|Home | About | Journals | Submit | Contact Us | Français|
In the past several decades, the therapeutic use of anabolic androgenic steroids (AAS) has been overshadowed by illicit use of these drugs by elite athletes and a growing number of adolescents to enhance performance and body image. As with adults, AAS use by adolescents is associated with a range of behavioral effects, including increased anxiety and altered responses to stress. It has been suggested that adolescents, especially adolescent females, may be particularly susceptible to the effects of these steroids, but few experiments in animal models have been performed to test this assertion. Here we show that chronic exposure of adolescent female mice to a mixture of three commonly abused AAS (testosterone cypionate, nandrolone decanoate and methandrostenolone; 7.5 mg/kg/day for 5 days) significantly enhanced anxiety-like behavior as assessed by the acoustic startle response (ASR), but did not augment the fear-potentiated startle response (FPS) or alter sensorimotor gating as assessed by prepulse inhibition of the acoustic startle response (PPI). AAS treatment also significantly increased the levels of corticotropin releasing factor (CRF) mRNA and somal-associated CRF immunoreactivity in the central amygdala (CeA), as well as neuropil-associated immunoreactivity in the dorsal aspect of the anterolateral division of the bed nucleus of the stria terminalis (dBnST). AAS treatment did not alter CRF receptor 1 or 2 mRNA in either the CeA or the dBnST; CRF immunoreactivity in the ventral BNST, the paraventricular nucleus (PVN) or the median eminence (ME); or peripheral levels of corticosterone. These results suggest that chronic AAS treatment of adolescent female mice may enhance generalized anxiety, but not sensorimotor gating or learned fear, via a mechanism that involves increased CRF-mediated signaling from CeA neurons projecting to the dBnST.
Anabolic androgenic steroids (AAS) comprise a large class of synthetic androgens developed for therapeutic purposes (for review, Basaria et al., 2001), but whose predominant use is now illicit self-administration to enhance athletic performance or body image (for review, Trenton and Currier, 2005; Kanayama et al., 2008). Adult men are reported to self-administer AAS at concentrations that reflect 10–100× therapeutic doses of testosterone, which are 10–40 mg/day (Wu, 1997; Daly et al., 2001; Trenton and Currier, 2005) and are prescribed to restore circulating testosterone to normal adult levels of 10 to 35 nmol/L (Perry et al., 2003; Matsumoto and Bremner, 2004). Girls and women are reported to take AAS at levels equivalent to or even exceeding those administered by men (Franke and Berendonk, 1997); thus the same self-administered doses that produce 10- to 100-fold higher than replacement levels in males may be expected to yield circulating levels of androgens that are orders of magnitude higher still than normal physiological levels of androgens in women and girls (< 2ng/L; Wu, 1997).
Laboratory experiments that assess the actions of a single AAS have provided excellent mechanistic insight into the actions of a given individual steroid (for review, Clark and Henderson, 2003; Clark et al., 2006). However, humans who use AAS nearly always do so in intricate and complex patterns that involve concurrent administration of combinations of different classes and types of AAS in a process called stacking (for review, Llewellyn, 2007; Kutscher et al., 2002). The rationale offered by athletes for combining AAS is that stacking activates multiple signaling pathways resulting in “synergistic actions” that augment the ratio of anabolic response to unwanted side effects (Sturmi and Diorio, 1998). Experimental data support the existence of such interactions. For example, recent data suggest that inclusion of 17α-alkylated AAS may inhibit endogenous aromatase, therefore altering the metabolic fate of 19-nortestosterone or testosterone ester AAS derivatives (Penatti et al., 2009b) that are substrates for aromatization (Ryan, 1959; Winters, 1990).
Approximately 0.5% of adolescent girls and ~2% of adolescent boys are estimated to abuse AAS (Bahrke et al., 2000; Miller et al., 2005; Johnston et al., 2009), engendering concern over the effects of exposure to high levels of synthetic steroids at a time when the regions of the brain important in affective behaviors are still developing and highly hormone-sensitive (Sato et al., 2008). Long-term risks associated with AAS use are reported to be higher in women than in men (Franke and Berendonk, 1997), and adolescents may be more sensitive than adults (O'Connor and Cicero, 1993). Sex-specific differences in sensitivity to AAS may also be compounded by sex-specific differences in susceptibility to some behavioral disorders, such as anxiety, which are more prevalent in women (for review, Zender and Olshansky, 2009). Despite the potential for greater untoward effects, few studies have focused on the effects of AAS use in females, and female adolescent subjects are particularly underrepresented (for review, Henderson et al., 2006).
