Prenatal stress may increase vulnerability to life events: Comparison with the effects of prenatal dexamethasone

Developmental Brain Research 159 (2005) 55 – 63 www.elsevier.com/locate/devbrainres

Research Report

Prenatal stress may increase vulnerability to life events: Comparison with the effects of prenatal dexamethasone Karin S. Hougaard a,*, Maibritt B. Andersen b, Sanna L. Kjær a, ˚ se M. Hansen a, Thomas Werge b, Søren P. Lund a A a

National Institute of Occupational Health, Lersø Parkalle´ 105, DK-2100 Copenhagen, Denmark b Sct. Hans Hospital, DK-4000 Roskilde, Denmark Accepted 27 June 2005 Available online 8 August 2005

Abstract Prenatal stress has been associated with a variety of alterations in the offspring. The presented observations suggest that rather than causing changes in the offspring per se, prenatal stress may increase the organism’s vulnerability to aversive life events. Offspring of rat dams stressed gestationally by chronic mild stress (CMS, a variable schedule of different stressors) or dexamethasone (DEX, a synthetic glucocorticoid, i.e., a pharmacological stressor) was tested for reactivity by testing their acoustic startle response (ASR). Two subsets of offspring were tested. One was experimentally naı¨ve at the time of ASR testing, whereas the other had been through blood sampling for assessment of the hormonal stress response to restraint, 3 months previously. Both prenatal CMS and dexamethasone increased ASR in the offspring compared to controls, but only in prenatally stressed offspring that had been blood sampled 3 months previously. In conclusion, similarity of the effects of maternal gestational exposure to a regular stress schedule and of exposure to a synthetic glucocorticoid suggests that maternal glucocorticoids may be a determining factor for changes in the regulatory mechanisms of the acoustic startle response. Further, a single aversive life event showed capable of changing the reactivity of prenatally stressed offspring, whereas offspring of dams going through a less stressful gestation was largely unaffected by this event. This suggests that circumstances dating back to the very beginning of life affect the individual’s sensitivity towards experiences in life after birth. The prenatal environment may thus form part of the explanation of the considerable individual variation in the development of psychopathology. D 2005 Elsevier B.V. All rights reserved. Theme: Development and regeneration Topic: Nutritional and prenatal factors, hormones and development Keywords: Acoustic startle response; Chronic mild stress; Corticosterone; Dexamethasone; Fear potentiation; Prenatal stress

1. Introduction Maternal stress during gestation may lead to a variety of behavioral, neuroendocrine, and neuroanatomical alterations in the offspring. Hence, cognitive function, social behavior, and levels and distribution of regulatory neurotransmitters Abbreviations: ASR, acoustic startle reaction; CMS, chronic mild stress; DEX, dexamethasone; PPI, prepulse inhibition * Corresponding author. Fax: +45 39 16 52 01. E-mail address: [email protected] (K.S. Hougaard). URL: www.ami.dk. (K.S. Hougaard). 0165-3806/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.devbrainres.2005.06.014

have shown susceptible to maternal stress during fetal life [37]. In the absence of a direct neural connection between the developing fetus and the dam, maternal hormones have been hypothesized to mediate the effects of prenatal stress, particularly through alterations in the maternal hypothalamic – pituitary– adrenal (HPA) axis. The trophic hormones of the HPA axis, i.e., adrenocorticotropic hormone and corticotropin releasing factor (CRF), do not traverse the placenta in the rat [5,42]. In contrast, the adrenal glucocorticoids passes from the maternal to the fetal compartment [48]. The role of glucocorticoids as mediators of prenatal stress effects has been investigated by administering

56

K.S. Hougaard et al. / Developmental Brain Research 159 (2005) 55 – 63

corticosterone (or synthetic analogues, e.g., dexamethasone) to pregnant dams and observing the offspring. Results from these studies suggest that fetal glucocorticoid exposure poses part of the mechanism by which prenatal stress programs the offspring (reviewed in [22,38]). Only few studies have investigated the consequences of prenatal stress for postnatal development in humans, and most are limited by the use of retrospective designs, small sample sizes, or lack of control for confounders. Further, the results from the few well-designed human studies are concordant, i.e., decreased adaptation to novelty, altered attention, and increased emotionality [15]. Although the range of behavioral abnormalities is much more limited in animals than in humans, animal models allow for control of environmental factors and hypothesis testing based on manipulation of not only the prenatal but also the postnatal environment [37]. Animal models are therefore of great value in order to identify which behavioral domains or physiological systems that are particularly vulnerable to prenatal stress, to investigate whether individual sensitivity plays a role and which physiological mechanisms that mediate the effects of prenatal stress [15]. Findings in prenatally stressed animals and humans show many similarities. Thus, changes associated with the Fprenatal stress syndrome_ point towards increased fearfulness in novel situations as well as reduced abilities to cope with stress [37,38]. The adult offspring of rat dams subjected to stressors during gestation displays an increase in anxietyrelated behaviors, e.g., suppressed exploration of the open areas of the elevated plus maze [27,49] and increased defensive withdrawal [35]. An important physiological characteristic of prenatally stressed offspring is a hyperactive HPA axis, most often observed as an enhanced and/or prolonged plasma corticosterone response to stressors [36]. Also neural systems related to CRF are susceptible to stress in fetal life. In the amygdala, there are reports of increased CRF concentration, release, and receptor density after prenatal stress [2,35,40]. Gestational stress has been observed to facilitate the development of CRF-containing neurons [6,29], and in prenatally stressed animals, CRF receptor antagonists normalize fearful behavior [35]. In a previous study, we found that prenatally stressed offspring displayed a persistently enhanced acoustic startle response (ASR, a characteristic sequential contraction of the skeletal musculature evoked by a sudden and intense acoustic stimulus) [14]. Specifically CRF has been shown to intensify the acoustic startle response [20,28,32]. The apparent up regulation of CRFergic systems in prenatally stressed animals therefore poses an attractive means of explaining our observation of enhanced ASR in prenatally stressed animals [14,41]. The ASR may also be enhanced by aversive expectation about potential dangers. In context aversive conditioning a repulsive event is linked to the experimental context of this event. Subsequent exposure of the animal to this or resembling contexts will trigger anxiety. In the case of

