K. Sakata, PhD, Department of Pharmacology/Psychiatry, College of Medicine, University of Tennessee Health Science Center, 874 Union Avenue, Room 430, Memphis, TN 38163, USA. E-mail: email@example.com
Transcription of Bdnf is controlled by multiple promoters, in which promoter IV contributes significantly to activity-dependent Bdnf transcription. We have generated promoter IV mutant mice [brain-derived neurotrophic factor (BDNF)-KIV] in which promoter IV-driven expression of BDNF is selectively disrupted by inserting a green fluorescent protein (GFP)-STOP cassette within the Bdnf exon IV locus. BDNF-KIV animals exhibited depression-like behavior as shown by the tail suspension test (TST), sucrose preference test (SPT) and learned helplessness test (LHT). In addition, BDNF-KIV mice showed reduced activity in the open field test (OFT) and reduced food intake in the novelty-suppressed feeding test (NSFT). The mutant mice did not display anxiety-like behavior in the light and dark box test and elevated plus maze tests. Interestingly, the mutant mice showed defective response inhibition in the passive avoidance test (PAT) even though their learning ability was intact when measured with the active avoidance test (AAT). These results suggest that promoter IV-dependent BDNF expression plays a critical role in the control of mood-related behaviors. This is the first study that directly addressed the effects of endogenous promoter-driven expression of BDNF in depression-like behavior.
Major depressive disorder (MDD) is a leading cause of disease burden characterized by a general loss of interest and by anhedonia (Wong & Licinio 2001). While an incomplete understanding of the molecular mechanisms underlying MDD has delayed the development of effective treatments, convergent evidence points out that both the pathophysiology of depression and the action of antidepressants involve brain-derived neurotrophic factor (BDNF), a major neuronal growth factor that promotes neurogenesis, neuronal maturation and synaptic plasticity (Duman et al. 1997; Nestler et al. 2002). BDNF expression is decreased in the serum, hippocampus (HIP) and prefrontal cortex (PFC) of patients with MDD (Dwivedi et al. 2003; Sen et al. 2008; Shimizu et al. 2003). In contrast, BDNF expression has been shown to be increased by chronic (but not acute) treatments with several classes of antidepressants following a time-course observed for the therapeutic effects in the serum of patients with MDD and in the HIP and frontal cortex of rodents (Duman & Monteggia 2006; Sen et al. 2008). Therefore, it has been hypothesized that depression may be, in part, caused by reduced expression of BDNF.
Although extensive studies have suggested the BDNF hypothesis of depression, the critical evidence that directly shows that reduced endogenous gene expression of Bdnf can lead to depression/depression-like behavior in vivo is still missing (Duman & Monteggia 2006; Groves 2007). So far, studies using heterozygous and conditional knockout (KO) mice and exogenous BDNF knockdown/infusion systems have produced mixed results. The major consensus is that rodents with reduced or increased BDNF levels are generally hyperactive and do not show depression-like behavior but rather a blunted behavioral response to antidepressants (Chan et al. 2006; Eisch et al. 2003; Monteggia et al. 2004, 2007). One critical issue in previous studies is that the manipulations used (e.g. removing the BDNF protein-coding region that is regulated by other gene promoters or exogenous BDNF knockdown/induction by viruses) may have reduced or induced BDNF expression to physiologically irrelevant levels. In addition, these manipulations may not have impacted endogenous BDNF levels in relevant cells and brain regions. Therefore, the outcomes may not reflect the situation in human patients.
To directly test the BDNF hypothesis of depression with physiological relevance, the present study focused on manipulating an endogenous promoter of BDNF. BDNF expression is regulated by at least nine promoters (Aid et al. 2007). Of the known promoters, promoter IV (previously classified as promoter III) is most responsive to neuronal activity and induces activity-dependent expression of BDNF in vitro and in vivo, containing calcium-responsive elements responsive elements (CaREs) and cyclic adenosine monophosphate (cAMP)/calcium response element (CRE) (Shieh et al. 1998; Tao et al. 1998; Timmusk et al. 1993). We hypothesize a potentially important feedback mechanism for sustaining neuronal activity; increased neuronal activity induces activity-dependent BDNF expression, which would induce neuronal activity to maintain active brain functions. Any disruption to activity-dependent BDNF expression would lead to a decrease in neuronal activity and function, which could lead to depression. Therefore, we focused on promoter IV as a dominant activity-dependent promoter and, here, tested whether deregulation of activity-dependent promoter IV-driven expression of BDNF leads to depression-like behavior.
