Stress-induced elevation of glucocorticoids is accompanied by structural changes and neuronal damage in certain brain areas. This includes reduced expression of brain-derived neurotrophic factor (BDNF) in the hippocampus which can be prevented by chronic electroconvulsive seizures and antidepressant drug treatment. In the last years we have bred two strains of rats, one which reacts with congenital helplessness to stress (cLH), and one which congenitally does not acquire helplessness when stressed (cNLH). After being selectively bred for more than 40 generations these strains have lost their behavioural plasticity including their sensitivity to antidepressant treatment. We show here that in cLH rats, acute immobilization stress does not induce a reduction of BDNF expression in the hippocampus which is observed in Sprague–Dawley and cNLH rats. All animals tested exhibited elevated corticosterone levels when stressed, an indication, that in cLH rats regulation of BDNF expression in the hippocampal formation is uncoupled from corticosterone increase induced through stress. This may explain the lack of adaptive responses in this strain.
Beside its classical function as a neurotrophin during development, BDNF (brain-derived neurotrophic factor) modulates synaptic plasticity1 and has been suggested to be involved in stress-induced hippocampal adaptation and pathogenesis of depression2 in the adult animal. This hypothesis is supported by both the fact that exogenous BDNF has an antidepressant effect,3 and that antidepressant treatment ameliorates stress-induced reduction of BDNF mRNA in the hippocampus.4 We have examined BDNF expression in rats with learned helplessness. Learned helplessness is a model for depression based on the hypothesis that depression is induced by uncontrollable stress in individuals with a predisposition. To enhance genetic predisposition Sprague–Dawley rats were selected for breeding based on their behavior in learned helplessness testing. Two strains evolved: rats with congenital helplessness (cLH) and rats with resistance to stress-induced helplessness (cNLH). cLH rats in particular show a markedly reduced behavioral variability to environmental stimuli: they have a deficit in learning under stressful conditions and furthermore do not respond to treatment with antidepressants.5 We were interested in molecular mechanisms underlying this deficit to adapt to stressful stimuli and, therefore, examined the stress-induced BDNF mRNA reduction of cLH rats with in situ hybridisation and compared it to the stress response of cNLH rats and outbred Sprague–Dawley rats as controls.
BDNF mRNA was most highly expressed in hippocampal region CA3 and dentate gyrus (DG); lower expression was found in frontal, parietal and pyriform cortex and amygdala (see Image section, p 358) as reported previously.6 There were no differences of the BDNF mRNA baseline levels between strains. This finding is consistent with the results in Flinders Sensitive Line, a different genetic animal model for depression where baseline expression of BDNF and NGF were similar in the hippocampal formation of Flinders Sensitive Line and their controls Flinders Resistant Line.7 This does not argue against a role of the neurotrophins in the pathogenesis of depression, since neurotrophins in the adult animal are known for their prominent role in plasticity8 and constitutive levels may be unchanged while differences become visible only after stress challenges. In Sprague–Dawley rats acute immobilisation stress led to a significant BDNF mRNA reduction of 26% in DG and to a lesser degree in CA3, confirming previously reported results.4,9 cNLH rats, whose learned helplessness response resembles the parental Sprague–Dawley strain, showed comparable BDNF mRNA reductions after stress: 23% (DG) and 19% (CA3). In contrast, in the cLH strain BDNF mRNA levels were not reduced after immobilisation stress (a slight increase of BDNF mRNA after stress was not significant) (Figure 1).
BDNF mRNA reduction after stress is dependent on corticosterone, which is highly elevated after immobilisation.9,10,11 We therefore determined corticosterone levels to find out if the missing BDNF response in cLH rats was due to a lack of corticosterone response after stress. As previously reported there were no statistically significant differences in baseline cortisol levels: (mean ± SD) 13 ± 14 μg dl−1 (n = 6) in cLH, 18 ± 25 μg dl−1 (n = 6) in cNLH, 57 ± 56 μg dl−1 (n = 4) in SD. After 45 min immobilisation, corticosterone was elevated more than 10-fold in all strains: 225 ± 59 μg dl−1 (n = 5) in LH, 261 ± 71 μl dl−1 (n = 5) in cNLH, 248 ± 130 μg dl−1 (n = 4) in SD (Figure 2). Thus, the absence of a decrease of BDNF mRNA in LH rats was not due to a smaller corticosterone elevation after immobilisation, but rather due to a lack of effect of corticosterone on BDNF mRNA regulation. As high concentrations of corticosterone are required to reduce BDNF mRNA, the effect is thought to be mediated by the glucocorticoid receptor (GR).12 Probably, there is no direct action of GR on BDNF mRNA transcription, since there is no glucocorticoid regulated element in the promotor region of BDNF. An indirect mechanism via transcription factor AP-1 or modulation of activity is assumed.12 Our findings suggest that the molecular mechanism of reduced plasticity in cLH rats may be a defective function of the GR which has been implicated in the pathogenesis of depression13 or a change in the signal transduction cascade modulated by the activated GR resulting in an altered BDNF mRNA transcription. Both possibilities are in agreement with previous findings in the cLH strain reporting an impaired stress response after early stress.14,15
A breeding program was established (Department of Psychiatry, SUNY Stony Brook, NY, USA) with Sprague–Dawley rats being selected on the basis of their susceptibility to develop learned helplessness. In the 29th generation two strains emerged: cLH, which exhibit helpless behavior (>95% showed escape deficit) without preshock, and cNLH, almost completely acquiring stress control (<5% demonstrated escape deficit). Our colony was derived from founder animals of the 35th Stony Brook generation, animals from the 8th, 9th and 10th subsequent generation were used for the experiments.
