic spines of LA neurons.
pts
nUmwr
tTss(
rsnMpmp
F
Cvc1ebaebBbrdltmm1e
gnotrSipLorcbt
ptir
tbmacetoep2cmMotaRaidkGtu
amiocaiodbccrhstwcscbn
oTpr
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299 295
reference for the remaining labeling. Thus we concludedhat the vast majority of GR-ir structures observed arepecific for GR.
The third strategy employed was to control for possibleon-specific labeling resulting from secondary antibodies.sing both an omission of primary antibody and a mis-atch of secondary antibodies in different experiments weere able to establish that the labeling detected is not the
esult of non-specific binding of the secondary antibody.The forth strategy employed was to ascertain whether
he GR-ir labeling at the PSD was an artifact of DAB use.hus we also performed GR-ir using the pre-embeddingilver– gold labeling technique. We found examples ofilver– gold GR-ir in dendritic spines and at the PSDFig. 2E).
In summary, while immunocytochemistry has inherentisks of false positive results, the use of these four controltrategies greatly minimizes this risk and suggests thaton-somatic GR and PSD localized membrane GR exist.oreover increasing evidence (discussed below) fromhysiological and behavioral studies indicates that GRay also act through membrane mechanisms. This studyrovides the first anatomical evidence for this possibility.
unctional implications
lassically, neurosteroids including GR exert their actionia intra-somatic receptors which translocate to the nu-leus and regulate gene transcription (McEwen et al.,988; Cintra et al., 1994; Sapolsky et al., 2000). Howevervidence is emerging for the possibility of membraneound active steroid receptors (Ramirez, 1996; Venerond Borrell, 1999; Falkenstein and Wehling, 2000; Schmidtt al., 2000; Losel et al., 2003). The existence of mem-rane bound GR has been controversial (Liposits andohn, 1993; Cintra et al., 1994; Ramirez, 1996) and haseen investigated to a lesser extent than other steroidseceptors (Losel et al., 2003). Our findings provide evi-ence that at least some of the membrane bound GR is
ocalized to the PSD in a population of asymmetric excita-ory synapses in the LA. Its location at the PSD, at asym-etrical synapses, is suggestive of a potential role in theodulation of excitatory synaptic transmission (Farb et al.,992; Smith and Pare, 1994; Pare and Smith, 1998; Smitht al., 2000).
While some studies have shown hippocampal CA1lucocorticoid signaling requires GR homodimerization,uclear translocation, and DNA binding (Karst et al., 2000),ther studies indicate that the rapid actions of adrenocor-ical hormones may be mediated by a G-protein-coupledeceptor at the cell surface (Orchinik et al., 1992, 1994).imilarly, the rapid actions of other steroid hormones may
nvolve cellular signaling pathways rather than DNA-rotein interactions (Pietras and Szego, 1975; Finidori-epicard et al., 1981; Levin, 2002). Recent evidence dem-nstrates the existence of a functional membrane estrogeneceptor (ER) (Revankar et al., 2005). The membrane ERan trigger a second messenger cascade and Ca2� mo-ilization (Revankar et al., 2005). Similar to the finding of
he localization of GR at the PSD in this study, ER� has p
reviously been shown to also be localized in spines and athe PSD (Adams et al., 2002) Evidence is therefore grow-ng for the possible existence of membrane bound steroideceptors.
Recent behavioral studies have also supported a po-ential role of membrane bound GR. This conclusion haseen derived by the dependence of GR action on secondessenger systems and secondly by the time course of thection. A wealth of evidence by McGaugh, Roozendaal andolleagues (McGaugh and Roozendaal, 2002; Roozendaalt al., 2002) supports a role of GR and noradrenergic recep-ors in the consolidation of memory. Their recent studiesn memory enhancement mediated by GRs show memorynhancement by GR antagonists can be inhibited by arotein kinase A inhibitor (Rp-cAMPS) (Roozendaal et al.,002). This finding that GR actions appear to act via alassic second messenger hints to potential non-classicalembrane bound GR mechanism. Moreover, Roozendaal,cGaugh and colleagues also found, in the same andther studies that the memory enhancing effects of post-raining GR activation also depend on coactivated nor-drenergic systems (McGaugh and Roozendaal, 2002;oozendaal et al., 2002). Antagonism of �-adrenoceptorslso blocked the memory enhancing effects of post-train-
ng GR activation (Roozendaal et al., 2002). Thus theseata provide the first evidence for an interaction between anown membrane bound neurotransmitter receptor andR and a second messenger system and strongly suggest
hat not all GR-mediated effects are necessarily via mod-lation of gene transcription.
