Variation in Serotonin Transporter Expression Modulates Fear-Evoked Hemodynamic Responses and Theta-Frequency Neuronal Oscillations in the Amygdala

Background Gene association studies detect an influence of natural variation in the 5-hydroxytryptamine transporter (5-HTT) gene on multiple aspects of individuality in brain function, ranging from personality traits through to susceptibility to psychiatric disorders such as anxiety and depression. The neural substrates of these associations are unknown. Human neuroimaging studies suggest modulation of the amygdala by 5-HTT variation, but this hypothesis is controversial and unresolved, and difficult to investigate further in humans. Methods We used a mouse model in which the 5-HTT is overexpressed throughout the brain and recorded hemodynamic responses (using a novel in vivo voltammetric monitoring method, analogous to blood oxygen level–dependent functional magnetic resonance imaging) and local field potentials during Pavlovian fear conditioning. Results Increased 5-HTT expression impaired, but did not prevent, fear learning and significantly reduced amygdala hemodynamic responses to aversive cues. Increased 5-HTT expression was also associated with reduced theta oscillations, which were a feature of aversive cue presentation in controls. Moreover, in control mice, but not those with high 5-HTT expression, there was a strong correlation between theta power and the amplitude of the hemodynamic response. Conclusions Direct experimental manipulation of 5-HTT expression levels throughout the brain markedly altered fear learning, amygdala hemodynamic responses, and neuronal oscillations.


Surgery
Mice were surgically implanted with one carbon paste electrode (CPE) into the basolateral amygdala (BLA) of one hemisphere (to measure tissue oxygen) and another electrode into the BLA of the contralateral hemisphere (to measure local field potentials (LFPs)) under isoflurane anesthesia. Approximately equal numbers of mice received left CPE / right LFP and right CPE / left LFP placements. CPEs were constructed from Teflon-coated 200 µm diameter silver wire (~270 µm coated diameter, Advent Research Materials, Oxon, UK) with the insulation pulled down the wire to produce a 2 mm cavity which was subsequently packed with carbon paste and smoothed (2). LFP electrodes were made from 125 µm diameter silver wire (~177 µm coated diameter, Advent). Coordinates for BLA implantations were -1.35 mm anterior/posterior, ±3.10 mm medial/lateral and -5.00 mm dorsal/ventral, relative to bregma.
Auxiliary and reference electrodes (200 µm diameter silver wire) were implanted into parietal cortex. Each electrode was soldered to a gold pin (E363/0, Plastics One, Roanoke, VA, USA), which was inserted into a pedestal plug (MS363, Plastics One) and secured with skull screws and dental cement ('Simplex Rapid', Associated Dental Products, Wilts, UK). Mice were allowed to recover for at least seven days after surgery.

Opposing-side Implantation of CPE and LFP Electrodes
We implanted LFP and T O2 electrodes into contralateral BLA sites, rather than into the same hemisphere, to minimize tissue damage to the BLA, which is a relatively small structure in a mouse (see Figure S1). It is possible, therefore, that differences in responses between left and right BLA could introduce variability when comparing neuronal oscillations with tissue oxygen (T O2 ) responses. However, we do not think that this impacts on our results for three reasons.
First, left/right placements were counterbalanced across mice so approximately equal numbers of T O2 and LFP electrodes were in a given hemisphere. Second, this counterbalancing was equivalent in WT and 5-HTTOE mice. Third, previous studies have shown that LFP responses (e.g. theta oscillations) show remarkably strong coherence across hemispheres (3).

Apparatus
T O2 signals were measured using constant potential amperometry, as described previously (4,5). A constant potential (-650 mV relative to a reference electrode) was applied to the CPEs using a low-noise potentiostat ('Biostat,' ACM Instruments, Cumbria, UK). Mice were connected to the potentiostat via a 6-channel rotating commutator (SL6C, Plastics One) held on a counter-weighted arm (PHM-110P1, Med Associates) using screened cables (363-363 6TCM, Plastics One). A Powerlab® 8/30 (AD Instruments Ltd, Oxon, UK) was used for analogue / digital conversion and data were collected on a Windows PC running Chart® v5 software (AD Instruments). LFPs were recorded using a differential amplifier (DP-301, Warner Instruments, CT, USA) or using the potentiostat with no potential applied, in which case the potentiostat acts as a differential amplifier. T O2 and LFPs were sampled continuously at 4 kHz.