While taken for performance and image-enhancing actions, AAS use is also associated with multiple behavioral effects, including mania, hypomania, somatization, increased anxiety, irritability, extreme mood swings, abnormal levels of aggression, and paranoia (Pagonis et al., 2006; for review, Gruber and Pope, 2000; Trenton and Currier, 2005). Due to ethical constraints, there have been few randomized clinical trials, however, in those that have been performed, using moderate doses of a single AAS in adult male subjects, increased levels of hostility and anxiety were reported (Su et al., 1993; Pope et al., 2000). The extended amygdala, comprising the CeA and adjoining structures including the BnST, is fundamental to the expression of anxious states (for review, Davis and Whalen, 2001; Davis, et al., 2009). Prior studies in rodents suggest that the CeA is paramount in the acquisition and expression of short-duration threat responses that reflect fear while the BnST is key in responses reflecting generalized (Walker et al., 2009a,b) and social (Lee et al., 2008) anxiety. Both changes to specific threats and generalized anxiety are noted in steroid abusers, but the latter may be particularly important with respect to the expression of inappropriate reactions (aggression) and increased negative perception of self and others (Pagonis et al., 2006). The acoustic startle response (ASR) is an involuntary, highly reproducible contraction of facial and skeletal muscles in response to sudden acoustic stimuli that is conserved across all mammalian species and is highly sensitive to changes in perceptual or emotional state (for review, Koch, 1999). Unconditioned startle responses provide an assay of generalized anxiety, while pairing of an aversive stimulus with an auditory or visual cue (fear-potentiated startle; FPS) provides an assay of conditioned fear (for review, Davis, 2006).
The CRF family of peptides acts to integrate sensory, endocrine and autonomic information and formulate an appropriate response to stress (for review, Korosi and Baram, 2008). Elevated levels of central CRF are implicated in anxiety and stress disorders in humans (for review, Reul and Holsboer, 2002; Holsboer and Ising, 2008) and central infusion of CRF in rodents stimulates anxiety-like behaviors (for review, Bale and Vale, 2004). The lateral portion of the CeA is the major extrahypothalamic site of CRF mRNA expression (Day et al., 1999; Asan et al., 2005). Neurons from the CeA send a strong projection to the anterolateral BnST (Sun and Cassell, 1993) where ~60 % of the CRF-immunolabeling in the BnST is associated with axonal profiles (Jaferi et al., 2009). CRF receptor 1 (CRF-R1) mRNA is expressed throughout the BnST, and CRF receptor 2 (CRF-R2) mRNA is expressed in the posterior, but not anterior, divisions of this nucleus (van Pett et al., 2000). CRF, acting primarily through CRFR1, augments the ASR, but not FPS, and the BnST is the critical site for CRF-dependent ASR enhancement (Swerdlow et al., 1986; 1989; Liang et al., 1992; Risbrough et al., 2003, 2004; Risbrough and Geyer, 2005; Walker et al., 2009a,b).
Few studies have assessed AAS effects on stress hormones or their receptors (Ahima and Harlan, 1992; Schlussman et al., 2000). The goal of the present study was to determine if chronic exposure of adolescent female mice to a mixture of commonly abused AAS would promote changes in either learned fear or generalized anxiety and whether such behavioral actions could be correlated with AAS-dependent alterations in the expression of CRF and its receptors in the CeA and/or BnST.
C57Bl/6 (Charles River Laboratory; Wilmington, MA, USA) or C57Bl/6J (Jackson Laboratories; Bar Harbor, ME, USA) mice were housed in groups of four in a temperature controlled room with 12 h light: 12 h dark and given free access to food and water. Care was taken to minimize the number of animals and their suffering and to utilize alternatives to in vivo techniques, if available. All procedures used were approved by the Dartmouth College Institutional Animal Care and Use Committee and were conducted in accordance with NIH guidelines. Human steroid abusers concurrently self-administer multiple AAS with different chemical and metabolic characteristics (Llewellyn, 2007). To mimic this human pattern of AAS “stacking”, adolescent female mice were injected intraperitoneally beginning at postnatal day (PN) 26 for 5 days per week for 4 weeks prior to experimentation with equal concentrations of a combination of three steroids that represent the major chemical groups of AAS: testosterone esters (testosterone cypionate; Sigma, St. Louis, MO, USA); 19-nortestosterone derivatives (nandrolone decanoate; Sigma) and 17α-alkylated testosterone derivatives (methandrostenolone; Steraloids, Newport, RI, USA) (for review, Clark and Henderson, 2003). AAS were administered at total concentrations of 7.5 of total AAS/kg/day for 6 days/week (37.5 mg/kg/week). This concentration in rodents was given to mimic the range of heavy use in human abusers (Pope and Katz, 1988; Perry et al., 1990; 2003; Kibble and Ross, 1987; Su et al., 1993; Clark and Barber, 2004; Blasberg et al., 2007). For example, a very high dose “12-Week Serious-Bulk (Pro Level) Stack” (Llewellyn, 2007) incorporates a schedule of 75 mg/kg/week for much of this cycle. In adult male mice, treatment with a mixture composed of comparable individual AAS to this “stack” (45 mg total AAS/kg/week) resulted in concentrations of testosterone in the forebrain that were 8× those observed in oil-injected control subjects (Penatti et al., 2009b). While we are assuming a simple conversion of mg/body mass between human subjects and mice, we recognize that species-specific variations in metabolism of each of these compounds may mean that alternative means of conversion (e.g., mg/body surface area; Regan-Shaw et al., 2007) may be superior once more data on the biochemical metabolism of each of these compounds in the mouse are more fully defined.
