Previous data have consistently demonstrated an inhibitory effect of androgens on stress-induced hypothalamic-pituitary-adrenal (HPA) responses. Several brain regions may influence androgen-mediated inhibition of the HPA axis, including the medial preoptic area. To test the role of the medial preoptic nucleus (MPN) specifically, we examined in high- and low-testosterone-replaced gonadectomized rats bearing discrete bilateral lesions of the MPN basal and stress-induced indexes of HPA function, and the relative levels of corticotropin-releasing hormone (CRH) and arginine vasopressin (AVP) mRNA in the amygdala. High testosterone replacement decreased plasma adrenocorticotropin hormone (ACTH) and paraventricular nucleus (PVN) Fos responses to restraint exposure in sham- but not in MPN-lesioned animals. AVP-, but not CRH-immunoreactivity staining in the external zone of the median eminence was increased by testosterone in sham animals, and MPN lesions blocked this increment in AVP. A similar interaction between MPN lesions and testosterone occurred on AVP mRNA levels in the medial nucleus of the amygdala. These findings support an involvement of MPN projections in mediating the AVP response to testosterone in both the medial parvocellular PVN and medial amygdala. We conclude that the MPN forms part of an integral circuit that mediates the central effects of gonadal status on neuroendocrine and central stress responses.
- corticotropin releasing hormone
- arginine vasopressin
- medial amygdala
- hypothalamic-pituitary-adrenal axis
the vast distribution of different types of sex steroid hormone receptors within the central nervous system place this class of hormones well beyond the realm of reproductive function (56). Indeed, testosterone and estrogen exert reliable inhibitory and stimulatory effects, respectively, on activity of the hypothalamic-pituitary-adrenal (HPA) axis, a neuroendocrine system essential for survival (55). Moreover, individual and gender-based differences in normal and abnormal HPA function can be attributed to variations in sex steroid hormone release (46, 51).
Threats to homeostasis activate the HPA axis by triggering the sequential release of a chain of hormones. This is initiated by the recruitment of neurosecretory neurons in the paraventricular nucleus (PVN) of the hypothalamus that secrete peptide stores from the median eminence, primarily corticotropin-releasing hormone (CRH) and arginine vasopressin (AVP) (44). CRH and AVP synergize on the release of adrenocorticotropin hormone (ACTH) from the anterior pituitary that, in turn, stimulates the release of glucocorticoids from the adrenal gland, cortisol in humans, and corticosterone in the rat (1, 3). A fine balance between glucocorticoid negative feedback inhibition and stress-induced drive to the HPA axis ultimately determines the magnitude of the ACTH response to stress (10, 54). Testosterone can operate on all of these elements by exerting inhibitory actions on the cellular and transcriptional activation of PVN motor neurons, ACTH secretagogue synthesis and release, as well as by enhancing glucocorticoid negative feedback efficacy (47, 49).
There are multiple sites and pathways mediating the central actions of androgens on the HPA axis. Various metabolites of testosterone, including 5α-dihydrotestosterone and its 3β-diol metabolite, are capable of acting locally to inhibit stress-induced levels of PVN Fos mRNA and plasma ACTH and corticosterone (26). Our connectional studies predict that androgens can also act within a very large assortment of brain regions relaying sensory and limbic information to the PVN region. The medial preoptic nucleus (MPN) stands out in this regard, as we found that it contains the highest concentration of androgen receptor (AR) positive efferents to the PVN (56). In line with this connectional data, testosterone implants in the vicinity of the MPN reduce the plasma ACTH and corticosterone responses to restraint (50). Moreover, high testosterone replacement levels in the periphery that normally suppress the magnitude of the HPA stress response fail to do so in rats bearing large electrolytic lesions of the medial preoptic area (50).
At this point, several uncertainties remain. First, while these electrolytic lesions in the medial preoptic area encompassed the MPN, they were also large enough to have damaged several neighboring subcortical relays to the PVN region. This would include the anterior hypothalamic area; the ventral noradrenergic ascending bundle, which travels through the lateral hypothalamic and lateral preoptic areas; and the posterior medial region of the bed nucleus of the stria terminalis (BST), which markedly inhibits HPA responses to acute stress (8, 9). Second, while systemic changes in testosterone can operate on CRH and AVP expression and stress-induced levels of Fos within PVN motor neurons (27, 48), the extent to which a functioning MPN is necessary for these synthetic and cellular stress responses to occur remains to be seen.
In the present study, we superimposed two levels of testosterone replacement in the periphery in animals receiving sham or small-volume, bilateral injections of ibotenic acid in the MPN. This allowed us to assess how testosterone acts and interacts with the MPN on ACTH secretagogue synthesis, HPA output, and intervening levels of Fos activation within different compartments of the PVN. Of note, the MPN receives and sends input to the medial nucleus of the amygdala (7, 39), one of many limbic regions expressing ARs (56) and regulating the HPA axis (11). Testosterone stimulates AVP mRNA and peptide in the medial amygdala (12), and several lines of evidence continue to relate the inhibitory influence of the gonadal axis on HPA function in males to testosterone-dependent increases in extrahypothalamic AVP (17, reviewed in Ref. 55). Based on the potential for the MPN to influence the HPA axis to and through AVP neurons in the medial amygdala, we gauged whether the AVP response to testosterone within this structure also depends on the integrity of the MPN. Our findings indicate that the MPN is integral for testosterone to act on the PVN, as well as its extended circuitries.