Materials and methods
We have generated mutant mice with selective disruption of promoter IV-driven BDNF expression as previously described (Sakata et al. 2009). Briefly, a green fluorescent protein (GFP) gene was inserted into BDNF exon IV, the DNA region located immediately downstream of promoter IV. In the BDNF gene, each promoter can drive transcription beginning with a small exon (protein noncoding exon) immediately downstream from the promoter, which is spliced onto a common, last exon that encodes the BDNF protein. This novel approach selectively precluded expression of BDNF protein through promoter IV-driven transcripts (BDNF exon IV-GFP gene-exon IX). The mutant mice (BDNF-knock-in IV: BDNF-KIV) retain intact promoter IV, which is critical to avoid compensatory expression of other BDNF exons and gene products, as was observed in mice with mutation or deletion of BDNF promoter IV (Hong et al. 2008). In BDNF-KIV mice, the reduction in expression of BDNF protein occurs in a brain region-specific manner under the regulation of BDNF promoter IV, which is active in areas of high synaptic plasticity such as the cerebral cortex and HIP (Malkovska et al. 2006; Metsis et al. 1993; Sakata et al. 2009). BDNF-KIV mice were backcrossed to a C57BL/6 background for >12 generations. All experiments were performed with male age-matched (2–6 months old) group-housed BDNF-KIV mice and their wild-type (WT) littermates from heterozygous parents. Three independant batches of animals were used. Batch A (n = 12 pairs; 12 BDNF-KIV and 12 WT mice) was used for the open field test (OFT). Batch B (n = 10 pairs) was used for a battery of tests that examined depression-related behavior. These tests were performed in the following order from the least to most stressful with 2–7 days off between tests, as indicated by the number in parentheses: tail suspension test (TST) (2), forced swim test (FST) (5), sucrose preference test (SPT) (4), active avoidance test (AAT) (7), and learned helplessness test (LHT). The batch B animals were also examined for comorbid anxiety-like behavior in the order of light and dark box test (LDT) (2), elevated plus maze test (EPMT) (3), novelty-suppressed feeding test (NSFT) (4), all which were conducted before AAT and LHT. The passive avoidance test (PAT) (7) was also conducted before the AAT to measure response inhibition. Batch C (n = 12 pairs) was used additionally for the TST and FST to ensure the statistic values obtained (the TST in batch B: WT = 155.5 ± 10.6, KIV = 184.8 ± 11.0, t18 = 1.9, P = 0.071; the TST in batch C: WT = 161.4 ± 10.4, KIV = 195.5 ± 13.4, t22 = 2.0, P = 0.057; the FST in batch B: WT = 117.7 ± 21.9, KIV = 120.3 ± 25.5, t18 = 0.077, P = 0.94; the FST in batch C: WT = 112.9 ± 14.1, KIV = 119.5 ± 19.3, t22 = 0.27, P = 0.79). Mice were habituated to testing facilities at least 1 h before behavioral assessment except in the OFT, EPMT and NSFT. All animals were housed in a normal 12:12 h dark–light cycle, and experiments were conducted in the light cycle except one of the LDT trials. All experiments were approved by the University of Tennessee laboratory animal care and use committee.
The OFT measures spontaneous activity in a novel open field (Walsh & Cummins 1976). Animals were individually placed in a novel open-field box (47 × 37 × 20 cm, 900 lux at the center; Accuscan Instruments, Columbus, OH, USA) equipped with 16 infrared beams that automatically measured total locomotor activity by beam breaks, for 10 min on two consecutive days. The time spent in the center (25% of the field) was measured as an anxiolytic indicator (Crawley 1999; Prut & Belzung 2003).
Tail suspension test
Mice were suspended by their tail, and the presence of immobility was assessed over a 6-min session by a trained observer (Cryan et al. 2005).
Forced swim test
Mice were placed individually in a clear Plexiglas cylinder (height, 25 cm; diameter, 10 cm) containing 15 cm of water kept at 25 ± 1°C, and behavior was videotaped for 6 min. The total period of immobility during the last 4 min was timed offline by an observer blind to the phenotype (Petit-Demouliere et al. 2005; Porsolt et al. 1977).
Sucrose preference test
Mice were habituated to two water bottles for 4 days before the preference test. The mice were deprived of water (from 1700 h) 16 h before initiation of the SPT. Mice were given a 3-h SPT (0900–1200 h) once a day for 3 consecutive days. During the test, each mouse was housed individually and was given two bottles with one containing water and the other containing a 1% sucrose solution. The amount of each solution consumed was determined by weighing the bottles before and after the 3-h consumption period (Pothion et al. 2004). Sucrose preference was calculated as total sucrose intake/total (sucrose + water) intake (Willner et al. 1987). The position of the bottles was alternated every day to account for bottle position preferences.
