Complementing its roles in cognitive and affective information processing, the medial prefrontal cortex (mPFC) is a nodal point of a limbic forebrain circuit that modulates stress-related homeostatic mechanisms, including the hypothalamic–pituitary–adrenal (HPA) axis. mPFC influences on HPA output are predominantly inhibitory and emanate from the prelimbic and/or dorsal anterior cingulate cortical fields (PL and ACd, respectively). mPFC projections do not target HPA effector neurons in the paraventricular hypothalamic nucleus (PVH) directly, distributing instead to nearby forebrain regions, including some that house GABAergic neurons implicated in inhibitory PVH control. To identify pathway(s) subserving HPA-inhibitory mPFC influences, an initial screen for sources of GABAergic input to PVH whose sensitivity to an acute emotional (restraint) stress was diminished by PL/ACd lesions identified a discrete region of the anterior bed nucleus of the stria terminalis (aBST) as a candidate for fulfilling this role. Anatomical tracing experiments confirmed projections from PL (but not ACd) to implicated aBST cell groups, and from these to PVH. Finally, selective immunotoxin-mediated ablation of GABAergic aBST neurons recapitulated the effects of PL/ACd lesions on acute stress-induced activation of HPA output. The identification of a proximate mediator of HPA-inhibitory limbic influences provides a framework for clarifying how inhibitory neural and hormonal controls of HPA output are integrated, adaptations of the axis to chronic stress are effected, and how endocrine abnormalities may contribute to stress-related psychiatric illnesses in which mPFC dysfunction is implicated.
Inhibitory mechanisms play important roles in limiting adverse catabolic and immunosuppressive effects of engaging the neuroendocrine (hypothalamic–pituitary–adrenal; HPA) stress response, particularly under conditions of repeated or prolonged stress exposure. Both neural (neurons using the inhibitory neurotransmitter GABA) and hormonal (glucocorticoid negative feedback) mechanisms are implicated in the negative regulation of the HPA axis (Sapolsky et al., 1984; Kovács and Makara, 1988; Herman et al., 2003). Cell groups in the limbic forebrain that are enriched in glucocorticoid receptor expression (Diorio et al., 1993; Akana et al., 2001) are activated in a stereotyped manner in response to acute emotional stress (Cullinan et al., 1995; Li and Sawchenko, 1998), and lesion studies have identified several of these as participating in the inhibitory control of neurosecretory neurons in the paraventricular hypothalamic nucleus (PVH) that express corticotropin-releasing factor (CRF) for the initiation of HPA axis responses to stress (Sapolsky et al., 1984; Diorio et al., 1993; Herman et al., 1995). Prominent among these is the medial prefrontal cortex (mPFC), which is involved in the modulation of attentional states and in regulating cognitive and emotional information processing (Whalen et al., 1998; Bush et al., 2000). These attributes are critical for comparing present with past experiences, and assigning valence to emotionally salient stimuli. Congruent with this, the mPFC also actively regulates neuroendocrine and autonomic responses to stress (Neafsey, 1990; Diorio et al., 1993; Radley et al., 2006a). mPFC impairment is associated with cognitive and affective symptoms of stress-related psychiatric illnesses (Shin et al., 2007; Liberzon and Sripada, 2008), to which HPA axis dysregulation may contribute in a feedforward manner.
None of the limbic cell groups implicated in negative HPA regulation, including mPFC, provide any substantial direct innervation of the PVH, leading to the widely held view that their stress-inhibitory influences are imparted via interposed GABAergic neurons (Herman et al., 2003). The fact that such GABAergic cell groups have not yet been identified has left open a host of issues as to how the corticosteroid-dependent and -independent inhibitory controls of the axis are organized and integrated. The present study sought to define this circuitry for the mPFC.