The potential time course for non-genomic-mediatedctions of GR activation has been proposed to be fromilliseconds (Ramirez, 1996; Losel et al., 2003) observed
n cellular studies to up to 1 h as proposed from behavioralbservations (Breuner et al., 1998). While behavioral andellular studies on the time course of rapid actions of GRctivation are scarce, Breuner et al. (1998) were able to
nduce a rapid behavioral effect via an oral administrationf corticosterone in birds. The time course included theigestion and absorption of corticosterone-loaded wormsy white-crowned sparrows. The increase in plasma corti-osterone was no longer observed after 1 h while in-reased locomotion was observed by 15 min. Moreover,estraint stress-induced potentiation of conditioned fearas been demonstrated with as little as 1 h of restrainttress (Radulovic et al., 1999). These times contrast withhe amount of time required for genomic action of steroidshich, in order to be detected as behavioral and cellularhange, are reported to take from hours to days (Falken-tein et al., 2000) (Losel et al., 2003). Thus, the timeourse, as well as second messenger activation and mem-rane localization, all suggest that GRs might functionongenomically in LA.
The most cogent evidence of rapid cellular responsesf corticosterone acting at the GR has been shown byakahashi et al. (2002) using hippocampal cell culturereparation. Takahashi and colleagues found corticoste-one, with less than 100 sec of incubation, dramatically
rolonged the NMDA mediated Ca2� influx (Takahashi et
amamtidmit
I
GlswL1oneettlFk2
tmiOsmLPitwaa
pdNwbsmttd(pwmru
igsaasRaca
AJM
A
A
B
B
B
B
B
B
B
B
C
C
C
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299296
l., 2002). Moreover they demonstrated this effect couldimic by the GR agonist RU 28362, albeit not as stronglyt the doses utilized, and the corticosterone effect could beaintained with bovine serum albumen conjugated to cor-
icosterone, a conjugation known to prevent the cellularnternalization of steroid molecules. These data support airect role of membrane GR modulation of EEA receptorodulation. We discussed below the potential behavioral
mplications of this finding in the mammalian fear condi-ioning circuit involving the lateral amygdala.
mplications for synaptic plasticity
Rs are known to directly modulate cellular models ofearning plasticity. Activation of GR causes an apparenthift in the level of afferent activity detected by synapses,hich results in an increased or decreased probability forTD and LTP induction, respectively (Coussens et al.,997; Kim and Diamond, 2002). Moreover, these effectsn plasticity are mediated via voltage-gated calcium chan-els (VGCC) and depend on NMDA channels (Coussenst al., 1997; Kim and Diamond, 2002). Aspects of this GRffect could be mediated via membrane bound receptors athe PSD. In addition to GR located at the PSD, shown inhis study, both NMDA receptors and VGCC channels areocated within dendritic spines in the LA (McDonald, 1996;arb and LeDoux, 1997; Pare and Smith, 1998) (Weiss-opf et al., 1999; Weisskopf and LeDoux, 1999; LeDoux,000; Bauer et al., 2002).
A major question in plasticity research is whether pro-ein translation can occur in dendrites (Steward and Schu-an, 2001, 2003), and specifically within individual spines
n a synapse specific manner (Nimchinsky et al., 2002).ur data demonstrating the localization of GR in dendriticpines suggest a potential rapid, non-genomic subcellularechanism for GR actions and plasticity regulation withinA spines. The identification of both GR-ir organelles andSD (Figs. 2 and 3) within individual dendritic spines may
ndicate the local synthesis of GR in spines prior to inser-ion into the postsynaptic membrane. This locates GRhere they could directly and independently interact withnd modulated mechanisms of plasticity (Nimchinsky etl., 2002).
