Loss of promoter IV-driven BDNF expression impacts oscillatory activity during sleep, sensory information processing and fear regulation

Posttraumatic stress disorder is characterized by hyperarousal, sensory processing impairments, sleep disturbances and altered fear regulation; phenotypes associated with changes in brain oscillatory activity. Molecules associated with activity-dependent plasticity, including brain-derived neurotrophic factor (BDNF), may regulate neural oscillations by controlling synaptic activity. BDNF synthesis includes production of multiple Bdnf transcripts, which contain distinct 5′ noncoding exons. We assessed arousal, sensory processing, fear regulation and sleep in animals where BDNF expression from activity-dependent promoter IV is disrupted (Bdnf-e4 mice). Bdnf-e4 mice display sensory hyper-reactivity and impaired electrophysiological correlates of sensory information processing as measured by event-related potentials (ERP). Utilizing electroencephalogram, we identified a decrease in slow-wave activity during non-rapid eye movement sleep, suggesting impaired sleep homeostasis. Fear extinction is controlled by hippocampal–prefrontal cortical BDNF signaling, and neurophysiological communication patterns between the hippocampus (HPC) and medial prefrontal cortex (mPFC) correlate with behavioral performance during extinction. Impaired fear extinction in Bdnf-e4 mice is accompanied by increased HPC activation and decreased HPC–mPFC theta phase synchrony during early extinction, as well as increased mPFC activation during extinction recall. These results suggest that activity-dependent BDNF signaling is critical for regulating oscillatory activity, which may contribute to altered behavior.

Supplemental Figure S3: Event-related potential (ERP) experimental design and additional ERP analysis results. (a) Trial design for the ERP experiment (S1=Sound 1 and S2=Sound 2). S1 and S2 consist of identical 70 dB intensity tones separated by a 500 ms inter-stimulus interval (ISI). Dashed line represents time of stimulus onset. Below, the physiological response as measured by the amplitude of electroencephalogram (EEG) is depicted. The response to the second tone is gated, i.e. the amplitude to S2 is smaller than to S1. (b-c) Grand average responses for WT (n=8) and Bdnf-e4 (n=8) following S1 and S2, obtained by averaging the EEG response obtained for each animal. (b) Grand average response for S1. (c) Grand average response for S2. (d-e) Post-S1 event-related spectral perturbation (ERSP) results depicted as a line graph for theta (3-8 Hz) and alpha (8- Post-hoc analysis did not reveal significant effects until the final 5 min of context extinction testing (1-5 min, and 5-10 min, n.s., 10-15 min 15, p=0.001). (e-g) Testing paradigm for the combined cue/context extinction experiments presented in Figure 3. (e) During conditioning, mice were first exposed to two 30 s, 1000 Hz neutral tones (NT) to assess any potential noise-induced freezing prior to conditioning. Conditioning consisted of 3 tone (4000 Hz)-foot-shock (0.6 mA) pairings, with the tone played for 30 s and the foot-shock taking place in the last 2 s of the tone. The conditioning session lasted 10 min. (f) 24 h following conditioning, extinction trials were started. On d 1 and d 2 mice were exposed to two sessions of extinction training (d1-Extinction 1, Extinction 2, and d 2-Extinction 3, Extinction 4, noted by white numbers) separated by a 1 h interval. 72 h post-conditioning on d 3 mice performed an Extinction Test. Hatched lines between sessions indicate a 24 h break between sessions. (g) Each extinction session consisted of a 2 min baseline in the conditioning context, followed by 20 CS tone exposures, which were 30 s in length with a 5 s ISI. Extinction sessions were 14.5 min long. Data are means ± SEM (**p<0.001, n.s.=not significant).