ASR, this will be displayed as an enhanced startle response, i.e., fear potentiated startle [9]. Prenatal stress has been shown to enhance conditioned fear and increase the sensitivity to environmental perturbations throughout life [8,30]. In our previous study, the increased ASR was observed in prenatally stressed animals that had been blood sampled for investigation of their hormonal stress response. This procedure involved 20 min of restraint and repeated blood sampling (described in [14]). Restraint is aversive to rats [7], and part of the experimental procedure of ASR measurement bears resemblance to aversive elements of the blood sampling procedure, i.e., confinement in a test tube. We therefore hypothesized that the aversiveness associated with blood sampling lead to the pronounced ASR in prenatally stressed animals, due to enhanced fear potentiated startle. To evaluate whether prenatal stress effects on ASR occurred as a consequence of fetal glucocorticoid exposure, the ASR was investigated in prenatally stressed offspring as well as offspring of dams injected gestationally with a potent synthetic glucocorticoid, dexamethasone.

2. Materials and methods 2.1. Animals 66 time-mated young adult nulliparous female Wistar rats (HanTac:WH, Taconic M&B, Denmark) were randomly distributed to white plastic cages (27  43  15 cm) with pine-bedding (Lignocel S8) in pairs, upon arrival at gestation day (GD) 3. Environmental conditions were automatically controlled with a 12-h light– dark cycle. Food (Altromin Standard Diet 1324) and tap water were provided ad libitum. Clean cages and new bedding were provided twice weekly. The animal welfare committee, appointed by the Danish Ministry of Justice, granted ethical permission for the studies. All procedures were carried out in compliance with the EC Directive 86/609/EEC and with the Danish law regulating experiments on animals. 2.2. Exposure The day after arrival (GD 4), the animals were weighed and assigned to three groups, control, dexamethasone (DEX), and chronic mild stress (CMS). Body weights were recorded at GD 4, 7, 10, 13, 15, 17, 19, and 21. Dexamethasone (100 Ag/kg per day, Sigma-Aldrich, Denmark) dissolved in 4% ethanol –0.9% saline (100 Ag/mL) was injected s.c. in the skin of the neck during the last week of gestation (GD 14– 21). Control and CMS animals received vehicle injections during the same period. The CMS model is a schedule of chronic stress, where various relatively mild stressors are presented in a random schedule, e.g., change of partner and isolation housing. The model was developed as an alternative to conventional animal models of chronic stress that often include physical and potentially painful stressors [31]. For

K.S. Hougaard et al. / Developmental Brain Research 159 (2005) 55 – 63

example, a much used chronic stress protocol for pregnant rats involves restraint thrice daily for 30 min (e.g., [19]), whereas others have used electric shock (e.g., 80 shocks of 0.5 mA, 5-s duration, every other day [34]). The CMS model was developed by Willner, based on the chronic varied procedure by Katz, who presented rats with several severe stressors (e.g., electric shock, tail pinch, and shaking) in random sequence [16]. The model of CMS adheres to Katz’ principle of variability and unpredictability, but the severe stressors have been exchanged for milder stressors, hence the name chronic mild stress [44]. As the variability and unpredictability maintains the stress-inducing properties of the model, CMS may be considered conceptually closer to the everyday stress encountered by humans in, e.g., occupational settings [10,24,46]. The specific schedule of CMS is shown in Table 1 and was based on the principles underlying Willner’s model of CMS [24,45] and studies performed in our laboratory [11,13], with duration of food and water deprivation kept short and the animals housed in pairs [3,24]. Chronic stress may take some time to build [4], exposure to CMS was therefore initiated earlier in gestation than exposure to dexamethasone, and took place at GD 9– 21. Each stressor was applied once or twice during the period of exposure. Stressors were evenly distributed throughout the 12 days of exposure, also with respect to the 24-h cycle (Table 1). 2.3. Parturition and lactation data After termination of exposures at GD 21, the females were singly housed. The expected day of delivery, GD 22, was designated postnatal day (PND) 0 for the pups. Litters were not culled [25] but were left with their dam throughout rearing to minimize stress on the dam as well as the offspring. On PND 3, each pup was weighed and the gender of each pup registered. Pup weights were also recorded on PND 10 and at weaning, on PND 22. Mated female rats that had not given birth, the dams, and offspring that did not participate in testing were decapitated in CO2 –anaesthesia Table 1 Schedule of chronic mild stress relative to gestation day

Vertical borders between gestation days correspond to 00:00 a.m. Horizontal bars indicate the period of time and time of day, when the individual stressors were applied. Specifics of stressors: smaller cage (25  20  14 cm; 1  7 h, 1  17 h); wire cage (1  7 h; 1  17 h); food and water deprivation (2  7 h); empty water bottle (1  1 h); damp bedding (1  9 h, 1  17 h); tilted cage (45-; 1  7 h, 1  17 h); crowding (8 dams to a standard cage; 2  7 h); isolation (1  17 h); new partner (1  24 h).