MATERIALS AND METHODS
Eighty-eight adult male Sprague-Dawley rats (Charles River, St. Constant, Canada) were used, weighing 250 g on arrival (56 days old) and 375 g when sampled (∼75 days old). Animals were pair housed under controlled temperature (23 ± 2°C) and lighting conditions (12:12-h light-dark cycle, lights on at 0600), with food (Labdiet; Rat diet 5012) and water available ad libitum. All experimental protocols were approved by the University of British Columbia Animal Care Committee.
To explore how testosterone interacts with the MPN, body weight-matched animals were gonadectomized (GDX), randomly assigned to one of four replacement-MPN treatment combinations [1) low testosterone-sham, 2) low testosterone-lesion, 3) high testosterone-sham, and 4) high testosterone-lesion], and sampled under basal and/or restraint stress conditions.
Testes were removed via a scrotal incision under ketamine-xylazine-acepromazine anesthesia (25, 5, and 1 mg/ml, respectively; 1 ml/kg sc). Each testis was delivered separately through the scrotal incision and removed by severing the vas deferens and spermatic artery, which was ligated to maintain hemostasis. Gonadectomy was completed by closing the scrotal incision with 4–0 nonabsorbable suture. Testosterone replacement was performed at the same time, using two subcutaneous Silastic implants (each 35 mm in length, 0.062 mm inner diameter, 1.25 mm outer diameter; Dow Corning, Midland, MI) that were packed with crystalline testosterone (Sigma Aldrich, Oakville, Canada) to a length of 10 or 60 mm for low and high testosterone replacement, respectively (22). As shown in results, these implants provide plasma testosterone concentrations in the physiological range observed in adult male rats (48, 50).
Comparing the effects of MPN lesions in gonadal-intact and GDX rats with or without testosterone replacement was initially considered. However, this is not the most informative approach for establishing that the MPN is an obligate target for the central actions of testosterone on the HPA axis to occur. As our previous connectional experiments suggest (56), there are likely several candidate structures in the brain supplying androgen-sensitive input to the PVN region, in addition to the MPN. Importantly, gonadotropin-releasing hormone neurons are diffusely distributed throughout a continuum of the medial basal forebrain, including within the region of the MPN (20, 37). Because lesions in the vicinity of the MPN can decrease circulating testosterone levels in animals with testes (23), this approach would not allow us to discern a role for the MPN independent from changes in testosterone release. GDX rats show increases in HPA activity similar to those bearing MPN lesions (48). Thus data gleaned from these animals would also lack the specificity for attributing the central effects of testosterone to the MPN. In light of these uncertainties, in the present study, we superimposed two levels of testosterone replacement in GDX animals bearing MPN lesions, both to control for testosterone and to specifically address whether the MPN is required for the HPA axis to register differences in testosterone in circulation.
After gonadectomy and testosterone replacement surgeries (3 days), rats were anesthetized by injection of pentobarbital sodium (50 mg/kg ip) and were placed in a Kopf stereotaxic apparatus. To produce discrete, axon-sparing lesions (see Ref. 23), rats received bilateral volume injections of ibotenic acid (100 nl, 5 μg/μl in 0.1 m PBS, pH 7.2; Sigma Chemicals, Oakville, Canada) or PBS (sham lesion; 0.1 m, pH 7.2) in each MPN sequentially (0.15 mm rostral to bregma, 0.4 mm lateral to the midsagittal sinus, 7.75 mm below the pial surface; bite bar set at 3 mm below the ear bars) using a Hamilton microsyringe (32 gauge blunt needle; Hamilton, Reno, NV). The syringe was lowered and kept in position for 5 min before injection, which was then delivered at a rate of 10 nl/min over a 10-min period. To reduce diffusion along the pipette track, the syringe was left in place for an additional 15 min before removal. One series of sections through the region of the MPN was counterstained with thionin to examine needle track location and neuron loss. Two additional adjacent series were processed to determine patterns of gliosis and AR staining. As described in greater detail in results, tissue and blood samples from animals bearing improper or ineffective MPN lesions were removed from the analysis.
Tissue and blood collection.
Animals were weighed daily and allowed to recover for 14 days before restraint exposure, which involved placing rats in flat-bottom Plexiglas restrainers (8.5 × 21.5 cm; Kent Scientific, Litchfield, CT) for 30 min and then returning them to their home cage for an additional 30-min period. Blood samples obtained from the tail vein were collected in ice-chilled aprotinin- and EDTA-treated tubes (3.75 mg EDTA/100 μl of blood), centrifuged at 3,000 g for 20 min, and stored at −20°C until assayed. Blood samples were obtained immediately following removal from the home cage (0 min), at the end of restraint (30 min), and 30 min following the termination of restraint (60 min). All testing was performed during the light phase of the cycle, beginning at 0800.