Tissue O 2 Voltammetry
Applying a potential (-650 mV) to an electrode results in the electrochemical reduction of dissolved O 2 on the surface of the electrode, inducing an electrical current that is measured by the potentiostat. concentration around the tip of the electrode produce directly proportional changes in the measured Faradaic current (6).
The area of sensitivity is estimated to be a sphere with diameter twice the electrode surface diameter, i.e. a 200 μm diameter electrode has ~400 μm diameter sphere of sensitivity (see also 7,8). The spatial resolution of the T O2 electrode is sufficient to detect T O2 differences between lamina in rat whisker barrel cortex, i.e. approximately ~400 μm (9). T O2 responses, like blood-oxygen-level dependent responses, are driven by changes in cerebral blood flow (CBF) and are dependent upon local neuronal activity (see (10), Figure 1).
Infusion of the AMPAR antagonist CNQX or the GABA A agonist muscimol lead to reductions in both CBF and T O2 response amplitude (11,12). Moreover, blocking sensory transduction in the peripheral nervous system (e.g. by applying the sodium channel blocker lidocaine to the whisker pad of rats) blocks both neuronal activity and T O2 responses in the somatosensory cortex (9).

Standard Rodent Fear Conditioning Paradigm
Fear conditioning in unoperated mice took place in an operant chamber (17 cm long × 11.5 cm wide × 20 cm high) located in a sound-insulated box. The walls and lid of the chamber (illuminated by a ceiling mounted light) were composed of clear Perspex whilst the floor of the chamber consisted of 19 stainless steel bars approximately 8.5 mm apart, through which scrambled shocks were delivered (0.3 mA, 0.5 s duration, San Diego Instruments shock generator). During training a plastic cube scented with artificial "apple pie" odor was placed next to the conditioning chamber inside the sound-insulated box to give it a distinct odor from the testing context. Between each trial all faeces/urine were removed and the boxes were cleaned.
Black-and-white video images of the mice were captured by a wide-angle video camera attached to the ceiling of the chamber, and relayed to a computer via a Panasonic video recorder (NVSD400). Video data were analysed using a Videotrack (vNT4.0) automated tracking system (Viewpoint, Champagne Au Mont D'Or, France) with a low and high activity threshold settings.
"Freezing" was defined as periods during which movement fell below the lower activity threshold. With this threshold breathing movements did not register as activity (i.e. absence of freezing), but small purposeful movements (e.g. sniffing) were detected as activity. The amount of time spent freezing per 30 s time bin was calculated. For measurement of the unconditioned response to the tone and the unconditioned burst activity response to the shock, time bins of 1 s were used.
The training session began with a 6 min acclimatization period, during which lowvolume white noise occurred. This was followed by a 30 s auditory tone. On tone offset mice received a 0.5 s footshock (0.3 mA). After a further 3 min, a second tone/shock pairing was delivered, followed by a further 3 min period before the mice were removed. Twenty four hours after the training session, fear memory recall for the cue was tested in a novel environmental context (a round plastic chamber with patterned walls, smooth floor, and a distinctive "chicken" odor). Mice experienced two 30 s presentations of the tone (without footshock) during the 5 min session.

Discriminative Fear Conditioning Paradigm
Discriminative fear conditioning in mice with recording electrodes was conducted in one of three operant chambers (ENV-307A, Med Associates Inc., Lafayette, IN, USA), each with distinct visual and olfactory cues to aid context discrimination. Stimulus delivery was controlled by a custom-written script in the MED-PC language. Timed TTL-pulses to the AD converter ensured that stimulus delivery was synchronized with the electrophysiological recordings at 1 ms resolution.
Discriminative fear conditioning was carried out over five consecutive days and the procedure on each day was virtually identical. First, the mouse was connected to the recording equipment and placed in the 'neutral context', i.e. a chamber in which the mouse had been previously habituated for at least 2 hours and in which they never received shocks. The potential was then applied to the T O2 electrode for 10 minutes before the experiment began to ensure that T O2 signals were stable. Day 1, pre-exposure, was performed entirely in the neutral context: five tone (2900 Hz) and five white noise stimuli (both 30 s duration, 80 dB) were presented in pseudorandom order with a mean inter-stimulus interval of 80 s (range 60-100 s), with no shocks administered. On training days 1-3, the mouse was placed into the neutral context for 10 minutes and then transferred to one of the conditioning chambers (e.g. context A). Mice then received five tone and five white noise stimuli (the same as during pre-exposure), but now one of the stimuli (counterbalanced across mice and across genotypes) was always paired with coterminating footshock (0.3 mA, 0.5 s). On day 5, the fear memory recall test, mice were placed first into the neutral context for 10 minutes and then placed into a novel conditioning chamber (e.g. context B if trained in context A) and the five tone and five white noise stimuli were played with no shocks administered. Behavior (freezing) was monitored via a video camera in the roof of the chamber.