57

the day after weaning, and the time-mated females were examined for the number of uterine implantation sites. 2.4. Post-weaning investigations At weaning at PND 22, two females per litter were selected at random for testing and housed in pairs of similar prenatal exposure and postnatal testing. One female from each litter had blood sampled for analysis of corticosterone before being tested for the acoustic startle response (ASR) whereas the other female was test naı¨ve at the time of ASR testing. From weaning, experimenters were kept unaware as to which prenatal exposure group the animals belonged. Investigations were performed during the light part of the diurnal cycle, and the time of testing on the day was counterbalanced between experimental groups. 2.5. Blood sampling and analysis for plasma corticosterone At the age of 3 months, tail blood was sampled thrice. Blood sampling took place during the first half of the light period. For the basal level, the rat was picked up by the body from its cage in the colony room, carried to a separate laboratory, restrained in an immobilizer (model IM/OH, Scanbur A/S, Denmark), and blood was collected (basal sample). After an additional 20 min of restraint, blood was again collected (stress sample), and the rat was returned to its cage. One hour later, the animal was again taken to the laboratory, restrained, and blood was withdrawn for a final sample (post-stress sample). Sampling was performed within 2 min from when the animal was picked up in its home cage, and cage mates were sampled on different days. Blood was collected directly into to iced Microtainer\ tubes with K2EDTA (Becton Dickinson No. 5973). After centrifugation (1202 RCF at 4 -C for 15 min) within 60 min from collection, plasma was aliquoted and stored in 500 AL polypropylene tubes at 80 -C until analyzed. Corticosterone in plasma was determined by competitive radioimmunoassay performed as previously described [26]. The method evaluation of a reference material in a buffer solution showed no bias of the method, i.e., recovery was 102% [CI: 100%; 104%]. Limit of detection (LOD) was 0.2 Ag/dL. Between-run CVs were 5.4% at 8.9 Ag/dL and 6.8% at 23.1 Ag/dL. Reference materials of two concentrations (2.2 Ag/dL, respectively, 10.0 Ag/dL corticosterone) were prepared from two rats. The measured concentrations were entered in a Westgaard control chart to ensure the precision and the trueness of the method at any time. Plasma was prepared as previously described and diluted in PBS (pH 7.4) [26]. 2.6. Acoustic startle (ASR) and prepulse inhibition (PPI) ASR testing was conducted as previously described [14], at the age of 6 months. At 4 months of age, the animals were transferred to the Research Institute of Biological Psychiatry,

58

K.S. Hougaard et al. / Developmental Brain Research 159 (2005) 55 – 63

Sct. Hans Hospital, where they were housed two to a cage in clear Macrolon Type IIIH cages in a temperature controlled room with a 12-h light – dark cycle. Food (Altromin Standard Diet 1324) and tap water were provided ad libitum. Briefly, testing was conducted in two sound insulated and ventilated startle chambers (San Diego Instruments). Each chamber contained a Plexiglas test tube (<: 8.2 cm, length 25 cm), mounted on a platform with a piezoelectric accelerometer attached beneath. The accelerometer detected and transduced displacement of the tube in response to movements of the rat. White background noise (70 dB(A)) was delivered continuously inside the chambers. The rats were transferred to the experimental room 1 h before the test. A 5-min acclimatization period commenced the test session that lasted approximately 20 min and consisted of 45 trials. Each session started and ended with 5 startle trials consisting of 120 dB(A) bursts of white noise, each lasting 40 ms. Following the 5 introductory startle trials, 35 test trials were delivered in a semi-randomized order (10 startle trials of 120 dB(A); 5 each of 4 prepulse + startle trials (PP72, PP74, PP78, and PP86); 5 trials with no stimulus except background noise). The inter-trial intervals were randomized between 10 and 20 s with a mean interval of 15 s. The movement of the tube was registered for 100 ms after onset of the startle stimulus (sampling frequency 1 kHz), amplified, and the average response over 100 ms (AVG) was calculated by computer. For each prepulse intensity, the AVGs were averaged and used for the analysis of PPI. PPI was expressed as the percent reduction in AVG compared to the average of the 10 middle startle trials: %PPI = 100 ((AVG at prepulse + startle trial) / (AVG at startle trial))  100%. Studies from other research groups include a single report of increased ASR in the initial startle trial on the first day of testing [41]. The average of the 5 initial startle trials