Our findings introduce the possibility that GR effects onlasticity, in the LA, can occur directly at the postsynapticensity, through a direct or indirect GR interaction withMDA receptors and VGCCs. Given that synaptic GRsere found to be selectively localized to a proportion of,ut not all, asymmetric synapses, GRs may regulate aubpopulation of excitatory afferents in LA. If the GRs wereore prevalent in the cortical, rather than thalamic afferent
argets, then during periods of extreme stress (such as inhe development of PTSD), a shift in the balance of amyg-ala-mediated fear memories toward indelible sub-corticalLeDoux, 2000) representations might occur. This pro-osed modulation of thalamic and cortical excitatory path-ays to the amygdala via GR at the PSD may also provideechanisms by which the coactivation of GR and norad-
energic receptors in the basolateral amygdala can mod-
late memory consolidation of fearful stimuli. The colocal-
zation and coactivation of glucocorticoid and noradrener-ic receptors, or GR and EAA receptors at the sameynapses in the same or adjacent dendritic spines mayccount for the ability of second messenger systems toffect the GR and noradrenergic receptor-mediated con-olidation of memory (McGaugh and Roozendaal, 2002;oozendaal et al., 2002). Future experiments need toddress the colocalization and coactivation of glucocorti-oid, noradrenergic and EAA receptors at synapses in themygdala.
cknowledgments—We thank Dr. Beth Stutzman for initial advice,ustine Barry for assistance with preliminary experiments andargaret Trafford for assistance. Supported by MH58911.
REFERENCES
dams MM, Fink SE, Shah RA, Janssen WG, Hayashi S, Milner TA,McEwen BS, Morrison JH (2002) Estrogen and aging affect thesubcellular distribution of estrogen receptor-alpha in the hippocam-pus of female rats. J Neurosci 22:3608–3614.
erni A, Traber R, Hock C, Roozendaal B, Schelling G, Papassotiro-poulos A, Nitsch RM, Schnyder U, de Quervain DJ (2004) Low-dose cortisol for symptoms of posttraumatic stress disorder. Am JPsychiatry 161:1488–1490.
auer EP, Schafe GE, LeDoux JE (2002) NMDA receptors and L-typevoltage-gated calcium channels contribute to long-term potentia-tion and different components of fear memory formation in thelateral amygdala. J Neurosci 22:5239–5249.
einfeld MC, Korchak DM, Roth BL, O’Donohue TL (1984) The distri-bution and chromatographic characterization of PHI (peptide histi-dine isoleucine amide)-27-like peptides in rat and porcine brain.J Neurosci 4:2681–2688.
lair HT, Schafe GE, Bauer EP, Rodrigues SM, LeDoux JE (2001)Synaptic plasticity in the lateral amygdala: a cellular hypothesis offear conditioning. Learn Mem 8:229–242.
onne O, Brandes D, Segman R, Pitman RK, Yehuda R, Shalev AY(2003) Prospective evaluation of plasma cortisol in recent traumasurvivors with posttraumatic stress disorder. Psychiatry Res 119:171–175.
remner JD (1999) Alterations in brain structure and function associ-ated with post-traumatic stress disorder. Semin Clin Neuropsychi-atry 4:249–255.
remner JD, Vythilingam M, Vermetten E, Afzal N, Nazeer A, New-comer JW, Charney DS (2004) Effects of dexamethasone on de-clarative memory function in posttraumatic stress disorder. Psychi-atry Res 129:1–10.
uchanan TW, Brechtel A, Sollers JJ, Lovallo WR (2001) Exogenouscortisol exerts effects on the startle reflex independent of emotionalmodulation. Pharmacol Biochem Behav 68:203–210.
intra A, Akner G, Covenas R, de Leon M, Wikstrom A-C, Agnati LF,Gustafsson J-A, Fuxe K (1994) Immunocytochemical studies onglucocorticoid receptor. In: Neurobiology of steroids (De Kloet RE,Sutanto W, eds), pp 143–160. New York: Academic Press.