AAS were dissolved in sesame oil. Control, age-matched subjects were injected with same volume (~20–30 μl) of sesame oil alone. Behavioral testing commenced at ~PN54, and treatment with either AAS or oil was continued during testing. Body weight for each animal was recorded weekly prior to treatment and during testing. Estrous cycle was determined by daily vaginal lavage (Cooper et al., 1993). High doses of single AAS (Blasberg and Clark, 1997; for review, Clark et al., 2006) or an AAS mixture (Penatti et al., 2009a) disrupts vaginal cyclicity in rodents, and the cytology of AAS-treated animals resembles a diestrous profile (Penatti et al., 2009a). Consistent with these prior studies, cytological profiles from AAS-treated animals in the present study were consistent with diestrus. Body mass and body mass index (body mass/anal to nasal length2) were measured at the end of 4 weeks of injections and just prior to acclimation.
Behavioral testing was carried out using the MED-ASR-PRO1/ MED-ASR-FPS apparatus and software equipped with a single chamber ENV-264C□animal holder (Med Associates; St. Albans, VT, USA) as illustrated by the timeline in Figure 1. Mice were acclimated in the chamber without stimuli for 8 min on two different days, with the final day preceding the first day of testing. Prior to each behavioral test, animals were allowed to acclimate without stimuli for 5 min. All ASR stimuli consisted of 50 ms white noise (WN) burst with a 1 ms rise/fall unless otherwise specified. Stimuli intensities and durations for behavioral tests were based on previously published studies defining criteria applicable to C57Bl/6 mice at ~ 2 months of age (Willott et al., 1994; Willott and Carlson, 1995; Falls et al., 1997). Habituation to the startle stimuli was performed in two sessions on consecutive days with 60 trials of a 100 dB stimulus presented at a 10 s interstimulus interval (ISI). Prepulse Inhibition: startle was elicited by 30, 100-dB stimuli on a 30-s ISI. Half of the stimuli, randomly, were preceded by a 20-ms, 70-dB WN prepulse with 100 ms from prepulse offset to startle stimulus onset. The percent of prepulse inhibition for each mouse was calculated by the formula: ((mean of trials of no prepulse)− (mean of trials with prepulse))/(mean of trials with no prepulse) × 100. Fear Potentiated Startle: Testing: Startle responses were elicited by either a WN stimulus of 110 dB or a series of WN stimuli of 100, 105 and 110 dB. Stimuli were presented as the WN burst alone or this stimulus preceded by a 70 dB/30 s WN (10 ms rise/fall) in 23 trials in randomized order (60 s ISI). The first 3 trials were 110 dB WN alone and were not included in the analysis. Training: Training began the day after an initial FPS test. Each subject was exposed to ten 70 dB/30 s WN conditioned stimuli (CS) paired with a 0.5 mA footshock (250 or 500 ms; variable ISI of between 90 and 150 s, unconditioned stimulus) with the shock delivered at the end of the 30 s WN. The paradigm of training followed on the next day by testing was then repeated (Figure 1). The percent FPS was calculated as the ASR amplitude elicited by presentation of the WN burst with the CS minus the ASR elicited by the WN burst alone; divided by the amplitude elicited by the WN burst alone times 100: %FPS = [(WN burst + CS − WN burst alone)/ WN burst alone) × 100]. In one experiment, a separate cohort of naïve male and female mice (~PN55 at the time testing began) was assessed for habituation and FPS on a separate apparatus. In brief, mice were tested in a locally manufactured 8 × 5 × 6 cm holding cage fabricated from aluminum strips with a grid floor of 2.5-mm brass rods fixed to plastic insulating material. The cage was sandwiched between compression springs attached to a rigid superstructure, housed within a sound-attenuating chamber, and illuminated by a 7.5 W white incandescent bulb. Vertical displacement of the chamber moved an attached magnet within a fixed coil, inducing a voltage, which was amplified, digitized (1 kHz), rectified, and integrated by a microcomputer system. Startle amplitude was measured as the integrated voltage of the 200-ms epoch beginning at the onset of the startle stimulus. Three pre-test trials were run on consecutive days followed by training and testing, using the same parameters as described above. Oil-injected control female subjects used for behavioral testing were not restricted by estrous cycle stage.
The hanging wire test (Crawley, 2000) was performed as a simple assessment of grip strength. In brief, the mouse is placed on top of a wire cage lid; the lid is shaken 3× to prompt the mouse to grab the cage wires, and the lid is then inverted. The lid remains suspended ~20 cm above a padded surface, and the time until the mouse falls off is recorded. A maximum time of 300 s was allowed for the assay.