An initial experiment aimed to localize identified GABAergic PVH-afferent cell groups whose engagement following acute emotional stress exposure was diminished by ablation of dorsal mPFC, implicated previously in HPA inhibition. This pointed to contiguous components of the anterior bed nucleus of the stria terminalis (aBST), notably the fusiform (BSTfu) and dorsomedial (BSTdm) subdivisions of Swanson and colleagues (Ju and Swanson, 1989; Dong et al., 2001; Dong and Swanson, 2006), as candidate mediators of mPFC influences on the stress axis. Complementary tract tracing experiments confirmed prelimbic area (PL) projections to BSTfu/dm (but not PVH, proper), and prominent aBST inputs to PVH (Dong et al., 2001; Dong and Swanson, 2006). Finally, the hypothesized functional involvement of an aBST relay in modulating stress-induced HPA activation was tested by the focal administration of an immunotoxin, anti-GABA-transporter-1-saporin (antiGAT-1-sap), which preferentially ablates GABAergic neurons expressing GAT-1. The results support a PL→BSTfu/dm (GABAergic)→PVH pathway effecting mPFC inhibition of HPA responses to acute emotional stress.
Materials and Methods
Animals and treatments.
Adult male Sprague Dawley albino rats, maintained under standard laboratory conditions, were used in all experiments. All experimental protocols were approved by the Institutional Animal Care and Use Committee of the Salk Institute. Restraint stress was performed in the morning (9:00 A.M.) in plastic restrainers (Braintree Scientific) for 30 min. Controls were handled comparably, but were not restrained. All animals remained in their home cages during and after restraint until the prescribed time of perfusion for histology (2 h after the termination of restraint), or during collection of repeated blood samples.
Axon-sparing excitotoxin lesions were produced via bilateral pressure ejection of ibotenic acid (10 mg/ml, Sigma) in sterile saline centered in PL, as previously described (Radley et al., 2006a). To selectively ablate GABAergic neurons in BSTfu/dm, antiGAT-1-sap (conjugates of saporin with a rabbit polyclonal antiserum raised against a peptide in the extracellular domain of the GABA transporter-1; Advanced Targeting Systems) were microinjected either unilaterally or bilaterally. A stock solution (1.9 mg/ml) was diluted 1:2 in artificial CSF (aCSF) at pH 7.4. Injections were placed via micropipettes (10–20 μm, inner diameter) using pressure ejection (World Precision Instruments) to deliver 120 nl per side at the following stereotaxic coordinates: anteroposterior (AP) = −0.25 mm; mediolateral (ML) = ±1.20 mm; dorsoventral (DV) = −7.20 mm (from dura). Control lesions involved either injecting saporin conjugated to IgG, or the aCSF vehicle. The extent and specificity of damage was estimated by reconstruction from material hybridized for the 67 kDa isoform of the GABA-synthetic enzyme, glutamic acid decarboxylase (GAD67), and CRF mRNA.
For retrograde labeling of PVH-projecting neurons, iontophoretic injections (+5 μA, 7 s on/off) of 2% Fluoro-Gold (Fluorochrome) (Schmued and Fallon, 1986) were made into PVH (AP = −1.4; ML = +0.4; DV = −7.4). Anterograde tracing experiments involved the following: (1) pressure ejection of 10% biotinylated dextran amine (BDA) (Sigma) made into PL (AP = +2.90; ML = +0.75; DV = −2.5); (2) unilateral iontophoretic injection (+5 μA, 7 s on/off) of 2.5% PHA-L (Vector Laboratories) centered in BSTfu. Both anterograde tracers provide high-resolution labeling of axons and terminals. Iontophoretic deposition of PHA-L yields very discrete injection sites (Gerfen and Sawchenko, 1984), which was considered optimal for targeting circumscribed subregions of the bed nucleus. The size of BDA deposits delivered by pressure ejection varies directly with injection volume, and this allowed more effective sampling of the much larger PL cortical field.
Histology and tissue processing.
Rats were anesthetized with chloral hydrate (350 mg/kg, i.p.) and perfused via the ascending aorta with 100 ml of 0.9% saline followed by 900 ml of ice-cold 4% paraformaldehyde in 0.1 m borate buffer, pH 9.5, at a flow rate of 55 ml/min. The brains were removed, postfixed for 3 h, and cryoprotected in 20% sucrose in 0.1 m phosphate buffer overnight at 4°C. Five one-in-five series of 30-μm-thick frozen coronal sections through the entire brain were cut and collected in cryoprotectant solution and stored at −20°C until processing.