was therefore included as the first block (AVGinitial), supplemented by the mean of the 5 concluding startle trials (AVGfinal). Each animal was tested twice to test for persistency of effect, with 1 week between sessions. 2.7. Statistics Analysis of variance (ANOVA) was used to analyze pregnancy and lactation data, except for loss of implantations (the Kruskal –Wallis test) and gestation length (the chisquared test). Post hoc analyses were performed by means of Tukey pairwise comparisons test, when appropriate. In order to avoid litter effects, the litter was considered the statistical unit. The average body weight of pups within a litter was therefore used for statistical analysis. Analysis of covariance (ANCOVA) was used to test for differences in maternal body weight gain and body weight of offspring during lactation, while controlling for litter size. Post hoc comparisons were likewise performed by ANCOVA. Startle data were analyzed by three-way ANOVA with repeated measures. Group (control, DEX, and CMS) and blood sampling (T) were the between-subject factors, and week (1, 2) was the within-subject factor. Tukey pairwise comparisons test was applied for post hoc analyses where appropriate. As the statistical approaches to avoid type I statistical errors after multiple comparisons increase the risk for type II errors, pairwise comparisons were only performed within triads (blood sampled control, DEX, and CMS groups, or non-sampled control, DEX, and CMS groups). For PPI, this analysis was performed separately for each prepulse intensity (SYSTAT Software Package version 9). Plasma corticosterone was analyzed in a variance component model with subject as random effect, by means of the mixed procedure in the SAS system. Group, sample (basal, stress, and post-stress), and day of blood sampling

Fig. 1. Average maternal body weight during gestation. Pregnant rats were either controls, exposed to dexamethasone (DEX), or chronic mild stress (CMS). Body weight gain throughout the gestation period was significantly lower in DEX dams compared to control and CMS dams, commencing as the exposure to dexamethasone was initiated (arrow). n = 15 – 22, see text for P values.

K.S. Hougaard et al. / Developmental Brain Research 159 (2005) 55 – 63 Table 2 Pregnancy and lactation data End point

Control

DEX

CMS

Number of time-mated dams Number of litters (n) Maternal weight gain GD 4 – 20 (g)** Gestation length (days) Implantations Implantation loss (%) Live pups per litter* Female pups (%) Pup weight, PND 3 (g)** Pup weight, PND 10 (g)* Pup weight, PND 21 (g)*

20 15 100.7 T 21a

22 18 64.4 T 14b

24 22 98.2 T 14a

22.8 T 0.5 11.3 T 3.0 24.1 T 19 8.6 T 3.1a 51.5 T 23 6.7 T 1.3b 16.5 T 3.1b 43.8 T 7.3a

22.8 12.6 13.4 10.9 54.6 8.0 18.0 46.0

22.5 T 11.4 T 14.4 T 9.5 T 53.1 T 8.4 T 18.8 T 45.7 T

0.5 4.6 17 3.3a,b 14 1.1a 2.3a 5.3a,b

T T T T T T T T

0.6 3.0 14 2.4b 15 1.1a 2.3a,b 6.0b

Values are mean T SD. *P < 0.05, **P < 0.001 (overall statistical analyses). Values in the same row not showing a common superscript (a, b) are significantly different.

were included in the initial model as classed independent variables (categorical variables). Time of sampling was included as a continuous independent variable (SAS Statistical Software v. 8.02). The accepted level of statistical significance was <0.05.

3. Results 3.1. Pregnancy and lactation data Gestational and lactational data were recorded from three groups of dams exposed to either chronic mild stress (CMS), dexamethasone (DEX), or control conditions. Maternal weight gain differed significantly between exposure groups [F(2,51) = 30.547; P < 0.001], as dams from the DEX group

59

gained significantly less weight during pregnancy than did dams from the control and the CMS groups (P < 0.001), commencing as exposure to dexamethasone was initiated (Fig. 1). No statistically significant differences were associated with the exposures for gestation length, sex ratio, or number or loss of implantations (Table 2). The number of live pups per litter varied significantly between groups [ F(2,52) = 3.223, P < 0.05] due to significantly smaller litters in the DEX group compared to the CMS group (P < 0.05). Offspring body weights on PND 3 varied significantly with prenatal exposure [F(2,51) = 11.683; P < 0.001]. Post hoc analysis revealed significantly reduced body weights in the DEX group compared to offspring from each of the other groups [Table 2; DEX < others; 0.001  P < 0.01]. The significant overall variability was still present at 10 days of age and at weaning [F(2,51) = 4.552; P < 0.05 and F(2,51) = 3.198, P < 0.05, respectively, refer to Table 2]. 3.2. Corticosterone To analyze the effect of the prenatal exposures on the acute stress response, corticosterone was determined in plasma from blood sampled prior to, during, and 60 min following restraint. No statistically significant differences owing to prenatal exposure were detected (Fig. 2), as only sample contributed significantly to the final variance component model [F(2,91) = 662.86; P < 0.0001]. 3.3. Acoustic startle response Reactivity was tested by means of the acoustic startle response. Visual inspection of Fig. 3 reveals a general pattern of increased ASR in the prenatally exposed groups

Fig. 2. Hormonal stress response to an acute stressor. At the age of 3 months, female offspring of dams exposed in gestation to control conditions, dexamethasone (DEX), or chronic mild stress (CMS) had blood sampled thrice for analysis of plasma corticosterone: at basal conditions, after 20 min of restraint (stress), and following 60 min of rest in the animal’s cage in the animal room (post-stress). The basal sample was collected within 2 min from the animal was picked up in its home cage. Plasma corticosterone did not vary significantly with prenatal exposure group, but values differed between acute stressor conditions. Values are presented as mean + SEM, n = 13 – 18, see text for P values.