Learned helplessness test
The shuttle box (Med Associate Inc., Burlington, VT, USA) consists of two equal-sized compartments (20 × 16 × 21 cm3, 10 lux) separated by a small automatic gate (9 cm wide and 11 cm high). In the learned helplessness paradigm, mice were exposed to 180 unpredictable foot shocks with varying durations (1–10 seconds) and inter-trial episodes (2–15 seconds) in the shuttle box on 2 consecutive days. Twenty-four hours after the second shock procedure, subjects were placed in the same shuttle box and received 30 shuttle escape trials (Chourbaji et al. 2005). Each trial started with a 5-second stimulus tone, announcing a subsequent foot shock of maximum 10-second duration. The trial ended when the mice escaped or when the foot shock was over and was followed by an inter-trial interval of 2–15 seconds. Escape latency was assessed by determining the time that took for the animal to escape after the initiation of each trial. Failures were recorded when no attempt to escape was made during the 15-second trial. Movements made between trials were also recorded by infrared beam breaks.
Novelty-suppressed feeding test
Twenty-four hours after food deprivation, mice were transferred to a new testing room in a new open clear cage (47 × 37 × 20 cm) without habituation. A small piece of regular mouse food (weighed) was placed in the center of the arena on a white square filter paper (6 × 6 × 6 cm). Each subject was placed in a corner of the testing area facing the wall, and the time to mouse's first consumption was recorded with a cutoff time of 10 min (Bodnoff et al. 1988). Immediately after the test subject began to eat the food, the animal was placed alone in its home cage with the same weighed piece of food for 5 min. Similarly, the first latency to consumption was recorded. At the end of this period, the amount of food consumed was determined by weighing the remaining piece of mouse food. This test was repeated for 2 days to examine habituation of the mice to the novel environment. The mouse was also given a piece of sweet food with fruit smell (Fruity-Gems™; Bio-Serve, Frenchtown, NJ, USA) in its home cage at the end of day 1 of the NSFT and in the novel environment on day 2. The first latency to consumption was recorded. Mice were food deprived and maintained at 85% of their original body weight during the NSFT.
Light and dark box test
The apparatus consisted of a dark chamber (two fifths: 21 × 20 × 38 cm, <6 lux) and an open illuminated white chamber (three fifths: 21 × 30 × 38 cm, 600 lux). The LDT was conducted on 2 consecutive days in the dark cycle (after 1800 h). On day 1, mice were initially placed in the light side, and on day 2, mice were initially placed in the dark side. The same animals were also subjected to the LDT in the light cycle (0900–1800 h), using another light and dark box (Med Associates Inc.), which consisted of a dark chamber (<6 lux) and a brightly illuminated chamber (600 lux) of equal size (20 × 16 × 21 cm3). Mice were initially placed in the dark side. The time spent in each chamber, total numbers of transitions and initial latency to enter the opposite chamber were measured for 10 min by video recording and the Med Associates automated system (Bourin & Hascoet 2003).
Elevated-plus maze test
The apparatus comprised two open arms (50 × 11 × 0 cm) across from each other and perpendicular to two closed arms (50 × 11 × 31 cm) with a center platform (11 × 11 cm), which were located 55 cm above the floor. The closed arms have a high (31 cm) wall enclosure allowing <5 lux of light intensity, whereas the open arms were completely open (100 lux). Mice were placed in the center of the apparatus and given free access to the four arms for 10 min. The number of entries into the open and closed arms and time spent in these arms were measured. An entry was defined as four paws inside an arm (Fernandez Espejo 1997).
Active avoidance test
Mice were subjected to 50-trial avoidance sessions in the same shuttle box as the LHT. Each trial started with a conditioned sound stimulus (accompanied by a light) of 5 seconds, announcing a subsequent foot shock (0.2 mA) of maximum 25-second duration. The trial ended when the mouse escaped or when the foot shock was over and was followed by an inter-trial interval of 15 seconds. Shuttle into the other chamber within 5 seconds of CS was defined as ‘avoid,’ while shuttle into the other chamber within the 25-second duration of foot shock was defined as ‘escape’ (Bovet et al. 1969). The latency for the mouse to escape/avoid was measured for each trial, and the average was measured across a block of five trials.
Passive avoidance test
The same shuttle box as the LHT was used with a modification that one compartment was lit and the other was covered in black. On day 1, the training day, mice were placed into the lit chamber for 10-second habituation, and then the door between the chambers was opened. When the mice entered the dark compartment with all four paws, the guillotine door was closed. The latency to enter the dark chamber was recorded (from the time the door was lifted). After 2 seconds, a single foot shock (0.2 mA) was delivered for 5 seconds. The mice were then kept in the dark chamber for an additional 10 seconds so that they could form an association between the context (the dark chamber) and the foot shock. The animals were then returned to their respective home cages. Twenty-four hours after the training, the mice were placed into the light compartment facing away from the dark compartment, and the guillotine door was lifted, and the latency to enter the dark compartment was recorded within the cutoff time of 180 seconds (Crawley 1999).