Acoustic Startle
To test for potential hearing impairments in 5-HTTOE mice, startle responses to acoustic stimuli of different intensities were measured in a separate cohort of mice (WT: n = 6; 5-HTTOE: n = 6) using the SR-Lab System (San Diego Instruments, San Diego, CA, USA). The test session began by placing a mouse in the Plexiglas cylinder where it was left undisturbed for 5 minutes. The test session consisted of eight 40 ms trial types at different sound intensities: 65, 70, 75, 80, 90, 100, 110, and 120 dB. Note that the 65 dB stimulus was identical to the background noise and was used to measure baseline movement in the cylinders. Five blocks of the eight trial types were presented in a pseudorandom order such that each trial type was presented once within a block of eight trials. The average inter-trial interval was 15 seconds (range: 10 to 20 seconds). The startle response was recorded for 65 ms (measuring the response every 1 ms), starting with the onset of the startle stimulus. The average startle amplitude recorded during the 65 ms sampling window was used as the measure of animals' reactivity to sounds. The data were analyzed using a general linear model with genotype as a between subjects factor, sound intensity as a within-subjects factor and body weight as a covariate.

Behavior
Freezing during the discriminative fear conditioning paradigm was measured using a script in NIH Image (13), which compared consecutive video frames (1 Hz sampling) for pixel changes and assigned a freezing score if the % pixel change was below a set threshold calibrated for an absence of movement except for breathing (14). The freezing difference score was calculated as follows: % freezing during the 30 s cue presentation minus % freezing during the 30s before cue presentation (i.e. positive freezing scores indicate increased freezing to the cue and negative freezing scores indicate decreased freezing to the cue relative to the pre-cue period).

Tissue oxygen (T O2 ) signals
T O2 signals were first down-sampled to 100 Hz. Cue-evoked T O2 responses were calculated by subtracting the mean T O2 signal in the 5 s before conditioned stimulus (CS) onset (i.e. baseline) from the T O2 signal during the 30 s CS presentation. This yielded a 30 s ΔT O2 signal, which was then divided into fifteen 2 s timebins (i.e. 0-2 s, 2-4 s, 4-6 s…28-30 s), with each data point equal to the mean value during each 2 s timebin (see Figure S2). Thus our T O2 data retain good temporal resolution (2 s), allowing us to analyze how the signals evolve during CS presentations, akin to event-related fMRI (15), rather than extracting a single response peak or area under the curve.
We have used an absolute ΔT O2 signal rather than a % change from baseline for the following reasons. First, there is occasionally considerable variation in the raw baseline signals (i.e. the background current) between animals. The precise reason for this variation is not known but could be due to differences in the active area of the electrode surface. Consider the raw T O2 responses presented in Figures S2C and S2D. These were recorded from two different WT mice to a CS+ presentation at exactly the same point in training (training day 3, CS+4). The T O2 response in Figure S2C has a raw baseline signal of 605 nA whereas the T O2 response in Figure   S2D has a raw baseline signal of 135 nA. Despite these baseline differences, the maximum signal changes evoked by the CS+ (i.e. the absolute ΔT O2 ) are similar in both cases (+13 nA in Figure S2C, +11 nA in Figure S2D) whereas the % changes are very different (+2% in Figure   S2C, +8% in Figure S2D). The second reason is that the raw T O2 signal can drift within a session, and this drift is not always linear. Under these circumstances, the same absolute signal change would yield different % signal changes for different trials within a single session. In short, subtracting a local baseline (e.g. the mean signal in the 5 s before CS onset), rather than using the % signal change from baseline, gives more reliable results in terms of the T O2 signal response.
During pre-exposure and training days, CS+ T O2 responses were averaged over the five CS+ trials and CS-T O2 responses were averaged over the five CS-trials of each session. On the fear memory recall day, we analyzed only the first CS+ versus the first CS-trial to mitigate the effects of extinction on subsequent presentations. For the correlational analysis with theta power, we calculated the mean CS+ evoked T O2 response (i.e. the mean value over the 15 time bins) for each mouse on each day of training and the fear memory recall day.