Localization of Fos protein and other antigens were performed by using an avidin-biotin peroxidase protocol (Sawchenko et al., 1990) to localize a primary antiserum raised against a fragment of rat Fos protein (residues 4–17), generously synthesized by J. Rivier (Salk Institute) and previously characterized (Radley et al., 2008). Dual immunoperoxidase labeling for Fos and Fluoro-Gold immunoreactivity was performed by sequentially localizing the antiserum against Fos using a nickel-enhanced diaminobenzidine method (black nuclear reaction product) (Shu et al., 1988), as above, and then a Fluoro-Gold antiserum (Chang et al., 1990), without nickel enhancement (brown cytoplasmic product). Specificity of the Fos antiserum was evaluated by direct colabeling for c-fos mRNA over a range of challenge conditions (data not shown). In addition, specific staining in experimental and control tissue was abolished by preadsorbing the antiserum overnight at 4°C with 50 μm of the synthetic peptide immunogen.
Techniques for probe synthesis, hybridization, and autoradiographic localization of mRNA signal were adapted from Simmons et al. (1989). In situ hybridization was performed using 35S-labeled sense (control) and antisense cRNA probes labeled to similar specific activities using a full-length probe for mRNA encoding CRF (1.2 kb; Dr. K. Mayo, Northwestern University, Evanston, IL), and the 67 kDa isoform of glutamic acid decarboxylase (GAD67, Dr. A. Tobin, University of California, Los Angeles, Los Angeles, CA) (Erlander et al., 1991). Probes for the vesicular glutamate transporter, types 1 and 2 (VGLUT1 and VGLUT2) were derived from mouse cDNAs (Open Biosystems) bearing a high degree of homology to rat (VGLUT1: 96% homology to nucleotides 65–437 of rat VGLUT1, GenBank accession number BE950784; Dr. H. Chin, National Institute of Mental Health, Bethesda, MD; VGLUT2: 94% homology to nucleotides 545–1070 of rat VGLUT2, accession number CB247147; Dr. R. Strausberg, National Institutes of Health, Bethesda, MD). Hybridization using antisense probes for VGLUT1 and VGLUT2 yielded labeling that conformed with the reported distributions in rat brain (Ziegler et al., 2002), and sense probes for each did not provide evidence of specific labeling. A GAT-1 probe (Dr. H. J. Lester, California Institute of Technology, Pasadena, CA) was derived from a cDNA encompassing the entire open reading frame of mouse GAT-1 (Chiu et al., 2002), and supported a localization pattern similar to that described in rat brain (Durkin et al., 1995). Combined immunoperoxidase and isotopic hybridization histochemical localization of Fos immunoreactivity and GAD67 mRNA, respectively, was performed using a modification of protocols detailed previously (Chan et al., 1993).
Separate groups of animals were implanted with indwelling jugular catheters 12 d after receiving bilateral immunotoxin or sham lesions in BSTfu/dm, and 2 d before stress exposure. The procedures for implanting catheters have been described previously (Ericsson et al., 1994). Blood samples (300 μl) were taken before restraint stress to estimate basal ACTH and corticosterone levels. Additional samples were collected at 0, 30, 60, and 90 min after the termination of restraint. ACTH was measured using a two-site immunoradiometric assay obtained in kit form (DiaSorin), with intra-assay and interassay coefficients of variation of 3 and 9%, respectively, and a sensitivity of 1.5 pg/ml. Plasma corticosterone was measured without extraction, using an antiserum raised in rabbits against a corticosterone-BSA conjugate, and 125I-corticosterone-BSA as tracer (MP Biomedicals). The sensitivity of the assay was 0.8 μg/dl; intra-assay and interassay coefficients of variation were 5 and 10%, respectively.
Stereological methods were used to quantify the number of Fos-immunoreactive neurons. These analyses were performed using a computer-assisted morphometry system consisting of a photomicroscope equipped with an X–Y–Z computer-controlled motorized stage, MicroFire camera (Optronics), Gateway microcomputer, and StereoInvestigator morphometry and stereology software (MBF Biosciences). For each analysis, boundaries defining the regions of interest were drawn at 25× using an adjacent series of Nissl-stained sections. In regions identified as PVH-projecting (i.e., retrogradely labeled following FG injections in PVH), labeled cells were used as a guide to further aid the delineation of anatomical boundaries. Analyses of Fos-immunoreactive cells were performed on every fifth section, avoiding cells in the outermost plane of focus. Counts were then multiplied by five to estimate the total number of labeled neurons in the defined region of interest. Volume estimates from cross-sectional area measurements were obtained using the Cavalieri method to probe for possible treatment effects on PVH volume, but no reliable effects were observed. FG- and GAD-labeled neurons (and those doubly labeled for Fos protein) were counted manually in every fifth section (experiment 1). In these analyses, the section thickness was too small relative to the average diameter of labeled neurons in order to permit stereologic approaches, and estimates of cell numbers were obtained using the Abercrombie correction (Abercrombie, 1949).