onrad CD, LeDoux JE, Magarinos AM, McEwen BS (1999) Repeatedrestraint stress facilitates fear conditioning independently of caus-ing hippocampal CA3 dendritic atrophy. Behav Neurosci 113:902–913.
onrad CD, MacMillan DD 2nd, Tsekhanov S, Wright RL, Baran SE,Fuchs RA (2004) Influence of chronic corticosterone and glucocor-ticoid receptor antagonism in the amygdala on fear conditioning.
Neurobiol Learn Mem 81:185–199.
C
C
C
C
D
d
d
d
D
D
D
D
F
F
F
F
F
F
F
G
H
K
K
K
K
K
K
L
L
L
L
L
L
L
M
M
M
M
M
M
M
M
M
M
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299 297
ordero MI, Kruyt ND, Sandi C (2003a) Modulation of contextual fearconditioning by chronic stress in rats is related to individual differ-ences in behavioral reactivity to novelty. Brain Res 970:242–245.
ordero MI, Venero C, Kruyt ND, Sandi C (2003b) Prior exposure to asingle stress session facilitates subsequent contextual fear condi-tioning in rats. Evidence for a role of corticosterone. Horm Behav44:338–345.
orodimas KP, LeDoux JE, Gold PW, Schulkin J (1994) Corticoste-rone potentiation of conditioned fear in rats. Ann N Y Acad Sci746:392–393.
oussens CM, Kerr DS, WC A (1997) Glucocorticoid receptor activa-tion lowers the threshold for NMDA-receptor-dependant homosyn-aptic long-term depression in the hippocampus through activationof voltage-dependant calcium channels. J Neurophysiol 78:1–9.
avis M, Whalen PJ (2001) The amygdala: vigilance and emotion. MolPsychiatry 6:13–34.
e Kloet ER, Oitzl MS, Joels M (1999) Stress and cognition: arecorticosteroids good or bad guys? Trends Neurosci 22:422–426.
e Quervain DJ, Roozendaal B, McGaugh JL (1998) Stress andglucocorticoids impair retrieval of long-term spatial memory. Na-ture 394:787–790.
e Quervain DJ, Roozendaal B, Nitsch RM, McGaugh JL, Hock C(2000) Acute cortisone administration impairs retrieval of long-termdeclarative memory in humans. Nat Neurosci 3:313–314.
elahanty DL, Nugent NR, Christopher NC, Walsh M (2005) Initialurinary epinephrine and cortisol levels predict acute PTSD symp-toms in child trauma victims. Psychoneuroendocrinology 30:121–128.
iamond DM, Fleshner M, Ingersoll N, Rose GM (1996) Psychologicalstress impairs spatial working memory: relevance to electrophysi-ological studies of hippocampal function. Behav Neurosci 110:661–672.
uval F, Crocq MA, Guillon MS, Mokrani MC, Monreal J, Bailey P,Macher JP (2004) Increased adrenocorticotropin suppression afterdexamethasone administration in sexually abused adolescentswith posttraumatic stress disorder. Ann N Y Acad Sci 1032:273–275.
alkenstein E, Tillmann HC, Christ M, Feuring M, Wehling M (2000)Multiple actions of steroid hormones-a focus on rapid, nongenomiceffects. Pharmacol Rev 52:513–556.
alkenstein E, Wehling M (2000) Nongenomically initiated steroidactions. Eur J Clin Invest. 30(Suppl 3):51–54.
anselow MS, LeDoux JE (1999) Why we think plasticity underlyingPavlovian fear conditioning occurs in the basolateral amygdala.Neuron 23:229–232.
arb C, Aoki C, Milner T, Kaneko T, LeDoux J (1992) Glutamateimmunoreactive terminals in the lateral amygdaloid nucleus: apossible substrate for emotional memory. Brain Res 593:145–158.