Animal testing and stress
Animals were housed in plastic cages 38 × 20 × 59 cm, four animals per cage in a room maintained at a constant temperature of 22°C and 12-h light-dark circle. Food and water were provided ad libitum. Animals were treated in accordance with the European Communities Council Directive, and the procedures were approved by Regierungspraesidium Karlsruhe. Experiments were performed in operant conditioning chambers with inside dimensions of 48.5 × 30 × 21.5 cm. The floor was constructed of steel rods 6 mm in diameter and 20 mm spaced apart. On one side of the boxes there was a lever of 35 × 35 mm only on testing days and a 12 W light bulb. Shocks of 0.8 mA intensity were randomly delivered to the grid floor and the walls by a shock generator. Shock was pulsating in intensity with a phase-duration of 200 μs. The experiment was controlled by a IBM compatible 4/86 computer. Boxes, shock generator and controlling program were obtained from TSE, Bad Homburg, Germany. Experiments were performed between 8 am and 12 am. Pretest day: cNLH animals received inescapable shocks (preshocks) of 0.8 mA for a total shock duration of 20 min in a 40-min session. The duration of single shocks and intershock times ranged from 5 to 15 s and was randomized by the computer. Testing: 24 h after preshock (cNLH) or without preshock (cLH), animals were tested individually. Testing consisted of 15 trials of 0.8 mA current lasting 60 s each, if the animal did not stop the current by pressing the lever.
Current was accompanied by a light clue. Intertrial time was 24 s. A trial not stopped or later than at 20 s was considered a failure. cNLH animals had to have less than five failures, cLH animals had to have more than 10 failures to be selected for experiments. Eighteen animals from each strain were selected. Two weeks after testing, five animals from cNLH and cLH respectively were immobilized in restraining tubes for 45 min. Fourteen outbred Sprague–Dawley rats purchased from Janvier, France were used as controls, four of them were immobilized. Animals were killed by decapitation immediately after stress or without stress at 8 am.
Plasma samples were stored at −20°C until assayed. Corticosterone was measured with specific radioimmunoassay after extraction according to the method published elsewhere.16,17 Briefly, 10–20 μl plasma was supplemented to 100 μl with 5% aqueous ethanol, tritium-labelled corticosterone (for the determination of procedural loss) was added, and the mixture was extracted with 500 μl of cyclohexane/dichloromethane (2:1, v/v). Organic extract was separated, evaporated to dryness, taken up in 1–2 ml of 5% aqueous ethanol and quantified with specific RIA. Intraassay variation was 12.4%, interassay variation was 14.3%. Each result was corrected for individually determined procedural loss.
In situ hybridisation
Brains were removed and frozen for in situ hybridization. Coronal sections of 20 μm (Bregma −3.30 mm) were cut on a cryostat, thaw mounted on coated slides (Superfrost Plus, Menzel Glaeser), fixed and dried. The BDNF cDNA clone, kindly provided by C Suter-Crazzolara, Heidelberg was linearized with HindIII (sense) and EcoRI (antisense) and uniformly labelled riboprobes corresponding to the antisense strand of 463 bp were generated with T7 (sense) and T3 (antisense) polymerase and 35S-CTP (NEN). The 35S-labelled riboprobes were hybridized with the brain sections (106 cpm per section) for 18 h at 55°C in hybridization buffer (50% formamide, 20 mM Tris pH 7.5, 1× Denhardt's solution, 1.25 mM EDTA pH 8.0, 100 mM DTT, 10% dextran sulfate, 2 × SSC, 0.1% SDS, 1 mg ml−1 yeast RNA). The sections were washed in 2 × SSC (SSC: 150 mM NaCl, 15 mM sodium citrate at pH 7.0) at 25°C and then treated with 10 μg ml−1 RNase A for 30 min in 1.5 × SSC. The sections were then washed twice in 2 × SSC and three times in 0.2 × SSC, dried and exposed to Biomax MR (Kodak). The specificity of our hybridization protocol was confirmed by demonstrating that 35S-labelled sense BDNF riboprobes did not show any significant hybridization signals (not shown).
Autoradiograms were quantified using the Analytical Imaging Station with the software AiS (InterFocus, Suffolk, UK). Dentate gyrus granule cell layer and CA3 pyramidal cell layer were outlined using a density slice function highlighting that area from neighboring tissue with less hybridisation signal. A series of 14C standards was used to correct the optical density for nonlinearity of the film. Additionally, the brain sections were exposed for 15 h each to a microimager (β-Vision, Biospace Instruments, Paris, France) to obtain the intensity (counts mm−2). The unspecific intensity of each slice was determined in the left thalamus, where no specific signal was visible, and subtracted from the specific intensity. Specific signals were corrected for variations of thickness between slices by dividing through the slice's individual unspecific intensity and multiplying with the mean unspecific intensity of all slices.
Baseline BDNF mRNA levels were subjected to one factorial analysis of variance. Stress effects within strains were compared using independent two-sided Student's t-tests with a significance level of P < 0.05. For comparison of stress effects between strains, contrasts of the interaction were determined for each strain and tested pairwise for significance with Student's t-tests.
FAH is supported by DFG grant HE 2857/2–2. We are grateful to Dr C Suter-Crazzolara for supplying BDNF cDNA and in situ protocol. We also wish to thank Dr R Spanagel for critical revision of the manuscript, Dr B Krumm for help with statistical analysis, S Keck for expert technical assistance and Dr N Gretz and Dr B Kränzlin for providing animal care facilities and valuable advice on breeding.