For CRF-immunoreactivity (CRF-ir) assays, oil-injected mice in diestrus and AAS-injected mice of the same age were euthanized by injection of a cocktail of ketamine (90mg/kg) and xylazine (10mg/kg), the heart was then injected with 0.1mL heparin (1000 U/mL; APP Pharmaceuiticals, Schaumburg, IL, USA), and animals were transcardially perfused with 0.1M phosphate buffer (PB) containing 0.9% (w/v) NaCl (PBS; pH = 7.2), followed by 4% (w/v) paraformaldehyde (Fisher, Fair Lawn, NJ, USA) in PB (PF; pH = 7.2). The brains were dissected and immersed in PF overnight at 4°C. The following day, 50 μm coronal brain sections were cut on an oscillating tissue slicer (Electron, Microscopy Sciences, Hatfield, PA, USA) and immunolabeled for CRF according to minor modifications of Asan et al. (2005): Free-floating sections were reacted with 10% H2O2 (Fisher) in PB containing 0.5% Tritron X-100 (PBT) (Acros, St. Louis, MO, USA) followed by a 90 min incubation in blocking buffer (5% normal goat serum in PBT; Colorado Serum, Denver, CO, USA) before reacting with primary antibody (rabbit anti-CRF antibody; Peninsula Labs, Belmont, CA; 1:4000, in PBT containing 1% normal goat serum) at 4°C for 48 hrs. Sections were then washed 3× for 15 min in PB followed by incubation in secondary antibody (goat anti-rabbit conjugated to Alexa-594;Invitrogen Corp, Carlsbad, CA; 1:500 in PBT) for 2 hrs. Labeled sections were mounted onto gelatin-coated slides and cover-slipped with Slow-Fade mounting medium (Invitrogen). Care was taken to limit photobleaching by incubating in secondary antibody inside foil-covered dishes.
Photomicrographs at 400× were taken bilaterally at two anteroposterior levels of each area using a Leica Confocal Laser Scanning Microscope from sections that included the dBnST and the ventral portion of the lateral aspect of the BnST (vBnST) (0.14 mm Bregma), the CeA and the PVN (−1.06 mm Bregma), and the ME (−1.70 mm Bregma) (Franklin and Paxinos, 1997). Images were analyzed using ImageJ (http://rsbweb.nih.gov/ij/) for the mean intensity above background. Somal-associated CRF-ir was subsequently determined according to minor modifications of an unbiased grid counting method (Oorscot, 1994; Casu et al., 2004; Petersen et al., 2005) in which a 9 × 9 grid pattern was overlaid on the image and the intensity of staining for somata beneath the intersections of grid lines (average of 5 –8 cells per image) was analyzed and averaged.
Gross brain dissections were made from oil-injected mice in diestrus and AAS-injected mice, as previously described (Penatti et al., 2009a,b; 2010). The dBnST was dissected from a 400 μm slice as a triangular-shaped region medial to the septum and dorsal to the anterior commissure (AC; −0.26 mm Bregma). The CeA was dissected from a 600 μm slice as the circular nucleus lying dorsal and lateral to the optic tract, and medial to the external capusle (−2.2mm Bregma). The PVN was dissected from the section between CeA and dBnST (400μm thick) and was defined as the area immediately lateral to the dorsal aspect of the third ventricle. Tissue was placed in lysis buffer and total RNA was extracted according to manufacturer's protocol for RNAqueous-Micro kit (Ambion Inc., Austin TX, USA). The concentration of the RNA was determined by measuring the optical density at 260 nm. Fifty ng of total RNA was reverse transcribed using RETROscript First-Strand Synthesis Kit for RT-PCR (Ambion) in a total reaction volume of 20 μl. RNA was denatured for 3 min at 75° C with 2 mM dNTPs and 5 μM random decamers. Ten units of RNase inhibitor, RT buffer to 1× and 100 units of M-MLV reverse transcriptase was added to this mixture, and the reaction was incubated for 1 hr at 42° C, followed by inactivation at 92° C for 10 min.
TaqMan Gene Expression Assays specific for mouse CRF, CRFR1, CRFR2 and 18S rRNA mRNAs were purchased from Applied Biosystems/Life Technologies (Carlsbad, CA, USA). Real time PCR was performed using an AB 7500 Sequence Detection System, and all cDNAs were analyzed in triplicate with the 18S rRNA as an internal standard. For each mRNA a PCR master mix was prepared, containing final concentrations of: 1× TaqMan Universal Master Mix (containing AmpliTaq Gold DNA Polymerase, AmpErase UNG, dNTPs with dUTP, Passive Reference 1, and optimized buffer components), 900 nM forward primer, 900 nM reverse primer, 250 nM probe in a total reaction volume of 25 μl. PCR for 18S rRNA was performed in a 25 μl reaction containing 1× TaqMan Universal Master Mix, 1× Eukaryotic 18S rRNA Endogenous Control (VIC/MGB Probe, Primer Limited; ABI), to which 1 μl cDNA was added. Thermocycling conditions included initial steps of 2 min at 50° C, 10 min at 95° C, and 40 cycles of PCR at 95° C for 15 s to denature cDNA, with 60° C for 1 min for primer/probe annealing and extension steps. Samples with reverse transcriptase omitted were used to control for genomic DNA contamination and reagent contamination was controlled for by omission of the primer template. Messenger RNA levels were determined as either 2−ΔCT or 2−ΔΔCT (Livak and Schmittgen, 2001; Peirson et al., 2003).
At the same time as decapitation for mRNA extraction, trunk blood was collected, allowed to clot for 90 min at room temperature, and the blood was then spun at 1000 ×g for 15 min. The serum was separated, frozen at −20°C and assayed for corticosterone in singlet via radioimmunoassay by the University of Virginia Ligand Assay Core Laboratory (http://www.healthsystem.virginia.edu/internet/crr/). Detection limits for the assay were 11.6 – 802.6 ng/ml.