arb CR, Aoki C, LeDoux JE (1995) Differential localization of NMDAand AMPA receptor subunits in the lateral and basal nuclei of theamygdala: A light and electron microscopic study. J Comp Neurol362:86–108.
arb CR, LeDoux JE (1997) NMDA and AMPA receptors in the lateralnucleus of the amygdala are postsynaptic to auditory thalamicafferents. Synapse 27:106–121.
oenjian AK, Pynoos RS, Steinberg AM, Endres D, Abraham K,Geffner ME, Fairbanks LA (2003) Hypothalamic-pituitary-adrenalactivity among Armenian adolescents with PTSD symptoms.J Trauma Stress 16:319–323.
ui GK, Figueroa IR, Poytress BS, Roozendaal B, McGaugh JL,Weinberger NM (2004) Memory enhancement of classical fearconditioning by post-training injections of corticosterone in rats.
Neurobiol Learn Mem 81:67–74.
arst H, Karten YJG, Reichardt ER, de Kloet ER, Schutz G, Joels M(2000) Corticosteroid actions in hippocampus require DNA bindingof glucocorticoid receptor homodimers. Nat Neurosci 3:977–978.
arst H, Nair S, Velzing E, Rumpff-van Essen L, Slagter E, Shinnick-Gallagher P, Joels M (2002) Glucocorticoids alter calcium conduc-tances and calcium channel subunit expression in basolateralamygdala neurons. Eur J Neurosci 16:1083–1089.
err DS, Campbell LW, Hao SY, Landfield PW (1989) Corticosteroidmodulation of hippocampal potentials: increased effects with ag-ing. Science 245:1505–1509.
im JJ, Diamond DM (2002) The stressed hippocampus, synapticplasticity and lost memories. Nat Rev Neurosci 3:453–462.
im JJ, Foy MR, Thompson RF (1996) Behavioral stress modifieshippocampal plasticity through N-methyl-D-aspartate receptor ac-tivation. Proc Natl Acad Sci U S A 93:4750–4753.
uhlmann S, Kirschbaum C, Wolf OT (2005) Effects of oral cortisoltreatment in healthy young women on memory retrieval of negativeand neutral words. Neurobiol Learn Mem 83:158–162.
eDoux JE (2000) Emotion circuits in the brain. Annu Rev Neurosci23:155–184.
eDoux JE, Farb CR (1991) Neurons of the acoustic thalamus thatproject to the amygdala contain glutamate. Neurosci Lett 134:145–149.
eDoux JE, Farb CR, Milner TA (1991) Ultrastructure and synapticassociations of auditory thalamo-amygdala projections in the rat.Exp Brain Res 85:577–586.
evin ER (2002) Cellular functions of plasma membrane estrogenreceptors. Steroids 67:471–475.
iposits Z, Bohn MC (1993) Association of glucocorticoid receptorimmunoreactivity with cell membrane and transport vesicles inhippocampal and hypothalamic neurons in the rat. J Neurosci Res35:14–19.
osel RM, Falkenstein E, Feuring M, Schultz A, Tillmann HC, Rossol-Haseroth K, Wehling M (2003) Nongenomic steroid action: contro-versies, questions, and answers. Physiol Rev 83:965–1016.
uecken LJ, Dausch B, Gulla V, Hong R, Compas BE (2004) Alter-ations in morning cortisol associated with PTSD in women withbreast cancer. J Psychosom Res 56:13–15.
aren S (2001) Neurobiology of Pavlovian fear conditioning. AnnuRev Neurosci 24:897–931.
aren S (2005) Building and burying fear memories in the brain.Neuroscientist 11:89–99.
cDonald AJ (1982) Neurons of the lateral and basolateral amygda-loid nuclei: a Golgi study in the rat. J Comp Neurol 212:293–312.
cDonald AJ (1985) Morphology of peptide-containing neurons in therat basolateral amygdaloid nucleus. Brain Res 338:186–191.
cDonald AJ (1992) Cell types and intrinsic connections of the amyg-dala. In: The amygdala: neurobiological aspects of emotion, mem-ory, and mental dysfunction (Aggleton JP, ed), pp 67–96. NewYork: Wiley-Liss, Inc.