Student's t-tests were performed with two data groups. Two-way repeated-measures (TW-RM) analyses of variance (ANOVA) were performed using PRISM (GraphPad Software Inc., La Jolla, CA, USA) for the SPT, NSFT, AAT, PAT and TST (with/without habituation) with groups or days as the independent variable. Significant effects were further evaluated using post hoc t-tests with Bonferroni corrections. The frequency distribution of failures in the LHT was analyzed by Fisher's exact test. Differences of P < 0.05 were considered significant. Data are presented as mean ± SE.
To evaluate the role of promoter IV-dependent expression of BDNF in the depression-related phenotype, we conducted a battery of tests using mutant mice, in which promoter IV-driven BDNF expression is selectively disrupted [BDNF-KIV mice (KIV) (Sakata et al. 2009)] and control WT littermates.
Open field activity test
First, we measured locomotor activity of the mice in a novel open field over 2 days. A TW-RM ANOVA showed a significant effect of genotype (F1,22 = 16.5, P < 0.05) and a genotype × day interaction (F1,22 = 3.7: P < 0.05). Post hoc t-tests showed that BDNF-KIV mice exhibited significantly reduced locomotor activity in a novel open field compared with WT mice on the first day (t22 = 3.3, P < 0.005; Fig. 1a). This difference in locomotor activity was not observed on the second day. This is likely because WT mice showed increased exploratory activity in the novel environment on the first day, and activity was subsequently reduced as they habituated to the environment, while BDNF-KIV mice did not show increased activity in the novel environment on the first day without changing the basal activity. The reduced explorative activity of the mutant mice could be because of novelty-induced anxiety-like behavior (Prut & Belzung 2003). However, reduced time spent in the center of the open field, an indicator of increased anxiety-like behavior (Crawley 1999), was not observed in BDNF-KIV mice compared with WT mice (WT = 92.4 ± 7.6 and KIV = 99.4 ± 14.9 seconds in center on day 1, t22 = 0.42, P = 0.68).
Tail suspension test
When mice are suspended by their tails, they become motionless, which is thought to reflect despair because of the inescapable stress (Cryan et al. 2005). BDNF-KIV mice showed significantly increased immobility compared to the control WT mice (t42 = 2.8, P < 0.01), suggesting a reduced ability to cope with suspension-induced stress (Fig. 1b).
Forced swim test
Similar to the TST, the FST is one of the most widely used stress-related despair tests, where longer immobility in water is interpreted as lack of coping action (Porsolt et al. 1977). In contrast to the results in the TST, BDNF-KIV mice and WT mice displayed similar immobility times (t22 = 0.27, P > 0.05, Fig. S1).
Sucrose preference test
Anhedonia, the loss of interest in normally pleasurable, rewarding activities, is a core symptom of depression and is a suitable behavioral indicator of a depression-like state in rodents (Pothion et al. 2004). To evaluate anhedonia-like behavior of the mutant mice, we measured their consumption of sweet solutions over 3 days. Interestingly, a main effect of genotype on sucrose preference (F1,18 = 9.9, P < 0.05) was found by a TM-RM ANOVA. Post hoc t-tests showed that BDNF-KIV mice displayed significantly reduced levels of sucrose preference compared with WT mice on day 2 (t18 = 2.5, P < 0.05) and day 3 (t18 = 2.1, P < 0.05), suggesting anhedonia-like behavior (Fig. 1c), whereas the difference on day 1 was not statistically significant. Even though mutant subjects had slightly lower levels of total liquid intake, the difference was not significant, indicating that the reduced sucrose preference was not due to the reduced liquid intake (Fig. 1d).
Learned helplessness test
Learned helplessness represents a key feature of depression in many species. Rodents that are susceptible to learned helplessness are known to show reduced coping activity after administration of inescapable stress (Chourbaji et al. 2005). The mutant mice showed significantly increased mean latency to escape the foot shock compared with the paired WT mice after exposure to inescapable foot shocks (t14 = 2.4, P < 0.05; Fig. 2a). The inter-trial movements were also significantly reduced in the mutant mice (t14 = 2.6, P < 0.01; Fig. 2b). The mutant mice displayed a distribution of frequencies in higher number of failures to escape (Fig. 2c) and an increased mean number of failures (Fig. 2d) compared with WT mice, although Fisher's exact test and t-test did not show a significant effect due to of the high variance in the number of failures (from 0 to 20 in WT mice).