60

K.S. Hougaard et al. / Developmental Brain Research 159 (2005) 55 – 63

compared to controls, and during the first week of testing, increased ASR in blood-sampled animals compared to their respective non-sampled counterparts. These observations were substantiated by the statistical analysis. For the initial trial block, the AVG varied significantly with prenatal exposure group [F(2,92) = 5.130; P < 0.01], i.e., startle amplitude increased with prenatal exposure. Further, there was an interaction of blood sampling and week [F(2,92) = 12.570; P = 0.001]. Post hoc comparison of the control groups to their respective DEX and CMS groups revealed significantly enhanced AVGinitial, but only in prenatally stressed animals that had previously undergone blood sampling. This was the case for DEX and CMS animals compared to controls during the first week of testing (P < 0.05) and during the second week of testing

in DEX animals compared to controls (P < 0.05). The significance of the interaction between blood sampling and week indicated that blood sampling exerted differential influence on ASR during the 2 weeks of testing. Separate statistical analysis of weeks 1 and 2 supported that blood sampling was only significant during the first week of testing (P < 0.01). Only blood-sampled CMS females exhibited a significantly enhanced startle reaction compared to their non-sampled counterparts (P < 0.05), for the DEX offspring this difference approached statistical significance (0.05 < P < 0.10). For the final trial block, AVG varied significantly with prenatal exposure group [F(2,86) = 7.157; P = 0.001]. Post hoc analyses comparing the control groups to their respective DEX and CMS groups found that, while prenatal

Fig. 3. The acoustic startle response (AVG). At the age of 6 months, female offspring of dams exposed in gestation to control conditions (CON), dexamethasone (DEX), or chronic mild stress (CMS) was tested in the acoustic startle test. Within each gestational exposure group, half of the females had had blood withdrawn thrice for investigation of the corticosterone response to an acute stressor, 3 months before the startle test was performed: minus ( ) indicates no blood sampling, plus (+) indicates blood sampling. Each animal was tested twice, with 1 week between sessions (weeks 1 and 2). The startle response was calculated as the average response over 100 ms. Top panel: the average of the five initial startle trials. Bottom panel: the average of the five final startle trials. The startle response was significantly increased with prenatal exposure. However, post hoc analyses established that the response was only significantly enhanced in animals that had undergone blood sampling 3 months previously. Data are shown as mean + SEM, n = 12 – 20. *P < 0.05 compared to respective control group; **P < 0.005 compared to respective control group.

K.S. Hougaard et al. / Developmental Brain Research 159 (2005) 55 – 63

exposure did not influence AVGfinal significantly in nonblood sampled animals, AVGfinal was significantly enhanced in both DEX (P < 0.05) and CMS animals (P < 0.005) compared to controls during the second week of testing. As in our previous study [14], prenatal stress was not associated with changes in PPI.

4. Discussion The present study assessed the influence of gestational stress, in the form of chronic mild stress or a pharmacological stressor (dexamethasone, a potent synthetic glucocorticoid) on the offspring. ASR was consistently enhanced in rats stressed during uterine development, but only statistically significant when ASR testing was preceded by blood sampling, 3 months previously. The effects of maternal stress and exposure to dexamethasone on the acoustic startle response were virtually indistinguishable. Several lines of research point towards fetal overexposure to glucocorticoids as an important determinant for the behavioral and neuroendocrine alterations observed in prenatally stressed offspring [1,34,39,40,43]. Specifically for prenatal exposure to dexamethasone, the offspring displayed a reduced birth weight. However, reduced birth weight does not seem to be associated with increased ASR per se [12]. In the present study, ASR was similarly enhanced in DEX and CMS offspring. Although other mediating substances as well as maternal behavior may be involved in the long-term effects of prenatal stress [17,22,33], fetal overexposure to glucocorticoids is a common feature of both prenatal exposures in the present study, suggesting that maternal glucocorticoids may be a determining factor for changes in the regulatory mechanisms of the acoustic startle response. Further, dexamethasone was administered during the last week of pregnancy only, indicating that the neural circuits of the startle response are sensitive to the programming properties of glucocorticoids during late gestation in the rat. The observation of enhanced ASR in offspring stressed during pregnancy confirms earlier observations from our laboratory, where offspring of dams stressed by a schedule of CMS exhibited significantly and persistently increased ASR [14]. Other research groups have investigated the effect of prenatal stress on ASR but report no or marginal (increased ASR in the initial startle trial on the first day of testing) elevations of the ASR after prenatal stress [18,19,41]. These studies involve qualitatively different maternal stressor schedules, for example maternal restraint thrice daily in a predictive schedule [19] and a variable stress paradigm involving among others restraint, swim stress, and cold exposure [18]. This, together with our finding of similar effects of CMS and dexamethasone, speaks against the specific stressor schedule itself as the explanatory factor for the discrepant findings. The timing of