Semiquantitative densitometric analysis of relative levels of GAD67 and CRF mRNA were performed on emulsion-coated slides using ImageJ software. The optical densities of hybridization signals were determined under dark-field illumination at 100× magnification. The hypophysiotropic PVH (i.e., dorsal medial parvicellular subdivision) was defined from Nissl staining pattern (Swanson and Kuypers, 1980) and aligned with corresponding dark-field images of hybridized sections by redirected sampling. The anterior BST was defined from a combination of Nissl staining and axonal fiber tracts (Dong et al., 2001). Optical density readings, corrected for background, were taken at regularly, 150 μm intervals and average values were determined through the extent of this cell group for each animal. Images from GAD67 and CRF mRNA densitometry were collected using a Hamamatsu Orca CCD camera under the control of OpenLab software (version 3.1.5). Images collected from each analysis were exported first to Adobe Photoshop (version 7) for adjustments to optimize/balance contrast and brightness, and then to Canvas (version 8) for assembly and labeling.
Grouped data from the immunoperoxidase and hybridization histochemical analyses (n = 4–5 per group) were compared with a one-way ANOVA for lesion and treatment (sham + control, sham + stress, lesion + stress) status followed by post hoc pairwise comparisons using a Bonferroni correction. Group data from the hormone assays (n = 5 per group) were compared with a mixed-design ANOVA with one within- (time) and one between- (lesion status) group variable, followed by individual pairwise comparisons as above. Data are expressed as the mean ± SEM.
Effects of dorsal mPFC lesions on stress-induced activation of PVH-projecting cell groups
We previously localized inhibitory influences of mPFC over PVH/HPA responses to acute emotional stress to the dorsal mPFC, including the PL and dorsal anterior cingulate area (ACd) (Radley et al., 2006a, 2008). These areas project weakly, if at all, to the PVH proper, distributing instead to nearby regions of the basal forebrain and hypothalamus (Sesack et al., 1989; Hurley et al., 1991), including some that house GABAergic neurons that could mediate inhibitory control imparted from mPFC (Cullinan et al., 1993; Roland and Sawchenko, 1993). An initial experiment sought to identify GABAergic cell groups that might subserve this role. To this end, excitotoxin lesions were placed in dorsal mPFC, a retrograde tracer (Fluoro-Gold) was injected into PVH, and we assayed PVH-afferent cell groups for lesion-induced reductions in sensitivity to acute restraint (Fig. 1A,B). Dual immunostaining was performed for Fluoro-Gold and Fos protein expression, a generic marker of neuronal activation, in sham-lesioned unstressed controls, and in sham- and dorsal mPFC-lesioned animals 2 h after the termination of restraint stress. We verified our previous observations that cellular activational responses (Fos) in PVH are reliably increased following 30 min of restraint stress, and are further enhanced (by 41%) following dorsal mPFC lesions, relative to sham-lesioned controls (p < 0.05) (Fig. 1C). Alterations in Fos activation in PVH have been found to be predictive of changes in pituitary–adrenal secretory responses under these conditions (Radley et al., 2006a).