Statistical significance was assessed by one-way analysis of variance (ANOVA) or repeated measures ANOVA using the general linear model procedure of SAS version 9.1.3 (Cary, NC, USA). Post-hoc comparisons were made using LS Means following one-way ANOVA and the Student's t test following repeated measures ANOVA. For real-time PCR analysis, outliers were removed when CT values were ± three standard deviations from the mean. Individual transcript mRNA levels were normalized to 18S rRNA and differences in the relative abundance of each mRNA for AAS-treated versus oil-injected subjects were assessed using Pair-Wise Fixed Reallocation Randomization t-test using the Excel-based Relative Expression Software Tool (REST). For all comparisons, the alpha level was set at p ≤ 0.05.
Habituation of the ASR to a 100 dB WN was assessed in eight separate cohorts of C57Bl/6 from Charles River and five cohorts of C57Bl/6J mice from Jackson Laboratories, each consisting of four mice injected with sesame oil and four mice injected with 7.5 mg/kg of the AAS mixture. The first 30 trials of habituation were analyzed as blocks of five trials. There was no effect of strain (C57Bl/6 versus C57Bl/6J), and results from all cohorts were combined. Treatment with AAS increased ASR amplitude on both days of testing (first day: F1,99 = 14.99, p < 0.0002; second day: F1,99= 14.24, p < 0.0003) and ASR amplitude decreased over trials on both days (F5,495= 18.89, p < 0.0001, and F5,495 = 19.34, p < 0.0001, respectively) (Figure 2), but there was no treatment × trial interaction on either day of habituation testing. Similar results were obtained when the data from both days were combined, indicating that AAS-treated subjects had greater startle amplitudes than did oil-injected subjects (F1,99= 16.49, p < 0.0001). There was no significant change in startle amplitude between days, and no treatment x day interaction.
ASR responses may be dependent on the mass of each animal or her strength. Body mass, body mass index and grip strength were assessed in 20 oil-injected and 20 AAS-treated C57Bl/6J female mice. AAS treatment did not significantly alter body mass (18.1 ± 0.3 g oil-injected vs. 18.5 ± 0.2 g AAS-treated), body mass index (0.32 ± 0.01 oil-injected vs. 0.33 ± 0.05 AAS-treated) or time on the hanging cage wire test (300 ± 0 oil-injected vs. 280.2 ± 11.5 AAS-treated). These results suggest that the enhanced ASR amplitude observed in AAS-treated subjects was not due to steroid-dependent increase in weight or strength.
To determine if the augmented ASR in AAS-treated animals might reflect drug-dependent changes in the saliency of the acoustic stimulus, experiments assessing PPI were performed in three cohorts of mice injected with oil alone or 7.5 mg/kg of the AAS mixture. For both oil-injected and AAS-treated mice, exposure to the 70 dB/20 ms WN prepulse diminished startle amplitudes when responses were compared to those elicited in the absence of the prepulse (F1,22 = 84.6, p < 0.0001), however, no significant differences were evident in the percent of PPI between oil-injected (60.9 ± 4.1%) and AAS-treated (57.3 ± 4.3%) animals. These data suggest that AAS exposure did not alter sensoriomotor gating at these startle and prepulse parameters.
Initial experiments to optimize FPS testing indicated that the percentage of FPS was greatest in AAS-treated animals when the startle stimulus was a 110 dB WN burst. Assays were therefore performed in four matched cohorts of C57Bl/6J mice treated with 7.5 mg/kg AAS or oil, and fear-potentiation of the ASR was determined following pairing of a 30 s/70 dB conditioned stimulus (CS) with a 250 ms/0.5 mA footshock. The amplitude of the ASR elicited by the 110 dB WN burst was greater in AAS-treated versus oil-injected animals (F1,30 = 22.71, p < 0.0001) (Figure 3A), as it was in animals tested for habituation to a 100 dB WN. Using this regime, however, we found a large variance in the potentiation to the CS observed prior to training, with some animals exhibiting an unexpectedly high degree of potentiation in the pretest trials. Moreover, with this paradigm, we observed a decrease in FPS following training (F2,60 = 5.43, p = 0.007; Figure 3B). This unexpected pattern of potentiation prior to training and a decrease following training was observed in both oil-injected and AAS-treated subjects, and results were not significantly different between the two treatment groups.
To further gauge if inherent characteristics of the C57Bl/6J mice or the sex of these mice might contribute to this pattern of response during FPS training/testing, we also assessed FPS in an additional cohort of naïve (uninjected) male and female C57Bl/6J mice tested on a separate apparatus (see Methods) at the same age that animals who had received injections were tested (~PN54). These naïve mice were subjected to three testing sessions prior to conditioning to footshock. On all three pretest days, mice exhibited greater startle on CS trials than on startle alone trials (F1,6 = 41.92, p <.001 for the mean of the three days). Significant potentiation to the CS in subsequent FPS testing was evident on each day (F1,6 = 13.78, p < .01 for the mean of the three test days), however, potentiation during FPS testing never exceeded the unconditioned potentiation during pretests. Finally, no significant sex-specific differences were evident in these naïve mice during either pre-test or FPS responses (data not shown).