cDonald AJ (1996) Glutamate and aspartate immunoreactive neu-rons of the rat basolateral amygdala: colocalization of excitatoryamino acids and projections to the limbic circuit. J Comp Neurol365:367–379.
cDonald AJ, Beitz AJ, Larson AA, Kuriyama R, Sellitto C, Madl JE(1989) Co-localization of glutamate and tubulin in putative excita-tory neurons of the hippocampus and amygdala: An immunohis-tochemical study using monoclonal antibodies. Neuroscience30:405–421.
cDonald AJ, Betette RL (2001) Parvalbumin-containing neurons inthe rat basolateral amygdala: morphology and co-localization ofcalbindin-D(28k). Neuroscience 102:413–425.
cDonald AJ, Mascagni F (2001) Colocalization of calcium-bindingproteins and GABA in neurons of the rat basolateral amygdala.Neuroscience 105:681–693.
cDonald AJ, Mascagni F (2002) Immunohistochemical characteriza-tion of somatostatin containing interneurons in the rat basolateral
amygdala. Brain Res 943:237–244.
M
M
M
M
M
N
N
O
O
P
P
P
P
P
P
P
P
P
R
R
RR
R
R
R
R
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
S
T
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299298
cEwen BS (2000) Effects of adverse experience for brain structureand function. Biol Psychiatry 48:721–731.
cEwen BS, Brinton RE, Sapolsky RM (1988) Glucocorticoid recep-tors and behavior: implications for the stress response. Adv ExpMed Biol 245:35–45.
cGaugh JL (2000) Memory: a century of consolidation. Science287:248–251.
cGaugh JL, Roozendaal B (2002) Role of adrenal stress hormonesin forming lasting memories in the brain. Curr Opin Neurobiol 12:205–210.
ilner TA, Lee A, Aicher SA, Rosin DL (1998) Hippocampal alpha2a-adrenergic receptors are located predominantly presynaptically butare also found postsynaptically and in selective astrocytes. J CompNeurol 395:310–327.
eylan TC, Brunet A, Pole N, Best SR, Metzler TJ, Yehuda R, MarmarCR (2005) PTSD symptoms predict waking salivary cortisol levelsin police officers. Psychoneuroendocrinology 30:373–381.
imchinsky EA, Sabatini BL, Svoboda K (2002) Structure and functionof dendritic spines. Ann Rev Physiol 64:313–353.
rchinik M, Moore FL, Rose JD (1994) Mechanistic and functionalstudies of rapid corticosteroid actions. Ann N Y Acad Sci 746:101–112; discussion 112–114.
rchinik M, Murray TF, Franklin PH, Moore FL (1992) Guanyl nucle-otides modulate binding to steroid receptors in neuronal mem-branes. Proc Natl Acad Sci U S A 89:3830–3834.
are D, Quirk GJ, Ledoux JE (2004) New vistas on amygdala net-works in conditioned fear. J Neurophysiol 92:1–9.
are D, Smith Y (1993) Distribution of GABA immunoreactivity in theamygdaloid complex of the cat. Neuroscience 57:1061–1076.
are D, Smith Y (1998) Intrinsic circuitry of the amygdaloid complex:common principles of organization in rats and cats. Trends Neu-rosci 21:240–241.
are D, Smith Y, Pare JF (1995) Intra-amygdaloid projections of thebasolateral and basomedial nuclei in the cat: Phaseolus vulgaris-leucoagglutinin anterograde tracing at the light and electron micro-scopic level. Neuroscience 69:567–583.
avlides C, Wantanabe Y, Margarinos AM, McEwen B (1995) Oppos-ing roles of type I and type II adrenal steroid receptors in hippocam-pal long term potentiation. Neuroscience 68:387–394.
helps EA, Anderson AK (1997) Emotional memory: what does theamygdala do? Curr Biol 7:R311–R314.
ico-Alfonso MA, Garcia-Linares MI, Celda-Navarro N, Herbert J,Martinez M (2004) Changes in cortisol and dehydroepiandros-terone in women victims of physical and psychological intimatepartner violence. Biol Psychiatry 56:233–240.