Rats were anesthetized for perfusion after a tail blood sample was taken, either following home cage removal or 60 min following the onset of restraint. Based on our previous time course studies, the 30-min poststress interval is optimal for detecting, within individual animals, both the relative differences in stress-induced indexes of HPA function and intervening levels of Fos in PVN attributable to differences in gonadal status (48). As verified by corneal and pedal pinch reflexes, deep anesthesia was reliably achieved within 1–2 min of chloral hydrate administration (40% wt/vol in dH20; 1 ml/100 g body wt ip). Rats were perfused via the ascending aorta with 0.9% saline, followed by 4% paraformaldehyde (pH 9.5), both at 4°C. Saline and fixative were delivered over 5 and 20 min, respectively, at a flow rate of 20 ml/min. Brains were postfixed for 4 h and cryoprotected overnight with 15% sucrose in 0.1 m potassium PBS (KPBS; pH 7.3) before slicing. Five 1-in-5 series of frozen 30-μm-thick coronal sections through the length of the brain were collected and stored in antifreeze (30% ethylene glycol and 20% glycerol in sterile diethyl pyrocarbonate-treated water) at −20°C until processing. Adjacent series of tissues from each animal were used for in situ hybridization- and immunohistochemical analyses. In all cases, an additional series was counterstained with thionin and alternately compared with dark- and bright-field illuminations for morphological and anatomical reference.
Plasma testosterone (25 μl), corticosterone (5 μl, diluted 1:200 as per kit instructions), and ACTH (50 μl) concentrations were measured using commercial RIA kits (MP Biomedical, Solon, OH). For corticosterone, the plasma samples were diluted 1:100 and 1:200 for prestress and poststress time intervals, respectively, to render hormone detection within the linear part of the standard curve. The intra- and interassay coefficients of variation for all of the assays typically ranged from 3 to 6 and 10 to 12%, respectively, and 125I-labeled ligands were used as tracer in all cases.
The testosterone antibody cross-reacts 100% with testosterone and slightly with 5α-dihydrotestosterone (3.40%), 5α-androstane-3β, 17β-diol (2.2%), and 11-oxotestosterone (2%). The standard curve ED50 for the testosterone RIA was 1.2 ng/ml, with a detection limit of 0.1 ng/ml. The corticosterone antibody cross-reacts 100% with corticosterone and slightly with desoxycorticosterone (0.34%), testosterone, and cortisol (0.10%) but does not react with the progestins or estrogens (<0.01%). The standard curve ED50 for the corticosterone RIA was 17 μg/dl, with a detection limit of 0.625 μg/dl. The ACTH antibody cross-reacts 100% with ACTH1–39 and ACTH1–24, but not with β-endorphin, α- and β-melanocyte-stimulating hormone, and α- and β-lipotropin (all <0.8%). The standard curve ED50 for the ACTH RIA was 82 pg/ml, with a detection limit of 20 pg/ml.
Lesion placement in the MPN was determined by analyzing patterns of Nissl staining, glial cell infiltration (glial fibrillary acidic protein, GFAP), and AR staining within adjacent tissue series. Glial cell infiltration was identified using a primary antiserum purified from bovine GFAP (AB5804, lot no. 0506002852; Millipore, Billerica, MA; 1:2,000). AR immunoreactive neurons were localized using a primary antiserum (0.025 μg/ml; 1:8,000) raised against the NH2-terminal amino acids 2–20 of the AR (sc-816, lot no. E1004; Santa Cruz Biotechnology, Santa Cruz, CA). Restraint-responsive neurons in the PVN were localized using Fos-immunoreactivity (ir) as a marker of cellular activation using a primary antiserum (1:45,000) raised against amino acids 4–17 of the human Fos protein (Ab-5, lot no. 4191–1-1; Oncogene Research Products, Boston, MA).
Free-floating sections were first rinsed in KPBS buffer to remove cryoprotectant and then pretreated with 0.3% hydrogen peroxide for 10 min to quench endogenous peroxidase activity. This was followed by four rinses in KPBS and then in sodium borohydride (1% wt/vol in KPBS) for 5 min to reduce free aldehydes. Sections were then incubated for 48 h at 4°C in a KPBS-Triton (0.3% Triton X; Sigma-Aldrich, Oakville, Ontario) solution containing 2% normal goat serum and the primary antiserum to detect AR or Fos, as described above. AR and Fos primary antisera were detected using a conventional nickel-intensified, avidin-biotin-immunoperoxidase (Vectastain Elite ABC kit; Vector laboratories, Burlington, CA) procedure (24). GFAP was detected by using a nonnickel variant of the procedure, as previously described (25). Control experiments, in which the primary antisera to AR or Fos were preadsorbed for 24 h at 4°C with excess synthetic peptide immunogen, corresponding to NH2-terminal amino acids 2–21 of the rat AR (0.25 μM, sc-816-P, EVQLGLGRVYPRPPSKTYRG; Santa Cruz Biotechnology) or to amino acids 4–17 of the human c-fos (50 μM, PP10, SGFNADYEASSSRC; Oncogene Research Products), failed to yield any evidence of specific AR or Fos staining. Additional control experiments involving the omission of either primary or secondary antibody yielded no specific labeling.