To rule out that the variability in FPS responsiveness might arise from the presentation of a single intensity 110 dB startle stimulus and that this paradigm might, in turn, mask an effect of treatment, three additional cohorts were assayed for FPS using a testing protocol with three different startle intensities (100, 105, 110 dB WN). Testing with a range of startle stimuli eliminated the most of the potentiation to the CS prior to training (~15% potentiation on the pre-test with the range of intensities versus ~45% on the pretest using 110 db, as described above). However, as with exposure to the single intensity startle stimulus alone (110db), there was no significant fear-potentiation of the startle response or a significant effect of AAS treatment on the FPS elicited in response to this combination of startle intensities (data not shown). We also assessed FPS responses for C57Bl/6 versus C57Bl/6J mice, a longer duration of footshock (250 versus 500 ms), and a lower concentration of AAS (5.0mg/kg versus 7.5mg/kg) to ascertain if any of these parameters might contribute to the variability in pre-test startle or unmask an effect of AAS treatment on startle. None of these variables was found to significantly influence the results. In all cases, even though AAS-treated animals consistently demonstrated higher startle amplitudes in FPS testing as in habituation testing, FPS was not reliably elicited. While these data suggest that adolescent C57Bl/6 or 6J mice may not be an optimal strain for FPS with WN startle stimuli, importantly, under no conditions was there a significant effect of AAS treatment on FPS under any of these permutations.
Steady state mRNA levels corresponding to CRF, CRF-R1 and CRF-R2 were assessed in the CeA and dBnST from seven C57Bl/6J diestrous oil-injected mice and nine C57Bl/6J mice treated with 7.5 mg/kg AAS for four weeks. Oil-injected subjects were euthanized in diestrus since vaginal epithelial cell profiles of AAS-treated animals are most closely matched to this stage (Penatti et al., 2009a). Levels of CRF mRNA were greater in the CeA (p < 0.05), but not in the dBnST, of AAS-treated versus oil-injected mice (Figure 4A). While CRF-R1 mRNA was evident in both regions, there was no significant effect of treatment on the levels of CRF-R1 mRNA in either region (Figure 4B). CRF-R2 mRNA was detected in only ~33% of subjects in either the CeA or the dBnST, expression in those samples was at very low levels (CT = 37), and AAS treatment did not significantly alter expression of this transcript. The levels of CRF, CRFR1, and CRF-R2 mRNAs in the PVN were also examined in a smaller cohort of mice (n = 3 mice per treatment condition). AAS treatment did not significantly alter levels of any of these transcripts in this brain region (data not shown).
CRF-ir was assessed in eight oil-injected (diestrous) and eight AAS-treated (7.5 mg/kg) C57Bl/6J mice. The mean intensity for CRF-ir was higher in AAS-treated versus oil-injected mice in the dBnST (F1,16 = 6.831, p = 0.019 (Figure 5A and D), but not in the CeA (Figure 5A and D), the vBnST (Figure 5A and D), the PVN (Figure 5B) or the ME (Figure 5B). While the overall intensity of CRF-ir was not higher in the CeA in AAS-treated versus oil-injected subjects, there was a redistribution of label such that the association of CRF-ir with cell somata in the CeA was greater in AAS-treated subjects (F 1,16= 4.917, p = 0.041) (Figure 5C and D). In contrast, while the overall CRF-ir intensity was higher in the dBnST of AAS-treated subjects, this increase was not preferentially associated with cell somata, but diffusely distributed throughout the neuropil (Figure 5C and D). As with overall mean intensity, no change in the distribution of somal-associated CRF-ir was evident in the vBnST with AAS treatment (Figure 5C).
The absence of an effect of AAS treatment on CRF-ir within the PVN or the ME suggested that chronic treatment with AAS might not have had a significant effect on activation of the hypothalamic-pituitary-adrenal (HPA) axis. Radioimmunoassays of peripheral levels of corticosterone from the mice from which real time PCR assays were made (n = 7 oil-injected and n = 6 AAS-treated C57Bl/6J females) supported this assertion. Serum corticosterone concentrations were comparable to those reported previously for both peri-adolescent and adult female mice (Laviola et al., 2002) and not significantly different in the two groups (72.3 ± 23.3 ng/ml for oil-injected diestrous vs. 91.4 ± 43.3 ng/ml for AAS-treated).
Greater than one hundred AAS have been identified to date (Llewellyn, 2007). The array of physiological and behavioral effects of these chemically disparate drugs is vastly compounded not only by complexity in the patterns of self-administration (Llewellyn, 2007), but also by the heterogeneity of the subjects that administer them (Graham et al, 2008). In particular, as predominant use of AAS has spread from adult, male professional athletes to encompass recreational, female and adolescent users (Kerr and Congeni, 2007), it is becoming increasingly apparent that information garnered from studies of adult males cannot be generalized across sex and age (for review, Sato et al., 2008; Clark et al., 2006). Here, we have assessed the effects of chronic exposure of adolescent female mice to a mixture of three chemically distinct and commonly abused AAS. Our data indicate that exposure of female mice to a high doses of a mixture of AAS during adolescents resulted in increased anxiety-like behavior; a result that correlated with enhanced CRF expression in the extended amygdala. AAS treatment did not alter the levels of peripheral corticosterone, a result that contrasts with a recent study demonstrating that a 4-week treatment of adult male rats with either nandrolone or stanozolol (5 mg/kg/day) resulted in elevated levels of plasma corticosterone during the “morning trough” (Matrisciano et al., 2010). The differences between these two studies may reflect age-, sex- or drug-specific effects (see Clark et al., 2006) or differences in the time of day for blood collection since our animals were killed in the afternoon. We note that CRF-R2 mRNA detection was negligible in the dBnST and the CeA; results consistent with prior studies in the mouse indicating a lack of detectable CRF-R2 mRNA in the CeA itself or in the anterior portions of the BnST that receive CeA projections (van Pett et al., 2000) and with data indicating that CRF-R1 is the predominant receptor mediating anxiogenic actions of CRF in the CNS (for review, Holsboer and Ising, 2008).