61

insult in relation to neural development is of importance for prenatally induced effects on startle behavior [23]. Another explanation for the discrepant findings therefore relates to the timing of stress in relation to fetal development. In the present study, the timing of exposure to dexamethasone corresponds to the periods of maternal exposure to stressors in the described studies. This explanation does therefore not seem plausible. In our previous study, observing enhanced ASR after prenatal stress, all animals had gone through blood sampling 3 months before ASR testing [14]. In other studies, observing no or marginal effect of prenatal stress on ASR, the animals were either test naı¨ve at the time of ASR testing [18,41] or some animals had been tested for latent inhibition in passive and/or active avoidance [19]. As outlined in the introduction, ASR testing involved confinement to a test tube. This resembles the experimental circumstances of blood sampling, involving 20 min of restraint flanked by blood withdrawal in either end, a period of rest in the animal’s own cage in the animal room, and rerestrainment for a final blood sample. We therefore hypothesized, that aversiveness associated with blood sampling would lead to enhanced ASR in prenatally stressed animals, due to enhanced capacity for context aversive conditioning in these animals, and therefore for fear potentiation of the startle reaction. Results from the present study support this hypothesis. Blood sampling was not associated with effects in control animals, and only during the AVGinitial does the blood sampled control group exhibit numerically higher values than the non-sampled control group. The ASR did not differ significantly between control and prenatally stressed offspring in the test-naı¨ve subset of female rats. However, in rats that had gone through blood sampling months previously the ASR was significantly enhanced in prenatally stressed animals compared to controls. Statistically significant differences appear primarily during AVGinitial, week 1 and AVGfinal, week 2. We have speculated if this pattern may be the result of two independent, opposing processes. One of extinction of fear potentiation related to the previous blood sampling procedure as the animals realize that startle testing is not as repulsive as blood withdrawal, and the other of sensitization to the startle test itself, again most prominent in previously blood sampled animals, as the startle test session appear to be a continuous event. However, the general pattern reflects increased ASR in the prenatally exposed blood sampled groups compared to controls. This result is consistent with previous observations, indicating that prenatally stressed offspring may display a heightened sensitivity to contextual cues after fear conditioning [8,29]. Griffin et al. observed increased freezing behavior in response to acute foot shock as well as increased freezing behavior the next day in the same context, without shock delivery, in prenatally stressed animals [8]. The finding of Louvart et al. extends even further, as they observed that an intense foot shock followed by weekly situational reminders increased immobility in

62

K.S. Hougaard et al. / Developmental Brain Research 159 (2005) 55 – 63

prenatally stressed animals 6 weeks after the initial foot shock [21]. A single aversive life event may therefore be capable of inducing changes in the reactivity of prenatally stressed offspring leaving offspring of dams going through a less stressful gestation largely unaffected [15]. Prenatal stress has been associated with elevated basal levels of plasma corticosterone as well as a protracted hormonal response to acute stressors [36]. In our previous study, no effect of prenatal stress on plasma corticosterone was observed. Blood sampling for plasma corticosterone was performed during the active (dark) period of the 24-h cycle, rather than during the light phase, and this was hypothesized to offer a possible explanation as to the lack of effect of prenatal exposure to CMS on plasma corticosterone [14]. In the present study, the hormonal stress response was therefore monitored during the light phase. However, plasma corticosterone was similar in control and prenatally stressed animals. Prenatally stressed offspring has exhibited increased sensitivity to sudden noise, in the absence of an altered endocrine stress response. This suggests that consequences of prenatal stress do not require set-point changes in the HPA axis to be displayed [27]. Further studies should investigate whether repetition of blood withdrawal changes the corticosterone levels in prenatally stressed offspring similarly to its effect on ASR, as well as if the increase in baseline startle in prenatally stressed animals is caused by aversive contexts or chronic arousal. In conclusion, prenatal stress was shown to increase the susceptibility to environmental manipulations profoundly. Stressful life events are considered an important factor in the development of human pathologies, for example posttraumatic disorders [47]. Our findings indicate that circumstances dating back to the very beginning of life affect the individual’s sensitivity towards experiences in life after birth. The prenatal environment may thus form part of the explanation of the considerable individual variation in the response to adverse life events.

Acknowledgments We thank Gitte B. Kristensen and Michael Guldbrandsen for laboratory assistance and care taking of the animals and Ulla Tegner for analysis of corticosterone, all done with great skill. The Danish National Institute of Occupational Health supported investigations taking place at the institute, and the Danish National Psychiatric Research Foundation supported the investigations at Sct. Hans Hospital.

References [1] A. Barbazanges, P.V. Piazza, M. Le Moal, S. Maccari, Maternal glucocorticoid secretion mediates long-term effects of prenatal stress, J. Neurosci. 16 (1996) 3943 – 3949.