ietras RJ, Szego CM (1975) Endometrial cell calcium and oestrogenaction. Nature 253:357–359.
ugh CR, Tremblay D, Fleshner M, Rudy JW (1997) A selective rolefor corticosterone in contextual-fear conditioning. Behav Neurosci111:503–511.
adulovic J, Ruhmann A, Liepold T, Spiess J (1999) Modulation oflearning and anxiety by corticotropin-releasing factor (CRF) andstress: differential roles of CRF receptors 1 and 2. J Neurosci 19:5016–5025.
ainnie DG (1999) Serotonergic modulation of neurotransmission inthe rat basolateral amygdala. J Neurophysiol 82:69–85.
amirez VD (1996) How do steroids act? Lancet 347:630–631.eul JMHM, de Kloet ER (1985) Two receptor systems for corticoste-
rone in rat brain: microdistribution and occupation. Endocrinology117:2505–2511.
evankar CM, Cimino DF, Sklar LA, Arterburn JB, Prossnitz ER(2005) A transmembrane intracellular estrogen receptor mediatesrapid cell signaling. Science 307:1625–1630.
oozendaal B (2000) 1999 Curt P. Richter award. Glucocorticoids andthe regulation of memory consolidation. Psychoneuroendocrinol-ogy 25:213–238.
oozendaal B, Hahn EL, Nathan SV, de Quervain DJ, McGaugh JL
(2004) Glucocorticoid effects on memory retrieval require concur-
rent noradrenergic activity in the hippocampus and basolateralamygdala. J Neurosci 24:8161–8169.
oozendaal B, Quirarte GL, McGaugh JL (2002) Glucocorticoids in-teract with the basolateral amygdala beta-adrenoceptor-cAMP/cAMP/PKA system in influencing memory consolidation. EurJ Neurosci 15:553–560.
ah P, Faber ES, Lopez De Armentia M, Power J (2003) The amyg-daloid complex: anatomy and physiology. Physiol Rev 83:803–834.
andi C, Merino JJ, Cordero MI, Touyarot K, Venero C (2001) Effectsof chronic stress on contextual fear conditioning and the hippocam-pal expression of the neural cell adhesion molecule, its polysialy-lation, and L1. Neuroscience 102:329–339.
apolsky RM (1992) Stress, the aging brain, and the mechanisms ofneuronal death. Cambridge, MA: MIT Press.
apolsky RM (2003) Stress and plasticity in the limbic system. Neu-rochem Res 28:1735–1742.
apolsky RM, Romero LM, Munck AU (2000) How do glucocorticoidsinfluence stress responses? Integrating permissive, suppressive,stimulatory, and preparative actions. Endocr Rev 21:55–89.
chelling G, Briegel J, Roozendaal B, Stoll C, Rothenhausler HB,Kapfhammer HP (2001) The effect of stress doses of hydrocorti-sone during septic shock on posttraumatic stress disorder in sur-vivors. Biol Psychiatry 50:978–985.
chelling G, Kilger E, Roozendaal B, de Quervain DJ, Briegel J,Dagge A, Rothenhausler HB, Krauseneck T, Nollert G, Kapfham-mer HP (2004) Stress doses of hydrocortisone, traumatic memo-ries, and symptoms of posttraumatic stress disorder in patientsafter cardiac surgery: a randomized study. Biol Psychiatry 55:627–633.
chmidt BM, Gerdes D, Feuring M, Falkenstein E, Christ M, WehlingM (2000) Rapid, nongenomic steroid actions: A new age? FrontNeuroendocrinol 21:57–94.
mith Y, Pare D (1994) Intra-amygdaloid projections of the lateralnucleus in the cat: PHA-L anterograde labeling combined withpostembedding GABA and glutamate immunocytochemistry.J Comp Neurol 342:232–248.
mith Y, Pare JF, Pare D (1998) Cat intraamygdaloid inhibitory network:ultrastructural organization of parvalbumin-immunoreactive elements.J Comp Neurol 391:164–179.