In sham-lesioned rats subjected to restraint, six distinct regions of the hypothalamus, including the preoptic area, anterior hypothalamic area, perifornical area, perisupraoptic area, dorsomedial (DMH) and ventromedial hypothalamic nuclei, and several subdivisions of aBST displayed colocalization of Fos and transported tracer, thus representing stress-sensitive PVH afferents (Table 1). Of these, only DMH and contiguous components of aBST, the BSTfu and BSTdm nuclei, exhibited reliable decrements (35% and 54%, respectively) in the number of cells labeled concurrently for Fos and tracer in acutely stressed rats bearing dorsal mPFC lesions compared with sham-lesioned animals (p < 0.05 for each region). Subsequent analyses in aBST and DMH involving combined histochemical localization of Fos protein and mRNA encoding the 67 kDa isoform of glutamate decarboxylase (GAD67; a marker of GABAergic neurons) revealed a significant decrement only in aBST (by 58%; p < 0.05) (Fig. 1D,E). This occurred despite the lack of a lesion effect on overall stress-induced Fos expression in aBST, and the lack of between-group differences in the number of retrogradely labeled cells in this region (supplemental Fig. 1, available at www.jneurosci.org as supplemental material).
Anterograde tracing experiments support aBST as a potential relay between PL and PVH
To determine whether implicated aspects of the aBST receive mPFC input, and thus represent candidate mediators of HPA-inhibitory prefrontal influences, complementary anterograde tracing experiments were performed involving pressure injections of BDA, into different mPFC subfields. Histochemical visualization of transported BDA revealed multiple descending pathways from mPFC. mPFC projections reach the BST by diverging medially at the level of the anterior commissure from fiber bundles coursing through the internal capsule to invade the anterodorsal, anteroventral, and anteromedial divisions of BST. The density of axonal/terminal innervation of aBST varied topographically in a dorsal-to-ventral manner. BDA injections centered in ACd (n = 4), the most dorsal aspect of mPFC, failed to label projections to aBST. More ventrally placed deposits labeled progressively more substantial inputs to BSTfu/dm, with deposits restricted to PL (n = 4) giving rise to an overall moderate innervation (Fig. 2A,B). Coupled with evidence from lesion studies (Diorio et al., 1993; Figueiredo et al., 2003; Radley et al., 2006a), these findings implicate PL, but not ACd, in the inhibition of stress-induced activation of PVH. It is noteworthy that large deposits of BDA into PL labeled few, if any, projections invading the medial parvicellular part of the PVH, which houses CRF-expressing neurons that govern HPA output (Fig. 2C). Separate experiments in which rats (n = 5) received BDA injections in PL and Fluoro-Gold injections centered in PVH, revealed PL projections that ramify extensively among PVH-projecting neurons in BSTfu/dm, with frequent juxtaposition of anterogradely labeled varicosities with retrogradely filled perikarya or dendrites (Fig. 2D–G).
A second set of tracing experiments using the anterogradely transported plant lectin Phaseolus vulgaris-leucoagglutinin (PHA-L) was performed to confirm (Dong et al., 2001; Dong and Swanson, 2006) that implicated aspects of the aBST project to the PVH. PHA-L injections centered in several aBST subregions (n = 4) produced robust labeling of axonal fibers and terminals throughout the anterior hypothalamus and in regions closely adjoining the PVH, with moderate terminal innervation throughout PVH proper, inclusive of both its magnocellular and parvicellular divisions. However, PHA-L injections centered in BSTdm and/or BSTfu (n = 4) labeled very prominent inputs to the PVH whose terminal ramifications included the hypophysiotropic zone, with minimal labeling of the magnocellular division and PVH-surround regions (Fig. 3A,B). In separate experiments, anterograde tracing from the aBST was performed in rats (n = 5) that were adrenalectomized, to upregulate CRF expression in stress-related neurosecretory neurons of the PVH. Confocal microscopic analysis of this material revealed frequent close appositions between labeled axonal elements and CRF-expressing parvicellular neurons (Fig. 3C–E).