In human subjects, the negative psychiatric side effects of chronic AAS abuse have been extensively documented for decades in both case reports and in controlled observational studies (Pope and Katz, 1988; Bahrke et al., 1996; Franke and Berendonk, 1997; Yesalis et al., 1997; Gruber and Pope, 2000; Kuhn, 2002; Pagonis et al., 2006). Symptoms of chronic use of supratherapeutic doses of AAS include mania, hypomania, somatization, increased anxiety, irritability, extreme mood swings, abnormal levels of aggression, depression, acute paranoia and even suicide. In particular, those administering the highest doses of AAS have been shown to have elevated scores on the Symptom Check List-90, a self-report system that includes a number of different dimensions of anxiety (Pagonis et al., 2006).
With respect to anxiety-like behaviors in animal models, previous reports examining the effects of AAS administration in rodents have yielded inconsistent and often contradictory results that are likely to reflect the variety in species, sex, and age of the subjects, as well as in environmental variables and the types and regimes of AAS administered. In adult males rodents, repeated exposure to a single AAS has been reported to be anxiolytic (Ågren et al., 1999; Bitran et al, 1993; Barreto-Estrada et al, 2004; Kouvelas et al., 2008; Steensland et al., 2005), anxiogenic (Minkin et al. 1993; Rocha et al., 2007; Ambar and Chiavegatto, 2009), or to have no effect (Rojas-Ortiz et al., 2006). Few studies have been performed in female rodents. Barreto-Estrada et al. (2004) report that treatment of adult female C57Bl/6J-NHsd mice with a high dose (7.5 mg/kg/day) of 17α-methyltestosterone had no effect on exploratory anxiety, as assessed in the elevated plus maze, but did product anti-conflict effects as assessed by a modified Vogel conflict test. In contrast, Pinna and colleagues have reported that a chronic, but non-anabolic, dose of testosterone propionate (~0.5 mg/kg/day for 3 weeks) enhanced contextual fear in adult Swiss-Webster female mice, but only in those animals that had been reared in social isolation prior to testing, not those that had been group-housed (Pibiri et al., 2006; 2008; Agis-Balboa et al., 2009; for review, Pinna et al., 2008).
The variability reflected in these studies may not only reflect variables mentioned previously, but also differences in the behavioral tests used to assess anxiety. In particular, many of the classical measures of anxiety-like behaviors depend on locomotor activity that may complicate interpretation of effects on anxiety. The ASR, in contrast, is relatively independent of locomotory paradigms and both the reflex and its underlying neuroanatomical substrates are generalizable across all mammals (for review Koch, 1999). Our data demonstrating that AAS treatment enhanced both the amplitude of the ASR and CRF peptide expression in the BnST, but was without effect on FPS, are consistent with a large body of previous work indicating that CRF augments the ASR, that the BnST is the critical site for CRF-dependent ASR enhancement, and that CRF-dependent actions do not contribute to FPS (Liang et al., 1992; Lee and Davis, 1997; Risbrough et al., 2003, 2004; Risbrough and Geyer, 2005; Walker et al., 2009a,b). Similarly, our data indicating that the AAS-induced increase in CRF mRNA in the CeA and a corresponding increase in CRF peptide expression in the dBnST are consistent not only with the behavioral effects on the ASR, but also with previous studies demonstrating that the lateral portion of the CeA is the major extrahypothalamic site of CRF mRNA expression (Day et al., 1999; Asan et al., 2005), that this region provides a major afferent input to the dorsolateral BnST (Sun and Cassell, 1993; Jaferi et al., 2009) and that CRF is released into the BnST from CeA projection neurons during periods of stress or anxiety (Sakanaka et al., 1986).
In sum, our data suggest that AAS treatment may augment CRF expression in the key projection from the CeA to the dBnST and that this increase may underlie the AAS-dependent enhancement of the ASR. We recognize that many variables, including the general level of anxiety, may influence the amplitude of the ASR, however, we also note that the connection between heightened ASR and anxiety has been established in many different paradigms. Specifically, the ASR is increased by anxiogenic drugs and decreased by anxiolytic drugs; increased by tonic stimulation of anxiety/fear related brain areas and increased in human anxiety disorders (for review, Koch, 1999). In addition, rats characterized as “anxious” by their tendency to freeze to an ASR-inducing stimulus show higher ASR amplitudes at the very first stimulus presentation (Plappert et al., 1993) as do rodents with increased anxiety induced by exposure to a predator (for review, Adamec et al. 2001). As we observed no significant effect of AAS treatment in the current study on the weight or strength of the mice, our results indicating AAS increase startle amplitude are most consistent with the interpretation that treatment augments anxiety in these subjects. Moreover, our findings indicating a parallel increase in CRF expression also suggest a neural mechanism that may contribute to a state of enhanced generalized anxiety in young females subjected to chronic AAS exposure.