mith Y, Pare JF, Pare D (2000) Differential innervation of parvalbumin-immunoreactive interneurons of the basolateral amygdaloid complexby cortical and intrinsic inputs. J Comp Neurol 416:496–508.
teward O, Schuman EM (2001) Protein synthesis at synaptic sites ondendrites. Annu Rev Neurosci 24:299–325.
teward O, Schuman EM (2003) Compartmentalized synthesis anddegradation of proteins in neurons. Neuron 40:347–359.
tutzmann GE, LeDoux JE (1999) GABAergic antagonists block theinhibitory effects of serotonin in the lateral amygdala: a mechanismfor modulation of sensory inputs related to fear conditioning. J Neu-rosci 19:RC8.
tutzmann GE, McEwen BS, LeDoux JE (1998) Serotonin modulationof sensory inputs to the lateral amygdala: dependency on cortico-sterone. J Neurosci 18:9529–9538.
ullivan GM, Apergis J, Bush DE, Johnson LR, Hou M, Ledoux JE(2004) Lesions in the bed nucleus of the stria terminalis disruptcorticosterone and freezing responses elicited by a contextual butnot by a specific cue-conditioned fear stimulus. Neuroscience128:7–14.
ullivan GM, Gorman JM (2002) Finding a home for post-traumatic stressdisorder in biological psychiatry. Is it a disorder of anxiety, mood,stress, or memory? Psychiatr Clin North Am 25:463–468, ix.
akahashi T, Kimoto T, Tanabe N, Hattori TA, Yasumatsu N, KawatoS (2002) Corticosterone acutely prolonged N-methyl-d-aspartatereceptor-mediated Ca2� elevation in cultured rat hippocampal
neurons. J Neurochem 83:1441–1451.
V
V
W
W
W
Y
Y
Y
Y
Z
L. R. Johnson et al. / Neuroscience 136 (2005) 289–299 299
enero C, Borrell J (1999) Rapid glucocorticoid effects on excitatoryamino acid levels in the hippocampus: a microdialysis study infreely moving rats. Eur J Neurosci 11:2465–2473.
yas A, Mitra R, Shankaranarayana Rao BS, Chattarji S (2002) Chronicstress induces contrasting patterns of dendritic remodeling in hip-pocampal and amygdaloid neurons. J Neurosci 22:6810–6818.
eisskopf MG, Bauer EP, LeDoux JE (1999) L-type voltage-gatedcalcium channels mediate NMDA-independent associative long-term potentiation at thalamic input synapses to the amygdala.J Neurosci 19:10512–10519.
eisskopf MG, LeDoux JE (1999) Distinct populations of NMDA re-ceptors at subcortical and cortical inputs to principal cells of thelateral amygdala. J Neurophysiol 81:930–934.
halen PJ, Kagan J, Cook RG, Davis FC, Kim H, Polis S, McLaren DG,Somerville LH, McLean AA, Maxwell JS, Johnstone T (2004) Human
amygdala responsivity to masked fearful eye whites. Science 306:2061
ehuda R, Golier JA, Halligan SL, Meaney M, Bierer LM (2004) TheACTH response to dexamethasone in PTSD. Am J Psychiatry161:1397–1403.
ehuda R, McFarlane AC, Shalev AY (1998) Predicting the develop-ment of posttraumatic stress disorder from the acute response to atraumatic event. Biol Psychiatry 44:1305–1313.
oung EA, Breslau N (2004) Saliva cortisol in posttraumatic stressdisorder: a community epidemiologic study. Biol Psychiatry 56:205–209.
oung EA, Tolman R, Witkowski K, Kaplan G (2004) Salivary cortisoland posttraumatic stress disorder in a low-income community sam-ple of women. Biol Psychiatry 55:621–626.
orawski M, Killcross S (2003) Glucocorticoid receptor agonist en-hances pavlovian appetitive conditioning but disrupts outcome-
specific associations. Behav Neurosci 117:1453–1457.
(Accepted 16 June 2005)(Available online 21 September 2005)