Active avoidance test
The AAT was conducted before the LHT to validate that the mutant mice were capable of sensing pain and learning to escape foot shocks. The AAT used the same shuttle-box and the same intensity of foot shocks as the LHT, but the shocks were escapable. All subjects showed escape behavior for all 50 trials (not shown). Both genotypes displayed learning curves showing gradually decreasing escape latencies (Ftrial (1,18) = 30.20, P < 0.0001; Fig. S2a). Significant differences were not detected between the genotypes in the learning curves (Fig. S2a) or in mean avoid and escape latencies in the 50 trials (Fig. S2b), indicating normal pain sensitivity and short-term learning to avoid foot shocks in BDNF-KIV mice.
Novelty-suppressed feeding test
The NSFT measures food consumption in a novel environment under food deprivation, where reduction is interpreted as anxiety-like behavior as well as depression-related behavior because the phenotype is reversed by chronic antidepressant treatment (Bodnoff et al. 1988). The NSFT was conducted over 2 days to examine the effect of habituation. Both genotypes displayed longer latency to consumption in a novel environment on day 1. The latency was reduced on day 2, as they became habituated to the novel environment (Fday (1,16) = 50.5, P < 0.001; Fig. 3a). A TW-RM ANOVA of latency found a genotype × day interaction (F1,16 = 3.3, P < 0.05). Post hoc t-tests showed that BDNF-KIV mice had a significantly increased latency compared with WT mice on day 2 (t16 = 2.5, P < 0.05), while they showed similar latency on day 1. This was a result of WT mice showing a dramatic decrease in consumption latency on day 2, while the mutant mice displayed a steady decline in the consumption latency (Fig. 3a), suggesting that the mutant mice are slower to habituate to an environment. The results echoed that of the initial latency to food consumption in the home environment, which was measured immediately after they were tested in the novel environment. A TW-RM ANOVA showed significant effects of genotype (F1,16 = 33.1, P < 0.001) and day (F1,16 = 15.7, P < 0.005) as well as a genotype × day interaction (F1,16 = 8.1, P < 0.05). BDNF-KIV mice displayed significantly longer latency to consume food compared with WT mice on day 1 (t16 = 5.2, P < 0.001), indicating that BDNF-KIV mice may exhibit slower adaptation to environmental changes (Fig. 3b). Total food consumption was also significantly reduced in BDNF-KIV mice compared with WT mice on day 1 (t16 = 2.7, P < 0.05) (Fig. 3c, Fgenotype (1,16) = 9.1, P < 0.05; Fday (1,16) = 32.3, P < 0.001).
NSFT with tasty gem
The results of the NSFT might raise a concern that the reduced latency and consumption of food by BDNF-KIV mice could be caused by reduced motivation for food despite food deprivation. To address this, a more attractive food, ‘Tasty Gem’ (TG), was used in the modified NSFT. Initially, we validated TG as an attractive food for both genotypes in the home environment. Both mutant and WT mice showed significantly reduced latency in consumption of TG compared with that of the regular food (t16 = 3.2, P < 0.01 and t16 = 3.1, P < 0.01 in BDNF-KIV and WT mice, respectively), verifying increased motivation for TG than for the regular food (Fig. 3d, ‘home’). No difference in latency between the genotypes was observed (t16 = 0.8, P > 0.05), suggesting that their motivation to the TG was similar. In the novel environment, both groups displayed significantly increased latency to consume TG compared with that in the home cage (P < 0.05 for BDNF-KIV and P < 0.05 for WT mice). A TW-RM ANOVA showed significant effects of genotype (F1,15 = 9.3, P < 0.05) and day (F1,15 = 22.7, P < 0.005) as well as a genotype × environment interaction (F1,15 = 11.1, P < 0.05). Post hoc t-test showed that the latency of BDNF-KIV mice to TG was significantly higher in the novel environment compared with that of WT mice (t16 = 3.2, P < 0.05; Fig. 3d, ‘novel’). These results confirmed the results of the NSFT and indicate that the mutant mice experienced increased depression and anxiety-like behavior in a novel environment.