Discrete localization of Fos-ir profiles to the medial dorsal parvocellular (mpd; neurosecretory, anterior pituitary-regulating) and to the dorsal and medial ventral parvocellular (dp and mpv, respectively; nonneurosecretory, autonomic regulating) populations of the PVN was assisted by redirected sampling of an adjacent series of thionin-stained sections. Total cell number estimates of Fos-positive cells were generated by counting bilaterally the number of Fos-positive cells through each region of interest, averaged by dividing the total number of cell counts by slice number, corrected for double counting errors (19), and multiplying this product by a factor of five to account for slice frequency (1 in 5 sections).
Characterization of CRH and AVP staining in the median eminence was performed using a dual immunohistochemical labeling technique, including a rabbit polyclonal antibody against CRH (T-4037, lot no. 970177-1; 1:2,000; Bachem, Torrance, CA) and a guinea pig antibody against AVP (T-5048, lot no. 061305; 1:25,000; Bachem). Free-floating tissue was prepared as described above, with slight modifications of these procedures to optimize double labeling for CRH- and AVP-ir, including 1) the elimination of hydrogen peroxide pretreatment, 2) using BSA as a blocking agent, and 3) incubating tissue sections in primary antisera for 24 h at 4°C. Primary antisera against CRH and AVP were detected using conjugated anti-rabbit IgG (Alexa 594; 1:500; Invitrogen) and anti-guinea pig IgG (1:500; Alexa 488, Invitrogen) fluorescent secondary antibodies, respectively. Concurrent immunofluorescence detection of CRH- and AVP-ir in the median eminence was achieved under appropriate fluorescence wavelength conditions, using a Texas red filter (Leica TX2 no. 513843) to detect Alexa 594 under 480 ± 40 nm excitation and 527 ± 30 emission and a fluorescein isothiocyanate filter (Leica L5 no. 513840) to detect Alexa 488 under 560 ± 40 nm excitation and 645 ± 75 emission wavelengths. Control experiments in which the primary antisera to CRH and AVP were preadsorbed with excess levels of their respective immunogens failed to yield any evidence of specific staining. Furthermore, experiments involving the cross adsorption of excess amounts of CRH and AVP, in addition to the omission of either primary or secondary antibody, provided no evidence of cross-reactivity of the CRH primary to detect AVP nor the AVP primary antibody to detect CRH.
A Leica 40X HCL PL Fluotar objective was used to quantify CRH- and AVP-ir localized to the external lamina of the median eminence, the anterior pituitary-directed part of the structure. Optical images from 10 regularly spaced (150 μm) sections through the median eminence were acquired and binarized using constant acquisition and threshold parameters. Binarized images were further skeletonized, and the total average density of pixels was recorded as a measure of staining intensity. Parvocellular PVN neurosecretory neurons are acknowledged as the principal source of CRH-ir terminals in the external zone of the median eminence (24, 60). Determination of AVP staining within terminals specific to this PVN cell population was achieved by redirected sampling of CRH and AVP staining and by quantifying only those AVP profiles superimposed by CRH-positive nerve terminals.
A hybridization approach was used under basal conditions to identify how testosterone acts independently or interacts with the MPN on the relative levels of CRH and AVP mRNA in the central and medial nucleus of the amygdala, respectively. CRH and AVP mRNA hybridization histochemistry were carried out using [33P]UTP-labeled (GE Healthcare Bio-Sciences, Baie d'Urfe, Canada) antisense cRNA probes transcribed from a full-length (1.2 kb) cDNA encoding the rat CRH gene and a 230-bp fragment from the 3′-end of exon C encoding the rat vasopressin gene. Techniques for riboprobe synthesis, hybridization, and the patterns of hybridization for these probes in the amygdala are described in greater detail elsewhere (41, 53). Based on the strength of the hybridization signal on X-ray film (β-max; Amersham), the hybridized slides were then coated with Kodak NTB2 liquid autoradiographic emulsion and exposed at 4°C in the dark with desiccant. Exposure time to emulsion was optimized to ensure that mRNA levels detected were within the linear range of the assay and could be quantified by making relative comparisons in optical density (OD) levels (12 days for CRH mRNA in the central amygdala; 24 days for AVP mRNA in the medial amygdala). Using a standard reference frame, average OD values were determined bilaterally on six and four regularly spaced (150 μm) intervals through the central and medial amygdala, respectively, and corrected by background subtraction. Hybridized tissue series were aligned using white matter morphology illuminated under darkfield conditions and the cytoarchitectonic features provided by an adjacent series of Nissl-stained material.
Parceling of the rat brain followed the mapping of Fos-ir in the PVN, and CRH and AVP mRNA in the amygdala, as defined by the morphological features provided by thionin staining of adjacent series of tissue, based on the terminology of Swanson (42), and of Swanson and Kuypers (43), Swanson and Simmons (45), and Viau and Sawchenko (52) to describe the PVN, of Dong and Swanson (14) to describe the BST, and of Canteras et al. (7) and Dong and colleagues (13) to describe the central and medial amygdala. Features and terminology to describe the MPN were based on Simerly et al. (38–40). Light- and dark-level images were captured using a Retiga 1300 CCD digital camera (Q-imaging, Burnaby, BC), analyzed using Macintosh OS X-driven, Open Lab Image Improvision version 3.0.9 (Quorum Technologies, Guelph, ON) and ImageJ version 1.38 software (NIH, Bethesda, MD), and exported to Adobe Photoshop (version 10.0; San Jose, CA), where standard methods were used to adjust contrast and brightness, and final assembly at a resolution of 300–600 dpi.