Supplemental Figure S6: Representative histology from animals with depth electrodes targeted to the infralimbic (IL) region of medial PFC (mPFC) and CA1 region of HPC. Following perfusion, brains were sectioned and cresyl violet stained.
Representative examples of IL and CA1, with arrows pointing to the site of the electrode track.
Supplemental Figure S7: Behavioral results acquired from local field potential (LFP) recordings conducted during extinction (a-c) Behavior results during electrophysiological recordings (WT n=11; Bdnf-e4 n=13). (a) Average time spent freezing per session, during Conditioning (C), Extinction 1 (Ext 1), Extinction 3 (Ext 3), and Extinction Test (Ext Test). During Extinction 1, the session in which HPC theta power is higher and cross channel coherence (CCC) is reduced in Bdnf-e4 mice, Bdnf-e4 mice freeze significantly more than p=0.0075 Supplemental Figure S8: Additional physiological data from recordings conducted during extinction. (a-c) ERSP and CCC representing electrophysiological results during the Habituation session. Data is averaged from 3-5 Hz (low theta), and is taken from 0-1000 ms following cessation of motion. (a) ERSP data from the mPFC (n=10 each genotype) and (b) ERSP data from the HPC (n=10 WT; n=12 Bdnf-e4) demonstrate that there are no differences between genotypes during Habituation in the low theta frequency range. (c) Additionally, there is not a significant difference between genotypes during habituation in HPC-mPFC phase synchrony, measured by CCC (n=9 WT; n=10 Bdnf-e4).
(d-f) Line graph representation of data averaged from 3-5 Hz during Extinction 3 from 0-1630 ms post-freezing demonstrates subtle differences in WT and Bdnf-e4 ERSP and CCC data across time.

Supplementary Methods
Quantitative PCR (qPCR) qPCR was conducted as previously described 1 . Briefly, WT and Bdnf-e4 adult mice (n=4-5 per genotype) were cervically dislocated and brain tissues were snap frozen with isopentane-this tissue was used for both qPCR and ELISA. Total RNA was isolated from hippocampus (HPC) and PFC (PFC) using TRIzol (Life Technologies, Carlsbad, CA). RNA was purified using an RNeasy minicolumn (Qiagen, Valencia, CA) and quantified by a NanoDrop spectrophotometer (Agilent Technologies, Savage, MD).
Using Superscript III (Life Technologies), RNA concentration was normalized and reverse transcribed into single-stranded cDNA (Life Technologies). Quantitative PCR was performed using a Realplex thermocycler (Eppendorf, Hamburg, Germany) using GEMM mastermix (Life Technologies) with 40 ng of synthesized cDNA. Individual mRNA levels were normalized for each well to Gapdh mRNA levels.

Homecage Behavior Monitoring
Homecage recordings were conducted as described previously 2 . Mice were placed in a home cage environment in a sound-attenuated, temperature-controlled chamber.
Recordings using digital cameras were begun at 6 pm, at the beginning of the dark cycle. Infrared light was used for illumination during the dark cycle. Automated video analysis of homecage behavior was completed using HomeCageScan (HCS) software (Cleversys Inc., Reston, VA). Using HCS, detection of behavior occurs by utilizing information about whole-body movements of the mouse-animal body parts such as head, tail, and limbs are identified, and then sequences of data are used to automatically analyze animal behavior in durations > 6 frames.

Testing of Baseline Behaviors
Prior to all behavioral testing, mice were habituated for at least one hour in a quiet room with low lighting. Open field, Light dark box testing, and elevated zero results were all assessed using Topscan behavioral software from CleverSys, Inc. Video was recorded from above, with a "bird's eye" view. Open field testing-Mice were placed in the center of an 18"x18" box. Distance travelled and duration spent in the corner and center within the arena of the box was tracked. Light dark box-Mice were placed into a 10"by17" box with an enclosed, dark space area and a light, open area to assess anxiety. Frequency and duration of time spent in both the light and dark areas of the box were recorded. swim test-mice were placed in a warm water tank for 6 minutes, and the bouts and duration of immobility were measured. The last 4 minutes was used for the total duration measurement, to examine escape-related behavior.

Cued and Contextual Extinction
For all experiments, WT and Bdnf-e4 -/-mice were conditioned as described in the main text, and in Fig S4e. During conditioning, mice were first exposed to two 30 s, 1000 Hz neutral tones (NT) to assess any potential noise-induced freezing prior to conditioning.
Conditioning consisted of 3 tone (4000 Hz)-foot-shock (0.6 mA) pairings, with the tone played for 30 s and the foot-shock taking place in the last 2 s of the tone. For cue and context extinction, mice were extinguished 24 h post-conditioning. For cue extinction, the interior of the conditioning chamber was altered with novel spatial cues and the animals were set in a plastic container, which covered the metal floor bars associated with foot shock. An identical auditory protocol was used for cue extinction that was used for the cue/context paradigm. For context extinction, mice were placed back in the conditioning chamber for 15 min and time spent freezing for each min was scored, and then averaged across 5 min.