In contrast to the consistent and significant enhancement by the AAS on the amplitude of the ASR, treatment with these steroids did not alter the stimulus processing associated with habituation, the perceived saliency or sensorimotor gating of acoustic stimuli as assessed by PPI (for review, Geyer et al., 2002; Willott et al., 2003) nor did it promote an enhancement of learned or conditioned fear as assessed by FPS. We qualify our conclusions with respect to the actions of AAS on sensorimotor gating by the fact that our assays were limited to PPI analysis with a single set of stimulus parameters. We qualify our conclusions with respect to AAS effects on FPS by noting that in our hands, C57Bl/6 mice could display high levels of pre-training potentiation to the CS and that FPS was not reliably generated in these mice, irrespective of dose of AAS, sub-strain of C57Bl mice, sex, or stimulus intensity. Despite the variability in FPS itself, no effect of AAS treatment on FPS was evident under any conditions. Thus, in adolescent female mice, this regime of AAS exposure appears to augment a generalized anxiety state, but not promote alterations in the perception of specific threats. It will be of interest to determine if comparable actions of these AAS are observed with adolescent male mice since inherent sex-specific differences in anxiety, impulsivity and aggression may predispose males to AAS-dependent modulation of PPI or FPS that were not observed in females.
While AAS-dependent increases in CRF mRNA and CRF peptide expression were observed in the CeA and dBnST, respectively, no effects of steroid exposure were observed in CRF-ir in the PVN or ME; areas corresponding to the primary site for synthesis and secretion, respectively, of CRF and thus key control regions of the HPA (for review, Herman et al., 2005). The lack of effect of AAS treatment on CRF-ir in the PVN and ME is consistent with our data also indicating an absence of steroid-induced changes in the levels of serum corticosterone. Together, these results support the conclusion that the predominant actions of the AAS are on central forebrain circuits involved in the expression of anxious states and stress-related behaviors.
The neural circuitry responsible for the expression of anxious states includes structures of the extended amygdala, including the dBnST and the CeA (for review, Davis and Whalen, 2001; Herman et al., 2005; Walker et al, 2009a). Both structures have a rich array of expression of neurotransmitters and neuropeptides that reflects the complexity of afferent integration and multisystem output inherent in generating homeostatic stress responses. While the interplay of these chemical mediators is likely to be intricate, neurotransmission mediated by γ-aminobutyric acid type A (GABAA) receptors is a particularly attractive candidate for being both a target for AAS modulation in the forebrain and a point of intersection with respect to CRF actions in the CNS (e.g., see Herman et al., 2002). Both the CeA and the BnST are densely populated with neurons containing GABA and CRF (Owens and Nemeroff, 1991; Veinante et al., 1997; Koob and Heinrichs, 1999; Asan et al., 2005). GABAergic transmission in the rodent forebrain is significantly altered by chronic exposure to the AAS in both male and female mice by mechanisms that involve classical androgen receptor signaling, indirect effects on classical estrogen receptor signaling by effects on aromatase, and direct allosteric modulation of GABAA receptors themselves (Penatti et al., 2009a,b; for review, Henderson, 2007; Pinna et al., 2007). CRF has been shown to augment GABAergic transmission in both the CeA and the BnST via CRF-R1-mediated signaling (Nie et al., 2004, 2009; Kash and Winder, 2006; Bajo et al., 2008; Roberto et al., 2010), raising the possibility that CRF may act as the mediator of AAS-dependent changes in forebrain GABAergic transmission. This possibility is particularly intriguing since CRF-dependent enhancement of GABA release has been postulated to be a key mechanism by which ethanol exerts its anxiogenic actions (for review, Koob, 2009; Nie et al., 2009; Silberman et al., 2009) and comorbidity of alcohol and AAS abuse has repeatedly been reported in adolescents, including adolescent females (DuRant et al., 1993; Kindlundh et al., 1999; Nilsson et al., 2001; Bahrke et al., 2000; Miller et al., 2005; Pallesen et al., 2006; Elliot et al., 2007). Thus, while increases in CRF expression in the extended amygdala may alter signaling mediated by a number of other neurotransmitter systems, as well as GABA (for review, Silberman et al., 2009), establishing that the AAS and ethanol may converge in regulating GABAergic transmission in these brain regions via their effects on CRF would provide an important and exciting advance in understanding not only the central actions of the AAS, but also the increased rate of concurrent use of alcohol in adolescent AAS abusers.
Supported by the NIH; DA022716 to LPH and BAC, DA14137 to LPH and training grant DK07508 for BAC and JGO. The NIH had no further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. The University of Virginia Ligand Assay Core Laboratory is supported by SCCPIR U54 HD28934 from the NIH.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.