Light and dark box test
The reduced feeding activity of BDNF-KIV mice in the NSFT raises a possibility that promoter IV-dependent BDNF expression plays a role in novelty-induced anxiety-like behaviors. Anxiety disorders are known to frequently co-occur with major depression, as 50–60% of individuals with MDD report a lifetime history of anxiety disorders (Kaufman & Charney 2000). To further examine anxiety-like and exploratory behavior of BDNF-KIV mice, we conducted the LDT, a widely used test for anxiety-like behavior in mice (Bourin & Hascoet 2003). The LDT is based on the innate aversion of rodents to brightly illuminated open areas. The time spent in the bright chamber was used as an index of anxiolytic and explorative activity. Both genotypes showed greater preference for the dark chamber than to the lit chamber (P < 0.001 for both genotypes). There was no significant difference observed in the percent time spent in the light and dark sides between both genotypes, indicating no anxiety-driven preference to the dark side of the mutant mice (Fig. 4a). The total number of crossings between the two chambers was also similar between both genotypes, suggesting similar activity levels for both genotypes in the LDT (Fig. 4b). Interestingly, BDNF-KIV mice showed significantly increased initial latency to enter the dark chamber compared to WT mice when placed in the lit side of the box (t18 = 3.4, P < 0.005; Fig. 4c). This delayed initial entrance of the mutant mice was not observed when the same animals were initially placed in the dark side, which was tested using both the familiar boxes (WT = 7.8 ± 2.4 seconds; KIV = 8.2 ± 1.0 seconds; t17 = 0.19, P = 0.85) and a new light and dark box (Fig. 4d). These results suggest that the increased latency of the mutant mice to enter the dark chamber was not likely a result of initial preference and explorative activity in the lit side but rather because of their inability to initiate preferable action in the novel bright environment. The LDT was conducted in both light and dark cycles because differences in basal activity may have affected its results (Bourin & Hascoet 2003), but the similar results were obtained (data not shown).
Elevated plus maze test
The results of the LDT suggest that BDNF-KIV mice may not display anxiety-like behavior measured by the dark/light preference. To further confirm this, we conducted the EPMT, which also measures anxiety-like behavior in mice based on their natural aversion toward open and elevated areas (Fernandez Espejo 1997). The number of entries into closed arms was used as an index of anxiety-like behavior against open space-induced exploration in mice. Both groups displayed significantly higher entry to the closed arms compared with the open arms (t18 = 11.0, P < 0.001 for WT and t18 = 4.0, P < 0.001 for KIV). BDNF-KIV mice showed similar percent entries into the open and closed arms, total entrance number and time spent on the closed arms compared with WT mice (Fig. S3). We also conducted the EPMT using naive animals (a new batch) to determine whether the use in other behavioral tests affected the results in the EPMT. Similar results were obtained as in the original EPMT (percent entries into open arms: WT = 16.9 ± 3.1; KIV = 17.9 ± 2.4; t16 = 0.26, P = 0.80).
Passive avoidance test
The results in the LDT and EPMT could suggest that the lack of promoter IV-dependent expression of BDNF did not alter anxiety-driven preference to the dark/closed side. Rather, it resulted in reduced attention and action toward rewarding stimuli (e.g. reduced food consumption in the NSFT, delayed initial entrance to the dark side in the LDT) as well as to aversive stimuli (e.g. reduced coping activity to tail suspension and foot shocks). These results further advanced the hypothesis that the defective phenotype of mutant mice may be deduced from their deficits in attention to proper actions. To test the hypothesis, we conducted the PAT, which measures response inhibition to aversive conditions. A TW-RM anova of latency showed a significant genotype × day interaction (F1,16 = 16.2, P < 0.01). Post hoc t-tests showed that BDNF-KIV mice had statistically higher latency to enter the dark chamber compared with WT mice on training day (day 1) (t16 = 2.3, P < 0.05; Fig. 5), which was similar to the LDT results (Fig. 4c). On test day, WT mice displayed significantly increased latency to enter the dark chamber (t16 = 5.3, P < 0.001, before and after the foot shock), which suggests that WT mice were able to associate the dark chamber with the aversive experience of foot shocks and, therefore, avoided the preferable dark chamber. Interestingly, BDNF-KIV mice did not display this inhibition in entrance, showing similar latency compared with that on day 1 (t16 = 0.67, P > 0.05, before and after the foot shock; Fig. 5). The latency of BDNF-KIV mice to enter the dark chamber on day 2 was significantly lower compared to that of WT mice (t16 = 3.6, P < 0.01; Fig. 5). These results suggest that the mutant mice exhibit deficiency in response inhibition, which is the ability to pay attention ‘not to go’ to the preferable dark side where the animal had received an aversive foot shock.