Data are expressed as means ± SE and were analyzed by using a two- and three-way ANOVA to detect testosterone replacement and MPN lesion effects on Fos under stress conditions, as well as CRH- and AVP-based data under basal conditions. ACTH and corticosterone were analyzed by ANOVA using one repeated measure (time of sample). Significant ANOVAs were followed using Newman-Keul's post hoc test. Immuno- and in situ hybridization-histochemical and hormone comparisons were made observer-blind by assigning coded designations to the data sets in advance.
Our pilot studies indicated that bilateral lesions that were centered within, but biased toward, the caudal part of the nucleus were most reliable in terms of inducing elevated plasma ACTH responses to stress. Pilot studies employing a unilateral lesion approach in gonadal intact animals demonstrated a complete loss of AR staining in the ipsilateral, but not in the contralateral, MPN. Furthermore, beyond sparing the amount of damage to fibers of passage (11), the volume and concentration of ibotenic acid used were effective in producing lesions that were histologically distinct and consistently uniform in the animals chosen for analysis.
The extent of the excitotoxin lesions was reliably demarcated by examining local patterns of GFAP induction and loss of AR staining (Figs. 1 and 2). The caudal part of the MPN is conspicuously composed of magnocellular neurons (38–40) and shows a high density of AR staining that almost completely mirrors its contours. Compared with shams, a lesion was deemed effective in animals showing a loss of AR staining and magnocellular material, as well as glial infiltration that was centered within the caudal part of the MPN. In general, cellular damage was restricted to the MPN; however, slight damage was occasionally observed within cells occupying the neighboring medial preoptic area and ventral portions of the posterior division of the BST, including the interfascicular nucleus (see Fig. 1). Although we cannot rule out completely the possible influence of these regions, the interfascicular nucleus has little direct input to the PVN (14). Furthermore, the principal nucleus, clearly spared in all of the animals examined, appears to be the major subnucleus of the posterior BST responsible for regulating the HPA axis (8, 9).
Rats showing unilateral or nonuniform bilateral lesions were removed from subsequent analysis. Furthermore, animals bearing lesions that were focused beyond the intended caudal part of the MPN or showing damage that extended well into the lateral preoptic and anterior hypothalamic areas were also excluded. Peptide-ir, mRNA, and stress hormone data were analyzed in only those animals showing proper lesions in the MPN, as independently verified by an observer who was blind to the experimental design. Based on the exclusion criteria described, a final “n” of six per group was achieved for each of the four testosterone replacement-MPN treatment combinations.
There was a main effect of lesion [F(1,44) = 4.3; P < 0.05] and a significant lesion × time interaction [F(2,88) = 15.0; P < 0.01] on body weight gain, but no effect of testosterone replacement. Differences in body weight gain during the first postsurgical week contributed to this interaction. During this interval, there was a significant effect of lesion [F(1,44) = 19.6; P < 0.01], as shams showed significantly higher body weight gains (P < 0.05) than did MPN-lesion animals, 4.6 ± 0.5 and 0.8 ± 0.6 g/day, respectively. However, there was no main effect of testosterone (P = 0.83) and no lesion × testosterone interaction (P = 0.98) on body weight gain during the first postsurgical week. By the second postsurgical week immediately before stress testing, body weight gains were comparable between sham and MPN-lesion animals, 6.1 ± 0.4 and 6.8 ± 0.5 g/day, respectively. Analysis of absolute body weights on the final day of testing indicated a main effect of testosterone replacement [F(1,44) = 5.5; P < 0.05] but no significant effect of lesion or a significant testosterone × lesion interaction. Low-testosterone-replaced animals showed higher body weights than their high-testosterone-replaced counterparts: 370.5 ± 4.4 and 355.5 ± 5.2 g, respectively. Taken together, these findings indicate that the destruction of the MPN did not contribute to changes in growth by the time of testing, whereas testosterone contributed to differences in body weight.
Testosterone replacement and HPA hormones.
In GDX, low- and high-testosterone-replaced rats, plasma testosterone concentrations were 0.48 ± 0.07 and 2.58 ± 0.08 ng/ml, respectively, validating the reliability of our testosterone replacement regimen.
Analysis of plasma ACTH revealed significant main effects of lesion [F(1,20) = 9.3; P < 0.01], testosterone [F(1,20) = 27; P < 0.01], and restraint [F(2,40) = 216; P < 0.01]. Lesion × testosterone [F(1,20) = 5.7; P < 0.05] and lesion × testosterone × restraint [F(2,20) = 6.5; P < 0.05] interactions were both significant. Post hoc analysis revealed no differences in prestress levels of ACTH. Compared with shams, rats with MPN lesions showed higher levels of ACTH under stress conditions, at both 30 and 60 min of restraint exposure (Fig. 3A).