Effects of unilateral ablation of GABAergic cells in aBST on PVH activational responses to acute restraint
While the foregoing data are consistent with the thesis that PL influences over stress-induced PVH/HPA activation are conveyed via a disynaptic inhibitory pathway involving GABAergic cell groups in BSTfu/dm, functional evidence to support this hypothesis is lacking. We sought to provide this by using a GABA neuron immunotoxin (antiserum to the GABA transporter, GAT-1, coupled to the ribosomal toxin, saporin; antiGAT-1-sap) to assess the effect of disrupting GABAergic elements in aBST on indices of basal and stress-induced HPA activation. GAT-1 is the most widely distributed GAT in rat brain, and is abundant in basal forebrain (Durkin et al., 1995). Preliminary experiments (n = 3) confirmed robust expression of GAT-1 in the aBST, including in identified PVH-projecting cells (supplemental Fig. 2, available at www.jneurosci.org as supplemental material). Since the initial experiments implicate the dorsomedial, and in particular, fusiform subdivisions of BST in relaying inhibitory influences of PL over stress-induced indices of PVH activation, immunotoxin injections were centered in these regions. The extent of unilateral immunotoxin lesions was evaluated by hybridization histochemical localization of GAD67. Representative immunotoxin lesions to aBST are shown in Figure 4. Selective targeting of inhibitory cell groups was facilitated by initial tests of the immunotoxin at varying concentrations, and assessing the extent of both GAD67 mRNA depletion and nonspecific damage, i.e., to different phenotypic markers (CRF, substance P, VGLUT1 and VGLUT2 mRNAs). The areas of damage common to all cases included, anterior aspects of the subcommissural, anteroventral, dorsomedial, and fusiform nuclei of the BST, portions of the parastrial nucleus, and at their caudalmost extent, portions of the rhomboid and anterolateral parts of the BST. Specificity of antiGAT-1-sap was evaluated by comparison of multiple mRNA transcripts indicative of distinct neuronal phenotypes within the BSTfu/dm in separate series of sections throughout the injection site. First, GAD67 and VGLUT mRNA were examined in adjacent series to evaluate the effect of the antiGAT-1-sap on inhibitory and excitatory neuron subpopulations in BSTfu/dm. The BSTfu/dm showed an abundance of GAD67-expressing cells on the sham-lesioned (intact) side, and were virtually absent on the injected side (Fig. 5). While VGLUT expression was relatively low in the aBST, no frank differences were evident in VGLUT1 and/or VGLUT2 mRNAs as assessed both densitometrically and by cell counts in BSTfu/dm on the lesioned, relative to the intact sides (p = 0.8) (supplemental Fig. 3, available at www.jneurosci.org as supplemental material).
Further characterization of immunotoxin specificity involved the analysis of CRF mRNA expression in another series of sections and comparison with GAD67 mRNA distribution (Fig. 6). This was also of functional interest, in view of evidence to support an excitatory influence of CRF-expressing cells in BSTfu/dm on HPA activity (Choi et al., 2007). CRF-expressing cells were moderately dense in BSTfu, and less so in the BSTdm on the intact side (Fig. 6C). Immunotoxin lesions to aBST resulted in a minimal loss of CRF-expressing cells in these subnuclei (Fig. 6D), supporting a preferential targeting of GABAergic neurons. Densitometric analyses revealed a significant decrease (by 80%, p < 0.01) in relative levels of GAD67 mRNA in BST fu/dm following immunotoxin lesions, but no reliable effect on CRF mRNA (p = 0.3).
Assessments were made of two indices of cellular activation in PVH following 30 min of restraint in animals bearing appropriately placed unilateral immunotoxin lesions of aBST. Whereas the number of cells expressing Fos protein in PVH were elevated following 30 min of restraint compared with unstressed controls regardless of lesion status (p < 0.05), measures of both Fos protein (Fig. 7A,B) and CRF mRNA (Fig. 7C,D) expression displayed significantly greater enhancements following 30 min of acute restraint (by 77% and 52%, respectively) on the side of PVH ipsilateral to the immunotoxin injection in aBST, compared with the intact side (p < 0.05 for each) (Fig. 7E).
Effects of ablating GABAergic cells in aBST on stress-induced hormone secretion
HPA secretory responses before and after 30 min restraint stress were examined in separate groups of sham and bilaterally immunotoxin-lesioned animals. Despite differences in mean peak ACTH levels between groups, data from the sham-lesioned animals were variable, and these did not differ significantly. Nonetheless, within-group comparisons revealed that in bilateral immunotoxin-lesioned animals, plasma ACTH remained significantly elevated above prestress levels for a longer duration, compared with sham-lesioned animals (at 60 min: p < 0.05 and p = 0.3, respectively) (Fig. 8A). Plasma corticosterone levels showed reliable main effects of lesion (p = 0.01) (Fig. 8B), with bilateral immunotoxin lesioned animals showing a 67% enhancement of plasma corticosterone values at the peak (30 min) time point compared with sham-lesioned animals (p < 0.05). Basal plasma levels of ACTH and corticosterone were not affected by the lesions (p = 0.6 and 0.2, respectively), suggesting that the effects of ablating GABAergic neurons in aBST are manifest during stress.