[2] M.S. Cratty, H.E. Ward, E.A. Johnson, A.J. Azzaro, D.L. Birkle, Prenatal stress increases corticotropin-releasing factor (CRF) content and release in rat amygdala minces, Brain Res. 675 (1995) 297 – 302. [3] M.F. Dallman, S.F. Akana, S. Bhatnagar, M.E. Bell, S. Choi, A. Chu, C. Horsley, N. Levin, O. Meijer, L.R. Soriano, A.M. Strack, V. Viau, Starvation: early signals, sensors and sequelae, Endocrinology 140 (1999) 4015 – 4023. [4] P.S. D’Aquila, J. Newton, P. Willner, Diurnal variation in the effect of chronic mild stress on sucrose intake and preference, Physiol. Behav. 62 (1997) 421 – 426. [5] J.P. Dupouy, A. Chatelain, P. Allaume, Absence of transplacental passage of ACTH in the rat: direct experimental proof, Biol. Neonate 37 (1980) 96 – 102. [6] T. Fujioka, Y. Sakata, K. Yamaguchi, T. Shibasaki, H. Kato, S. Nakamura, The effects of prenatal stress on the development of hypothalamic paraventricular neurons in fetal rats, Neuroscience 92 (1999) 1079 – 1088. [7] G.B. Glavin, W.P. Pare, T. Sandbak, H.K. Bakke, R. Murison, Restraint stress in biomedical research: an update, Neurosci. Biobehav. Rev. 18 (1994) 223 – 249. [8] W.C. Griffin, H.D. Skinner, A.K. Salm, D.L. Birkle, Mild prenatal stress in rats is associated with enhanced conditioned fear, Physiol. Behav. 79 (2003) 209 – 215. [9] C. Grillon, Startle reactivity and anxiety disorders: aversive conditioning, context, and neurobiology, Biol. Psychiatry 52 (2002) 958 – 975. [10] S.E. Hahn, C.S. Smith, Daily hassles and chronic stressors: conceptual and measurement issues, Stress Med. 15 (1999) 89 – 101. [11] K.S. Hougaard, Neurobehavioral teratology of maternal stress in combination with chemical exposure in rats. PhD thesis. National Institute of Occupational Health, Copenhagen, Denmark, 2003. [12] K.S. Hougaard, U. Hass, S.P. Lund, L. Simonsen, Effects of prenatal exposure to toluene on postnatal development and behavior in rats, Neurotoxicol. Teratol. 21 (1999) 241 – 250. ˚ .M. Hansen, U. Hass, S.P. Lund, Toluene depresses [13] K.S. Hougaard, A plasma corticosterone in pregnant rats, Pharmacol. Toxicol. 92 (2003) 148 – 152. [14] K.S. Hougaard, M.B. Andersen, A.M. Hansen, U. Hass, T. Werge, S.P. Lund, Effects of prenatal exposure to chronic mild stress and toluene in rats, Neurotoxicol. Teratol. 27 (2005) 153 – 167. [15] A.C. Huizink, E.J. Mulder, J.K. Buitelaar, Prenatal stress and risk for psychopathology: specific effects or induction of general susceptibility? Psychol. Bull. 130 (2004) 115 – 142. [16] R.J. Katz, M. Sibel, Further analysis of the specificity of a novel animal model of depression—effects of an antihistaminic, antipsychotic and anxiolytic compound, Pharmacol. Biochem. Behav. 16 (1982) 979 – 982. [17] G.I. Keshet, M. Weinstock, Maternal naltrexone prevents morphological and behavioral alterations induced in rats by prenatal stress, Pharmacol. Biochem. Behav. 50 (1995) 413 – 419. [18] J.I. Koenig, G.I. Elmer, P.D. Shepard, P.R. Lee, C. Mayo, B. Joy, E. Hercher, D.L. Brady, Prenatal exposure to a repeated variable stress paradigm elicits behavioral and neuroendocrinological changes in the adult offspring: potential relevance to schizophrenia, Behav. Brain Res. 156 (2005) 251 – 261. [19] J. Lehmann, T. Stohr, J. Feldon, Long-term effects of prenatal stress experiences and postnatal maternal separation on emotionality and attentional processes, Behav. Brain Res. 107 (2000) 133 – 144. [20] K.C. Liang, K.R. Melia, M.J. Miserendino, W.A. Falls, S. Campeau, M. Davis, Corticotropin-releasing factor: long-lasting facilitation of the acoustic startle reflex, J. Neurosci. 12 (1992) 2303 – 2312. [21] H. Louvart, S. Maccari, M. Darnaudery, Prenatal stress affects behavioral reactivity to an intense stress in adult female rats, Brain Res. 1031 (2005) 67 – 73. [22] S. Maccari, M. Darnaudery, S. Morley-Fletcher, A.R. Zuena, C. Cinque, O. Van Reeth, Prenatal stress and long-term consequences:

K.S. Hougaard et al. / Developmental Brain Research 159 (2005) 55 – 63

[23]

[24]

[25] [26]

[27]

[28]

[29]

[30]

[31]

[32]

[33] [34]

[35]

[36]