Analysis of plasma corticosterone revealed significant main effects of lesion [F(1,20) = 9.3; P < 0.01] and restraint [F(2,40) = 233; P < 0.01]. There was no main effect of testosterone and no significant lesion × testosterone interaction. Thus, in contrast to the ACTH response within shams, there was no apparent inhibitory effect of testosterone on corticosterone levels at 30 and 60 min of restraint. However, both lesion × restraint [F(2,40) = 6.2; P < 0.01] and lesion × testosterone × restraint [F(2,40) = 3.6; P < 0.05] interactions were significant, signifying a capacity for the MPN lesions to influence corticosterone. Indeed, post hoc analysis confirmed a stimulatory effect of MPN lesions on plasma corticosterone levels immediately before and during restraint, regardless of testosterone replacement (Fig. 3B).
Parvicellular PVN Fos-ir.
Quantitative assessment of the number of Fos-positive cells within regions of the parvicellular division of the PVN (Figs. 4 and 5) revealed significant main effects of lesion [F(1,40) = 6.2; P < 0.05], testosterone replacement [F(1,40) = 10.7; P < 0.01], and restraint [F(1,40) = 1,763; P < 0.01], as well as a significant lesion × testosterone × restraint × region [F(2,80) = 9.4; P < 0.01] interaction. There were no main effects of lesion or testosterone under basal conditions, neither between or within the medial parvocellular regions analyzed. As a function of stress, there was no effect of restraint on the numbers of cells recruited to express Fos protein in the dp part of the PVN [F(1,40) = 2.4; P = 0.1] (Fig. 5A), whereas both the mpd [F(1,40) = 1,550; P < 0.01] and mpv [F(1,40) = 493; P < 0.01] parts showed elevations in stress-induced Fos-ir (Fig. 5, A and C). For the mpv part of the PVN, there was a significant lesion × testosterone × restraint interaction [F(1,40) = 8.9; P < 0.01]. Sham, high-testosterone-replaced animals contributed to this interaction, showing the highest levels of Fos cell numbers under stress conditions (Fig. 5C). Significant effects of lesion [F(1,40) = 9.8; P < 0.01] and testosterone [F(1,40) = 16; P < 0.01] and a significant lesion × testosterone interaction [F(1,40) = 8.0; P < 0.01] were revealed in the mpd part of the PVN. Post hoc analysis confirmed that the inhibitory effect of testosterone on the number of Fos-ir cells in the mpd occurred in sham but not in MPN lesion animals (Fig. 5B).
Median eminence CRH- and AVP-ir.
Our previous findings indicated that the inhibitory effect of testosterone on stress-induced ACTH release is associated with changes in AVP, but not CRH, content in the median eminence (50). Content measures of median eminence AVP provide, for the most part, an index of magnocellular activity, and both magnocellular and parvocellular neurosecretory neurons contribute to AVP in pituitary portal plasma (3). To provide a more precise index of parvocellular activity, in the current study, we used a dual immunohistochemical approach to detect CRH staining in the external zone of the median eminence and to assess the relative levels of AVP contained by these CRH terminals of medial parvocellular origin (Fig. 6). For CRH-ir, there were no main effects of lesion and testosterone. The lesion × testosterone interaction approached significance [F(1,20) = 3.8; P = 0.06], as reflected by a tendency for MPN lesion rats to show lower levels of CRH staining under high testosterone replacement (Fig. 7A). Analysis of AVP staining revealed significant effects of lesion [F(1,20) = 6.2; P < 0.05], testosterone [F(1,20) = 10.7; P < 0.01], and a significant lesion × testosterone interaction [F(1,20) = 10.4; P < 0.01]. As revealed by post hoc analysis, the lesion × testosterone interaction was based within the high testosterone replacement group, where AVP staining was significantly greater in sham than in MPN lesion animals (Fig. 7B). In low-testosterone rats, AVP staining was comparable between rats bearing sham and MPN lesions (P = 0.15).
Amygdala CRH and AVP mRNA.
Densitometric analyses of CRH mRNA through the extent of the central nucleus of the amygdala (CeA) under basal conditions (Fig. 8) indicated a significant effect of testosterone [F(1,20) = 4.6; P < 0.05] but revealed no significant effect of lesion and no significant lesion × testosterone interaction. The effect of testosterone was attributed to an overall inhibitory effect of high testosterone replacement on CRH expression in both sham and MPN lesion groups (Fig. 8D). Analysis of AVP mRNA in the anterodorsal part of the medial amygdala (Fig. 9) revealed significant effects of lesion [F(1,20) = 15.1; P < 0.01] and testosterone [F(1,20) = 62; P < 0.01] and a significant lesion × testosterone interaction [F(1,20) = 16.8; P < 0.01]. Post hoc analysis revealed that the AVP response to high testosterone replacement was significantly higher in shams compared with MPN lesion animals (Fig. 9D).
Our previous experiments showed that lesioning a large extent of the medial preoptic area blocked the inhibitory effects of a single dose of testosterone replacement on HPA function (50). In the current study, we used four treatment groups, encompassing two background levels of testosterone replacement in GDX adult male rats bearing sham and MPN lesions specifically. Thus our current design allowed us to make new inroads on how central and peripheral components of the HPA axis responds to differences in circulating testosterone levels and whether the MPN is required for the dose-related effect of testosterone to occur.