In an initial study, acute restraint-induced cellular activation of PVH was exaggerated following excitotoxin lesions of dorsal mPFC, verifying involvement of this cortical region in HPA inhibition during emotional stress exposure (Diorio et al., 1993; Figueiredo et al., 2003; Radley et al., 2006a). Inferences drawn from comparison of regional dual labeling patterns in separate series of sections, suggested that acute restraint-induced activation of PVH-projecting GABAergic cell groups in BSTfu/dm was markedly diminished following excitotoxin lesions of dorsal mPFC. Tracing experiments confirmed that projections of PL, which minimally innervate PVH proper, provide a moderate input to BSTfu/dm, which in turn, densely innervates the CRF-rich neurosecretory zone of the PVH (Fig. 9). The provision of an interposed inhibitory relay from PL to PVH is consistent with evidence that extrinsic mPFC projections are overwhelmingly glutamatergic (Ottersen et al., 1995). In the final experiment, targeted ablation of inhibitory cell groups in BSTfu/dm resulted in the enhancement of multiple indices of PVH and HPA activation in response to restraint stress, closely recapitulating the effects of excitotoxin lesions of PL reported here and previously (Radley et al., 2006a).
This is the first published report endorsing antiGAT-1-sap as a tool for focally ablating GABAergic cell groups in the CNS. GAT-1 is the most widely distributed of the plasma membrane-associated GABA transporters (Durkin et al., 1995; Borden, 1996), and histochemical analyses of cellular GAT-1 expression support its predominant localization to GABAergic neurons, although expression in some regions is limited to subsets of GABAergic cells (Durkin et al., 1995; Borden, 1996). The immunotoxin comprises an antiserum raised against an extracellular GAT-1 sequence to foster targeted internalization of the ribosome-inactivating protein, saporin, to which it is coupled. GAT-1 protein localizes principally to axons and terminals, consistent with predominantly presynaptic sites of action (Borden, 1996), leaving open questions as to the mechanism of immunotoxin targeting and how generally applicable this tool may be. GABAergic neuron loss was apparent only within the area of aBST to which the immunotoxin was focally administered, whereas no frank effects were seen in GABAergic cell groups known to project to aBST (data not shown), notably the lateral part of the central nucleus of the amygdala (Petrovich and Swanson, 1997; Day et al., 1999). This would indicate that the immunotoxin is not effective in lesioning at a distance on the basis of retrograde transport. As our preliminary work indicated that the efficacy and specificity of immunotoxin effects were concentration-dependent, similar groundwork and controls for specificity will be required to employ the method effectively in other systems.
In the present studies, immunotoxin efficacy was indicated by marked decrements in expression of an enzyme marker for GABAergic neurons (GAD67 mRNA), while selectivity was supported by the failure of the toxin to significantly impact mRNA markers of glutamatergic (VGLUT1 and VGLUT2) and peptidergic (substance P, CRF) neurons. The BSTfu neurons display a similar complement of neurochemical phenotypes as the more dorsally situated oval nucleus of the BST complex, including GABA and CRF (Ju and Swanson, 1989; Day et al., 1999; Dong et al., 2001). Although CRF and GABA are coexpressed in a substantial subpopulation of oval nucleus neurons (Day et al., 1999; Dong et al., 2001), our data suggest that these phenotypes are largely distinct in BSTfu/dm. Furthermore, the possibility remains that the tendency toward reduction of the CRF complement following immunotoxin injection, although not statistically reliable, could be indicative of a small subset of BSTfu neurons that coexpress GAD and CRF.