implications of glucocorticoid hormones, Neurosci. Biobehav. Rev. 27 (2003) 119 – 127. M. Mintz, G. Yovel, A. Gigi, M.S. Myslobodsky, Dissociation between startle and prepulse inhibition in rats exposed to gamma radiation at day 15 of embryogeny, Brain Res. Bull. 45 (1998) 289 – 296. R. Muscat, P. Willner, Suppression of sucrose drinking by chronic mild unpredictable stress: a methodological analysis, Neurosci. Biobehav. Rev. 16 (1992) 507 – 517. A.K. Palmer, B.C. Ulbrich, The cult of culling, Fundam. Appl. Toxicol. 38 (1997) 7 – 22. ˚ .M. Hansen, S.P. Lund, A.H. Garde, Validation of a T.K.H. Pedersen, A radioimmunoassay for the determination of total corticosterone in rat plasma, Anal. Chim. Acta 413 (2000) 63 – 69. R. Rimondini, G. Agren, S. Borjesson, W. Sommer, M. Heilig, Persistent behavioral and autonomic supersensitivity to stress following prenatal stress exposure in rats, Behav. Brain Res. 140 (2003) 75 – 80. V.B. Risbrough, R.L. Hauger, M.A. Pelleymounter, M.A. Geyer, Role of corticotropin releasing factor (CRF) receptors 1 and 2 in CRFpotentiated acoustic startle in mice, Psychopharmacology (Berlin) 170 (2003) 178 – 187. A.K. Salm, M. Pavelko, E.M. Krouse, W. Webster, M. Kraszpulski, D.L. Birkle, Lateral amygdaloid nucleus expansion in adult rats is associated with exposure to prenatal stress, Brain Res. 148 (2004) 159 – 167. J.W. Smythe, C.M. McCormick, M.J. Meaney, Median eminence corticotrophin-releasing hormone content following prenatal stress and neonatal handling, Brain Res. Bull. 40 (1996) 195 – 199. W. Sutanto, E.R. De Kloet, The use of various animal models in the study of stress and stress-related phenomena, Lab. Anim. 28 (1994) 293 – 306. N.R. Swerdlow, M.A. Geyer, W.W. Vale, G.F. Koob, Corticotropinreleasing factor potentiates acoustic startle in rats: blockade by chlordiazepoxide, Psychopharmacology (Berlin) 88 (1986) 147 – 152. H.H. Szeto, Dynorphin and the hypothalamo – pituitary – adrenal axis during fetal development, Life Sci. 73 (2003) 749 – 758. L.K. Takahashi, J.G. Turner, N.H. Kalin, Prolonged stress-induced elevation in plasma corticosterone during pregnancy in the rat: implications for prenatal stress studies, Psychoneuroendocrinology 23 (1998) 571 – 581. H.E. Ward, E.A. Johnson, A.K. Salm, D.L. Birkle, Effects of prenatal stress on defensive withdrawal behavior and corticotropin releasing factor systems in rat brain, Physiol. Behav. 70 (2000) 359 – 366. M. Weinstock, Does prenatal stress impair coping and regulation of hypothalamic – pituitary – adrenal axis? Neurosci. Biobehav. Rev. 21 (1997) 1 – 10.

63

[37] M. Weinstock, Alterations induced by gestational stress in brain morphology and behaviour of the offspring, Prog. Neurobiol. 65 (2001) 427 – 451. [38] L.A. Welberg, J.R. Seckl, Prenatal stress, glucocorticoids and the programming of the brain, J. Neuroendocrinol. 13 (2001) 113 – 128. [39] L.A. Welberg, J.R. Seckl, M.C. Holmes, Inhibition of 11betahydroxysteroid dehydrogenase, the foeto-placental barrier to maternal glucocorticoids, permanently programs amygdala GR mRNA expression and anxiety-like behaviour in the offspring, Eur. J. Neurosci. 12 (2000) 1047 – 1054. [40] L.A. Welberg, J.R. Seckl, M.C. Holmes, Prenatal glucocorticoid programming of brain corticosteroid receptors and corticotrophinreleasing hormone: possible implications for behaviour, Neuroscience 104 (2001) 71 – 79. [41] D.A. White, D.L. Birkle, The differential effects of prenatal stress in rats on the acoustic startle reflex under baseline conditions and in response to anxiogenic drugs, Psychopharmacology (Berlin) 154 (2001) 169 – 176. [42] M.T. Williams, H.N. Davis, A.E. Mccrea, M.B. Hennessy, The distribution of radiolabeled corticotropin-releasing factor in pregnant rats: an investigation of placental transfer to the fetuses, Int. J. Dev. Neurosci. 16 (1998) 229 – 234. [43] M.T. Williams, H.N. Davis, A.E. Mccrea, S.J. Long, M.B. Hennessy, Changes in the hormonal concentrations of pregnant rats and their fetuses following multiple exposures to a stressor during the third trimester, Neurotoxicol. Teratol. 21 (1999) 403 – 414. [44] P. Willner, Validity, reliability and utility of the chronic mild stress model of depression: a 10-year review and evaluation, Psychopharmacology 134 (1997) 319 – 329. [45] P. Willner, A. Towell, D. Sampson, S. Sophokleous, R. Muscat, Reduction of sucrose preference by chronic unpredictable mild stress, and its restoration by a tricyclic antidepressant, Psychopharmacology (Berlin) 93 (1987) 358 – 364. [46] P. Willner, J.L. Moreau, C.K. Nielsen, M. Papp, A. Sluzewska, Decreased hedonic responsiveness following chronic mild stress is not secondary to loss of body weight, Physiol. Behav. 60 (1996) 129 – 134. [47] R. Yehuda, Biology of posttraumatic stress disorder, J. Clin. Psychiatry 61 (Suppl. 7) (2000) 14 – 21. [48] M.X. Zarrow, J.E. Philpott, V.H. Dennenberg, Passage of 14 C-4Corticosterone from the rat mother to the foetus and neonate, Nature 226 (1970) 1058. [49] B. Zimmerberg, L.G. Blaskey, Prenatal stress effects are partially ameliorated by prenatal administration of the neurosteroid allopregnanolone, Pharmacol. Biochem. Behav. 59 (1998) 819 – 827.