The data implicate testosterone-sensitive pathways from the MPN in mediating both the activational response to stress and biosynthetic capacity of PVN neurosecretory neurons. Testosterone exerted a dose-related inhibitory effect on restraint-induced levels of Fos within the mpd part of the PVN as well as ACTH levels in plasma. Within the high testosterone replacement group, rats bearing MPN lesions showed higher levels of stress-induced Fos in the mpd PVN and plasma ACTH compared with shams. These findings suggest that testosterone inhibition of HPA effector neurons in the PVN is mediated by ARs located outside the nucleus and that the MPN is required for this mechanism to occur. The extent to which the lesions reflect the removal of testosterone actions that normally occur within or distal to the MPN cannot be ascertained at this point. However, microimplants of the AR antagonist hydroxyflutamide in the MPN can increase the ACTH response to restraint (57).
MPN-lesioned animals showed significantly higher levels of corticosterone both under basal and stress conditions than did sham-lesioned rats (Fig. 3B). However, we found no interaction between lesions and testosterone. This was obviously a consequence of the sham group of animals, showing only a slight testosterone-dependent decrement in corticosterone levels at 30 min of restraint exposure. This departure between ACTH and corticosterone disagrees with our previous experiments showing negative relations between testosterone and stress-induced ACTH and corticosterone in animals with testes (48, 50). Our previous GDX + testosterone experiments also show that stress-induced levels of ACTH and corticosterone vary strongly and negatively with testosterone (50), in animals replaced over the entire range of naturally occurring differences in plasma testosterone (∼0.2–7 ng/ml). These findings could explain why we were unable to detect an inhibitory effect on corticosterone using a single, high dose of testosterone replacement. One may still argue that testosterone is of limited significance to the glucocorticoid response, at least in the current study. However, because of our restricted replacement regimen, we can only interpret our data in so far as determining the relative capacity of different central and peripheral components of the HPA axis to respond to a unique level of testosterone.
Our current design, nevertheless, exposed a potential autonomic involvement, perhaps at the level of the PVN, on how testosterone contributes to the net glucocorticoid response. In support of this possibility, sham animals bearing high testosterone replacement showed higher numbers of Fos-ir neurons in the mpv part of the PVN than low-testosterone-replaced animals under stress conditions (Fig. 5B). Because the mpv cells contribute to the long-descending influences of the PVN on the preganglionic spinal cord neurons controlling the adrenal response to ACTH (reviewed in Ref. 55), this increment in restraint-induced mpv Fos could account for the dissociation observed between ACTH and corticosterone release in sham animals with high testosterone. The mpv Fos response to high testosterone replacement was muted, however, in animals bearing MPN lesions, despite showing higher restraint-induced levels of corticosterone. This paradoxical finding is rescued, perhaps, when considering that the magnitude of the corticosterone response to stress occurs as a function of autonomic outflow in addition to ACTH release as executed by the recruitment of the mpd PVN motor neurons (5, 15). The connective properties of the cellular populations in the PVN that are differentially recruited as a function of testosterone and MPN lesions have not yet been defined. Thus the extent to which testosterone acts and interacts with the MPN on PVN mpv cells identified as projecting to the preganglionic spinal cord neurons requires further clarification. Nevertheless, it should be noted that the AR and the estrogen receptor-β isoform are uniquely distributed within autonomic-related cells of the PVN, including the dorsal, lateral, and ventral components of the medial parvocellular division (4). Thus the mpv PVN neurons appear as ideal candidates for mediating the actions of testosterone on autonomic function directly, whereas the influence of testosterone on hypophysiotropic function appears to be indirect. We propose that testosterone normally acts on both the autonomic and neuroendocrine arms of the PVN and that a balance between these systems defines the net glucocorticoid response to stress.
In response to high testosterone replacement, unstressed sham animals showed no change in CRH but a substantial increase in AVP staining localized to CRH-positive terminals in the external lamina of the median eminence (Figs. 6 and 7). AVP is a weak ACTH secretagogue but potently enhances the stimulatory effects of CRH on ACTH release (3). Thus the stimulatory effect of testosterone and the opposing influence of MPN lesions on AVP content in the median eminence under basal conditions would appear contradictory to effects observed on ACTH under stress conditions. It is generally conceived that resting state levels of CRH- and AVP-ir in the external zone of the median eminence reflect the capacity of the mpd neurons of the PVN to synthesize these peptides (3). As several previous studies have indicated, however, the relative release patterns and contributions of CRH and AVP to the ACTH response are stressor and context specific and cannot be inferred solely on the basis of basal measures of peptide content in the median eminence alone (2, 3). Our current findings indicate that testosterone requires a functioning MPN to inhibit the stress-induced activation of mpd neurons, as well as to redirect the capacity of these neurons to express AVP in favor of CRH. Taken together, this suggests that the inhibitory effect of testosterone on stress-induced ACTH does not occur as a consequence of the capacity of mpd neurons to express AVP but may be functionally coupled to the number of mpd neurons recruited to release peptide stores. The extent to which AVP release actually contributes to testosterone regulation of the HPA axis remains to be determined, and worthy of pursuit, since AVP is thought to be the key variable imparting situation-specific alterations in the magnitude of the ACTH response to stress (2, 3).