A focus on the BSTfu/dm was mandated by results of initial experiments implicating these cell groups as sources of GABAergic input to PVH whose sensitivity to acute emotional stress was diminished by dorsal mPFC lesions. Preferential ablation of GABAergic cell groups in BSTfu/dm enhanced central and hormonal indices of stress-induced HPA output in a manner that closely recapitulated the effects of excitotoxin lesions of dorsal mPFC on acute restraint-induced HPA output (Radley et al., 2006a). An inhibitory role for BSTfu/dm was unanticipated in light of recent studies reporting attenuation of HPA responses to acute restraint following indiscriminate excitotoxin lesions to aBST (Spencer et al., 2005; Choi et al., 2007), from which it was concluded that CRF-expressing cell groups in BSTfu/dm exert a stimulatory influence on stress-induced HPA activation (Choi et al., 2007). The present finding of an effect opposite in sign following ablations that target GABAergic neurons preferentially while sparing CRF-expressing ones is consonant with this interpretation. Other studies have shown that stimulation of the aBST can facilitate or inhibit HPA activity (Dunn, 1987; Casada and Dafny, 1991), supporting the idea of distinct HPA-regulatory influences arising from neurochemically differentiated BST subpopulations.
The identification of a relatively discrete cell cluster that is necessary and sufficient for the exertion of stress-inhibitory mPFC influences should provide a framework for addressing lingering organizational issues in central HPA control. First, other stress-sensitive limbic forebrain cell regions, notably the hippocampal formation and septal region, have been implicated in HPA inhibition (McEwen et al., 1968; Dobrakovová et al., 1982), and it is unclear if these effects are mediated independently, or may share a common arbiter in the aBST. There already exists strong evidence that this general region of the BST mediates influences of the ventral subiculum (Cullinan et al., 1993). It is also noteworthy that implicated regions of the aBST receive prominent inputs from the central nucleus of the amygdala, among other cell groups involved in central autonomic control (Dong et al., 2001; Dong and Swanson, 2006), likely defining an additional layer of integration of stress-related effector mechanisms. Second, and related, is the fact that these same limbic structures, including mPFC, comprise candidate sites of corticosteroid negative feedback regulation of HPA output (Diorio et al., 1993; Sánchez et al., 2000; Akana et al., 2001). Knowledge of an inhibitory relay of mPFC influences, at least, will facilitate designing experiments to parse steroid-dependent and -independent components of limbic control. Third, adaptations such as habituation or facilitation are important determinants of adverse health consequences of repeated stress exposure, yet there remain gaps in understanding the neural substrates of these phenomena. Either can be eliminated by lesioning a midline thalamic cell group, the posterior paraventricular thalamic nucleus (PVTp) (Bhatnagar and Dallman, 1998; Bhatnagar et al., 2002), but the proximate mediators of PVTp influences are unclear. In this light, it is of interest that the PVTp issues a moderate projection to BSTfu/dm (Li and Kirouac, 2008), while not targeting the PVH at all, and innervating preferentially the ventral (infralimbic) part of mPFC (Berendse and Groenewegen, 1991).
PFC dysfunction is implicated in a range of psychiatric disorders that may be precipitated (e.g., posttraumatic stress disorder) or exacerbated (forms of depression) by stress (Hains and Arnsten, 2008). These conditions are associated with disordered HPA regulation (Mason et al., 1986; Plotsky et al., 1998), which is apt to contribute to pathology, as repeated emotional stress exposure in animal models gives rise to regressive alterations in mPFC morphology and function (Radley et al., 2006b; Cerqueira et al., 2007; Liu and Aghajanian, 2008). In defining a proximate link between the mPFC and HPA control, the present findings allow assessment of the effects of perturbing key stress-modulatory pathways on affective disease mechanisms, with the promise of highlighting novel targets for pharmacotherapeutic intervention.
This work was supported by National Institutes of Health Grant DK-26741 and was conducted in part by the Clayton Medical Research Foundation. P.E.S. is an Investigator of the Clayton Medical Research Foundation. J.J.R. is supported by a National Alliance for Research on Schizophrenia and Depression Young Investigator award. We thank J. Vaughn for generating the Fos antiserum and Drs. J. Serrats and R. Rissman for antiserum characterization. We also thank Casey Peto for help in imaging, Kris Trulock for help in the preparation of the illustrations, Linda Feighery and Belle Wamsley for editorial assistance, Carlos Arias and Marty Mortrud for technical assistance, and Yaira Haas for assistance with radioimmunoassay.
- Correspondence should be addressed to Dr. Jason J. Radley, Laboratory of Neuronal Structure and Function, The Salk Institute, 10010 North Torrey Pines Road, La Jolla, CA 92037.