Local Field Potentials
LFPs were band-pass filtered between 1 Hz and 45 Hz. We calculated power spectra for the first 10 s of CS presentation. Spectra were then averaged over the five CS+ versus the five CS-trials for each mouse for each day. Spectra were computed in MATLAB (The Mathworks, Natick, MA, USA) using a fast Fourier transform size of 2000 samples with a Hamming window (50% overlap) at a sampling rate of 1 kHz, and a frequency resolution of ~0.5 Hz. To compare across mice, the raw power spectra were normalized by expressing the power in each frequency bin (Φ i, ) as a proportion of the total power between 1 and 45 Hz: where P i = normalized power, Φ i = raw power (mV 2 ) An example of this normalization procedure is shown in Figure S2F-G. To compare theta power across genotypes, we calculated the sum of CS+ and CS-evoked theta power between 7-10 Hz for each mouse on each day. For the correlational analysis between T O2 response amplitude and theta power, we calculated a theta:delta ratio by dividing the sum of theta power between 7-10 Hz by the sum of delta power between 1-4 Hz. Spectrograms were generated in MATLAB, using a sliding time-window of 1 s, with 500 ms overlap, and a frequency resolution of ~0.25 Hz.

Determination of Electrode Placements
At the end of the fear conditioning experiment, mice implanted with recording electrodes were injected with sodium pentobarbitone; 200 mg⁄kg) and perfused transcardially with physiological saline (0.9% NaCl), followed by 10% formol saline (10% formalin in 0.9% NaCl).
Their brains were removed and placed in 10% formol saline for 3 days, and then transferred to a 30% sucrose-formalin solution for 24 h and frozen. Coronal sections (40 μm) were then cut on a freezing microtome and stained with Cresyl violet to enable visualization of the recording sites.
Only mice with confirmed electrode placements in the basolateral amygdala were used in the T O2 and LFP analyses (see Figure S1).

Quantification of Serotonin Transporter Binding Levels
Serotonin transporter binding levels in the amygdala were assessed in a separate group of mice (OE: n = 6; WT: n = 5; aged 3-6 months). Briefly, brains were snap frozen and sectioned, as described in Jennings et al. (1), and prepared for autoradiography using 2 nM [ 3 H]citalopram to determine serotonin transporter binding. Slides were exposed to [ 3 H]-sensitive Hyperfilm (Amersham Biosciences) and densitometric quantification of autoradiograms was performed, calibrated to tritiated tissue equivalents (Amersham Biosciences) and corrected for nonspecific signals. Optical densities from the basolateral amygdala of each hemisphere in three sections per animal were averaged and expressed as femtomoles per milligram of tissue (fmol/mg).

Neurochemical Analysis of 5-HT and 5-HIAA in Amygdala Tissue
High performance liquid chromatography with electrochemical detection was used to measure amygdala tissue levels of 5-HT. In a separate cohort of mice (n = 5 per group), tissue micropunches were obtained from the amygdala of WT and 5-HTTOE mice (16). Mice were euthanized by cervical dislocation and brains were rapidly removed, and frozen before being cut (1 ml/min flow rate). Because of the low weight of tissue punches, data were expressed as picomoles/sample rather than picomoles/mg tissue.

Statistical Procedures
Behavioral analyses for the standard rodent fear conditioning paradigm were performed