Our lesions targeted the caudal half of the MPN, which houses primarily γ-aminobutyric acid (GABA)-ergic neurons in addition to the peptide galanin (6, 30). Although galanin has been implicated in the neuroendocrine regulation of reproduction and energy balance (22), its involvement in HPA regulation has not been approached. Although a dependence of cellular activation and peptide expression in the PVN on the integrity of MPN GABA inputs have yet to be established directly, several lines of evidence support this possibility. Testosterone induces GABA activity in the MPN (18, 59); and several GABA-rich projections to the PVN, including the MPN, are recruited to express Fos protein and glutamic acid decarboxylase mRNA during stress exposure (6, 34).
The MPN shows strong bidirectional connections with the medial amygdala (7, 39), and AVP expression in this region is extremely sensitive to changes in circulating testosterone levels (12). Several lines of evidence suggest an involvement of extrahypothalamic AVP neurons in mediating the central actions of testosterone on the HPA axis (17, reviewed in Ref. 55). Furthermore, the medial amygdala is critical for the HPA response to stressful stimuli, particularly emotional stressors, such as restraint (11, 16, 21, 28). Thus we wondered whether the MPN lesions could influence the AVP response to testosterone within this region of the amygdala (Fig. 9). As expected, sham animals displayed an increase in AVP mRNA levels in the medial nucleus of the amygdala in response to high testosterone replacement. This increment in AVP expression, however, was significantly reduced in animals bearing MPN lesions. Assessment of the relative levels of CRH mRNA in the CeA revealed no interactions between lesions and testosterone (Fig. 8), consistent with the fact that the MPN shows no direct projections to this region of the amygdala (39). Unlike the medial amygdala, the central nucleus (CeA) appears to be less important for HPA activation in response to restraint (11, 16, 32). Thus our findings argue against a role for the CeA in mediating the central actions of testosterone, at least in response to acute forms of psychological stressors. However, since we observed a main negative effect of testosterone on CRH expression in the CeA, variations in testosterone may differentially prepare the HPA response under more physical or systemic types of challenges (11, 32, 58).
Virtually all AVP cells in the medial amygdala of the rat are immunoreactive for ARs (12). Although this signifies a local mode of action, our current findings challenge the notion that testosterone regulates AVP neurons in the medial amygdala directly, subject to the influences of the MPN. The medial amygdala, like most limbic regions, has little or no projections to the hypophysiotropic zone of the PVN (8, 21, 31). The functional influences of the medial amygdala nuclei on the HPA axis, if at all mediated by AVP neurons, may instead involve potential relays in the BST and various hypothalamic structures, including the MPN (8, 21, 55). The extent to which any of these projections contain AVP, rely on testosterone, and depend on a functioning MPN remains worthy of pursuit. AVP pathways originating from the medial amygdala have been shown to contribute to a broad, but linked, array of behaviors associated with autonomic, emotional, and coping responses to stress (25). Taken together with our findings, the MPN stands out as an important neural substrate for harmonizing the central actions of testosterone on behavior and neuroendocrine stress responses.
Perspectives and Significance
It is interesting that the functional effects of MPN lesions were discriminated, for the most part, in high- but not in low-testosterone-replaced rats. Circulating levels of testosterone vary as a function of age, sexual experience, social status, time of day, and in response to stress (55). Insofar as the MPN is recruited to modulate neuroendocrine, autonomic, and behavioral responses to stress, this may very well depend, therefore, on gonadal status and situation-specific changes in testosterone secretion (17). It is important to note, at least in the current study, that, despite massive changes in central stress pathways and at the pituitary, a single dose of high testosterone did not have a large impact on the net corticosterone response. Previous studies have implicated a critical role for AVP in sustaining corticotroph responsiveness during chronic stress (2). Taken together, our findings may be relevant to understanding how testosterone determines normal adaptation under repeated stress conditions, in addition to affective disease states associated with changes in adrenal function and gonadal status. Indeed, genuine gender differences in depression and anxiety (33) and the association of depressive illness with hypogonadism in males (35, 36), suggest a potential role for testosterone in the predisposition of mood disorders related to HPA dysfunction. Taken together with the instability of testosterone levels in rodents and humans, and the strength to which the MPN and testosterone interact on the HPA axis, the MPN may be integral to individual differences in HPA function attributed to variations in testosterone release.
Our present findings underscore how testosterone can bridge several independent, yet converging influences on the PVN. The anatomic specificity by which the MPN influences the inhibitory effect of testosterone on HPA axis function still remains unsettled given the connectivity of the MPN with several other extended circuitries of the PVN, also rich in ARs (39, 56). This does not indict the utility of our present design, since it provides several tenable bases for revisiting how changes in endogenous testosterone levels are registered within the circuits described, and how this impacts the stress-induced activation of these projections to the PVN.
This study was supported by the Canadian Institutes of Health Research (V. Viau).
We thank Brenda Bingham, Megan Gray, and Jenny Wu for technical assistance.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society