Results of the present study showed that selective deficiency in promoter IV-dependent expression of BDNF in mice leads to depression-like behavior in several behavioral paradigms, showing reduced locomotor activity in the OFT, increased immobility in the TST, reduced sucrose preference in the SPT, reduced food consumption in the NSFT and decreased escape activity in the LHT. The results of the TST and LHT indicate increased stress-induced despair, while reduced sucrose preference suggests an anhedonia-like phenotype. Intact active avoidance in BDNF-KIV mice verified that the reduced shuttle activity in the LHT was not because of decreased pain sensitivity or learning deficits. The only test that did not detect the depression-related phenotype of the mutant mice was the FST. This is consistent with other studies showing that the BDNF mutant mice do not display increased immobility in the FST compared with controls (Duman & Monteggia 2006). The FST was developed as a primary screening test for antidepressants in rats and may not be sensitive enough to detect depression-like behavior in mice (Cryan & Mombereau 2004; West 1990) because mice might show increased anxiety-driven movements when not habituated to water. Supporting this idea, similar to the FST results, when the TST was conducted in a novel environment without a habituation process, both genotypes showed increased movements, and a significant difference between the genotypes was not observed (Fgenotype × habituation interaction (1,15) = 6.2, P < 0.05; Fhabituation (1,15) = 48.1: p < 0.001; Fgenotype (1,15) = 0.4; P = 0.67 by a TW anova, t15 = 1.1, P > 0.05 between genotypes without habituation, Fig. S4). Compared with the FST, the TST may be less affected by anxiety in mice when provided a proper habituation process, as supported by studies using mice with anxiety-like behavior, such as GABAB KO mice that displayed less immobility in the FST but not in the TST (Mombereau et al. 2004).
Depression is often comorbid with anxiety (Kaufman & Charney 2000). Thus, reduced exploratory activity and food consumption, seen in BDNF-KIV mice in the OFT and NSFT, can be interpreted as anxiety-like behavior as well as depression-like behavior. On the other hand, BDNF-KIV mice did not present increased anxiety-driven preference to the dark/closed side in the LDT or EPMT. This is distinct from studies with other lines of BDNF mutant mice: heterozygous BDNF-KO mice whose BDNF levels are lowered by 50% (Chourbaji et al. 2008) and BDNFMet/Met mice whose activity-dependent secretion of BDNF is reduced by 30% (Chen et al. 2006) show anxiety-like behavior in the LDT and EPMT. A study reported that BDNF antisense oligodeoxynucleotide infusions into the central and medial amygdala provoked anxiety-like behavior in rats tested by the EPMT, which suggests that reduction in BDNF expression in the central and medial amygdala might produce a preference for the closed arms in the EPMT (Pandey et al. 2006). Promoter I-driven as well as promoter IV-driven BDNF expression is observed in the amygdala (Metsis et al. 1993; Rattiner et al. 2004). Thus, one reason that BDNF-KIV mice displayed no preference to the dark/closed sides in the LDT and EPMT might have been due to sufficient BDNF expression by promoter I in the amygdala.
One compelling interpretation of the results of this study is that lack of promoter IV-dependent BDNF expression may cause impaired attention and reduced action to both rewarding stimuli (e.g. food, dark/closed sides and sucrose) and aversive stimuli (e.g. tail suspension and foot shocks) in a new or stressful environment. In addition to defective attention of ‘what to do,’ the mutant mice appear to have defective attention to ‘what not to do’ as suggested by the additional results in the PAT. BDNF-KIV mice did not develop response inhibition toward entering the dark side where an aversive foot shock had been previously experienced. While this could be caused by a deficit in context memory in the mutant mice, it is unlikely because their 24-h memory was intact when examined with the contextual fear-conditioning test (Sakata & Lu 2007; unpublished data). Interestingly, BDNF-KIV mice showed increased initial latency to enter the dark side in the LDT and PAT, which also suggests delayed initiation of action or delayed recognition of the entrance upon exposure to a novel bright chamber. This was unlikely caused by sensory and motor deficits of the mutant mice because they showed similar latency to escape from foot shocks in the AAT (Fig. S2), similar latency to eat food with fruit smell (TG) in the home environment (Fig. 3d) and similar number of crossings in the LDT (Fig. 4b) compared with WT mice. The reduced activity of BDNF-KIV mice was observed only when they were exposed to a novel environment or to stress but not basal levels (e.g. locomotor activity on day 2 in the OFT; Fig. 1a), which indicates that slow initiation of action in a novel environment was unlikely caused by basal locomotor deficits. Motor and visual functions of the mutant mice were also verified to be intact by the rotarod test, suspended wire test, dowel test, vertical pole test and visual cliff test (Sakata et al., unpublished data).
Overall, the results strongly suggest that defective behavior observed in BDNF-KIV mice might be derived from impaired attention that disturbs initiation of proper action to the given conditions, either to act on benefit or to avoid aversive stimuli. Interestingly, these defective behavioral phenotypes of BDNF-KIV mice may mimic the symptoms of patients with depression, which include markedly diminished interest, impaired attention (Moritz et al. 2002), cognitive and response slowness (O’Connor et al. 2005) and impaired response inhibition (Gohier et al. 2009). Our previous research has found that BDNF-KIV mice display decreased GABAergic function in the PFC (Sakata et al. 2009), which has been suggested to be a possible mechanism underlying impaired higher-order cognitive processes (Rossi et al. 2009). Further studies will be required to understand the mechanisms underlying the phenotypes of BDNF-KIV mice, which may show new insights on some depression-related symptoms such as reduced attention.