Are Amygdala Theta Oscillations Locally Generated or Volume Conducted?
Several sources of evidence argue that theta oscillations are generated by amygdala neurons and do not simply reflect volume conduction from other brain regions. Pare and Collins have shown that single-unit activity is modulated by theta oscillations in the lateral amygdala of cats during discriminative fear conditioning (18). Intra-cellular recordings in guinea pigs and cats also confirm that theta oscillations can be induced in lateral amygdala neurons by near-threshold membrane depolarization via intracellular current injection (19,20). Moreover, if amygdala theta was volume conducted, the most likely source would be the hippocampus as it exhibits high amplitude theta activity. In mice, Pape and colleagues have shown that theta activity in the BLA is synchronized with hippocampal theta during fear acquisition and recent retrieval of fear memories but not during fear extinction or remote fear retrieval or spatial exploration (21)(22)(23)(24).
Thus hippocampal-BLA theta-synchronization is not always present, which argues against passive volume conduction from the hippocampus. Also, hippocampal theta oscillations are strongly correlated with speed of movement (i.e. higher amplitude and higher frequency theta as running speed increases (25)), whereas theta oscillations in the BLA are strongest when mice are freezing (i.e. stationary). One interpretation of these datasets is that theta oscillations are intrinsically generated by amygdala neurons and are not volume conducted from the hippocampus (23).
However, against this, most models of neuronal oscillation assume that there must be spatial separation of current sinks and sources (in order to produce a dipole), and that these must be in coherently aligned neurons for oscillations to be observed at a population level in the LFP.
In non-laminar structures, such as the amygdala, it is unclear whether the neuronal architecture allows for the generation of theta oscillations at all. Resolution of this issue is not straightforward and would require multiple single unit and LFP recordings simultaneously from the amygdala and any candidate regions of theta generation; followed by silencing of those regions (e.g. by lesions) to see if amygdala theta oscillations persisted (for further discussion of this problem as it relates to the striatum, see Burke, 2005 (26)).  from (A) transformed into ΔT O2 signal by subtracting the baseline (mean T O2 signal during 5s before CS onset) and binning the mean T O2 signal into fifteen 2 s epochs. (C-D) Raw T O2 data illustrating the difference in the baseline signal between two WT mice. Despite these baseline differences, the magnitude of the CS+ evoked ΔT O2 response is similar in each mouse (max: +13nA in (C); +11 nA in (D)). These data demonstrate that subtracting a local baseline is the appropriate method for analysis, rather than using % signal change from baseline. (E) Bandpassfiltered (1-45 Hz) LFP recordings from the BLA in the 10 s periods before CS+ onset (blue) or after CS+ onset (red). (F) Data from (E) expressed as a power spectrum. In this example, CS+ onset evoked an increase in theta power compared to the pre-CS+ period. (G) Data from (F) normalized by expressing the power in each frequency bin as a proportion of the total power between 1-45 Hz. BLA, basolateral amygdala; CS, conditioned stimulus; US, unconditioned stimulus; WT, wild-type. Freezing responses (% change in freezing compared to the baseline) to the 'to-be-allocated' CS+ and CS-stimuli (no effect of genotype, CS type, or interaction; all F < 1, p > 0.4). (B) T O2 responses (ΔT O2 change compared to the baseline) to the 'to-be-allocated' CS+ and CS-stimuli (no effect of genotype, CS type, or interaction; all F < 2.1, p > 0. 15). (C) Example spectrograms from a WT mouse during 'to-beallocated' CS+ (left) and CS-(right) trials. The 30 s before and after stimulus onset is shown, with stimulus onset at 0 as indicated by the white vertical line. The T O2 responses are superimposed in red (CS+) and green (CS-). Note, this is the same WT mouse as shown in the spectrogram in Figure 4A, which shows responses after fear conditioning. (D) Normalized power spectral density (PSD) plots for 'to-be-allocated' CS+ evoked responses in WT versus 5-HTTOE mice. The left panel shows spectra between 1-20 Hz; the middle panel shows spectra between 5-12 Hz (the boxed section in the left panel); the right panel shows the summed theta power between 7-10 Hz (the boxed section in the middle panel). There were no statistical differences between WT and 5-HTTOE mice (F < 1.    Figure S5. Peri-stimulus responses to the CS+ on training day 3 in four different WT mice. (A-D) The top panel shows LFP amplitude (blue trace, in mV) for the 5 s before and 5 s after CS+ onset. The lower panel shows power spectra for the pre-CS+ period (blue) and the CS+ period (red) shown in the top panel. In each case, CS+ onset led to lower delta (1-4 Hz) power and higher peak theta power (5-10 Hz) compared to the pre-CS+ period. CS, conditioned stimulus; LFP, local field potential; WT, wild-type. The top panel shows LFP amplitude (blue trace, in mV) for the 5 s before and 5 s after CS+ onset. The lower panel shows power spectra for the pre-CS+ period (blue) and the CS+ period (red) shown in the top panel. In B-D (but not A), CS+ onset led to slightly higher peak theta power (5-10 Hz) compared to the pre-CS+ period. However, the marked increases in theta power and the concomitant reductions in delta activity seen in WT mice were much less evident in 5-HTTOE mice. 5-HTTOE, serotonin transporter over-expressing; CS, conditioned stimulus; CSP, DEFINE; LFP, local field potential; WT, wild-type.  Table S1. 5-HTT expression levels and 5-HT tissue levels in the amygdala of wild-type and 5-HTT over-expressing mice.