The results strongly support the BDNF hypothesis of depression. Previous studies using genetically manipulated mice of BDNF and rodents injected with exogenous BDNF blockers are limited and rather have not detected depression-like behavior. For example, heterozygous BDNF-KO mice and conditional BDNF-KO mice are generally hyperactive and do not show depression-like behavior (Duman & Monteggia 2006). Region-selective knockdown of BDNF using virus systems in either the CA1 or the dentate gyrus of the HIP did not alter depression-related behaviors (Adachi et al. 2008), while BDNF knockdown in the ventral tegmental area (VTA) and nucleus accumbens (NAc) caused increased activity (Krishnan et al. 2007). A limited number of studies have reported depression-like behavior in BDNF mutant mice: (1) heterozygous BDNF-KO mice displayed a depressive phenotype in the LHT (Macqueen et al. 2001), although this effect could be because of, in part, decreased pain sensitivity; (2) Nestin- and CaMKII-promoter-dependent conditional BDNF-KO mice exhibited depression-like phenotypes in the TST (Chan et al. 2006); and (3) female (but not male) glial fibrillary acidic protein (GFAP)-dependent conditional BDNF-KO mice and Ca2+/calmodulin-dependent protein kinase (CaMKII)-dependent BDNF-KO mice showed depression-related behaviors as measured by the SPT and TST (Monteggia et al. 2007); and (4) female (but not male) enolase promoter-dependent rTA-inducible BDNF-KO mice showed decreased locomotor activity only when they were stressed (Autry et al. 2009). These manipulations (i.e. deletion of the BDNF protein-coding region regulated by other gene promoters) can be useful tools for understanding the region/cell-specific role of BDNF, although the outcomes may not reflect the conditions in human patients because they may lack physiological levels of BDNF in the relevant regions. The biggest difference in this study is that BDNF-KIV mice lack endogenous promoter-driven BDNF expression. In contrast to knockout/knockdown of the BDNF protein-coding region, decreased function of BDNF promoter IV may occur in real life via reduced neuronal stimuli, mutations in the promoter region and epigenetic processes through stress. Supporting this, recent studies have shown that social dominant stress and immobilization stress decrease the function of promoter IV through epigenetic regulation processes (Fuchikami et al. 2009; Tsankova et al. 2006). On the other hand, stimulating promoter IV activity is physiologically possible by increasing neuronal activity and antidepressant treatments through (1) membrane depolarization, elevated intracellular calcium and activation of CaRFs (calcium-responsive transcription factors) (Tao et al. 1998) and (2) activation of the cAMP cascades and CREB (Nibuya et al. 1996). In addition, chronic antidepressant treatments could induce and compensate for reduced BDNF expression by activating other BDNF promoters. For example, chronic administration of desipramine induces promoter I-driven BDNF expression in the frontal cortex and HIP, while fluoxetine induces promoter II-driven BDNF expression in the HIP (Dwivedi et al. 2006). Although BDNF induction by other promoters may not be as robust as that by activity-induced promoter IV, accumulation of a small increase in BDNF protein may provide an antidepressant effect, and this might be one of the reasons why antidepressant treatments take a prolonged time (>3 weeks) to become clinically effective. We hypothesize that depression-like behavior of BDNF-KIV mice can be reversed by chronic antidepressant treatments that induce BDNF expression levels by other promoters (e.g. promoter II by fluoxetine), which should be tested in the future. It is plausible that, by having nine different promoters, BDNF gene expression might be tightly regulated by environmental factors and treatments in temporal and brain region-specific manners. Region-specific bidirectional actions of BDNF have been reported: infusion of BDNF into the HIP leads to an antidepressant-like effect while infusion into the VTA or NAc produces depression-like behavior (Eisch et al. 2003). While it is unclear whether the infused BDNF levels represent endogenous expression levels and which promoter would be physiologically responsible for the BDNF expression in the VTA and NAc, it is important to consider that different promoter-specific expression of BDNF in different brain regions may explain various behavioral phenotypes. Details in understanding the mechanisms of how endogenous gene regulation of BDNF contributes to depression and recovery from mood disorders would be crucial for future therapies of depression.
In summary, this study showed that mice with disrupted promoter IV-dependent BDNF expression exhibited depression-like behavior in the OFT, TST, SPT, NSFT and LHT and impaired response inhibition in the PAT. These results provide critical evidence that supports the hypothesis that the suppressed endogenous BDNF gene regulation may be one of the causes of depression.
We thank Dr Jeff Steketee for thoughtful comments and advice.